Corrosion behavior of silicon nitride bonding silicon carbide in molten magnesium and AZ91 magnesium alloy

Transcription

1 Materials Science and Engineering A 45 (2006) Corrosion behavior of silicon nitride bonding silicon carbide in molten magnesium and AZ9 magnesium alloy Hukui Chen, Jianrui Liu, Weidong Huang State Key Laboratory of Solidification Processing, Northwestern Polytechnical University, Xi an, Shaanxi 70072, People s Republic of China Received in revised form 26 September 2005; accepted 29 September 2005 Abstract The corrosion behaviors of silicon nitride bonding silicon carbide (Si 3 N 4 /SiC) composites in molten magnesium and AZ9 magnesium alloy were investigated through immersion tests. The microstructure and the component of the surface layer of the composites were characterized by scanning electron microscopy (SEM), energy dispersive spectroscopy (EDS) and X-ray diffraction (XRD). The experimental results show that there were slight corrosion phenomena on the surface of the composites when it contacted the liquid metals at initial stage. With increasing time, the reaction went to equilibrium gradually between the interface of composites and magnesium alloy. After about 7 h, the composites did not further react with liquid magnesium because of forming a resistant layer on the composites surface. The corrosion mechanism of the composites in molten magnesium and AZ9 magnesium alloy was also discussed Elsevier B.V. All rights reserved. Keywords: Silicon nitride bonding silicon carbide; Magnesium; Magnesium alloy; Corrosion. Introduction Melting, refining and casting of magnesium alloys are generally carried out using iron-base crucible. Solid iron exhibits an excellent chemical inertness towards liquid magnesium at the temperature of which alloys are processed [,2]. There are some advantages using iron crucible as the smelting crucible of magnesium alloys, such as low cost, easy to be made and regeneration. It was found in actual production of magnesium alloys, however, that the surface iron element of iron-base crucibles was dissolved partly into molten magnesium under the action of the high temperature liquid magnesium, which resulted in the iron content of magnesium alloys increasing and the quality of the alloys reducing. The main reason was that the excellent chemical inertness between iron and liquid magnesium was effectively changed by the presence of impurities or alloy elements in the two base metals [2 5]. The iron content was strictly limited to a very low level for magnesium and magnesium alloys. The increasing of iron content would accelerate the corrosion of magnesium alloys and affects its mechanical properties [6,7]. There Corresponding author. Tel.: ; fax: address: cgs@nwpu.edu.cn (J. Liu). were two kinds of ways, which could avoid the pollution of iron from iron-base crucible in the smelting processes of magnesium and magnesium alloys. One way was to coat a protective layer on the surface of iron crucible. Another way was to use some materials, which had good stability to molten magnesium and magnesium alloys as the substitute materials of iron-base crucible. The selection of materials, which do not react with magnesium and alloy elements at high temperature, is a key for above two ways. The composites of Si 3 N 4 /SiC have high hardness, chemical inertness, thermal stability and good dielectric properties [8 0]. It has been paid much attention in high temperature applications and has already been used in metallurgy of the nonferrous metals, such as in the aluminum melting furnace, zinc and copper melting furnace. There are, however, few reports on the application of the composites in the process of magnesium and magnesium alloys melting. In the present study, in order to study the possibility of the composites being used as a crucible material for the melting of magnesium and magnesium alloy, the corrosion behaviors of silicon nitride bonding silicon carbide (Si 3 N 4 /SiC) composites in molten magnesium and AZ9 magnesium alloy were investigated through immersion tests. The microstructure and the component of the surface layer of the composites were /$ see front matter 2005 Elsevier B.V. All rights reserved. doi:0.06/j.msea

2 292 H. Chen et al. / Materials Science and Engineering A 45 (2006) characterized by SEM, EDS and XRD. The reaction mechanism between the composites and the elements of magnesium alloy was also discussed. 2. Experimental 2.. Materials and method The composites used for the tests were supplied by Luoyang Institute of Refractories Research. Its chemical composition is SiC > 75%, Si 3 N 4 > 20% and a trace amount of SiO 2. Its bulk density and apparent porosity values of the sample are 2.60 g/cm 3 and 4%, respectively. The composites were cut down to size of 20 mm 0 mm 0 mm as test samples. The composition of magnesium and AZ9 magnesium alloy that the experiment used is shown in Table. The tests were carried out in a resistance furnace with an iron-base crucible. The volume of crucible could melt approximately kg of magnesium and magnesium alloy. Magnesium or magnesium alloy was put into the crucible, heated to melt and continuously heated up to the temperature of 750 C. Then the sample of silicon nitride silicon carbide being warmed-up was immersed into the melt for 20 h at the temperature. In the process of heat preservation, the content of the silicon in the melt was analyzed by drawing a small amount of metal liquid from the crucible at regular time intervals. In order to avoid any risks of fracture or cracking when the sample was immersed into the melt, the sample was heated in advance to approach the temperature of the melt. The tests were conducted under SF 6 /air atmosphere for fear oxidation of the magnesium. After the immersion test completed, the sample used for surface analysis was taken out of the melt and cooled in the air to room temperature. The sample used for cross-section analysis was cooled down together with the melt under SF 6 /air atmosphere, and then cut in cross-section and polished Analysis The content of silicon in molten magnesium and magnesium alloy during immersion tests at regular time intervals was analyzed by plasma atom emission spectrum (ICP). The microstructure of the surface and the cross-section of the samples, before and after immersed in liquid of pure magnesium and AZ9 magnesium alloy, were examined by scanning electron microscopy (SEM). The composition of the surface of the samples and the concentration distribution of elements across the composites/alloy interface were determined by an energy dispersive spectroscopy (EDS). The interfacial reaction products were identified by X-ray diffraction (XRD). Fig.. Relationship curves of the changes of Si content in molten Mg and Mg alloy to time. 3. Results and discussion 3.. Corrosion kinetics The corrosion kinetics of the composites in molten Mg and AZ9 Mg alloy were determined by analyzing the content of silicon in the melt at regular time intervals during immersion tests. The relationship curves between change of Si content in molten Mg and Mg alloy and immersion time are shown in Fig.. It can be seen from Fig. that the content of silicon in liquid of magnesium and AZ9 magnesium alloy increased rapidly with increasing time at initial stage of immersion tests. This indicated that corrosion had taken place at this stage and the corrosion of the composites may be caused by reactions between the composites and the melt. A part of reaction products containing silicon dissolved in liquid of magnesium and AZ9 magnesium alloy. After immersion time was over 7 h, the silicon content did not remarkably increase with time prolonging, which indicated that the rate of the corrosion reactions was very slow at this stage. Since the corrosion kinetics curves of the composites in molten Mg and AZ9 Mg alloy followed approximately a parabolic law at this stage, it was highly probable that the corrosion could be controlled by the diffusion of the melt through a reaction layer. It can also be seen from Fig. that the change of content for silicon in magnesium was obvious higher than that in AZ9 magnesium alloy at the same period Surface morphology and microstructure In order to investigate the reaction between the composites and molten Mg and AZ9 Mg alloy, the surface and the cross-section of the composites immersed in the liquid metal were observed by scanning electron microscopy with Table Chemical composition of magnesium and magnesium alloys, in wt.% Material Al Zn Cu Fe Ni Mn Si Mg Mg Balance AZ Balance

3 H. Chen et al. / Materials Science and Engineering A 45 (2006) Fig. 2. SEM micrographs of the surface of samples (a) before immersion test, (b) after immersed in molten Mg and (c) in molten AZ9 Mg alloy for 20 h. energy dispersive spectroscopy. The SEM micrographs for the surface and the cross-section of the composites are shown in Figs. 2 and 3a and b, respectively. The EDS line scanning results for concentration profiles of the elements at the cross-section of the composites are shown in Fig. 3c and d. It can be found from Fig. 2a that the surface microstructure of Si 3 N 4 /SiC composites was that silicon carbide particle (large crystal particle which was concluded by EDS point analyses) act as a base framework and silicon nitrogen (fine particle which was also concluded by EDS point analyses) distributed around silicon carbide particle, forming a netted composites structure. After the composites were immersed in the liquid of magnesium or magnesium alloy for 20 h, the surface microstructures of the composites were not the same as its original structure being seen from the Fig. 2. The obvious interface between silicon nitrogen and silicon carbide became illegibility. The fine particle of silicon nitrogen disappeared, and the silicon carbide particles were detected. There was a layer of non-crystal compounds formed (Fig. 2b and c). This might be that the silicon nitrogen existing in the composites surface reacted with Mg and alloy elements at high temperature, which resulted in silicon nitrogen disappearing and the formation of a reaction product layer on the surface of the composites. In order to identify the composition of surface of the immersed composites, EDS analysis was preformed. The results of EDS analysis confirmed that the main elements on the surface of composites immersed in molten Mg were Mg, O, Si, C and N, and the surface of composites immersed in AZ9 magnesium alloy contained Mg, O, Si, C, N and Al. The results of EDS analysis showed that the elements of Mg and Al might come from the reaction between the composites and liquid metal, Si, C and N might come from the composites itself and O might be produced by the little free SiO 2 existing on the surface of the composites. Fig. 3a and b shows that after the immersion test, there was an obvious transition layer between the metal and the composites. The thickness of the transition layer was about m. It can be seen from Fig. 3c and d that in the transition layer, there had much Mg and O elements (Fig. 3c), or Mg and O, Al elements (Fig. 3d). The transition layer resulted from the reaction of liquid magnesium or magnesium alloy and the composites. The reaction products filled the hole of composites surface. There was no metal element penetrating into the composites base, which was attributed to the transition layer forming X-ray diffraction analysis In order to identify the reaction products on the composites surface before and after it was immersed in molten Mg and Mg alloy, X-ray diffraction analysis was carried out. The results were shown in Fig. 4. As could be found from Fig. 4a, before the composites immersed, the majority of diffraction peaks in the pattern

4 294 H. Chen et al. / Materials Science and Engineering A 45 (2006) Fig. 3. SEM micrographs of the cross-section of sample (a) after immersed in molten Mg, (b) after in molten AZ9 Mg alloy for 20 h, (c) EDS line scanning results corresponding to (a) and (d) EDS line scanning results corresponding to (b). (The arrows indicate the starting location and the direction of EDS line scanning.) could be explained by the characteristic peaks of SiC (consistent with JCPDS card: ) and Si 3 N 4 (consistent with JCPDS card: ), which indicated its main composition was SiC and Si 3 N 4. But there were a few of diffraction peaks consistent with the characteristic peaks of SiO 2 (consistent with JCPDS card: ), which indicated that a little of free SiO 2 impurity existed on the surface of the composites. After immersion test, the characteristic peaks of Si 3 N 4 and SiO 2 disappeared (Fig. 4b and c). The relative intensity of SiC characteristic peaks after being immersed was almost the same as that before immersion test comparing with Fig. 4a c. New weak diffraction peaks, which were consistent with characteristic, peaks of MgO (consistent with JCPDS card: ) and Mg 2 Si (consistent with JCPDS card: ) appeared on the surface of the composites after being immersed in liquid magnesium (Fig. 4b). On the surface of the sample being immersed in AZ9 magnesium alloy, the characteristic peaks of MgO, Mg 2 Si and AlN (consistent with JCPDS card: ) appeared in the diffraction pattern (Fig. 4c). The appearance of MgO, Mg 2 Si and AlN was attributed to the reactions of the composites Fig. 4. XRD patterns of the surface of the composites (a) before immersion tests, (b) after immersed in molten Mg and (c) after in molten AZ9 Mg alloy for 20 h.

5 H. Chen et al. / Materials Science and Engineering A 45 (2006) and magnesium or magnesium alloy at high temperature. The results show that after the composites were immersed in the liquid of magnesium and magnesium alloy, Si 3 N 4 and SiO 2 on the surface completely reacted with liquid metal elements. 4. Discussion Silicon nitride silicon carbide composites consist mainly of SiC, Si 3 N 4 and a little amount of free SiO 2. Reactions between SiC, Si 3 N 4, SiO 2 and Mg might happen as follows: 2 SiC + Mg 2 Mg 2 Si + 2 C () 6 Si 3N 4 + Mg 3 Mg 3 N Si (2) 2 SiO 2 + Mg MgO + 2 Si (3) 2 Si + Mg 2 Mg 2Si (4) The Gibbs free energies of reaction Eqs. () (4) within the temperature range of C were calculated using thermodynamic data. The results are shown in Table 2, which indicates that there is no chemical reaction between SiC and magnesium at the temperature from 650 to 900 C. This result could be confirmed by the fact that the relatively intensities of diffraction peaks of SiC on the surfaces of composites before and after immersed in the liquid of magnesium have almost no change (Fig. 4). Previous studies [,2] had also proved the result. The values of G 0 (2) and G0 (3) are negative, which suggests that Si 3 N 4 and SiO 2 might react with liquid magnesium. According to the analysis of the dynamics of the reaction of SiO 2 with Mg Li melt, Yu et al. [3] found that the reaction of SiO 2 with Mg was carried out by two steps. MgO and Si were formed firstly, and then MgO reacted with Si to form Mg 2 Si. Bochenek and Braszczynska [4] analyzed the structure of the Mg 5%Al matrix cast composites containing SiC particles, and found that the SiO 2 film covering the silicon carbide particles had been reduced by molten magnesium according to the above two-stage reaction. Shi et al. [5] reported that SiO 2 on the surface of SiC could react with Mg in Al Mg melt within 20 min, producing a layer of MgO. According to the viewpoint of above authors, it could be concluded that the molten Mg could easily react with SiO 2. This result could explain the presence of characteristic peaks of MgO and Mg 2 Si, and the increase of silicon content in liquid magnesium at the initial stage of immersion test. For reaction of Si 3 N 4 with liquid magnesium, Jeong et al. [6] found that the Al-matrix alloy would react with Si 3 N 4 crystals when added magnesium in Al Cu alloy reinforced with Si 3 N 4 particles by studying the interface structure of Al Cu alloy and Al Cu Mg Table 2 The calculation results of Gibbs free energies for different temperature for the equations T ( C) G 0 () (J) G 0 (2) (J) G 0 (3) (J) G 0 (4) (J) alloy composites. This implied that the reaction between Si 3 N 4 and liquid Mg could take place. However, Mg 3 N 2 formed by the reaction of Si 3 N 4 with liquid magnesium had not been found in X-ray diffraction pattern of sample surface after immersed in molten Mg (Fig. 4b). The reason is, in course of the composites being cooled down in the air, that Mg 3 N 2 easily reacted with H 2 O of air to form Mg(OH) 2. Mg(OH) 2 decomposed into MgO at once. So, the characteristic peaks of Mg 3 N 2 did not appear in the XRD pattern. The primary alloy elements in AZ9 magnesium alloy are Al and Zn. The study [7,8] reported Zn did not react with SiC and Si 3 N 4. The amount of Zn is relative lower in AZ9 magnesium alloy, the reactions between Zn and SiC and Si 3 N 4 were not considered. Here were only discussed the reactions between Al and SiC and Si 3 N 4. The possible reactions were as followings: 3 4 SiC + Al 4 Al 4C Si (5) 4 Si 3N 4 + Al AlN Si (6) The Gibbs free energies of reactions (4) and (5) within the temperature range of C were shown in Table 3, which indicated that SiC did not react with Al within the temperature range of C because of the bigger positive values of G 0 (5). This was in agreement with the result of XRD analysis (Fig. 4). The values of G 0 (6) was negative, which indicated that Si 3 N 4 might react with Al. Mouradoff et al. [9] reported that there was a interfacial reaction between liquid aluminum and silicon nitride and a protective dense AlN layer on the surface of silicon nitride by studying of the interaction between liquid aluminum and silicon nitride. This result could be proved by the existence of AlN and Mg 2 Si in XRD patterns (Fig. 4c) and the increase of Si content in AZ9 magnesium alloy (Fig. ). AlN has very high hardness, high density and an intrinsic inertness to aluminum liquid [20]. AlN layer adhered to the surface of Si 3 N 4 /SiC composites, which prevented Si 3 N 4 from further reacting with Al, and thus made the composites resistant to corrosion of aluminum liquid. The calculation results of the Gibbs free energies of the reaction of AlN with magnesium showed the reaction could not happen between AlN and magnesium liquid within the temperature range of C. This result was also confirmed with the experimental results of He et al. [2]. So, AlN layer could prevent Si 3 N 4 /SiC composites from reacting further with the magnesium in AZ9 magnesium alloy liquid too. Because the reaction layer formed on the composites surface in molten AZ9 magnesium contained AlN besides MgO and Mg 2 Si, the layer was more compact than that formed on the composites surface in molten magnesium. Therefore, the corrosion rate of reaction product layer for the composites in molten AZ9 magnesium was lower than that in molten magnesium Table 3 The calculation results of Gibbs free energies for different temperature for the equations T ( C) G 0 (5) (J) G 0 (6) (J)

6 296 H. Chen et al. / Materials Science and Engineering A 45 (2006) during the same period. This might be reason that the change of content for silicon in magnesium is obvious higher than that in AZ9 magnesium alloy at the same period. Based on the results of above analyses, it could be found that the corrosion processes of the composites in the liquid of magnesium and magnesium alloy consisted of two stages. When the composites were immersed in liquid magnesium or magnesium alloy, SiO 2 and Si 3 N 4 on the surface of the composites reacted with liquid metal elements firstly. At the initial stage of immersion tests, the amount of corrosion products formed by the reaction of the composites with liquid metal was very little and the reaction layer was porous and discontinuous. So, the content of silicon in the liquid of metal increased rapidly with the increasing of immersion time. When the immersion time exceeded 7 h, the interspaces of the porous and discontinuous surface layer were filled by the reaction products, and the reaction layer became dense and compact, which could prevent liquid metal elements from reacting with the composites further. The content of silicon in magnesium liquid did not obviously increase with the time increase. 5. Conclusions The corrosion attack of liquid magnesium or AZ9 magnesium alloy on Si 3 N 4 /SiC composites mainly took place at initial stage of immersion. There was a reaction layer formed on the composites surface when the composites contacted with liquid magnesium or magnesium alloy. The composites showed a better corrosion resistance to the liquid magnesium. The layers acting as a barrier prevented the composites from reacting with magnesium and magnesium alloy further at high temperature. It could be concluded from the results of the slight corrosion and the formation of a reaction layer on the composites surface that the composites would be a better candidate as a coating material of iron crucible for the handling or the melting of magnesium and magnesium alloy, or as a crucible material to make into crucible to melt magnesium and magnesium alloy directly. References [] C.S. Robert, Magnesium and its Alloys, Wiley, New York, 960. [2] J.C. Viala, D. Pierre, F. Bosselet, Scripta Mater. 40 (999) 85. [3] D. Pierre, M. Peronnet, F. Bosselet, J.C. Viala, J. Bouix, Mater. Sci. Eng. B 94 (2002) 86. [4] D. Pierre, C. Viala, J.M. Peronnet, J.C. Viala, J. Bouix, Mater. Sci. Eng. A 349 (2003) 256. [5] D. Pierre, F. Bosselet, M. Peronnet, J.C. Viala, J. Bouix, Acta Mater. 49 (200) 653. [6] A. Proats, T.K. Aune, D. Hawke, W. Unsworth, J. Hillis, Metals Handbook, vol. 3, ninth ed., ASM Metals Park, Ohio, 987, p [7] R. Zeng, W. Ke, Y. Xu, E. Han, Z. Zhu, Acta Metall. Sin. 37 (200) 673. [8] M. Sternitze, J. Eur. Ceram. Soc. 7 (997) 06. [9] M. Herrmann, C. Schuber, A. Rendtel, H. Hübner, J. Am. Ceram. Soc. Bull. 8 (998) 095. [0] M.F. Gozzi, E. Radovanovic, I.V.P. Yoshida, Mater. Res. 4 (200) 3. [] Z.J. Liu, S.H. Zhou, X.X. Xi, Z.K. Liu, Physica C 379 (2003) 87. [2] R.A. Saravanan, M.K. Surappa, Mater. Sci. Eng. A 276 (2000) 08. [3] H.S. Yu, G.H. Min, Y.X. Liu, X.C. Chen, Chin. J. Nonferrous Met. 9 (999) 785 (in Chinese). [4] A. Bochenek, K.N. Braszczynska, Mater. Sci. Eng. A290 (2000) 22. [5] Z.L. Shi, M.Y. Gu, J.Y. Liu, G.Q. Liu, J.C. Lee, Chin. Sci. Bull. 46 (200) 6 (in Chinese). [6] H.G. Jeong, K. Hiraga, M. Mabuchi, K. Higashi, Acta Mater. 46 (998) [7] Z.Y. Chen, Naihuo Cailiao 34 (2000) 224 (in Chinese). [8] Z.Y. Chen, Naihuo Cailiao 32 (998) 4 (in Chinese). [9] L. Mouradoff, A.L. Durand, J. Desmaison, J.C. Labbe, O. Grisot, R. Rezakhanlou, J. Eur. Ceram. Soc. 3 (994) 323. [20] M. Yan, Z. Fan, J. Mater. Sci. 36 (200) 285. [2] T. He, R.J. Cava, J.M. Rowell, Appl. Phys. Lett. 80 (2002) 29.