The appearance of the FSPed composite specimens

taken into consideration. The composition of particle reinforced Mg based metal matrix composites can remain closer to the original base alloy and not be shifted much. The applicable properties of this kind of composite are more comparable with the original base alloy.

3.2.2.2 Microstructure of Mg matrix composites made by FSP

The OM observation is limited due to its depth of field and magnification. Therefore, SEM is conducted to characterize the composite specimens with the magnification greater than 1000 times and the observed results are analyzed by the Optimas^® image analysis software.

Microstructure characterization in this study was mainly focused on the distribution and local clustering of the nano-ZrO2 particles, as well as the matrix grain structures in the stirred zone that have undergone dynamic recrystallization. The frictional heating and severe plastic deformation are simultaneously introduced into the stirred material during FSP by the rotating tool. Therefore, it is expected that both the frictional heating and plastic strain would lead to the formation of dynamically recrystallized grains. These two effects would also help in dispersing the inserted nano-ZrO2 and nano-SiO2 particles in the stirred zones.

After one-pass (1P) FSP, the dispersion of the nano-particles within the central cross-sectional area of the friction stir zone (FSZ) is basically uniform, as shown in Fig.

3-28(a). The observed clustered particle size is frequently 0.1-2 μm, much larger than the individual nano-particle size (~20 nm). In addition, some local inhomogeneous areas of the particles can be found in the 1P FSP sample, as shown in Fig. 3-28(b). The clustered size of the particles after two to four passes (2P to 4P) appears to have further reduced, as shown in Fig. 3-28(c). It means that the composite made by FSP with further stirring passes; the clustering phenomenon could be eliminated gradually. The statistical measurements for the volume fraction of amorphous SiO2 nano particles in matrix yield the values around 5 vol%

and 10 vol% for the specimens with one groove (1G) and two grooves (2G), respectively. In

contrast, the volume fractions of the ZrO2 nano-particles in matrix are estimated to be around 10 vol% and 20 vol% for 1G and 2G, respectively. The higher ZrO2 content than the SiO2

case is due to the fact that ZrO2 particles are much heavier, leading the insertion of particles in the groove more efficient.

Figures 3-29 and 3-30 show the SEM/SEI images of the SiO2 and ZrO2 composites specimens with different volume fractions under different magnifications, respectively. The SEM/BEI images of ZrO2 composite specimens with different volume fractions are also shown in Fig. 3-31. The bigger particles are the clustered ZrO2 nano particles, as shown in Fig. 3-32. Because of the small atomic number difference between Mg and Si, it is difficult to distinguish SiO2 particles from the Mg alloy matrix under SEM/BEI images, but it is much more readily for the ZrO2/Mg composites. Meanwhile, with increasing FSP passes, the average grain size, dg, of the AZ31 alloy matrix is also significantly reduced from around 75 μm in the initial billet to 2~4 μm in the 4P FSP samples. At higher magnifications, these clustered particles are generally located on the grain boundaries or triple junctions, and some are embedded inside the grains, as shown in Fig. 3-33. The summary of the clustered size of the particles and the average grain size of the AZ31 alloy matrix in the 4P FSP samples are listed in Table 3-5.

Provided that all of the nano-particles are individually and uniformly dispersed in the alloy, the theoretically estimated particle interspacing Ls (=(d/2)(2π^/3f)1/2, where f is the particle volume fraction) [64], and thus the approximate grain size dg, should be less than 0.1 μm. It follows that a certain level of local clustering is inevitable, and not all nano-particles can restrict grain boundary migration. Note that the typical grain size of AZ31 alloy (without nano-particles) after the same 4P FSP was measured to be around ~6 μm. The grain size in the FSP composite samples with nano-particles can be refined to 2~4 μm, indicating that the

nano-particles or clusters in the matrix did play an effective rule in restricting grain boundary migration.

3.2.2.3 XRD results

Figure 3-34 shows the XRD patterns of the modified AZ31 Mg alloys with different advancing speeds. It could be seen that the (0002) peak in the FSPed AZ31 Mg alloy with a 90 mm/min advancing speed is significantly stronger than the (1011) peak, as compared with the random Mg. It implies that the (0002) plane tends to lie on the transverse plane of the FSP specimen (or perpendicular to the pin travel direction). However, the XRD patterns of the FSPed AZ31 Mg alloy with 45 mm/min show random-like distributions despite in the modified AZ31 alloy or AZ31 composites, as shown in Figs. 3-34 and 3-35. It is suggested that the AZ31 Mg alloy is softer, and is easily to be affected by the heat generated during the process. As FSPed at a much lower advancing speed of 45 mm/min, the induced higher temperature rise results in more complete dynamic recrystallization, as well as more random orientation, almost approaching to the random pattern shown in Fig. 3-13(a).

The XRD patterns for the transverse cross-sectional plane of the ZrO2 and SiO2 FSP composites are presented in Fig. 3-34. It can be seen from Fig. 3-35(a) that in the Mg-AZ31/ZrO2 composite there is no new phase except for a small amount of the ZrO2

reinforcement phase (weak peaks) and the Mg matrix. This indicates that the crystalline ZrO2

phase is very stable, no reaction between the ZrO2 phase and Mg-AZ31 matrix occurred during FSP. However, some additional weak peaks, identified as Mg2Si and MgO, can be found in the FSP Mg-AZ31/SiO2 composite, as shown in Fig. 3-35(b). It is evident that the chemical reaction between the SiO2 phase and Mg matrix has occurred during the FSP mixing. The reaction in the Mg - SiO2 system can be described by the following reaction of

4Mg + SiO2→ 2MgO + Mg2Si. Our previous study also confirmed the presence of the Mg2Si and MgO phases in the Mg-AZ61/nano-SiO2 composite fabricated by the FSP route [171].

3.2.2.4 Hardness measurements

The typical Vickers hardness readings, Hv, measured along the central cross-sectional zones of the FSP samples are depicted in Fig. 3-36. Compared with the AZ31 alloy without the ZrO2 powders reinforcements, almost a double increment of the hardness, 105 Hv, was achieved for the present composites, especially for the 2G4P sample with ~20 vol% ZrO2

particles, as seen in Table 3-6. After FSP, the scattering of Hv within the FSP nugget zone is considered to be relatively minor, implying that the pin stirring has efficiently dispersed the nano-ZrO2 particles in a reasonably uniform manner, especially after more than one pass. In comparison, the SiO2 containing composites show lower Hv, mainly a result of the lower particle volume fraction. For the AZ31 alloy without any ZrO2 reinforcement, after four passes FSP, the Hv could also increase from ~50 for the AZ31 billet up to ~70, due to the grain refinement from ~75 μm down to ~6 μm via dynamic recrystallization. This hardness result is also consistent with the grain size observation in OM/SEM. The grain size in composites specimens could be refined to 2 μm, compared with the AZ31 alloy specimen (~6 μm) without any ZrO2 or SiO2 particle added into matrix.

3.2.2.5 Mechanical properties

All tensile samples were machined perpendicular to the processing direction from the central region of the FSP nugget. Table 3-6 also lists the tensile properties of the AZ31 FSP alloy and composites, taking the average from two or three samples. For the AZ31 billet without FSP, the room-temperature yield strength (YS), ultimate tensile strength (UTS), and

tensile elongation are ~100 MPa, 160 MPa, and 9%, respectively. After 4P FSP for the AZ31 billet, they are improved to ~120 MPa, 204 MPa, and 18%. The increase of YS and UTS as well as elongation for the FSP AZ31 sample was mainly contributed by the grain refinement.

For the FSP composites, the YS and UTS were improved also by the nano-particle reinforcements, in addition to the apparent grain refinement. For example, the yield stress of the Mg-AZ31/ZrO2 FSP composites was improved to 143 MPa in the 1G4P (~10% ZrO2) and to 167 MPa in the 2G4P (~20% ZrO2) samples. The ultimate tensile strength is also appreciably improved in parallel. The increment of YS and UTS for the SiO2 containing composites was slightly lower, due to the lower particle volume fraction.

The differences in the fracture behavior after tensile tests between the FSP Mg-AZ31 alloy and the particle reinforced composites can be seen from the SEM fractographs in Fig.

3-37. The fracture surface of the FSP AZ31 alloy exhibits elongated uniform dimples, as shown in Fig. 3-37(a), which indicate the overall ductile fracture mode with a tensile elongation of 18%. In contrast, the fracture behavior of the present composite, for example, Mg-AZ31/10%ZrO2, is very different, as shown in Fig. 3-37(b). Some dimples with the clusters of ZrO2 particles, and some round and shallow small dimples in the matrix area can be seen on the fracture surface of the composite. The more shallow dimples indicate a relatively more brittle fracture mode with a tensile elongation of 6%.

3.2.2.6 Mg based composites with tetragonal phase nano-ZrO2 particles fabricated by FSP

The tetragonal phase nano-ZrO2 particles, yttria-stabilized zirconia (YSZ), with an average size around 40 nm are also chosen as the reinforcement here. The processing parameters and fabrication method are both the same as other AZ31/monoclinic phase ZrO2

composites during FSP. The amount of added tetragonal phase ZrO2 particles is also the same with the monoclinic phase ZrO2 particles for 1G and 2G, respectively. The XRD patterns for the transverse cross-sectional plane of the tetragonal phase ZrO2 FSP composites are shown in Fig. 3-38. It can be seen that there is also no new phase except for a small amount of the ZrO2 reinforcement phase (weak tetragonal phase ZrO2 peaks) and the Mg matrix. There is no reaction between the tetragonal phase ZrO2 and Mg-AZ31 matrix during FSP which is the same as the previous monoclinic phase ZrO2 results and also can be proven from the Mg-Zr phase diagram. From the phase diagram, there is no reaction occurring or intermetallic compound produced during the specific temperature period and composition range in the FSP process. Figure 3-39 shows the Vickers hardness readings, Hv, measured along the central cross-sectional zones of the FSP samples. It can be seen that average hardness values of both 1G4P and 2G4P tetragonal nano-ZrO2 particles reinforced Mg AZ31 composites are higher than 1G4P and 2G4P Mg AZ31 composites which are reinforced by monoclinic phase nano-ZrO2, respectively.

The mechanical properties, strength and especially toughness are enhanced by the presence of transformation toughened zirconia particles. The increasing hardness of Mg/tetragonal-ZrO2 Mg composite is suggested to be induced by the transformation of the tetragonal zirconia particles. It is well-known that tetragonal phase zirconia is able to transform to the monoclinic phase if a stress field is developed around the particles [173].

Because of the volume expansion (> 3%) and shear strain (1-7%) developed in the transformed particles (Fig. 3-40), a resultant compressive stress is generated in the matrix which can prevent the crack from moving, thus accounting for the increase in hardness.

Therefore, the increment of hardness in the Mg/tetragonal-ZrO2 Mg composites is caused from the transformation toughened zirconia.

3.2.2.7 The XRD and hardness analysis for the Mg/tetragonal phase ZrO2 composites after subsequent compression

Lee et al. [174] have proposed a route for improving mechanical properties of friction stirred Mg-Al-Zn alloys by subsequent compression along the normal direction which could improve the unfavorable texture by inducing deformation twins, thus raising the yield stress and hardness significantly. Therefore, the Mg/tetragonal phase ZrO2 composite was chosen to be subjected to the subsequent compression along the normal direction. The XRD diffraction patterns of the 1G4P and 2G4P Mg/tetragonal phase ZrO2 composites after 6% strain of subsequent compression along the normal direction are shown in Fig. 3-41. The grain orientations are changed after compression. Fig. 3-42 shows the hardness profiles for both 1G4P and 2G4P composites after 6% compression, respectively. After 6% compression, the hardness values for both 1G4P and 2G4P are slightly increased from 103 Hv and 113.5 Hv to 105 Hv and 120 Hv, respectively. Unlike the results of compressed pure AZ61 alloys in Lee’s report [174], the increment of hardness after compression here is much smaller. This may because the grain sizes here are much smaller than previous reported AZ61 alloy. The activation of twins is more difficult and the effect of twins is less obviously.

3.2.2.8 Brief conclusions for Mg-AZ31 based composites with nano-ZrO2 and nano-SiO2 particles

After a series of systematic experiments on the Mg AZ31/ZrO2 and SiO2 composites, it can be concluded briefly as follows:

(1) Friction stir processing successfully fabricated bulk Mg-AZ31 based composites with 10

~ 20 vol% of nano-ZrO2 particles and 5 ~10 vol% of nano-SiO2 particles. The distribution of the 20 nm nano-particles after four FSP passes resulted in satisfactorily uniform distribution.