A physical approach to the direct recycling of Mg-alloy scrap by the rheo-diecasting process

Materials Science and Engineering A 472 (2008) 251 257 A physical approach to the direct recycling of Mg-alloy scrap by the rheo-diecasting process G. Liu, Y. Wang, Z. Fan Brunel Centre for Advanced Solidification Technology (BCAST), Brunel University, Uxbridge UB8 3PH, UK Received 7 December 2006; received in revised form 8 March 2007; accepted 9 March 2007 Abstract Due to the rapidly growing market for magnesium diecastings in the automotive industry in the past decade, there is a steady increase in magnesium scrap sourced from both end-of-life vehicles (old scrap) and the manufacturing processes (new scrap). In addition, the energy required to recycle magnesium is only 5% of that for extraction of primary magnesium from Mg-containing minerals. This makes magnesium recycling extremely beneficial for cost reduction, preservation of natural resources and protection of the environment. In this paper, a physical approach is presented to magnesium recycling using the rheo-diecasting (RDC) process. The RDC process was applied to process AZ91D alloy sourced from both primary alloy ingots and diecast scrap. The experimental results showed that the RDC process could be used to produce recycled AZ91D alloy with fine and uniform microstructure and very low level of porosity. The intermetallic compounds containing the impurity elements were fine and spherical particles distributed uniformly in the alloy matrix. No oxide particle clusters and oxide films were found in the RDC microstructure. The tensile properties of the recycled AZ91D alloy were comparable to those produced from the primary alloy ingots. 2007 Elsevier B.V. All rights reserved. Keywords: Recycling; Rheo-casting; Mg-alloy; Microstructure; Mechanical property 1. Introduction Due to the increasing concerns for environmental protection and ever-tightening government regulations for sustainable economic development, there is an increasing pressure on automakers to improve fuel consumption to reduce greenhouse gas emissions. The automakers have studied the relationship between vehicle mass and fuel economy for decades. The majority of the studies to date conclude that for every 10% reduction in vehicle weight there will be a corresponding 6 8% decrease in fuel consumption. Magnesium alloys, as the lightest of all structural metallic materials, find increasing applications in the automotive industry for vehicle weight reduction. Since the early 1990s we have seen a 15% average annual growth rate of diecast Mg-alloy components in the automotive industry [1]. It is predicted that the usage of light alloys in cars will continue to rise at an even faster pace in the next decade. This fast growth rate means a fast increase in Mg-alloy scrap from both manufacturing source (new scrap) and end-of-life vehicles (old scrap). Corresponding author. Fax: +44 1895 269758. E-mail address: guojun.liu@brunel.ac.uk (G. Liu). A magnesium inventory analysis has been conducted for Germany up to 2020 [2], which concluded that the major source of magnesium scrap is end-of-life vehicles. In Germany alone, the stored automotive magnesium will reach 51 kt in 2020, representing an average annual growth rate of 80%. The general picture of magnesium inventory is expected to be similar in the developed countries. In addition, the source for new scrap is also growing fast. In typical magnesium diecasting operations only around 50% of the material input ends up as finished products, and the remaining 50% of magnesium alloys is accumulated as scrap [3]. Furthermore, unlike other materials for engineering applications, metals, such as aluminium and magnesium, can be recycled repeatedly without loss of their inherent properties. Recycled light alloys from a scrap source only consume 5% of the energy required to produce the same amount of primary alloys. The application of recycled light alloys in cars can lead to a substantial reduction of greenhouse gas emissions. Therefore, recycling Mg-alloy scrap is becoming an important technical and economical challenge. A major barrier to the recycling of Mg-alloys is the existence of substantial amount of inclusions and impurity elements in the scrap (both new and old), the former causes severe lose of ductility and strength, and the latter reduces significantly the corrosion 0921-5093/$ see front matter 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2007.03.026

252 G. Liu et al. / Materials Science and Engineering A 472 (2008) 251 257 resistance. The major challenges in reprocessing light alloy scrap are dealing with the increased inclusions and impurity elements. Conventional wisdom is to reduce the amount of such inclusions and impurities by a chemical approach [4 6], e.g., flux or fluxless refining, and hydrometallurgy process. However, there are still some problems in these recycling technologies, such as process complexity, low productivity and high energy consumption. Therefore, there is a growing need for more effective recycling processes for Mg-alloy scraps. In recent years, the semisolid metal processing (SSM) has been shown to possess a number of technological advantages over the traditional casting processes. It is proved to be particularly suitable for light alloy components with improved mechanical performance [7,8]. Among these SSM processes, the rheo-diecasting process (RDC), recently developed by BCAST at Brunel University, is a promising technology in terms of the process flexibility, component integrity, microstructural homogeneity and cost-effectiveness [9 11]. In the RDC process, the intensive shearing and turbulence are imposed to the solidifying liquid by the high speed rotating twin screws, resulting in a fine and uniform microstructure. The aim of the present study is to investigate the feasibility to use the RDC process as an enabling technology for direct recycling of magnesium alloy scrap. In this paper, we present the RDC process as a physical approach to direct recycling of Mg-alloy scrap. The experimental results presented in this paper include variation of chemical compositions, microstructural evolution, and mechanical properties of RDC recycled AZ91D alloy processed under different conditions. The discussion will be focused on the effects of processing on microstructure and mechanical properties of recycled Mg-alloys and the advantages of RDC as a physical approach to direct recycling of Mg-alloy scrap. 2. Experimental The primary AZ91D alloy used in the present study was provided by MEL (Magnesium Elektron Ltd., Manchester, UK), and its chemical compositions are listed in Table 1 in comparison with the ASTM specifications for the same alloy [12]. The AZ91D alloy scraps used in the present study were from diecasting and consisted of biscuits and gates (70 wt%), runners (20 wt%), overflows (9 wt%), and dross (1 wt%). The primary AZ91D ingot with or without either 50 or 100 wt% scrap was melted in a steel crucible, using an electrical resistance furnace at a holding temperature of 675 C. A mixed gas of N 2 containing 0.4 vol.% SF 6 was used to protect the alloy melt from oxidising during the melting. The alloy samples were prepared by the RDC process, developed recently by BCAST at Brunel University. This novel technology is an innovative one-step SSM processing technique for manufacturing near net shape components of high integrity directly from liquid alloys. Detailed description of the RDC process can be found elsewhere [13]. The RDC equipment consists of two basic functional units: a twin-screw slurry maker (TSSM) and a standard cold chamber high-pressure diecasting (HPDC) machine. The TSSM has a pair of screws rotating inside a barrel. The fluid flow inside the slurry maker is characterised by high shear rate and high intensity of turbulence. In the present study, the rotation speed of the TSSM was fixed at 800 rpm, and the shearing time at 30 s. A 280-t standard HPDC machine was used to produce the standard tensile test samples. For all the experiments in this investigation, the die temperature was kept constant at 235 C. The dimensions of the tensile test samples were 6 mm in gauge diameter, 60 mm in gauge length and 150 mm in total length. The as-cast samples were then subjected to various heat treatments with the optimised conditions identified previously [14]. The samples were solution treated at 413 C for 5 h and quenched in water (T4), following by aging at 216 C for 5.5 h (T6). T5 treatment was carried out at 216 C for 5 h. The evaluation of mechanical properties under as-cast and various heat treated conditions was carried out at room temperature on an Instron tensile test machine, with a strain rate of 6.7 10 4 s 1. The microstructures of the RDC samples with and without heat treatment were examined by optical microscopy (OM) with quantitative metallography. The specimens for OM were prepared by the standard technique of grinding with SiC abrasive paper and polishing with an Al 3 O 2 suspension solution, followed by etching using a solution of 5 vol.% concentrated HNO 3 in 95 vol.% ethanol. A Zeiss AxioVision optical imaging system was utilised for the OM observations and the quantitative measurements of the microstructure features. Scanning electron microscopy (SEM) was carried out using a Zeiss Supera 35 machine with a field emission gun, equipped with an energy dispersive spectroscopy (EDS) facility and operated at an accelerating voltage of 15 kv. The chemical composition analysis was conducted on a WAS Foundry Master with at least five burns on each of the three different areas of each sample. 3. Results 3.1. Microstructures Fig. 1 shows the general microstructure of AZ91D Mg-alloy produced by the RDC process from both primary alloy ingot and 100% diecast scrap. It can be seen that both RDC samples exhibit the typical microstructural features of AZ91D alloy produced by the RDC process. There is no obvious difference in microstructure between the two alloy samples. As shown in Fig. 1, there is a certain volume fraction of the spherical primary -Mg particles Table 1 Chemical compositions of the primary alloy used in this study in comparison with the ASTM standard specifications for AZ91D Mg-alloy (all in wt%) Element Zn Al Si Cu Mn Fe Ni Be Others Primary alloy 0.67 8.6 0.03 0.01 0.22 0.002 0.0007 0.001 <0.01 Alloy specifications [12] 0.45 0.9 8.5 9.5 0.05 0.015 >0.17 0.004 0.001 0.01

G. Liu et al. / Materials Science and Engineering A 472 (2008) 251 257 253 Fig. 2. SEM micrograph showing the detailed microstructure of RDC recycled AZ91D alloy from 100% scrap. Fig. 3 is a backscattered electron SEM image of the RDC recycled AZ91D alloy showing the inclusion particles. It can be seen that these particles have a spherical morphology and fine particle size (1 3 m), and are well dispersed in the remaining liquid matrix. EDS analysis of such particles (see Fig. 3) has confirmed that such inclusion particles contain mainly Al, Mn Fig. 1. Microstructure of RDC AZ91D Mg-alloy manufactured from (a) primary alloy ingot and (b) 100% scrap. being uniformly distributed in the entire section of the sample. The relatively large and spherical particles are formed inside the TSSM under high shear rate and high intensity of turbulence. The average particle size and the volume fraction of the primary -Mg globules were measured to be 37.8 m and 19.4% for the primary alloy. Quantitative metallography showed a very similar result for the RDC recycled alloy, with the average particle size and volume fraction of the primary -Mg phase being 37.6 m and 19.5%, respectively. Both alloy samples show a very high casting quality and high integrity with the total porosity being 0.2 0.5 vol.%. The remaining liquid of the semisolid slurry produced by the twin-screw slurry maker will undergo a further solidification in the shoot sleeve and inside the die cavity of the HPDC machine, which has been referred as to secondary solidification [15]. A few of -Mg dendrite fragments can be found randomly in both alloy samples, as shown in Fig. 1. Such dendrites were formed in the shot sleeve and then fragmented when they were passing through the narrow gate of the die, giving rise to the observed dendrite fragments in the final microstructure. The remaining liquid in the semisolid slurry finally solidified in the die cavity under high cooling rate, and the resulting microstructure is very fine -Mg particles (less than 10 m) defined by the eutectic -Mg 17 Al 12 phase, as shown in Fig. 2. Fig. 3. SEM backscattered electron image of RDC recycled AZ91D alloy showing the size and morphology of intermetallic particles (marked by the arrow). Also shown here are the EDS results from an intermetallic.

254 G. Liu et al. / Materials Science and Engineering A 472 (2008) 251 257 Fig. 4. Chemical compositions of AZ91D alloy processed under different conditions: (a) compositions of elements Al, Zn and Mn; (b) compositions of elements Si, Fe and Be. and O. It is more likely that such inclusion particles are Al 8 Mn 5 particles associated with O. Further quantitative EDS analysis revealed that some of the bright particles contained small amount of Fe. Detailed microstructural analysis has confirmed that neither oxide films nor clusters of oxide particles existed in the RDC recycled AZ91D alloy samples. 3.2. Chemical compositions There are three main alloying elements (Al, Zn, and Mn) and five minor detectable elements (Be, Si, Fe, Cu, and Ni) in the AZ91D Mg-alloy. Fig. 4 presents the chemical compositions of AZ91D alloy at different processing states and with different starting materials. It can be seen that for both alloying elements and impurity elements (except Fe), the slight changes in compositions occurred after melting of the primary alloy, and there were no obvious changes in compositions after RDC processing with different starting materials. After melting of the primary alloy ingot Al content increased by 0.31 wt% and Zn content decreased by 0.04 wt%. It should be noted that all these changes in composition are well within the alloy specifications (see Table 1). The standard deviation of chemical analysis is tabulated in Table 2. Therefore, it can be concluded that RDC processing of AZ91D diecast scrap (even with 100% scrap) produces samples within the ASTM specifications. Table 2 Chemical compositions of AZ91D Mg-alloy under different processing conditions (all in wt%) Element Al Zn Si Cu Mn Fe Ni Be Primary ingot 8.38 ± 0.75 0.78 ± 0.02 0.0425 ± 0.0006 0.041 ± 0.002 0.456 ± 0.032 0.0038 ± 0.0001 0.0030 0.0011 ± 0.0001 Primary molten 8.69 ± 0.78 0.74 ± 0.02 0.0428 ± 0.0005 0.046 ± 0.002 0.473 ± 0.023 0.0040 ± 0.0001 0.0040 0.0010 ± 0.0001 50% Scrap molten 8.75 ± 0.40 0.76 ± 0.02 0.0428 ± 0.0005 0.046 ± 0.003 0.476 ± 0.016 0.0042 ± 0.0001 0.0038 ± 0.0005 0.0010 100% Scrap molten 8.76 ± 0.46 0.75 ± 0.02 0.0432 ± 0.0005 0.048 ± 0.002 0.482 ± 0.015 0.0043 ± 0.0001 0.0040 0.0011 ± 0.0001 Primary RDC 8.66 ± 0.31 0.77 ± 0.02 0.0429 ± 0.0007 0.047 ± 0.003 0.472 ± 0.021 0.0041 ± 0.0001 0.0038 ± 0.0005 0.0010 ± 0.0001 50% Scrap RDC 8.8 ± 0.27 0.75 ± 0.04 0.044 ± 0.0019 0.047 ± 0.007 0.471 ± 0.02 0.0042 ± 0.0001 0.0045 ± 0.0005 0.0010 ± 0.0001 100% Scrap RDC 8.83 ± 0.34 0.75 ± 0.02 0.0452 ± 0.0016 0.050 ± 0.006 0.475 ± 0.012 0.0042 ± 0.0001 0.0045 ± 0.0005 0.0010

G. Liu et al. / Materials Science and Engineering A 472 (2008) 251 257 255 Fig. 5. Iron content of RDC AZ91D alloy processed from different starting materials. Fig. 6. Mechanical properties of RDC AZ91D alloy processed from different starting materials. The iron concentration of the RDC recycled AZ91D alloy is slightly higher than that in the primary alloy, as shown in Fig. 5. However, from Fig. 5, it is important to point out that, the RDC process does not increase Fe content in the final RDC castings compared with former melting stage. As shown in Table 2, for the primary alloy, Fe content increased from 38 ppm in the ingot to 40 ppm in the molten alloy and 41 ppm in the RDC sample; while for the scrap alloy, Fe content had no change after the RDC process. In general, Fe is picked up mainly from the use of iron equipment during melting. Table 2 also shows that, after RDC of AZ91D Mg-alloy, there was no obvious change in chemical concentrations for other impurity elements compared to that after melting, although Ni has increased from 40 ppm in the molten alloy to 45 ppm in the RDC recycled samples. 3.3. Mechanical properties of RDC recycled AZ91D alloy Table 3 summarises the tensile properties of the RDC recycled AZ91D alloy under various processing conditions. The mechanical properties for the primary alloy prepared at the same conditions are also given for comparison. Fig. 6 shows a direct comparison of the tensile properties of RDC AZ91D alloy produced with primary alloy, 50% Scrap and 100% scrap. It can be seen that both yield strength (YS) and ultimate tensile strength (UTS) for the RDC recycled Mg-alloy are comparable with that of the primary AZ91D alloy. However, recycling of the AZ91D scraps does cause decrease in ductility, but the reduction is comparably small, being 0.5% compared with that of the primary alloy. The RDC alloy recycled from 100% scraps has a considerably high elongation level of 6.4% at as-cast state, which is much higher than 3.0 3.3%, achieved by the normal HPDC process using primary alloy ingot [12,16]. Fig. 7 shows the strain stress curves for the RDC AZ91D alloy at different heat treatment conditions. It can be seen that T6 treatment improves the UTS significantly, whilst T5 treatment results in the highest YS but with some sacrifice of the elongation. It is important to observe that the solution treatment without aging (T4) provides a superior combination of elongation and UTS. In the present study, the achieved maximum UTS and elongation for the RDC AZ91D alloy were 264 MPa and 8.1% under as-cast state, and 278 MPa UTS and 13.9% under T4 treated condition. Table 3 Tensile properties of the RDC AZ91D alloy processed under different conditions Processing conditions Yield stress (MPa) UTS (MPa) Elongation (%) RDC as-cast Primary 139.8 ± 4.1 250.1 ± 7.6 6.86 ± 0.68 50% Scrap 139.6 ± 2.3 249.3 ± 8.2 6.67 ± 0.92 100% Scrap 139.6 ± 2.7 249.2 ± 6.6 6.39 ± 0.59 RDC+T4 Primary 96.9 ± 4.0 263.8 ± 10.7 11.61 ± 1.57 100% Scrap 95.9 ± 2.1 260.9 ± 9.0 11.33 ± 1.32 RDC+T5 Primary 142.8 ± 1.9 242.4 ± 8.9 5.27 ± 0.51 100% Scrap 144.9 ± 1.3 243.0 ± 7.3 4.54 ± 0.53 RDC+T6 Primary 134.0 ± 6.0 259.2 ± 8.5 5.59 ± 0.82 100% Scrap 134.1 ± 6.4 260.6 ± 12.6 4.76 ± 0.88 Fig. 7. Strain stress curves for RDC AZ91D alloy obtained under different processing conditions. Also shown here are the strength vs. elongation plot for some of the tested samples.

256 G. Liu et al. / Materials Science and Engineering A 472 (2008) 251 257 4. Discussion 4.1. Microstructural evolution Solidification of the AZ91D alloy during the RDC process has two distinct stages, i.e., the primary solidification under intensive forced convection inside the TSSM and the secondary solidification under high cooling rate inside the die cavity [15]. As shown in Fig. 1, the relatively larger primary -Mg particles, 38 m on an average, are formed inside the slurry maker during the primary solidification. The secondary solidification occurs when the semisolid slurry is transferred into the shot sleeve and is responsible for the solidification of the remaining liquid in the semisolid slurry. The high cooling rate (about 10 3 K/s) inside the die cavity is obviously responsible for the fine primary -Mg grains (less than 10 m in diameter) and the fine inter-granular eutectic -Mg 17 Al 12 phase, as shown in Fig. 2. In addition to the fine and homogeneous microstructure, the recycled alloy exhibits high integrity. The total porosity in the RDC alloy produced from 100% scrap was measured to be 0.2 0.5 vol.%, which is significantly less than the usual porosity levels observed for the conventional HPDC AZ91D alloy [12]. In addition, no oxide clusters or oxide films were found in the RDC recycled AZ91D alloy using the SEM/EDS analysis. Apart from the eutectic -Mg 17 Al 12 phase, it was found that the Mn-containing intermetallic particles observed in the recycled alloy are usually associated with oxygen and have an extremely fine size (1 3 m in diameter). This is a significant advantage of the RDC technology compared to the conventional diecasting technologies, where the oxide particle clusters, oxide films and intermetallic particles containing impurity elements were found in the cast microstructure usually with a large size and a non-uniform distribution [17]. 4.2. Mechanical properties The properties of magnesium alloy can be considerably affected by the incorporation of impurities within the recycled alloys. These chemical impurities can lead to a severe reduction in ductility. The ductility decrease up to 50% has been found in some magnesium alloy systems [4]. Therefore, the quality of recycled magnesium alloys has to be controlled carefully through chemical refining before diecasting. In this study, however, there is no considerable reduction in ductility for the RDC recycled alloy in comparison with the primary RDC alloy. The elongation can be as high as 6.4% for the 100% recycled alloy, which is more than double of that achievable by HPDC process using the same primary alloy [12]. It is also interesting to note that the YS and UTS of the recycled alloy are essentially kept on the same level as that for the primary alloy. The good combination of strength and ductility of the RDC recycled AZ91D alloy can be attributed to the following factors: (1) Much reduced porosity in the as cast microstructure. The porosity level has been reduced from 2 5 vol.% in the HPDC sample to 0.2 0.5 vol.% in the RDC sample. (2) Oxide and intermetallic compounds in the RDC samples have a fine size, spherical morphology and uniform distribution, and therefore, their harmful effect to mechanical properties has been effectively eliminated. (3) Fine and uniform microstructure throughout the entire RDC sample can promote uniform deformation, and hence eliminating stress concentration. 4.3. RDC process for recycling magnesium scrap A major challenge in recycling Mg-alloy scrap is to deal with the increased inclusions and impurity elements. In this work, we have developed a physical approach to recycle magnesium scrap using the RDC process. Instead of eliminating the inclusions and impurity elements themselves in the conventional chemical approaches, here we eliminate the harmful effects of the inclusions and impurity elements. Alloy melt prepared from scrap is physically treated under intensive forced convection to eliminate the detrimental effects of both inclusions and impurity elements, so that higher grade light alloy products can be produced from their scrap. Since the RDC recycled Mg-alloy has better mechanical properties than that of the primary alloy processed by the conventional technologies (e.g., HPDC process), this new approach is termed as upcycling, which is in contrast to the current concept of recycling where the scrap is converted into an existing grade of alloy often with inferior quality compared with the corresponding primary alloy grade. Compared with the conventional chemical approach to recycling, the physical approach developed here has the following advantages: (1) Through enhanced nucleation (known as effective nucleation [15]), the primary intermetallic compounds containing impurity elements will have a much finer particle size and a compact morphology (Fig. 3), resulting in improved ductility (Fig. 6). Therefore, intensive melt shearing can increase the tolerance of light alloys to impurity elements. (2) Inclusions, such as oxide particles and oxide films, will be pulverised (1 3 m in size) and dispersed uniformly throughout the alloy matrix, and become strengthening phases, rather than detrimental factors, leading to a significant improvement of mechanical properties by elimination of the stress concentration points. (3) Due to the chemical uniformity and fine and uniform microstructure provided by the RDC process, it is expected that the RDC recycled Mg-alloy will assist to improve corrosion resistance, due to the well dispersed intermetallic compounds containing Fe, Ni, or Cu. (4) The RDC process can be used to produce high quality castings directly from Mg-alloy scrap. Compared with the conventional chemical approach, it is more efficient, less energy intensive, lower cost and less environmental impact. Finally, it should be pointed out that, RDC process as a physical approach to recycling of Mg-alloy scrap does have its own limitations. It is more suitable for direct recycling of highergrade scrap, such as diecast scrap and well sorted end-of-life vehicle components. However, for lower grade scraps, such as furnace dross and old scrap mixed with other materials, the con-

G. Liu et al. / Materials Science and Engineering A 472 (2008) 251 257 257 ventional chemical approaches have to be used to purify the alloy to certain level before the RDC process can be used for direct recycling. 5. Conclusions (1) Rheo-diecasting (RDC) process can be used to recycle high grade Mg-alloy scrap to produce high integrity Mg-alloy castings. (2) The RDC recycled Mg-alloy has fine and uniform microstructure and extremely low porosity. (3) The RDC recycled AZ91D alloy has chemical compositions well within the ASTM specifications for the same alloy. (4) No oxide particle clusters and oxide films were found in the RDC recycled Mg-alloy. (5) Intermetallic compound particles containing the impurity elements have a fine size, spherical morphology and a uniform distribution in the alloy matrix. (6) The RDC recycled AZ91D alloy exhibits superior mechanical properties, particular elongation, over the primary AZ91D alloy processed by the conventional HPDC process. Acknowledgement The financial support from EPSRC, UK is acknowledged with gratitude. References [1] A. Lou, JOM 54 (2002) 42 48. [2] C. Scharf, C. Blawert, A. Ditze, in: K.U. Kainer (Ed.), Magnesium, Viley- VCH Verlag GmbH Co. KgaA, 2004, pp. 980 987. [3] H. Antrekowitsch, G. Hanko, P. Ebner, in: H.I. Kaplan (Ed.), Magnesium Technology 2002, TMS, Washington, 2002, pp. 43 49. [4] A. Javaid, E. Essadiqi, S. Bell, B. Davis, in: A.A. Luo, N.R. Neelameggham, R.S. Beals (Eds.), Magnesium Technology 2006, TMS, Texas, 2006, pp. 7 12. [5] C. Scharf, A. Ditze, in: B.L. Mordike, K.U. Kainer (Eds.), Magnesium Alloys and Their Applications, Werkstoff-Informationsgesellschaft GmbH, Wolfsburg, Germany, 1998, pp. 685 690. [6] R. Lisabeth, JOM 48 (1996) 44 46. [7] M.C. Flemings, Metall. Trans. A 22A (1991) 957 980. [8] Z. Fan, Int. Mater. Rev. 47 (2002) 49 85. [9] Z. Fan, M.J. Bevis, S. Ji, PCT Patent, WO 01/21343 A1, 1999. [10] Z. Fan, X. Fang, S. Ji, Mater. Sci. Eng. A A412 (2005) 298 306. [11] Z. Fan, G. Liu, Y. Wang, J. Mater. Sci. 41 (2006) 3631 3644. [12] M.M. Avedesian, H. Baker (Eds.), Magnesium and Magnesium Alloys, ASM International, Materials Park, 1999, pp. 230 231. [13] Z. Fan, Mater. Sci. Eng. A A413/A414 (2005) 72 78. [14] Y. Wang, G. Liu, Z. Fan, Acta Mater. 54 (2006) 689 699. [15] Z. Fan, G. Liu, Acta Mater. 53 (2005) 4345 4537. [16] C. Pitsaris, T. Abbott, C.H.J. Davies, G. Savage, in: K.U. Kainer (Ed.), Proceedings of the 6th International Conference on Mg-alloys and their Applications, Wiley-VCH, Weinheim, 2003, pp. 694 698. [17] L. Liu, F. Samuel, J. Mater. Sci. 33 (1998) 2269 2281.