- Email: [email protected]
AZ91D composites by liquid–solid extrusion directly following vacuum infiltration technique
Materials Science and Engineering A 531 (2012) 164–170
Contents lists available at SciVerse ScienceDirect
Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea
Compressive behavior of Csf /AZ91D composites by liquid–solid extrusion directly following vacuum infiltration technique J. Liu a , L.H. Qi a,b,∗ , J.T. Guan a , Y.Q. Ma a , J.M. Zhou a a b
School of Mechatronic Engineering, Northwestern Polytechnical University, Xi’an 710072, PR China Education Ministry Key Laboratory of Modern Design and Integrated Manufacturing Technology, Northwestern Polytechnical University, Xi’an 710072, PR China
a r t i c l e
i n f o
Article history: Received 6 August 2011 Received in revised form 18 October 2011 Accepted 18 October 2011 Available online 3 November 2011 Keywords: Magnesium matrix composites Liquid–solid extrusion Compressive behavior Microstructure Strengthening mechanism
a b s t r a c t 10 vol. % short carbon fiber reinforced AZ91D composites (Csf /AZ91D) were fabricated by liquid–solid extrusion directly following vacuum infiltration (LSEVI) technique. Liquid–solid extrusion of the composite induced reasonably uniform distribution and oriented arrangement of the carbon fibers. Compressive behaviors of the composites were investigated in the temperature range from room temperature to 300 ◦ C. The shapes of the compressive stress–strain curves at temperatures below and above 200 ◦ C are very different, which can be attributed to the combined influence of matrix work hardening and strain softening induced by the rotation of the fibers. The ultimate compressive strength (UCS) and compressive yield strength (CYS) of the composites are enhanced by 86.5% and 123% than those of matrix alloy at room temperature, respectively. The composites are thermal stable up to 200 ◦ C, where the CYS is approximately 2.8 times as high as that of the AZ91D matrix. However, both the UCS and CYS of the composites are slightly less than those of monolithic AZ91D at 300 ◦ C. The plastic deformation of the Csf /AZ91D composites mainly localizes in a shear band along the diagonal axis, 45◦ to the loading axis at the center of samples. The main failure mechanism of the composite samples is shear fracture or plasticity instability induced by shear deformation, and the failure strain increases with the increasing test temperature. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved.
1. Introduction Carbon fiber reinforced magnesium composites are gaining increasing interest as structural materials for applications in high-precision aerospace system, automotive industry and sports equipment owing to their low density, high specific strength, high damping capacity and excellent dimensional stability [1–3]. Short carbon fiber reinforced magnesium composites (Csf /Mg) are generally fabricated by stir casting [4], vacuum pressure infiltration method [5], squeeze casting method [6] and liquid–solid extrusion process following vacuum infiltration (LSEVI) developed by our team [7]. LSEVI is a special forming technique that combines the principles of vacuum infiltration, squeeze casting and liquid–solid extrusion. The major advantages of this process are elimination of porosity and shrinkage, low deformation resistance at liquid–solid state and near-net forming [8]. The Csf /AZ91D composites produced by this technique exhibited significantly improved room-temperature tensile properties with reasonably
∗ Corresponding author at: School of Mechatronic Engineering, Northwestern Polytechnical University, Xi’an 710072, PR China. Tel.: +86 29 88460447; fax: +86 29 88491982. E-mail address: [email protected] (L.H. Qi).
uniform distribution of fibers, good fiber–matrix interfacial bonding, and minimal amount of porosity [9]. In recent years, mechanical behavior of discontinuously reinforced magnesium matrix composites has been primarily investigated for their tension behavior, but a relatively little amount of research is conducted to investigate their compressive behavior, which is crucial in some engineering applications where compressive loads are dominant. Ataya and El-Magd [10] showed that the compressive yield stress of the magnesium matrix composites reinforced with 23 vol.% carbon short fibers increased approximately 3.5 times compared to AE42 matrix and approximately 2.5 times compared to AZ91 matrix. Towle and Friend [11] reported that an increase in compressive strength up to 92% and reduction of failure strain up to 79% was realized for magnesium matrix composites reinforced with saffil fibres (95 wt% ␦-alumina and 5 wt% silica; 3 m diameter and 150 m length). Trojanová et al. [12] reported that the short Saffil fiber reinforcing phase increased both the yield and maximum stresses, but the difference between the matrix and composite decreased with the increasing deformation temperature. Trojanová et al. [13] also showed that there was remarkable strength improvement at temperatures up to 200 ◦ C for (SiCp + Si)/AZ91 composites, and enhancement in ductility was found at temperatures from 200 ◦ C. These researches on the compressive behavior of magnesium matrix composites are
0921-5093/$ – see front matter. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2011.10.051
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
165
Table 1 Physical properties of T300 short carbon fibers. Fiber
Fiber diameter (m)
Ultimate strength (MPa)
Young’s modulus (GPa)
Elongation (%)
Density (g/cm3 )
T300
6–8
3500
230
1.5
1.76
mainly limited to the magnesium matrix composites produced using general liquid metal infiltration technique. While a systematic investigation on the effect of LSEVL process on the compressive behavior of the Csf /Mg composites has not yet been conducted, especially on the compressive properties at the elevated temperatures. In the present work, the room-temperature and elevatedtemperature compressive strength of the Csf /AZ91D composites by LSEVI were evaluated, and compared with AZ91D matrix. The deformation behaviors of the Csf /AZ91D composites during compression at different temperatures were investigated with the combination of the microstructure observation. Furthermore, the related failure mode and strengthening mechanism of the composites were also studied.
2. Experimental procedure 2.1. Preparation of composites The AZ91D alloy (Mg–9 wt%Al–1 wt%Zn–0.2 wt%Mn) was used as matrix alloy in the experiments. T300 short carbon fibers (Table 1 for physical properties) were selected as the reinforcement, and fabricated into preforms with a volume fraction of 10% by wet forming method without any binder. Then, the surfaces of carbon fibers were coated with pyrolytic carbon (PyC) using isothermal chemical vapor deposition (ICVD) process [14]. PyC layer can effectively reduce the degree of interfacial reaction by inhibiting the formation of brittle phase such as Al4 C3 and Al2 CMg2 , because the chemical activity of PyC is lower than that of carbon fiber. The schematic illustration of the LSEVI technique used to fabricate the Csf /AZ91D composites is shown in Fig. 1. Prior to infiltration, the preform and the melt of the matrix alloy were preheated to 580–620 ◦ C and 780–820 ◦ C under argon atmosphere,
respectively. After the preset temperatures were reached and maintained for 30–60 min, the container chamber was evacuated using a vacuum pump, and the melt was poured into container from crucible in melting unit (Fig. 1a). Then, argon gas pressure (0.3–0.6 MPa) was applied to the surface of melt to force the melt penetrate into the carbon fiber perform completely (Fig. 1b). After infiltration, the matrix melt was forced to solidify under high squeezing pressure of punch (Fig. 1c). Finally, the infiltrated composites containing a small fraction of liquid phase were extruded out of the female die when the die was cooled to the preset temperature of 380–420 ◦ C (Fig. 1d). 2.2. Compression tests Compression tests were performed on a computer-controlled servo-hydraulic testing machine in the temperature range of 20–300 ◦ C, at a constant crosshead speed giving an initial strain rate of 6.9 × 10−4 s−1 . Cylindrical specimens of 8 mm diameter and 12 mm length were machined along the extrusion direction. Before the test, the samples were heat-treated at 180 ◦ C for 24 h. The graphite powder was sprayed on two ends of the compression samples so that the effect of friction could be minimized. For comparison purpose, compression tests of monolithic AZ91D were also carried out under the same condition. 3. Results and discussion 3.1. Microstructure of the composites Fig. 2 shows the microstructure of matrix AZ91D alloy and obtained Csf /AZ91D composites by LSEVI. It can be seen that the microstructure of the matrix AZ91D alloy is characterized by coarse grains. By comparison, the grains in Csf /AZ91D composites are
Fig. 1. Schematic illustration of the liquid–solid extrusion directly following vacuum infiltration technique: (a) vacuum casting, (b) gas pressure infiltration, (c) squeeze casting, and (d) liquid–solid extrusion.
166
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
Fig. 3. X-ray diffraction results of Csf /AZ91D composites.
Fig. 2. Optical micrograph of the matrix AZ91D alloy (a) and Csf /AZ91D composites (b).
refined significantly. It can be attributed to severe deformation in extrusion and the addition of carbon fibers, which can serves either nucleation sites or obstacle to grain growth during solidification. According to the phase diagram of AZ91D alloy, the main constituents of the matrix alloy are solid solution (␣-Mg) and lamellar eutectic (composed of ␣-Mg and -Mg17 Al12 ) [15]. The interfacial region includes precipitated phase (-Mg17 Al12 ) and reactive phase (Al4 C3 ) according to the XRD pattern of Csf /AZ91D composites as shown in Fig. 3. It was observed that fiber agglomeration or pores seldom presented in the composites. Fibers were preferentially aligned along the direction of extrusion with an average aspect ratio of 10.
by flow softening, where stress gradually drops to about 175 MPa with the increasing strain. When the temperature rises to 300 ◦ C, compression tests show a more obvious strain softening behavior after a peak stress occurring at a very small strain (about 0.05). The shape of the stress–strain curves above their peak stresses generally depends on the failure mode of the compression samples. In order to correlate the compressive deformation behavior with failure feature and the microstructure of the composites, the samples after compression were sectioned into two parts for the microscopic observations as shown by the arrow in Fig. 5. Then, the mid-sections of the samples were ground and polished with diamond grinding paste. The metallographic samples were also observed by SEM to study the microstructure evolution and crack propagation during compression deformation at different test temperatures. It can be seen from Figs. 4 and 5 that after a little deformation, the samples directly split into two parts at room temperature and 100 ◦ C, and failed primarily via one main crack which was inclined at about 60◦ and 45◦ to the compressive axis, respectively. The failure was generally initiated from the sample ends or equator and occurred under the maximum stress. From 200 ◦ C, at higher strains, work softening was evident and this can be ascribed to the onset of plastic instability. Both samples deformed non-symmetrically, and failed by buckling not by barreling without obvious crack at 30% reduction of specimen height.
3.2. Compressive deformation behavior The typical compressive stress–strain curves of the Csf /AZ91D composites compressed at temperatures ranging from room temperature to 300 ◦ C are shown in Fig. 4. It can been seen that during initial work hardening after yielding, there are two similar concave regions in the curves at room temperature and 100 ◦ C. Beyond the concave regions, more rapid work hardening occurs until reaching their failure stress without further deformation. The similar compressive deformation behavior was reported in references [16–19], in which the deformation materials were extruded magnesium alloy or magnesium matrix composites with fine equiaxed grains. At 200 ◦ C, the curve is characterized by convex nature, and the work hardening steadily increases to a peak value (313.6 MPa) followed
Fig. 4. Compressive stress–strain curves of Csf /AZ91D composites at different temperatures.
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
167
Fig. 5. Composite samples before and after compression at different temperature, showing the failure mode and the sections for the microscopic observations.
Fig. 6 presents the polished longitudinal section of failed samples tested at different temperatures. It can be seen that matrix in composites hardly flowed before fracture due to low plasticity at room temperature (Fig. 6a). There are some subcracks on the sides of fracture surface. The subcracks may be induced during the propagation of main crack as the high stress concentration is ahead of the crack tip. When the test temperature rises to 100 ◦ C, the main crack propagated at an angle of 45◦ (Fig. 6b). The crack was accompanied by the formation of kink region with deflected fibers. The fibers within this region were broke due to the shear strain. At 200 ◦ C, an internal crack inclined at an angle of 45◦ to the compressive axis emerged after larger deformation (Fig. 6c). The fibers near the crack readjusted their orientation to the direction parallel to the shear direction during compressive deformation. At 300 ◦ C, no internal crack and fiber breakage were observed (Fig. 6d). As is well-known, aligned fibers parallel to the compressive direction in composites have the best load carrying ability, and the load carrying ability decreases with increasing the fiber orientation angle. Fibers rotation and breakage during compression deformation generally induce strain softening [20]. The higher the test temperature is, the
lower the deformation resistance and hardening rate of matrix are, and the more easily fibers can readjust themselves. Consequently, the effect of strain softening on the flow stress of composites would be more and more remarkable [21]. Strain softening induced mainly by the fiber rotation was dominant over the matrix work hardening at 300 ◦ C, resulting in the lower peak stress at small strain in the stress–strain curve of the composites as shown in Fig. 4. 3.3. Compressive properties Table 2 shows the compressive properties of Csf /AZ91D composites and monolithic AZ91D alloy at room temperature. The UCS and CYS are 512.2 MPa and 251.7 MPa, which are enhanced by about 86.5% and 122.5% than those of the matrix alloy, respectively. The compressive failure strain is similar to that of matrix alloy, does not reduce significantly like other fiber reinforced composites [11]. The improvement in ductility can be attributed to the reasonably uniform distribution of reinforcement and minimal amount of porosity, suitable interfacial bonding between matrix and fibers due to the PyC coating on the fibers as well as grain refinement
Fig. 6. Polished longitudinal section of failed samples tested at (a) RT, (b) 100 ◦ C, (c) 200 ◦ C, and (d) 300 ◦ C. and represent compressive stress and shear stress, respectively.
168
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
Table 2 Compressive properties of AZ91D and Csf /AZ91D composites samples at room temperature. Materials
UCS (MPa)
CYS (MPa)
UCS /a
CYS /
Total strain to failure (%)
AZ91D Csf /AZ91D composites
274.7 512.2
113.1 251.7
152.6 285.2
62.8 140.1
16.99 16.63
a
represents density.
occurred at 200 ◦ C. By comparison, the improvement ratio of UCS is lower than that of CYS of the composites. According to the strengthening theory of metal matrix composites, the improvement of Csf /AZ91D composites in CYS can be mainly attributed to several possible factors as follows [22–24]: (1) Hall–Petch effect due to grain size refinement, (2) load transfer from matrix to fibers, and (3) enhanced dislocation density due to the CTE (coefficient of thermal expansion) mismatch between the matrix and the carbon fibers. The improvement in CYS regarding above factors can be expressed as follows:
=
Fig. 7. Ultimate compressive strength (UCS) of the Csf /AZ91D composites compared to the corresponding matrix alloy.
2
2
(grain ) + (load ) + (thermal )
(2)
where Hall–Petch slope k varies over a wide range with different deformation mechanisms for extruded magnesium alloy, and it is a function of temperature and grain size. The values of k for twinning-dominated flow at low temperature is a constant (6.7 MPa mm1/2 ), which is higher by a factor of 2–10 times than that in slip-dominated and transition regions [16]. The grain size (d) of Csf /AZ91D was measured by a mean linear intercept method, equal 10 m. Hence, grain is about 67 MPa at room temperature according to Eq. (2). With the modified shear-lag model, the improvement associated with load transfer is given by load = ym
(L + t)A 4L
f + ym (1 − f )
(3)
where ym is the yield stress of the unreinforced matrix, f is the fiber volume friction, L is the fibre size in the direction of the applied stress, t is the fiber size in the perpendicular direction and A is the fiber aspect ratio (L/t). The strengthening component load at room temperature gives a value of 132.9 MPa (taking for A = 10, f = 0.1, and ym = 113.1 MPa) according to Eq. (3). A typical TEM image of the Csf /AZ91D composites reveals intense dislocation density around the fiber/matrix interface as shown in Fig. 9. The dislocation is related to the large difference in the CTE between the matrix and carbon fibers. The strengthening due to the thermal mismatch can be expressed as
thermal =
Fig. 8. Compressive yield strength (CYS) of the Csf /AZ91D composites compared to the corresponding matrix alloy.
(1)
where grain is the improvement associated with grain size refinement, load is the improvement associated with load transfer, and thermal is the improvement associated with the increase in dislocation density due to thermal mismatch. The contribution to the yield stress by grain refinement can be estimated using Hall–Petch relation as grain = kd−1/2
induced by LSEV [9]. Fig. 7 shows the UCS of the composites at different temperatures compared to the corresponding matrix alloys. Both UCS of Csf /AZ91D composites and monolithic AZ91D alloy decrease with the increasing test temperature. The UCS of composites is 1.86, 1.72 and 1.20 times as high as that of AZ91D matrix alloy at room temperature, 100 ◦ C and 200 ◦ C, respectively. A large strength improvement is confirmed for Csf /AZ91D composites at temperature up to 200 ◦ C. At 300 ◦ C, the UCS is slightly smaller than that of the matrix alloy. The CYS of Csf /AZ91D composites as a function of temperature are shown in Fig. 8. Similar trend is found that the CYS of Csf /AZ91D composite decreases with the increasing test temperature, and it is slightly smaller than that of the matrix alloy at 300 ◦ C. However, the improvement for CYS is not decrease monotonically with the increasing test temperature. The CYS of Csf /AZ91D composites are 2.23, 2.15 and 2.81 times as high as that of AZ91D matrix alloy at RT, 100 ◦ C and 200 ◦ C, respectively. The highest improvement ratio
2
˛• Gm • b•
Bf˛T bd(1 − f )
(4)
where ˛ is a constant, G is the shear modulus, b is the magnitude of the Burgers vector, B is a geometrical constant, ˛ is the difference between the two coefficients of thermal expansion, and T is the difference between the processing temperature and the test temperature. Taking into account the thermal dislocations, we obtain a contribution to the CYS of about 48.9 MPa (with B = 10, ˛ = 20 × 10−6 /K, ˛ = 0.35, b = 3.2 × 10−10 m, and G = 16.6 Gpa [12,25]). Finally, inserting above strengthening terms in Eq. (1), the improvement in CYS can be calculated as =
672 + 132.92 + 48.92 = 156.7MPa. The calculated value is
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
169
Mg17 Al12 has a nonnegligible influence on the mechanical behavior of Csf /AZ91D composites. Mg17 Al12 mainly distributes near the grain boundary and the C/Mg interface [15], and it is a hard and brittle phase at room temperature, which can play an effective role in pinning the movement of dislocations during deformation. Thereby, the mechanical properties of Csf /AZ91D composites are improved significantly. However, Mg17 Al12 is thermally unstable at temperature above 120–130 ◦ C, where it becomes soft and stimulates the softening of the matrix at elevated temperature [10]. Both the UCS and CYS of the Csf /AZ91D composites are smaller than those of the matrix alloy at 300 ◦ C, it may be attributed to a different deformation mechanism, which causes weakening of the composites. The similar observation was also reported by Syu and Ghosh in 2014A1/15 vol.% A12 O3 composite under a hot tensile test [26]. 4. Conclusions
Fig. 9. Dislocations in the matrix close to a carbon fibre interface.
slightly higher than the measured one of 138.6 MPa. The difference may be attributed to the hypothesis that all the fibers are aligned directionally along the compression direction. In fact, the parallel fibers in the compressive samples are not completely aligned along the load direction, there are a few fibers randomly oriented as shown in Fig. 10. They lost load transfer ability due to interface debonding under shear stress. In summary, the main mechanism determining the strengthening of Csf /AZ91D is mainly the stress transfer. The CYS of matrix alloy deceases with the increasing temperature as shown in Fig. 8, consequently, load decreases according to Eq. (3) as well. Furthermore, thermal stress in the interface region of composites decreases with the increasing temperature according to Eq. (4). Hence, the CYS of Csf /AZ91D composites decreases with the increasing temperature consequently. The differences of the CYS between Csf /AZ91D composites and matrix alloy are 138.6 MPa, 126.7 MPa and 121.2 MPa at room temperature, 100 ◦ C, and 200 ◦ C, respectively. At 300 ◦ C, the CYS of Csf /AZ91D composites is even smaller than that of the AZ91D alloy. Moreover, massive precipitate of
(1) Well-aligned fibers with uniform distribution and porosityfree microstructure were obtained in Csf /AZ91D composite fabricated by LSEVI. The room-temperature UCS and CYS of the composites were improved by 86.5% and 123% than those of the monolithic AZ91D alloy, respectively. (2) The Csf /AZ91D composite fabricated by LSEVI exhibited desirable thermal stability up to 200 ◦ C, and the CYS was approximately 2.15–2.81 times as high as that of the AZ91D matrix. Both the UCS and CYS of the Csf /AZ91D composites were smaller than those of the matrix alloy at 300 ◦ C. (3) The compression curves of Csf /AZ91D composites were characterized by high work hardening behavior at temperatures below 200 ◦ C. After reaching the peak stresses, the composites shown a sudden failure due to their low plasticity. From 200 ◦ C, strain softening induced mainly by the fiber rotation was dominant over the matrix work hardening, and the deformability of the composites was significantly improved. (4) The deformation of Csf /AZ91D composite fabricated by LSEVI mainly localized to the shear band region in sample center during compression. The main failure mode was shear fracture or plasticity instability. Acknowledgements The authors are grateful for the financial supports from the National Nature Science Foundation of China (No. 50972121), Technological Innovations Foundation of Northwestern Polytechnical University (No. 2010KJ0202). References
Fig. 10. Fractograph of Csf /AZ91D composites compressed at room temperature.
[1] H.Z. Ye, X.Y. Liu, J. Mater. Sci. 39 (2004) 6153–6171. [2] M. Russell-Stevens, R. Todd, M. Papakyriacou, Mater. Sci. Eng. A 397 (2005) 249–256. [3] H.B. Ou-Yang, H.J. Li, L.H. Qi, Z.J. Li, L.Z. Su, J.F. Wei, Rare Metal Mat. Eng. 38 (2009) 931–935. [4] Y. Huang, Master Thesis, Harbin Institute of Technology, 2007. [5] W. Hufenbach, M. Andrich, A. Langkamp, A. Czulak, J. Mater. Process. Technol. 175 (2006) 218–224. [6] M. Russell-Stevens, D.C. Plane, J. Summerscales, P. Schulz, M. Papakyriacou, Mater. Sci. Technol. 18 (2002) 501–506. [7] L.H. Qi, J.M. Zhou, L.Z. Su, H.B. Ou-Yang, H.J. Li, Solid State Phenom. 141-143 (2008) 91–96. [8] J.T. Guan, L.H. Qi, L.Z. Su, J. Liu, Adv. Mater. Res. 97-101 (2010) 382–385. [9] J.M. Zhou, L.H. Qi, H.B. Ou-Yang, H.J. Li, Adv. Mater. Res. 89-91 (2010) 692–696. [10] S. Ataya, E. El-Magd, Theor. Appl. Fract. Mech. 47 (2007) 102–112. [11] D.J. Towle, C.M. Friend, Mater. Sci. Technol. 9 (1993) 35–41. [12] Z. Trojanová, V. Gärtnerová, P. Lukáˇc, Z. Drozd, J. Alloys Compd. 378 (2004) 19–26. ´ M. Chalupová, P. Palˇcek, P. [13] Z. Trojanová, V. Gärtnerová, A. Jäger, A. Námeˇsny, Lukáˇc, Compos. Sci. Technol. 69 (2009) 2256–2264. [14] H.B. Ou-Yang, H.J. Li, L.H. Qi, Z.J. Li, T. Fang, J.F. Wei, J. Mater. Sci. 43 (2008) 4618–4624.
170
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
[15] M. Russell-Stevens, R. Todd, M. Papakyriacou, Surf. Interface Anal. 37 (2005) 336–342. [16] M.R. Barnett, Z. Keshavarz, A.G. Beer, D. Atwell, Acta Mater. 52 (2004) 5093–5103. [17] Q.B. Nguyen, M. Gupta, Compos. Sci. Technol. 68 (2008) 2185–2192. [18] K.S. Tun, M. Gupta, Mater. Sci. Eng. A 527 (2010) 5550–5556. [19] A.K. Khanra, H.C. Jung, K.S. Hong, K.S. Shin, Mater. Sci. Eng. A 527 (2010) 6283–6288. [20] L. Geng, A.B. Li, Q.Y. Meng, Mater. Sci. Eng. A 386 (2004) 212–221.
[21] A.B. Li, Q.Y. Meng, L. Geng, C.F. Deng, Rare Metals 26 (2007) 182–192. [22] Q.Q. Li, A. Viereckl, C.A. Rottmair, R.F. Singer, Compos. Sci. Technol. 69 (2009) 1193–1199. [23] Z. Száraz, Z. Trojanová, M. Cabbibo, E. Evangelista, Mater. Sci. Eng. A 462 (2007) 225–229. [24] C.S. Goh, J. Wei, L.C. Lee, M. Gupta, Scripta Mater. 55 (2007) 5115–5121. [25] W.J. Kim, H.G. Jeong, H.T. Jeong, Scripta Mater. 61 (2009) 1040–1043. [26] D.-G.C. Syu, A.K. Ghosh, Mater. Sci. Eng. A 184 (1994) 27–35.
Contents lists available at SciVerse ScienceDirect
Materials Science and Engineering A journal homepage: www.elsevier.com/locate/msea
Compressive behavior of Csf /AZ91D composites by liquid–solid extrusion directly following vacuum infiltration technique J. Liu a , L.H. Qi a,b,∗ , J.T. Guan a , Y.Q. Ma a , J.M. Zhou a a b
School of Mechatronic Engineering, Northwestern Polytechnical University, Xi’an 710072, PR China Education Ministry Key Laboratory of Modern Design and Integrated Manufacturing Technology, Northwestern Polytechnical University, Xi’an 710072, PR China
a r t i c l e
i n f o
Article history: Received 6 August 2011 Received in revised form 18 October 2011 Accepted 18 October 2011 Available online 3 November 2011 Keywords: Magnesium matrix composites Liquid–solid extrusion Compressive behavior Microstructure Strengthening mechanism
a b s t r a c t 10 vol. % short carbon fiber reinforced AZ91D composites (Csf /AZ91D) were fabricated by liquid–solid extrusion directly following vacuum infiltration (LSEVI) technique. Liquid–solid extrusion of the composite induced reasonably uniform distribution and oriented arrangement of the carbon fibers. Compressive behaviors of the composites were investigated in the temperature range from room temperature to 300 ◦ C. The shapes of the compressive stress–strain curves at temperatures below and above 200 ◦ C are very different, which can be attributed to the combined influence of matrix work hardening and strain softening induced by the rotation of the fibers. The ultimate compressive strength (UCS) and compressive yield strength (CYS) of the composites are enhanced by 86.5% and 123% than those of matrix alloy at room temperature, respectively. The composites are thermal stable up to 200 ◦ C, where the CYS is approximately 2.8 times as high as that of the AZ91D matrix. However, both the UCS and CYS of the composites are slightly less than those of monolithic AZ91D at 300 ◦ C. The plastic deformation of the Csf /AZ91D composites mainly localizes in a shear band along the diagonal axis, 45◦ to the loading axis at the center of samples. The main failure mechanism of the composite samples is shear fracture or plasticity instability induced by shear deformation, and the failure strain increases with the increasing test temperature. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved.
1. Introduction Carbon fiber reinforced magnesium composites are gaining increasing interest as structural materials for applications in high-precision aerospace system, automotive industry and sports equipment owing to their low density, high specific strength, high damping capacity and excellent dimensional stability [1–3]. Short carbon fiber reinforced magnesium composites (Csf /Mg) are generally fabricated by stir casting [4], vacuum pressure infiltration method [5], squeeze casting method [6] and liquid–solid extrusion process following vacuum infiltration (LSEVI) developed by our team [7]. LSEVI is a special forming technique that combines the principles of vacuum infiltration, squeeze casting and liquid–solid extrusion. The major advantages of this process are elimination of porosity and shrinkage, low deformation resistance at liquid–solid state and near-net forming [8]. The Csf /AZ91D composites produced by this technique exhibited significantly improved room-temperature tensile properties with reasonably
∗ Corresponding author at: School of Mechatronic Engineering, Northwestern Polytechnical University, Xi’an 710072, PR China. Tel.: +86 29 88460447; fax: +86 29 88491982. E-mail address: [email protected] (L.H. Qi).
uniform distribution of fibers, good fiber–matrix interfacial bonding, and minimal amount of porosity [9]. In recent years, mechanical behavior of discontinuously reinforced magnesium matrix composites has been primarily investigated for their tension behavior, but a relatively little amount of research is conducted to investigate their compressive behavior, which is crucial in some engineering applications where compressive loads are dominant. Ataya and El-Magd [10] showed that the compressive yield stress of the magnesium matrix composites reinforced with 23 vol.% carbon short fibers increased approximately 3.5 times compared to AE42 matrix and approximately 2.5 times compared to AZ91 matrix. Towle and Friend [11] reported that an increase in compressive strength up to 92% and reduction of failure strain up to 79% was realized for magnesium matrix composites reinforced with saffil fibres (95 wt% ␦-alumina and 5 wt% silica; 3 m diameter and 150 m length). Trojanová et al. [12] reported that the short Saffil fiber reinforcing phase increased both the yield and maximum stresses, but the difference between the matrix and composite decreased with the increasing deformation temperature. Trojanová et al. [13] also showed that there was remarkable strength improvement at temperatures up to 200 ◦ C for (SiCp + Si)/AZ91 composites, and enhancement in ductility was found at temperatures from 200 ◦ C. These researches on the compressive behavior of magnesium matrix composites are
0921-5093/$ – see front matter. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved. doi:10.1016/j.msea.2011.10.051
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
165
Table 1 Physical properties of T300 short carbon fibers. Fiber
Fiber diameter (m)
Ultimate strength (MPa)
Young’s modulus (GPa)
Elongation (%)
Density (g/cm3 )
T300
6–8
3500
230
1.5
1.76
mainly limited to the magnesium matrix composites produced using general liquid metal infiltration technique. While a systematic investigation on the effect of LSEVL process on the compressive behavior of the Csf /Mg composites has not yet been conducted, especially on the compressive properties at the elevated temperatures. In the present work, the room-temperature and elevatedtemperature compressive strength of the Csf /AZ91D composites by LSEVI were evaluated, and compared with AZ91D matrix. The deformation behaviors of the Csf /AZ91D composites during compression at different temperatures were investigated with the combination of the microstructure observation. Furthermore, the related failure mode and strengthening mechanism of the composites were also studied.
2. Experimental procedure 2.1. Preparation of composites The AZ91D alloy (Mg–9 wt%Al–1 wt%Zn–0.2 wt%Mn) was used as matrix alloy in the experiments. T300 short carbon fibers (Table 1 for physical properties) were selected as the reinforcement, and fabricated into preforms with a volume fraction of 10% by wet forming method without any binder. Then, the surfaces of carbon fibers were coated with pyrolytic carbon (PyC) using isothermal chemical vapor deposition (ICVD) process [14]. PyC layer can effectively reduce the degree of interfacial reaction by inhibiting the formation of brittle phase such as Al4 C3 and Al2 CMg2 , because the chemical activity of PyC is lower than that of carbon fiber. The schematic illustration of the LSEVI technique used to fabricate the Csf /AZ91D composites is shown in Fig. 1. Prior to infiltration, the preform and the melt of the matrix alloy were preheated to 580–620 ◦ C and 780–820 ◦ C under argon atmosphere,
respectively. After the preset temperatures were reached and maintained for 30–60 min, the container chamber was evacuated using a vacuum pump, and the melt was poured into container from crucible in melting unit (Fig. 1a). Then, argon gas pressure (0.3–0.6 MPa) was applied to the surface of melt to force the melt penetrate into the carbon fiber perform completely (Fig. 1b). After infiltration, the matrix melt was forced to solidify under high squeezing pressure of punch (Fig. 1c). Finally, the infiltrated composites containing a small fraction of liquid phase were extruded out of the female die when the die was cooled to the preset temperature of 380–420 ◦ C (Fig. 1d). 2.2. Compression tests Compression tests were performed on a computer-controlled servo-hydraulic testing machine in the temperature range of 20–300 ◦ C, at a constant crosshead speed giving an initial strain rate of 6.9 × 10−4 s−1 . Cylindrical specimens of 8 mm diameter and 12 mm length were machined along the extrusion direction. Before the test, the samples were heat-treated at 180 ◦ C for 24 h. The graphite powder was sprayed on two ends of the compression samples so that the effect of friction could be minimized. For comparison purpose, compression tests of monolithic AZ91D were also carried out under the same condition. 3. Results and discussion 3.1. Microstructure of the composites Fig. 2 shows the microstructure of matrix AZ91D alloy and obtained Csf /AZ91D composites by LSEVI. It can be seen that the microstructure of the matrix AZ91D alloy is characterized by coarse grains. By comparison, the grains in Csf /AZ91D composites are
Fig. 1. Schematic illustration of the liquid–solid extrusion directly following vacuum infiltration technique: (a) vacuum casting, (b) gas pressure infiltration, (c) squeeze casting, and (d) liquid–solid extrusion.
166
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
Fig. 3. X-ray diffraction results of Csf /AZ91D composites.
Fig. 2. Optical micrograph of the matrix AZ91D alloy (a) and Csf /AZ91D composites (b).
refined significantly. It can be attributed to severe deformation in extrusion and the addition of carbon fibers, which can serves either nucleation sites or obstacle to grain growth during solidification. According to the phase diagram of AZ91D alloy, the main constituents of the matrix alloy are solid solution (␣-Mg) and lamellar eutectic (composed of ␣-Mg and -Mg17 Al12 ) [15]. The interfacial region includes precipitated phase (-Mg17 Al12 ) and reactive phase (Al4 C3 ) according to the XRD pattern of Csf /AZ91D composites as shown in Fig. 3. It was observed that fiber agglomeration or pores seldom presented in the composites. Fibers were preferentially aligned along the direction of extrusion with an average aspect ratio of 10.
by flow softening, where stress gradually drops to about 175 MPa with the increasing strain. When the temperature rises to 300 ◦ C, compression tests show a more obvious strain softening behavior after a peak stress occurring at a very small strain (about 0.05). The shape of the stress–strain curves above their peak stresses generally depends on the failure mode of the compression samples. In order to correlate the compressive deformation behavior with failure feature and the microstructure of the composites, the samples after compression were sectioned into two parts for the microscopic observations as shown by the arrow in Fig. 5. Then, the mid-sections of the samples were ground and polished with diamond grinding paste. The metallographic samples were also observed by SEM to study the microstructure evolution and crack propagation during compression deformation at different test temperatures. It can be seen from Figs. 4 and 5 that after a little deformation, the samples directly split into two parts at room temperature and 100 ◦ C, and failed primarily via one main crack which was inclined at about 60◦ and 45◦ to the compressive axis, respectively. The failure was generally initiated from the sample ends or equator and occurred under the maximum stress. From 200 ◦ C, at higher strains, work softening was evident and this can be ascribed to the onset of plastic instability. Both samples deformed non-symmetrically, and failed by buckling not by barreling without obvious crack at 30% reduction of specimen height.
3.2. Compressive deformation behavior The typical compressive stress–strain curves of the Csf /AZ91D composites compressed at temperatures ranging from room temperature to 300 ◦ C are shown in Fig. 4. It can been seen that during initial work hardening after yielding, there are two similar concave regions in the curves at room temperature and 100 ◦ C. Beyond the concave regions, more rapid work hardening occurs until reaching their failure stress without further deformation. The similar compressive deformation behavior was reported in references [16–19], in which the deformation materials were extruded magnesium alloy or magnesium matrix composites with fine equiaxed grains. At 200 ◦ C, the curve is characterized by convex nature, and the work hardening steadily increases to a peak value (313.6 MPa) followed
Fig. 4. Compressive stress–strain curves of Csf /AZ91D composites at different temperatures.
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
167
Fig. 5. Composite samples before and after compression at different temperature, showing the failure mode and the sections for the microscopic observations.
Fig. 6 presents the polished longitudinal section of failed samples tested at different temperatures. It can be seen that matrix in composites hardly flowed before fracture due to low plasticity at room temperature (Fig. 6a). There are some subcracks on the sides of fracture surface. The subcracks may be induced during the propagation of main crack as the high stress concentration is ahead of the crack tip. When the test temperature rises to 100 ◦ C, the main crack propagated at an angle of 45◦ (Fig. 6b). The crack was accompanied by the formation of kink region with deflected fibers. The fibers within this region were broke due to the shear strain. At 200 ◦ C, an internal crack inclined at an angle of 45◦ to the compressive axis emerged after larger deformation (Fig. 6c). The fibers near the crack readjusted their orientation to the direction parallel to the shear direction during compressive deformation. At 300 ◦ C, no internal crack and fiber breakage were observed (Fig. 6d). As is well-known, aligned fibers parallel to the compressive direction in composites have the best load carrying ability, and the load carrying ability decreases with increasing the fiber orientation angle. Fibers rotation and breakage during compression deformation generally induce strain softening [20]. The higher the test temperature is, the
lower the deformation resistance and hardening rate of matrix are, and the more easily fibers can readjust themselves. Consequently, the effect of strain softening on the flow stress of composites would be more and more remarkable [21]. Strain softening induced mainly by the fiber rotation was dominant over the matrix work hardening at 300 ◦ C, resulting in the lower peak stress at small strain in the stress–strain curve of the composites as shown in Fig. 4. 3.3. Compressive properties Table 2 shows the compressive properties of Csf /AZ91D composites and monolithic AZ91D alloy at room temperature. The UCS and CYS are 512.2 MPa and 251.7 MPa, which are enhanced by about 86.5% and 122.5% than those of the matrix alloy, respectively. The compressive failure strain is similar to that of matrix alloy, does not reduce significantly like other fiber reinforced composites [11]. The improvement in ductility can be attributed to the reasonably uniform distribution of reinforcement and minimal amount of porosity, suitable interfacial bonding between matrix and fibers due to the PyC coating on the fibers as well as grain refinement
Fig. 6. Polished longitudinal section of failed samples tested at (a) RT, (b) 100 ◦ C, (c) 200 ◦ C, and (d) 300 ◦ C. and represent compressive stress and shear stress, respectively.
168
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
Table 2 Compressive properties of AZ91D and Csf /AZ91D composites samples at room temperature. Materials
UCS (MPa)
CYS (MPa)
UCS /a
CYS /
Total strain to failure (%)
AZ91D Csf /AZ91D composites
274.7 512.2
113.1 251.7
152.6 285.2
62.8 140.1
16.99 16.63
a
represents density.
occurred at 200 ◦ C. By comparison, the improvement ratio of UCS is lower than that of CYS of the composites. According to the strengthening theory of metal matrix composites, the improvement of Csf /AZ91D composites in CYS can be mainly attributed to several possible factors as follows [22–24]: (1) Hall–Petch effect due to grain size refinement, (2) load transfer from matrix to fibers, and (3) enhanced dislocation density due to the CTE (coefficient of thermal expansion) mismatch between the matrix and the carbon fibers. The improvement in CYS regarding above factors can be expressed as follows:
=
Fig. 7. Ultimate compressive strength (UCS) of the Csf /AZ91D composites compared to the corresponding matrix alloy.
2
2
(grain ) + (load ) + (thermal )
(2)
where Hall–Petch slope k varies over a wide range with different deformation mechanisms for extruded magnesium alloy, and it is a function of temperature and grain size. The values of k for twinning-dominated flow at low temperature is a constant (6.7 MPa mm1/2 ), which is higher by a factor of 2–10 times than that in slip-dominated and transition regions [16]. The grain size (d) of Csf /AZ91D was measured by a mean linear intercept method, equal 10 m. Hence, grain is about 67 MPa at room temperature according to Eq. (2). With the modified shear-lag model, the improvement associated with load transfer is given by load = ym
(L + t)A 4L
f + ym (1 − f )
(3)
where ym is the yield stress of the unreinforced matrix, f is the fiber volume friction, L is the fibre size in the direction of the applied stress, t is the fiber size in the perpendicular direction and A is the fiber aspect ratio (L/t). The strengthening component load at room temperature gives a value of 132.9 MPa (taking for A = 10, f = 0.1, and ym = 113.1 MPa) according to Eq. (3). A typical TEM image of the Csf /AZ91D composites reveals intense dislocation density around the fiber/matrix interface as shown in Fig. 9. The dislocation is related to the large difference in the CTE between the matrix and carbon fibers. The strengthening due to the thermal mismatch can be expressed as
thermal =
Fig. 8. Compressive yield strength (CYS) of the Csf /AZ91D composites compared to the corresponding matrix alloy.
(1)
where grain is the improvement associated with grain size refinement, load is the improvement associated with load transfer, and thermal is the improvement associated with the increase in dislocation density due to thermal mismatch. The contribution to the yield stress by grain refinement can be estimated using Hall–Petch relation as grain = kd−1/2
induced by LSEV [9]. Fig. 7 shows the UCS of the composites at different temperatures compared to the corresponding matrix alloys. Both UCS of Csf /AZ91D composites and monolithic AZ91D alloy decrease with the increasing test temperature. The UCS of composites is 1.86, 1.72 and 1.20 times as high as that of AZ91D matrix alloy at room temperature, 100 ◦ C and 200 ◦ C, respectively. A large strength improvement is confirmed for Csf /AZ91D composites at temperature up to 200 ◦ C. At 300 ◦ C, the UCS is slightly smaller than that of the matrix alloy. The CYS of Csf /AZ91D composites as a function of temperature are shown in Fig. 8. Similar trend is found that the CYS of Csf /AZ91D composite decreases with the increasing test temperature, and it is slightly smaller than that of the matrix alloy at 300 ◦ C. However, the improvement for CYS is not decrease monotonically with the increasing test temperature. The CYS of Csf /AZ91D composites are 2.23, 2.15 and 2.81 times as high as that of AZ91D matrix alloy at RT, 100 ◦ C and 200 ◦ C, respectively. The highest improvement ratio
2
˛• Gm • b•
Bf˛T bd(1 − f )
(4)
where ˛ is a constant, G is the shear modulus, b is the magnitude of the Burgers vector, B is a geometrical constant, ˛ is the difference between the two coefficients of thermal expansion, and T is the difference between the processing temperature and the test temperature. Taking into account the thermal dislocations, we obtain a contribution to the CYS of about 48.9 MPa (with B = 10, ˛ = 20 × 10−6 /K, ˛ = 0.35, b = 3.2 × 10−10 m, and G = 16.6 Gpa [12,25]). Finally, inserting above strengthening terms in Eq. (1), the improvement in CYS can be calculated as =
672 + 132.92 + 48.92 = 156.7MPa. The calculated value is
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
169
Mg17 Al12 has a nonnegligible influence on the mechanical behavior of Csf /AZ91D composites. Mg17 Al12 mainly distributes near the grain boundary and the C/Mg interface [15], and it is a hard and brittle phase at room temperature, which can play an effective role in pinning the movement of dislocations during deformation. Thereby, the mechanical properties of Csf /AZ91D composites are improved significantly. However, Mg17 Al12 is thermally unstable at temperature above 120–130 ◦ C, where it becomes soft and stimulates the softening of the matrix at elevated temperature [10]. Both the UCS and CYS of the Csf /AZ91D composites are smaller than those of the matrix alloy at 300 ◦ C, it may be attributed to a different deformation mechanism, which causes weakening of the composites. The similar observation was also reported by Syu and Ghosh in 2014A1/15 vol.% A12 O3 composite under a hot tensile test [26]. 4. Conclusions
Fig. 9. Dislocations in the matrix close to a carbon fibre interface.
slightly higher than the measured one of 138.6 MPa. The difference may be attributed to the hypothesis that all the fibers are aligned directionally along the compression direction. In fact, the parallel fibers in the compressive samples are not completely aligned along the load direction, there are a few fibers randomly oriented as shown in Fig. 10. They lost load transfer ability due to interface debonding under shear stress. In summary, the main mechanism determining the strengthening of Csf /AZ91D is mainly the stress transfer. The CYS of matrix alloy deceases with the increasing temperature as shown in Fig. 8, consequently, load decreases according to Eq. (3) as well. Furthermore, thermal stress in the interface region of composites decreases with the increasing temperature according to Eq. (4). Hence, the CYS of Csf /AZ91D composites decreases with the increasing temperature consequently. The differences of the CYS between Csf /AZ91D composites and matrix alloy are 138.6 MPa, 126.7 MPa and 121.2 MPa at room temperature, 100 ◦ C, and 200 ◦ C, respectively. At 300 ◦ C, the CYS of Csf /AZ91D composites is even smaller than that of the AZ91D alloy. Moreover, massive precipitate of
(1) Well-aligned fibers with uniform distribution and porosityfree microstructure were obtained in Csf /AZ91D composite fabricated by LSEVI. The room-temperature UCS and CYS of the composites were improved by 86.5% and 123% than those of the monolithic AZ91D alloy, respectively. (2) The Csf /AZ91D composite fabricated by LSEVI exhibited desirable thermal stability up to 200 ◦ C, and the CYS was approximately 2.15–2.81 times as high as that of the AZ91D matrix. Both the UCS and CYS of the Csf /AZ91D composites were smaller than those of the matrix alloy at 300 ◦ C. (3) The compression curves of Csf /AZ91D composites were characterized by high work hardening behavior at temperatures below 200 ◦ C. After reaching the peak stresses, the composites shown a sudden failure due to their low plasticity. From 200 ◦ C, strain softening induced mainly by the fiber rotation was dominant over the matrix work hardening, and the deformability of the composites was significantly improved. (4) The deformation of Csf /AZ91D composite fabricated by LSEVI mainly localized to the shear band region in sample center during compression. The main failure mode was shear fracture or plasticity instability. Acknowledgements The authors are grateful for the financial supports from the National Nature Science Foundation of China (No. 50972121), Technological Innovations Foundation of Northwestern Polytechnical University (No. 2010KJ0202). References
Fig. 10. Fractograph of Csf /AZ91D composites compressed at room temperature.
[1] H.Z. Ye, X.Y. Liu, J. Mater. Sci. 39 (2004) 6153–6171. [2] M. Russell-Stevens, R. Todd, M. Papakyriacou, Mater. Sci. Eng. A 397 (2005) 249–256. [3] H.B. Ou-Yang, H.J. Li, L.H. Qi, Z.J. Li, L.Z. Su, J.F. Wei, Rare Metal Mat. Eng. 38 (2009) 931–935. [4] Y. Huang, Master Thesis, Harbin Institute of Technology, 2007. [5] W. Hufenbach, M. Andrich, A. Langkamp, A. Czulak, J. Mater. Process. Technol. 175 (2006) 218–224. [6] M. Russell-Stevens, D.C. Plane, J. Summerscales, P. Schulz, M. Papakyriacou, Mater. Sci. Technol. 18 (2002) 501–506. [7] L.H. Qi, J.M. Zhou, L.Z. Su, H.B. Ou-Yang, H.J. Li, Solid State Phenom. 141-143 (2008) 91–96. [8] J.T. Guan, L.H. Qi, L.Z. Su, J. Liu, Adv. Mater. Res. 97-101 (2010) 382–385. [9] J.M. Zhou, L.H. Qi, H.B. Ou-Yang, H.J. Li, Adv. Mater. Res. 89-91 (2010) 692–696. [10] S. Ataya, E. El-Magd, Theor. Appl. Fract. Mech. 47 (2007) 102–112. [11] D.J. Towle, C.M. Friend, Mater. Sci. Technol. 9 (1993) 35–41. [12] Z. Trojanová, V. Gärtnerová, P. Lukáˇc, Z. Drozd, J. Alloys Compd. 378 (2004) 19–26. ´ M. Chalupová, P. Palˇcek, P. [13] Z. Trojanová, V. Gärtnerová, A. Jäger, A. Námeˇsny, Lukáˇc, Compos. Sci. Technol. 69 (2009) 2256–2264. [14] H.B. Ou-Yang, H.J. Li, L.H. Qi, Z.J. Li, T. Fang, J.F. Wei, J. Mater. Sci. 43 (2008) 4618–4624.
170
J. Liu et al. / Materials Science and Engineering A 531 (2012) 164–170
[15] M. Russell-Stevens, R. Todd, M. Papakyriacou, Surf. Interface Anal. 37 (2005) 336–342. [16] M.R. Barnett, Z. Keshavarz, A.G. Beer, D. Atwell, Acta Mater. 52 (2004) 5093–5103. [17] Q.B. Nguyen, M. Gupta, Compos. Sci. Technol. 68 (2008) 2185–2192. [18] K.S. Tun, M. Gupta, Mater. Sci. Eng. A 527 (2010) 5550–5556. [19] A.K. Khanra, H.C. Jung, K.S. Hong, K.S. Shin, Mater. Sci. Eng. A 527 (2010) 6283–6288. [20] L. Geng, A.B. Li, Q.Y. Meng, Mater. Sci. Eng. A 386 (2004) 212–221.
[21] A.B. Li, Q.Y. Meng, L. Geng, C.F. Deng, Rare Metals 26 (2007) 182–192. [22] Q.Q. Li, A. Viereckl, C.A. Rottmair, R.F. Singer, Compos. Sci. Technol. 69 (2009) 1193–1199. [23] Z. Száraz, Z. Trojanová, M. Cabbibo, E. Evangelista, Mater. Sci. Eng. A 462 (2007) 225–229. [24] C.S. Goh, J. Wei, L.C. Lee, M. Gupta, Scripta Mater. 55 (2007) 5115–5121. [25] W.J. Kim, H.G. Jeong, H.T. Jeong, Scripta Mater. 61 (2009) 1040–1043. [26] D.-G.C. Syu, A.K. Ghosh, Mater. Sci. Eng. A 184 (1994) 27–35.