Removal of Magnesium from Molten Aluminum Scrap by Compound-Separation Method with Shirasu

Transcription

1 Materials Transactions, Vol. 51, No. 5 (2010) pp. 838 to 843 Special Issue on Growth of Ecomaterials as a Key to Eco-Society IV #2010 The Japan Institute of Metals Removal of Magnesium from Molten Aluminum Scrap by Compound-Separation Method with Shirasu Tomokazu Hashiguchi* and Hidekazu Sueyoshi Graduate School of Science and Engineering, Kagoshima University, Kagoshima , Japan The removal of magnesium from the molten aluminum scrap containing magnesium was examined by compound-separation method using Shirasu as an additive. When Sirasu is added to the molten aluminum scrap, which is followed by agitation, MgAl 2 O 4 and MgO are formed by the reactions among SiO 2 and Al 2 O 3 in Shirasu and magnesium in the molten scrap. It is feasible to remove magnesium by separating these reaction products as dross. The effect of addition temperature of Shirasu on magnesium removal is smaller than that of the agitation time. In the case of the addition of flaky Shirasu with mm of particle size, the amount of magnesium removed increases linearly with agitation time because the over-all reaction is mainly controlled by a surface-controlled reaction. In the case of the addition of granular Shirasu with mm of particle size, the amount of magnesium removed increases parabolically with agitation time because the over-all reaction is mainly controlled by a diffusion-controlled reaction. The amount of magnesium removed can be controlled by changing the amount of Shirasu added. [doi: /matertrans.mh200914] (Received November 4, 2009; Accepted December 11, 2009; Published February 3, 2010) Keywords: aluminum scrap, recycling, magnesium removal, shirasu additive, compound-separation method 1. Introduction The amount of energy required to manufacture aluminum ingots from aluminum scrap (about 7 MJ/t) is significantly lower than that required when from ores (about 214 MJ/t). 1) Thus, aluminum products have a high recycling efficiency. However, various scrap alloys are contaminated because there is no well-established scrap screening system. Therefore, most of the recycled products are of a cascade recycle because of the quality degradation due to impurity contamination. 2) In order to promote a horizontal recycling, it is desirable to remove impurity elements from the recycled aluminum ingots. Magnesium is one of the main impurity elements associated with cascade recycling. A large amount of magnesium is added to aluminum alloys to improve their mechanical properties, deformability and corrosion resistance. In the automobile industry, aluminum and magnesium alloys are now being used because the parts are of much lighter weight. One can predict that the magnesium contamination of recycled aluminum ingots will become more worse. In the refining process for magnesium removal, Cl 2 gas injection and flux treatment have been performed. 3 5) However, Cl 2 gas injection results in the generation of HCl gas which causes the production of dioxins. Maximum Achievable Control Technology (MACT) standards were established in the United State in 2000, and the use of Cl 2 gas is now regulated. 6) The reduction of Cl 2 gas emission has also been established as an object of the refining industry in Japan. 7) Flux treatment as an alternative method has various problems such as the use of deleterious substances (chloride or fluoride), environmental damage due to the toxic gas produced during treatment and high cost. Therefore, the development of a new magnesium removal technology is desired. For the removal of impurity silicon from the molten aluminum scrap a compound-separation method has been studied. 8) However, there have been no reports regarding *Graduate Student, Kagoshima University removing impurity magnesium from the molten aluminum scrap. In the present study, magnesium removal from the molten aluminum scrap by compound-separation method, which has low environment impact and low cost, was examined. As an additive, Shirasu that exists in large quantities as a natural resource in South Kyushu, Japan, and is harmless and inexpensive, was used. When Shirasu was added to the molten aluminum scrap containing magnesium, MgAl 2 O 4 and MgO were formed by the reaction between Shirasu and magnesium in the molten scrap. By removing the dross consisting of these compounds, it was feasible to remove magnesium from the molten scrap. In this paper, these results are reported. 2. Experimental Procedure As an aluminum scrap, aluminum alloy containing magnesium (JIS: A5083, Mg: 4.63 mass%, Si: 0.11 mass%, Fe: 0.25 mass%, Cu: 5 mass%, Mn: 7 mass%, Ti: 2 mass%, Cr: 0.12 mass%, Zn: 2 mass%, Al: Bal.) was used. The pieces of scrap had a plate shape ( mm). As an additive, two types of Shirasu (Seishin Co. Ltd., Japan) with different particle sizes were used. Chemical compositions of Shirasu obtained by X-ray fluorescence analysis (XRF; ZSX100e, Rigaku Co. Ltd.) are shown in Table 1. The primary component of Shirasu was SiO 2 and the secondary component was Al 2 O 3. The chemical composition of large-size Shirasu was somewhat similar to that of small-size Shirasu. Figure 1 shows secondary electron (SE) and back scattered electron (BSE) images of Shirasu obtained by scanning electron microscopy (SEM; XL30CP, FEI Co. Ltd.) with an energy dispersive X-ray analyzer (EDX). As shown in Fig. 1, the shape of small-size Shirasu (AS-100) with mm particle size was a flake. As shown in BSE image of the cross section of Shirasu mounted with a resin, a thickness of the Shirasu (white part) was about 20 mm. Large-size Shirasu (KESB-03) with mm particle size

2 Removal of Magnesium from Molten Aluminum Scrap by Compound-Separation Method with Shirasu 839 Table 1 Chemical compositions of Shirasu (mass%). Shirasu Particle size Chemical compositions (mm) SiO 2 Al 2 O 3 K 2 O Na 2 O Fe 2 O 3 CaO MgO TiO 2 P 2 O 5 AS KESB Fig. 1 SE and BSE images of Shirasu. SE image of small size Shirasu (AS-100), BSE image of the cross section of small-size Shirasu mounted with a resin, SE image of large-size Shirasu (KESB-03) and BSE image of the cross section of large-size Shirasu mounted with a resin. was a mixture of flake and nodule. X-ray diffraction (XRD; X pert Pro MPD, PANalytical Co. Ltd.) analysis revealed that Shirasu is amorphous. The aluminum scrap was dissolved in a carbon crucible under air ambient. Shirasu wrapped with aluminum foil was added to the molten scrap, which was followed by agitation for various lengths of time. Dross floating on the surface of the molten scrap was skimmed off and then the molten scrap was poured into a steel mold. The reaction products which settled on the crucible bottom (settled dross) were also collected. The magnesium and silicon contents of the ingot were examined by XRF. The microstructures of the ingot and dross were observed by SEM with EDX. In order to identify the reaction products, the dross was examined by XRD. 3. Results Figure 2 shows the relationship between magnesium content in the ingot and agitation time. Because the experiments were performed under air ambient, one can predict that magnesium content decreases following the reaction between magnesium in the molten scrap and oxygen gas in the atmosphere, that is, the formation of MgO. Therefore, the Mg content of ingot, C Mg /mass% Addition temperature 973 K Particle size µm : Agitation only at Agitation time, t/min Fig. 2 Relationship between magnesium content in the ingot and agitation time. change in the magnesium content of the ingot produced by agitation only without Shirasu addition is also shown in Fig. 2. In the case of agitation only without Shirasu addition, there was little reduction in the magnesium content. On the

3 840 T. Hashiguchi and H. Sueyoshi Si content of ingot, C Si /mass% Fig. 3 Addition temperature 973 K Particle size µm : Agitation only at Agitation time, t/min Relationship between silicon content in the ingot and agitation time. Mg content of ingot, C Mg / mass% Calculated value for Mg Calculated value for Si : Mg Amount of Shirasu, M / mass% Fig. 4 Relationships between magnesium and silicon contents in the ingot produced by the addition of small-size Shirasu at, which was followed by agitation for 40 min and the amount of small-size Shirasu. : Si Si content of ingot, C Si /mass% 50µm (e) (f) Fig. 5 The result of EDX analysis of the ingot produced by the addition of small-size Shirasu at, which was followed by agitation for 6 min. BSE image, Al K-ray image, Mg K-ray image, Si K-ray image, (e) Fe K-ray image and (f) Mn K-ray image. other hand, Shirasu addition resulted in a decrease in magnesium content. In the case of the addition of smallsize Shirasu, magnesium content decreased linearly with agitation time for each addition temperature of Shirasu. As the addition temperature of Shirasu was high, the amount of magnesium removed was large. However, the effect of addition temperature of Shirasu on magnesium removal was smaller than that of the agitation time. For the addition of large-size Shirasu, magnesium content decreased parabolically with agitation time. Figure 3 shows the relationship between silicon content in the ingot and agitation time. All the silicon content increased until 20 min of agitation time but saturated when the agitation time exceeded 20 min. There was little different in the effect of addition temperature of small-size Shirasu on silicon content. For the addition of large-size Shirasu, the silicon content saturated was smaller than that for the addition of small-size Shirasu. Figure 4 shows the relationship between magnesium and silicon contents in the ingot produced by the addition of small-size Shirasu at, which was followed by agitation for 40 min and the amount of small-size Shirasu. The magnesium content decreased linearly with the amount of small-size Shirasu. On the other hand, the silicon content increased with the amount of small-size Shirasu. The dotted lines (calculated values for magnesium and silicon) are explained later. Figure 5 shows the result of EDX analysis of the ingot produced by the addition of small-size Shirasu at, which was followed by agitation for 6 min. Concentrated magnesium and silicon were detected at cell boundaries. A point analysis (quantitative analysis) of this part revealed that this phase is Mg-Si compound (Mg 2 Si), 9) which crystallizes upon cooling to room temperature in the steel mold. There was a large phase (white part) as shown in BSE image. The K-ray images and point analysis revealed that this white

4 Removal of Magnesium from Molten Aluminum Scrap by Compound-Separation Method with Shirasu (e) (f) µm Fig. 6 The result of EDX analysis of the ingot produced by the addition of small-size Shirasu at, which was followed by agitation for 40 min. BSE image, Al K -ray image, Mg K -ray image, Si K -ray image, (e) Fe K -ray image and (f) Mn K -ray image. Al-Mg oxide 100µm (e) Mg oxide Fig. 7 The result of EDX analysis of the settled dross formed by the addition of large-size Shirasu at, which was followed by agitation for 40 min. BSE image, Al K -ray image, O K -ray image, Mg K -ray image and (e) Si K -ray image. phase is an Al-Si-Fe-Mn compound. Although the existence of magnesium in the aluminum matrix was not clear in MgK ray image (Fig. 5), point analysis revealed that magnesium exists in the matrix. Figure 6 shows the result of EDX analysis of the ingot produced by the addition of small-size Shirasu at, which was followed by agitation for 40 min. There was little magnesium at cell boundaries. Although the existence of magnesium in the aluminum matrix was not clear in Mg-K ray image (Fig. 6), point analysis revealed that small amount of magnesium exists in the matrix. Thus, the magnesium content was very small compared with that shown in Fig. 5. These results correspond approximately to the magnesium content in the ingot (Fig. 2). There was a marked increase in silicon at cell boundaries. This part corresponds to a white phase in BSE image. The K ray images and point analysis revealed that this phase is an Al-SiFe-Mn compound. On the other hand, there was an area in which only silicon was detected. These results suggest that most of the silicon produced by the dissociation of SiO2 dissolves into the molten scrap, and then precipitates as monolithic silicon or an Al-Si-Fe-Mn compound. Figure 7 shows the result of EDX analysis of the settled dross formed by the addition of large-size Shirasu at, which was followed by agitation for 40 min. As shown in BSE image, the shape of reaction product was the same as Shirasu. As shown in K -ray images, there was an area in which aluminum, magnesium and oxygen coexisted. This phase may be Al-Mg oxide (Fig. 7). Also, there was an area in which magnesium and oxygen coexisted. This phase

5 842 T. Hashiguchi and H. Sueyoshi Intensity(Arb.unit) : Al : Si : MgAl 2 O 4 : MgO θ Fig. 8 XRD pattern of the settled dross formed by agitation for 40 min after Shirasu addition. The addition of small-size Shirasu at 973 K, the addition of small-size Shirasu at and the addition of large-size Shirasu at. Standard free energy G 0 /kj. mol -1 T /K Mg(l)+SiO 2 (s)=2mgo(s)+si(s) 2Mg(l)+Al 2 O 3 (s)+sio 2 (s)=mgo(s)+mgal 2 O 4 (s)+si(s) -500 Si(l)+O 2 (g)=sio 2 4/3Al(l)+O 2 (g)=2/3al 2 O Mg(l)+O 2 (g)=2mgo MgO(s)+2Al(l)+3/2O 2 (g)=mgal 2 O 4 (s) may be Mg oxide (Fig. 7). As shown in Fig. 7(e), silicon was locally detected in the reaction product. This is because silicon placed inside the large-size reaction product may be more difficult to dissolve into the molten scrap. These results suggest that part of silicon produced by the dissociation of SiO 2 remains in the reaction product without dissolving into the molten scrap. This is the reason why the silicon content in the ingot saturated when the agitation time exceeded 20 min, why the silicon content of the ingot for the addition of large-size Shirasu was lower than that for the addition of small-size Shirasu (Fig. 3). Figure 8 shows XRD pattern of the settled dross formed by agitation for 40 min after Shirasu addition. Peaks of MgAl 2 O 4 and MgO were detected in each profile. This suggests that Al-Mg and Mg oxides observed in Fig. 7 are MgAl 2 O 4 and MgO, respectively. It has been reported in preparing aluminum matrix composites that the reactions among aluminum, magnesium and SiO 2 results in the formation of MgAl 2 O 4 and MgO ) A peak of silicon was also detected in each profile. This indicates that the silicon which remains in the reaction product (Fig. 7) is a monolithic metal and not a compound. 4. Discussion As mentioned above, the dross consists mainly of MgAl 2 O 4 and MgO. The reactions between Shirasu and the molten scrap can be summarized as 2Mg(l) þ Al 2 O 3 (s) þ SiO 2 (s) ¼ MgO(s) þ MgAl 2 O 4 (s) þ Si(s) ð1þ 2Mg(l) þ SiO 2 (s) ¼ 2MgO(s) þ Si(s) ð2þ MgO(s) þ 2Al(l) þ 3/2O 2 (g) ¼ MgAl 2 O 4 (s) ð3þ where l = liquid, s = solid and g = gas. The standard Gibbs free energy (G ) for these reactions was calculated using HSC Chemistry Version 5. The results are shown in Fig. 9. The results for oxide formation are also shown in Fig. 9 for reference. The values of G for reactions (1), (2) and (3) were negative at the addition temperature of Shirasu. This suggests that the likelihood of achieving these reactions is high. As shown in Table 1, Shirasu had about 14 mass% of Fig. 9 The standard Gibbs free energies for reactions (1), (2) and (3) and oxide formation as a reference. Al 2 O 3. The reaction (1) among the SiO 2 and Al 2 O 3 in Shirasu and magnesium in the molten scrap may occur in preference to the reaction (2) between SiO 2 in Shirasu and Magnesium in the molten scrap because the value of G for reaction (1) is lower than that for reaction (2). However, the amount of Al 2 O 3 in Shirasu is small. Therefore, the reaction shifts from reaction (1) to reaction (2) when all the Al 2 O 3 in Shirasu is consumed by the reaction (1). Although part of the MgO produced by reactions (1) and (2) is removed from the molten scrap together with floating dross, residual MgO reacts with O 2, which is dissolved into the molten scrap by agitation, resulting in the formation of MgAl 2 O 4 (reaction (3)). Because the density of Shirasu (0.6 Mg/m 3 ) is smaller than that of Aluminum (2.7 Mg/m 3 ), Shirasu added floats on the surface of the molten scrap. By agitation local reaction occurs on the surface of Shirasu because of low wettability between Shirasu and the molten scrap. Shirasu with a large amount of unreacted parts floats on the surface of the molten scrap because the density is low compared with that of the molten scrap. On the other hand, most of the reaction products settle down to the crucible bottom because their densities (MgAl 2 O 4 : 3.6 Mg/m 3 and MgO: 3.58 Mg/m 3 ) are high compared with that of the molten scrap. Assuming that all the Shirasu added reacted with magnesium in the molten scrap by reactions (1) and (2) and all the silicon produced by the dissociation of SiO 2 dissolved into the molten scrap without remaining in the reaction product, magnesium and silicon contents of the ingot, which are calculated from reactions (1) and (2), are shown by dotted lines in Fig. 4. The calculated value for magnesium content is in agreement with the measured one. This indicates that all the Shirasu reacted with magnesium in the molten scrap. Thus, the amount of magnesium removed can be controlled by changing the amount of Shirasu added. On the other hand, the measured value for silicon was lower than the calculated one. This is because part of silicon produced by the dissociation of SiO 2 remains in the reaction product without dissolving into the molten scrap as shown in Fig. 7. In Fig. 2, for the addition of small-size Shirasu, magnesium content of the ingot decreased linearly with agitation

6 Removal of Magnesium from Molten Aluminum Scrap by Compound-Separation Method with Shirasu 843 time. This suggests that reactions (1) and (2) depend on a surface-controlled reaction. This is because the reaction in the transverse direction of flaky Shirasu is achieved at a short time, Shirasu/molten scrap contact area increases with increasing agitation time though small-size Shirasu agglutinates on the surface of molten scrap, and the over-all reaction is mainly controlled by the rate of the reaction along the surface of the flaky Shirasu. On the other hand, for the addition of large-size Shirasu, the magnesium content of the ingot decreased parabolically with agitation time. It is well known that parabolic relationship between the amount of reaction and reaction time holds in a diffusion-controlled reaction. 13) This suggests that these reactions depend on a diffusion-controlled reaction. This is because the over-all reaction is mainly controlled by the diffusion rate in the transverse direction of granular Shirasu. As shown in Fig. 7, part of silicon produced by the dissociation of SiO 2 remained in the reaction product without dissolving into the molten scrap. This is the reason why the change in silicon content with agitation time (Fig. 3) was independent of the different reaction mechanism such as a surface-controlled and diffusion controlled reactions. 5. Conclusions The removal of magnesium from the molten aluminum scrap containing magnesium was examined by compoundseparation method using Shirasu as an additive. The results obtained are as follows: (1) When Sirasu is added to the molten aluminum scrap, which is followed by agitation, MgAl 2 O 4 and MgO are formed by the reactions among SiO 2 and Al 2 O 3 in Shirasu and magnesium in the molten scrap. It is feasible to remove magnesium by separating these reaction products as dross. (2) As the addition temperature of Shirasu is high, the amount of magnesium removed becomes large. However, the effect of addition temperature of Shirasu on magnesium removal is smaller than that of the agitation time. (3) In the case of the addition of flaky Shirasu with a small particle size, the amount of magnesium removed increases linearly with agitation time because the over-all reaction is mainly controlled by a surfacecontrolled reaction. In the case of the addition of granular Shirasu with a large particle size, the amount of magnesium removed increases parabolically with agitation time because the over-all reaction is mainly controlled by a diffusion-controlled reaction. (4) The amount of magnesium removed can be controlled by changing the amount of Shirasu added. REFERENCES 1) T. Ohnishi: J. JILM 46 (1996) ) A. R. Khoei, I. Masters and D. T. Gethin: J. Mater. Process. Thechnol. 127 (2002) ) K. Nishina: J. JILM 41 (1991) ) R. R. Roy and Y. Sahai: Light Metals 1998, ed. by B. Welch, (The Minerals, Metals and Materials Society, 1998) pp ) T. A. Utigard, K. Friesen, R. R. Roy, J. Lim, A. Silny and C. Dupuis: J. Miner. Met. Mater. Soc. 50 (1998) ) K. A. Kitzman: Light Metals 2002, ed. by W. Schneider, (The Minerals, Metals and Materials Society, 2002) pp ) M. Tsunekawa: J. JILM 54 (2004) ) M. Nagao and T. Nakamura: J. JILM 46 (1996) ) Y. L. Liu, S. B. Kang and H. W. Kim: Mater. Lett. 41 (1999) ) M. Hanabe and P. B. Aswath: Acta Mater. 45 (1997) ) D. A. Weirauch, Jr.: J. Mater. Res. 3 (1988) ) V. M. Sreekumar, R. M. Pillai, B. C. Pai and M. Chakraborty: J. Alloy. Compd. 461 (2008) ) J. D. Verhoeven: Fundamentals of Physical Metallurgy, (John wiley & Sons, New York, 1975) pp