Effect of Vanadium Reinforcement on the Microstructure and Mechanical Properties of Magnesium Matrix Composites

: In this work, vanadium particles (V P ) were utilized as a novel reinforcement of AZ31 magnesium (Mg) alloy. The nanocrystalline (NC) AZ31–V P composites were prepared via mechanical milling (MM) and vacuum hot-press sintering. During the milling process, the presence of V P contributed to the cold welding and fracture mechanism, resulting in the acceleration of the milling process. Additionally, increasing the V P content accelerated the grain reﬁnement of the matrix during the milling process. After milling for 90 h, the average grain size of AZ31-X wt % V p (X = 5, 7.5, 10) was reﬁned to only about 23 nm, 19 nm and 16 nm, respectively. In the meantime, V P was reﬁned to sub-micron scale and distributed uniformly in the matrix, exhibiting excellent interfacial bonding with the matrix. After the sintering process, the average grain size of AZ31-X wt % V P (X = 5, 7.5, 10) composites still remained at the NC scale, which was mainly caused by the pinning effect of V P . Besides that, the porosity of the sintered composites was no more than 7.8%, indicating a good densiﬁcation effect. As a result, there was little difference between the theoretical and real density. Compared to as-cast AZ31 Mg alloy, the microhardness of sintered AZ31-X wt % V P (X = 5, 7.5, 10) composites increased by 65%, 87% and 96%, respectively, owing to the strengthening mechanisms of grain reﬁnement strengthening, Orowan strengthening and load-bearing effects.


Introduction
The rapid development of high-strength, lightweight alloys is urgently needed for improving energy efficiency in modern industry [1,2]. Magnesium (Mg), as the most promising lightweight metal, has a wide perspective application in aircraft industries, newenergy vehicles, electronic appliances, etc., due to its high specific strength, low density (1.738 g/cm 3 ) and excellent recyclability [3][4][5]. However, the low modulus, mechanical strength and poor plasticity of Mg alloys has seriously limited their large-scale industrial applications [6]. In order to improve their poor mechanical performance, particle-reinforced Mg matrix composites (PRMMCs) have been developed to overcome the disadvantages of Mg alloys. At present, ceramic particles, such as SiC, TiB 2 , and SiO 2 [7][8][9], are the common reinforcements of PRMMCs, which can significantly improve thermal stability, yield strength (YS), modulus, microhardness, fatigue and wear resistance of the Mg matrix. However, the wettability between ceramic reinforcements and the Mg matrix is so poor that there is a bottleneck in the further improvement of performance for PRMMCs. Meanwhile, the grain size of many PRMMCs is not fine enough, meaning the advantages of grain The preparation process of AZ31-VP composites was as follows: Firstly, the AZ31 Mg alloy chips and VP powders were mixed together in a stainless-steel cup filled with argon. In this way, three kinds of mixtures with different VP contents (5 wt %, 7.5 wt % and 10 wt %) were prepared. Then, the mixtures were mechanically milled on a QM-QX4 type ball mill for 90 h. The mill rotation speed was chosen as 200 rpm, and the ball (Φ5 mm, Φ8 mm, Φ10 mm) to powder weight ratio was 60:1. Moreover, in order to avoid oxidation, all the handling steps were performed under argon atmosphere. Finally, the as-milled composite powders (50 g) were put into a graphite mold (Φ35 mm), and consolidated using the vacuum hot-press sintering method. The processing pressure, temperature and holding time were chosen as 300 °C, 40 MPa and 1 h, respectively. Figure 2 shows the schematic illustration of vacuum hot-press sintering, and the image of sintered AZ31-VP composites. After sintering process, the theoretical density of samples was calculated by using the rule of mixture, which was illustrated in Equation (1) [7] t m m r r where ρt is the sintered theoretical density. ρr and ρm represent the theoretical densities of VP and Mg matrix. Vr and Vm are the volume fractions of VP and Mg matrix. The preparation process of AZ31-V P composites was as follows: Firstly, the AZ31 Mg alloy chips and V P powders were mixed together in a stainless-steel cup filled with argon. In this way, three kinds of mixtures with different V P contents (5 wt %, 7.5 wt % and 10 wt %) were prepared. Then, the mixtures were mechanically milled on a QM-QX4 type ball mill for 90 h. The mill rotation speed was chosen as 200 rpm, and the ball (Φ5 mm, Φ8 mm, Φ10 mm) to powder weight ratio was 60:1. Moreover, in order to avoid oxidation, all the handling steps were performed under argon atmosphere. Finally, the as-milled composite powders (50 g) were put into a graphite mold (Φ35 mm), and consolidated using the vacuum hot-press sintering method. The processing pressure, temperature and holding time were chosen as 300 • C, 40 MPa and 1 h, respectively. Figure 2 shows the schematic illustration of vacuum hot-press sintering, and the image of sintered AZ31-V P composites.
Crystals 2021, 11, x FOR PEER REVIEW 3 of 16 these chips were cleaned by ultrasonic. Figure 1 exhibits the morphology of initial materials. It can be clearly seen that the chips of AZ31 Mg alloy showed laminar morphology, and the VP powders exhibited polygonal shape. The preparation process of AZ31-VP composites was as follows: Firstly, the AZ31 Mg alloy chips and VP powders were mixed together in a stainless-steel cup filled with argon. In this way, three kinds of mixtures with different VP contents (5 wt %, 7.5 wt % and 10 wt %) were prepared. Then, the mixtures were mechanically milled on a QM-QX4 type ball mill for 90 h. The mill rotation speed was chosen as 200 rpm, and the ball (Φ5 mm, Φ8 mm, Φ10 mm) to powder weight ratio was 60:1. Moreover, in order to avoid oxidation, all the handling steps were performed under argon atmosphere. Finally, the as-milled composite powders (50 g) were put into a graphite mold (Φ35 mm), and consolidated using the vacuum hot-press sintering method. The processing pressure, temperature and holding time were chosen as 300 °C, 40 MPa and 1 h, respectively. Figure 2 shows the schematic illustration of vacuum hot-press sintering, and the image of sintered AZ31-VP composites. After sintering process, the theoretical density of samples was calculated by using the rule of mixture, which was illustrated in Equation (1) [7] t m m r r where ρt is the sintered theoretical density. ρr and ρm represent the theoretical densities of VP and Mg matrix. Vr and Vm are the volume fractions of VP and Mg matrix. After sintering process, the theoretical density of samples was calculated by using the rule of mixture, which was illustrated in Equation (1) [7] where ρ t is the sintered theoretical density. ρ r and ρ m represent the theoretical densities of V P and Mg matrix. V r and V m are the volume fractions of V P and Mg matrix. Additionally, the real density of sintered samples was measured by using Archimedes' principle, as expressed in Equation (2) [23] where ρ e represents the real density of sintered sample, ρ w is the density of distilled water at room temperature, W e and W w are the weight of sample in air and water. Moreover, the porosity (P) of sintered samples can be calculated via Equation (3) [24] In addition, the powder morphography was analyzed by a FEI APREO scanning electron microscope (SEM) (FEI Company, Hillsboro, OR, USA), and the elemental analysis was conducted by energy dispersive spectrometer (EDS). X-ray diffractometer (XRD) (Rigaku Ultima IV, Rigaku Corporation, Tokyo, Japan) with a Cu-Kα radiation (40 Kv,40 mA) was employed to analyze the phase composition. The scan range and scan speed were chosen as 20-90 • and 4 • /min. Furthermore, transmission electron microscope (TEM) characterization was carried out on a JEM-200 EX TEM (JEOL Ltd., Tokyo, Japan) to further investigate the microstructure of AZ31-V P composites. High-resolution TEM (HRTEM) was also utilized to analyze the Mg/V interface of the composites.
The microhardness of AZ31-V P composites was examined by using HVS-1000 tester (Shanghai Optical Instrument Factory, Shanghai, China). The load and dwell time were chosen as 200 g and 15 s, respectively. Meanwhile, at least ten tests for each sample were performed to obtain normalized values. Figure 3 exhibits the evolution of powder morphology for AZ31-5 wt % V P during the milling process. In the initial stage of milling, many lamellas fractured from the Mg alloy chips, as shown in Figure 3a,b. Simultaneously, some lamellas piled up together as a result of the cold welding mechanism. In other words, there was a competitive relationship between cold welding and fracture mechanism [25]. With the extension of milling time, the particle size of powders decreased continuously, as shown in Figure 3c-e. This indicated that the fracture mechanism played a predominant role in the evolution of powder morphology. In other words, the particles became hardened and tended to fracture after milling for 50 h, owing to the repeated severe plastic deformation. As a result, the formation of large particles was limited, while the particle size decreased. As the milling time increased to 90 h, it was observed that some equiaxed particles were formed, and the ultimate average particle size was refined to about 635 µm, as shown in Figure 3e,f. There are two main reasons for the formation of equiaxed particles. Firstly, it is widely proposed that the bigger and harder powder particles were more prone to break, thus inducing the formation of equiaxed particles [26]. Secondly, the occurrence of cold welding could contribute to the formation of equiaxed particles [10]. Similar to the evolution of particle size with the extended milling time, there was also a decreasing trend for the length to width (L/W) ratio of particles, as indicated in Figure 3a-e. Such phenomenon was mainly attributed to the repeated plastic deformation of powders, which was caused by the high energy impact between the balls and the cup [27]. Through the above discussion, it can be concluded that the finer particle size and the more regular powder shape could be achieved with the increasing milling time.  In order to research the influence of VP content on the morphology evolution of the composites, the powder morphology of AZ31-X wt % VP (X = 7.5, 10) after milling for 50 h was also analyzed, as shown in Figure 4. By comparison with Figure 3c, it can be observed that the particle size decreased with the increasing content of VP, illustrating that the VP accelerated the milling process. Zhou et al. [27] reported a similar phenomenon that the milling process was accelerated by Ti particles. Such phenomenon was mainly attributed to the following reason. The local deformation of the Mg matrix near the reinforcement particles was enhanced, and the larger local deformation could not only promote the occurrence of cold welding but also induce the fracture [28]. In order to research the influence of V P content on the morphology evolution of the composites, the powder morphology of AZ31-X wt % V P (X = 7.5, 10) after milling for 50 h was also analyzed, as shown in Figure 4. By comparison with Figure 3c, it can be observed that the particle size decreased with the increasing content of V P , illustrating that the V P accelerated the milling process. Zhou et al. [27] reported a similar phenomenon that the milling process was accelerated by Ti particles. Such phenomenon was mainly attributed to the following reason. The local deformation of the Mg matrix near the reinforcement particles was enhanced, and the larger local deformation could not only promote the occurrence of cold welding but also induce the fracture [28].

Phase Composition and Matrix Structure
The XRD patterns of AZ31-X wt % VP (X = 5, 7.5, 10) powders are illustrated in Figure  5, in which the peaks of the Mg and V phases can be observed, and no other phases were detected. Meanwhile, there was no Mg17Al12 peak in these XRD patterns, which was due to the decomposition of Mg17Al12 and the dissolving of Al in the Mg matrix [29]. Moreover, the peaks of the Mg phase, such as (002) and (101), significantly decreased in diffraction intensity with the increasing milling time. Such phenomenon was mainly caused by the decrease in grain size and accumulation of microstrain. The grain size and microstrain of Mg matrix composites can be calculated via Equations (4) and (5) [30] where D is the grain size, e is the microstrain, K is the Scherrer constant, λ is the X-ray wavelength, θ is the Bragg angle. and can be obtained from the ratio of the Cauchy integral breadth ( / ) and the Gauss integral breadth ( / ).

Phase Composition and Matrix Structure
The XRD patterns of AZ31-X wt % V P (X = 5, 7.5, 10) powders are illustrated in Figure 5, in which the peaks of the Mg and V phases can be observed, and no other phases were detected. Meanwhile, there was no Mg 17 Al 12 peak in these XRD patterns, which was due to the decomposition of Mg 17 Al 12 and the dissolving of Al in the Mg matrix [29]. Moreover, the peaks of the Mg phase, such as (002) and (101), significantly decreased in diffraction intensity with the increasing milling time. Such phenomenon was mainly caused by the decrease in grain size and accumulation of microstrain. The grain size and microstrain of Mg matrix composites can be calculated via Equations (4) and (5) where D is the grain size, e is the microstrain, K is the Scherrer constant, λ is the X-ray wavelength, θ is the Bragg angle. β f C and β f G can be obtained from the ratio of the Cauchy integral breadth (β f C /β) and the Gauss integral breadth (β f G /β). Figure 6 shows the variations in grain size and microstrain of AZ31-X wt % V P (X = 5, 7.5, 10) powders as a function of the milling time. As shown in Figure 6a, it can be observed that the grain size of Mg matrix decreased continuously with the increase of milling time. Meanwhile, the refinement process can be divided into three stages. In the initial stage, the dislocation density gradually increased, due to repeated severe plastic deformation of composite powders. As the increasing dislocation density reached the critical value, the coarse grains split into the sub-grains separated by low angle grain boundaries, resulting in rapid grain refinement [31]. With the formation of many new grain boundaries, the critical value became larger, and thus more energy was required. However, the input mechanical energy was constant during the milling process. As a result, the refinement rate slowed down in the middle stage. As the milling time went over 70 h, there was almost no more refinement for grain size, which reached a relatively stable stage. This was associated with the dynamic balance between grain refinement and grain growth caused by thermal effect [32].
where D is the grain size, e is the microstrain, K is the Scherrer constant, λ is the X-ray wavelength, θ is the Bragg angle. and can be obtained from the ratio of the Cauchy integral breadth ( / ) and the Gauss integral breadth ( / ).  Figure 6 shows the variations in grain size and microstrain of AZ31-X wt % VP (X = 5, 7.5, 10) powders as a function of the milling time. As shown in Figure 6a, it can be observed that the grain size of Mg matrix decreased continuously with the increase of milling time. Meanwhile, the refinement process can be divided into three stages. In the initial stage, the dislocation density gradually increased, due to repeated severe plastic deformation of composite powders. As the increasing dislocation density reached the critical value, the coarse grains split into the sub-grains separated by low angle grain boundaries, resulting in rapid grain refinement [31]. With the formation of many new grain boundaries, the critical value became larger, and thus more energy was required. However, the input mechanical energy was constant during the milling process. As a result, the refinement rate slowed down in the middle stage. As the milling time went over 70 h, there was almost no more refinement for grain size, which reached a relatively stable stage. This was associated with the dynamic balance between grain refinement and grain growth caused by thermal effect [32].  Figure 6 shows the variations in grain size and microstrain of AZ31-X wt % VP (X 5, 7.5, 10) powders as a function of the milling time. As shown in Figure 6a, it can b observed that the grain size of Mg matrix decreased continuously with the increase o milling time. Meanwhile, the refinement process can be divided into three stages. In th initial stage, the dislocation density gradually increased, due to repeated severe plasti deformation of composite powders. As the increasing dislocation density reached the crit ical value, the coarse grains split into the sub-grains separated by low angle grain bound aries, resulting in rapid grain refinement [31]. With the formation of many new grai boundar   Additionally, it should be noticed that the refinement rate increased with the increasing V P content. After milling for 90 h, the average grain size of AZ31-X wt % V P (X = 5, 7.5, 10) powders was refined to about 23 nm, 19 nm and 16 nm, respectively, which reached the NC scale. This was partly because of the pinning effect of V P , which could hinder the migration of grain boundaries and restrain the grain growth. Another primary reason was that the dislocation density would be higher in the composites with the larger V P content, which could lead to the finer grain size. In order to verify the relationship between dislocation density and V P content, the TEM observations of AZ31-V P powders with different V P contents were carried out, as exhibited in Figure 7. Obviously, the dislocation density increased with the increasing V P content under the same milling time. Figure 6b exhibits the microstrain of AZ31-V P powders at different milling times. Contrary to the grain size, the microstrain exhibited an upward trend with the extended milling time, as well as the increasing V P content. Such a phenomenon was closely related to the increasing dislocation density, which was caused by the work-hardening and fracture mechanism of powders. It has been reported that the pile up of dislocations could promote the accumulation of microstrain [27]. Additionally, it should be noticed that the refinement rate increased with the increasing VP content. After milling for 90 h, the average grain size of AZ31-X wt % VP (X = 5, 7.5, 10) powders was refined to about 23 nm, 19 nm and 16 nm, respectively, which reached the NC scale. This was partly because of the pinning effect of VP, which could hinder the migration of grain boundaries and restrain the grain growth. Another primary reason was that the dislocation density would be higher in the composites with the larger VP content, which could lead to the finer grain size. In order to verify the relationship between dislocation density and VP content, the TEM observations of AZ31-VP powders with different VP contents were carried out, as exhibited in Figure 7. Obviously, the dislocation density increased with the increasing VP content under the same milling time. Figure 6b exhibits the microstrain of AZ31-VP powders at different milling times. Contrary to the grain size, the microstrain exhibited an upward trend with the extended milling time, as well as the increasing VP content. Such a phenomenon was closely related to the increasing dislocation density, which was caused by the work-hardening and fracture mechanism of powders. It has been reported that the pile up of dislocations could promote the accumulation of microstrain [27].   Figure 8a, it can be confirmed that the matrix was Mg. Combined with Figure 8b, it was revealed that the grain size of AZ31-5 wt % VP powders were mainly distributed between 10 and 30 nm, and the average grain size was approximately 20 nm. As shown in Figure 8c-f, the average grain size of AZ31-X wt % VP (X = 7.5, 10) powders was refined to only about 18 nm and 14 nm, respectively. By comparison with Figure 6a, it can be indicated that the measured values of grain size were very close to the above calculation results. This verified the accuracy of the calculation results.  Figure 8 displays the TEM images of AZ31-X wt % V P (X = 5, 7.5, 10) powders after milling for 90 h, as well as the corresponding distribution maps of grain size. According to the selected area diffraction (SAED) pattern in Figure 8a, it can be confirmed that the matrix was Mg. Combined with Figure 8b, it was revealed that the grain size of AZ31-5 wt % V P powders were mainly distributed between 10 and 30 nm, and the average grain size was approximately 20 nm. As shown in Figure 8c-f, the average grain size of AZ31-X wt % V P (X = 7.5, 10) powders was refined to only about 18 nm and 14 nm, respectively. By comparison with Figure 6a, it can be indicated that the measured values of grain size were very close to the above calculation results. This verified the accuracy of the calculation results.  Figure 9 shows the interfacial structure of AZ31-5 wt % VP powders after milling for 90 h. The Mg/V interface can be clearly observed in Figure 9a, which was demonstrated by the SAED patterns and energy dispersive X-ray spectroscopy (EDX) analysis, as given in Figure 9b,c. On the basis of the SAED patterns, it can be concluded that the region A was V and region B was the Mg matrix. In addition, the d-spacings of regions C, D and E were measured, which were 0.276151, 0.259451 and 0.249874 nm, respectively, as given in Figure 9d-f. Combining the XRD data, the corresponding crystal planes were determined to be (100), (002) and (101) of Mg. Through the above analysis, it can be deduced that the Mg-V interface formed a close relationship on the above crystal planes.  Figure 9 shows the interfacial structure of AZ31-5 wt % V P powders after milling for 90 h. The Mg/V interface can be clearly observed in Figure 9a, which was demonstrated by the SAED patterns and energy dispersive X-ray spectroscopy (EDX) analysis, as given in Figure 9b,c. On the basis of the SAED patterns, it can be concluded that the region A was V and region B was the Mg matrix. In addition, the d-spacings of regions C, D and E were measured, which were 0.276151, 0.259451 and 0.249874 nm, respectively, as given in Figure 9d-f. Combining the XRD data, the corresponding crystal planes were determined to be (100), (002) and (101) of Mg. Through the above analysis, it can be deduced that the Mg-V interface formed a close relationship on the above crystal planes.

Evolution of VP During the Milling Process
It is widely recognized that the size and distribution of particle reinforcement was important for improving the mechanical performance of PRMMCs [33,34]. Figure 10a-e exhibits the back scattered electron (BSE) images of AZ31-5 wt % VP powders milled for different times. Based on the EDS results of Figure 10g,h, it can be concluded that the gray area of point A in Figure 10c was the Mg matrix, while the white particles of point B were VP. After milling for 10 h, a large number of VP clusters stuck on the surface of the Mg matrix, due to the inadequate milling. Meanwhile, some VP existed in the gap between lamellas, as shown in Figure 10a. At 30 h, the quantity and area of VP clusters gradually decreased, but the small VP clusters, marked with the broken circles, still could be observed in the Mg matrix, as shown in Figure 10b. This was caused by the high energy collisions between balls and powders. With the extended milling time, VP clusters were continuously disintegrated and more and more VP gradually entered into the matrix, as shown in Figure 10c,d. Similar phenomena were also observed in the other PRMMCs during the milling process [10,25], which was attributed to the cold welding and cracking of particles. With the milling time of 90 h, VP had distributed uniformly in the matrix, as observed in Figure 10e. Throughout the milling process, the particle size of VP was also gradually refined. The average particle size of VP was only 0.58 μm after milling for 90 h, which had reached sub-micron scale, as shown in Figure 10f.

Evolution of V P during the Milling Process
It is widely recognized that the size and distribution of particle reinforcement was important for improving the mechanical performance of PRMMCs [33,34]. Figure 10a-e exhibits the back scattered electron (BSE) images of AZ31-5 wt % V P powders milled for different times. Based on the EDS results of Figure 10g,h, it can be concluded that the gray area of point A in Figure 10c was the Mg matrix, while the white particles of point B were V P . After milling for 10 h, a large number of V P clusters stuck on the surface of the Mg matrix, due to the inadequate milling. Meanwhile, some V P existed in the gap between lamellas, as shown in Figure 10a. At 30 h, the quantity and area of V P clusters gradually decreased, but the small V P clusters, marked with the broken circles, still could be observed in the Mg matrix, as shown in Figure 10b. This was caused by the high energy collisions between balls and powders. With the extended milling time, V P clusters were continuously disintegrated and more and more V P gradually entered into the matrix, as shown in Figure 10c,d. Similar phenomena were also observed in the other PRMMCs during the milling process [10,25], which was attributed to the cold welding and cracking of particles. With the milling time of 90 h, V P had distributed uniformly in the matrix, as observed in Figure 10e. Throughout the milling process, the particle size of V P was also gradually refined. The average particle size of V P was only 0.58 µm after milling for 90 h, which had reached sub-micron scale, as shown in Figure 10f. In order to further verify the distribution of VP in the matrix, the EDS map analysis was carried out, as shown in Figure 11. According to Figure 11b,c, it can be observed that In order to further verify the distribution of V P in the matrix, the EDS map analysis was carried out, as shown in Figure 11. According to Figure 11b,c, it can be observed that the V element was uniformly dispersed in the Mg matrix. From Figure 11d-f, it can be observed that most of the Al, Mn and Zn elements were dissolved uniformly in the Mg matrix. Meanwhile, it should be noticed that the intermetallic compound containing Mn existed in the Mg matrix, which was highlighted by the broken circle in Figure 11e. the V element was uniformly dispersed in the Mg matrix. From Figure 11d-f, it can be observed that most of the Al, Mn and Zn elements were dissolved uniformly in the Mg matrix. Meanwhile, it should be noticed that the intermetallic compound containing Mn existed in the Mg matrix, which was highlighted by the broken circle in Figure 11e.

Density Evolution and Microstructure of Sintered AZ31-VP Composites
After the vacuum hot-press sintering process, the AZ31-VP composite billets were prepared. According to Equations (1)-(3), the theoretical density, real density and porosity of sintered composites were calculated, as listed in Table 1. Obviously, it can be observed that the porosity of sintered composites increased with the increasing VP content. This was because the high content level of VP contributed to the increase in porosity, resulting in the occurrence of local cracking [35]. It has been reported [36] that the porosity would significantly influence the quality and the mechanical properties of sintered billets. The high porosity could cause the inhomogeneous distribution of stress across the crosssection, thus reducing the effective load-bearing area [37]. Moreover, both the theoretical density and real density of the sintered composites also increased with the increasing VP content, because the density of V was higher than that of Mg. The similar phenomenon has been reported in the other PRMMCs [7,17,35]. Besides that, the maximum theoretical density (1.915 g/cm 3 ) of sintered composites increased by just 7.6% compared with the density of the AZ31 Mg alloy, which maintained the advantage of being lightweight.  Figure 12 shows the TEM images of sintered AZ31-X wt % VP (X = 5, 7.5, 10) compo-

Density Evolution and Microstructure of Sintered AZ31-V P Composites
After the vacuum hot-press sintering process, the AZ31-V P composite billets were prepared. According to Equations (1)-(3), the theoretical density, real density and porosity of sintered composites were calculated, as listed in Table 1. Obviously, it can be observed that the porosity of sintered composites increased with the increasing V P content. This was because the high content level of V P contributed to the increase in porosity, resulting in the occurrence of local cracking [35]. It has been reported [36] that the porosity would significantly influence the quality and the mechanical properties of sintered billets. The high porosity could cause the inhomogeneous distribution of stress across the cross-section, thus reducing the effective load-bearing area [37]. Moreover, both the theoretical density and real density of the sintered composites also increased with the increasing V P content, because the density of V was higher than that of Mg. The similar phenomenon has been reported in the other PRMMCs [7,17,35]. Besides that, the maximum theoretical density (1.915 g/cm 3 ) of sintered composites increased by just 7.6% compared with the density of the AZ31 Mg alloy, which maintained the advantage of being lightweight.  Figure 12 shows the TEM images of sintered AZ31-X wt % V P (X = 5, 7.5, 10) composites, and the average grain size of different composites. After the sintering process, the average grain size of AZ31-X wt % V P (X = 5, 7.5, 10) composites increased to 56, 54 and 49 nm, respectively, due to the heat effect. Obviously, the average grain size of sintered AZ31-V P composites was below 100 nm, which still remained at the NC scale. The reason for such a phenomenon was the pinning effect of the sub-micron V P during the sintering process [38]. Razavi et al. [39] proposed that the hard reinforcement particles hindered the movement of grain boundaries in the metal matrix during high temperature treatment. Moreover, a high content of reinforcement particles would result in greater inhibition of grain boundary movements and grain growth. AZ31-VP composites was below 100 nm, which still remained at the NC scale. The reason for such a phenomenon was the pinning effect of the sub-micron VP during the sintering process [38]. Razavi et al. [39] proposed that the hard reinforcement particles hindered the movement of grain boundaries in the metal matrix during high temperature treatment. Moreover, a high content of reinforcement particles would result in greater inhibition of grain boundary movements and grain growth.  Figure 13 shows the microhardness of as-cast AZ31 Mg alloy and AZ31-VP composites. Compared to as-cast AZ31 Mg alloy, the microhardness of AZ31-X wt % VP (X = 5, 7.5, 10) composites increased by 65%, 87% and 96%, respectively. Such phenomenon was mainly attributed to the following reasons: Firstly, the average grain size of composites were of NC scale, inducing remarkable grain refinement strengthening. Secondly, Orowan strengthening mechanism was another reason for the increase of microhardness [40]. The well-dispersed sub-micron VP in the matrix obstructed the movement of dislocations, thus causing significant strengthening. Thirdly, good interfacial bonding was formed between the Mg matrix and VP, resulting in effective load transfer [41,42]. This strengthening mechanism is generally known as load-bearing effects [43][44][45].  Figure 13 shows the microhardness of as-cast AZ31 Mg alloy and AZ31-V P composites. Compared to as-cast AZ31 Mg alloy, the microhardness of AZ31-X wt % V P (X = 5, 7.5, 10) composites increased by 65%, 87% and 96%, respectively. Such phenomenon was mainly attributed to the following reasons: Firstly, the average grain size of composites were of NC scale, inducing remarkable grain refinement strengthening. Secondly, Orowan strengthening mechanism was another reason for the increase of microhardness [40]. The well-dispersed sub-micron V P in the matrix obstructed the movement of dislocations, thus causing significant strengthening. Thirdly, good interfacial bonding was formed between the Mg matrix and V P , resulting in effective load transfer [41,42]. This strengthening mechanism is generally known as load-bearing effects [43][44][45]. Crystals 2021, 11, x FOR PEER REVIEW 14 of 16 Figure 13. The microhardness of as-cast AZ31 Mg alloy and AZ31-X wt % VP (X = 5, 7.5, 10) composites.

Microhardness Analysis
Besides that, the microhardness of AZ31-VP composites increased with the increasing weight fraction of VP. There were three main reasons for such phenomenon. Firstly, the decrease in grain size enhanced the grain refinement strengthening effect. Secondly, the higher the VP content, the greater the hindrance to dislocation motion. Thirdly, the dislocation density increased with the increasing VP content. In other words, the average distance between dislocations decreased. Therefore, the motion of dislocation was impeded by the other dislocations, thus improving the microhardness [7]. On the other hand, Huang et al. [9] fabricated AZ61 Mg alloy reinforced with 10 Vol % SiO2, which showed the microhardness of 105 HV and the average grain size of 0.8 μm. Du et al. [46] prepared the AZ31-0.1 vol % CNTs composite with the microhardness of 88 HV, and the average grain size of 1.5 μm. By comparison, AZ31-10 wt % VP composite in this work possessed a competitive microhardness of 106 HV, and the smallest grain size of 49 nm.
Through the above analysis of microstructure and mechanical property, it can be concluded that VP was an excellent novel reinforcement for PRMMCs, while the P/M technique was a feasible method for preparing NC AZ31-VP composites. In future researches, other performance tests of AZ31-VP composites, including tensile property, compression performance and wear resistance, will be investigated for the large-scale industrial application of Mg alloys.

Conclusions
In this work, nanocrystalline (NC) AZ31-X wt % VP (X = 5, 7.5, 10) composites were successfully prepared via the powder metallurgy method. The powder morphology, microstructure evolution and microhardness of AZ31-VP composites were also investigated. The main conclusions were drawn as follows: (1) The presence of VP could contribute to the cold welding and fracture mechanism, resulting in the acceleration of the milling process. Simultaneously, VP was refined to sub-micron scale (0.58 μm) and distributed uniformly in the matrix, which also exhibited excellent interfacial bonding with the matrix, after milling for 90 h.
(2) During the milling process, the matrix grain size decreased with the increasing milling time and VP content. After milling for 90 h, the average grain size of AZ31-X wt % VP (X = 5, 7.5, 10) was refined to only 23 nm, 19 nm and 16 nm, respectively. Besides Figure 13. The microhardness of as-cast AZ31 Mg alloy and AZ31-X wt % V P (X = 5, 7.5, 10) composites.
Besides that, the microhardness of AZ31-V P composites increased with the increasing weight fraction of V P . There were three main reasons for such phenomenon. Firstly, the decrease in grain size enhanced the grain refinement strengthening effect. Secondly, the higher the V P content, the greater the hindrance to dislocation motion. Thirdly, the dislocation density increased with the increasing V P content. In other words, the average distance between dislocations decreased. Therefore, the motion of dislocation was impeded by the other dislocations, thus improving the microhardness [7]. On the other hand, Huang et al. [9] fabricated AZ61 Mg alloy reinforced with 10 Vol % SiO 2 , which showed the microhardness of 105 HV and the average grain size of 0.8 µm. Du et al. [46] prepared the AZ31-0.1 vol % CNTs composite with the microhardness of 88 HV, and the average grain size of 1.5 µm. By comparison, AZ31-10 wt % V P composite in this work possessed a competitive microhardness of 106 HV, and the smallest grain size of 49 nm.
Through the above analysis of microstructure and mechanical property, it can be concluded that V P was an excellent novel reinforcement for PRMMCs, while the P/M technique was a feasible method for preparing NC AZ31-V P composites. In future researches, other performance tests of AZ31-V P composites, including tensile property, compression performance and wear resistance, will be investigated for the large-scale industrial application of Mg alloys.

Conclusions
In this work, nanocrystalline (NC) AZ31-X wt % V P (X = 5, 7.5, 10) composites were successfully prepared via the powder metallurgy method. The powder morphology, microstructure evolution and microhardness of AZ31-V P composites were also investigated. The main conclusions were drawn as follows: (1) The presence of V P could contribute to the cold welding and fracture mechanism, resulting in the acceleration of the milling process. Simultaneously, V P was refined to sub-micron scale (0.58 µm) and distributed uniformly in the matrix, which also exhibited excellent interfacial bonding with the matrix, after milling for 90 h. (2) During the milling process, the matrix grain size decreased with the increasing milling time and V P content. After milling for 90 h, the average grain size of AZ31-X wt % V P (X = 5, 7.5, 10) was refined to only 23 nm, 19 nm and 16 nm, respectively. Besides that, the average grain size of AZ31-V P composites increased to 56, 54, 49 nm, respectively, after the sintering process, which still remained at NC scale due to the pinning effect of sub-micron V P . (3) The porosity of the AZ31-V P composites was no more than 7.8%, indicating a good densification effect. Moreover, the effect of small porosity on mechanical properties was little. Compared to as-cast AZ31 Mg alloy, the microhardness of AZ31-X wt % V P (X = 5, 7.5, 10) composites increased by 65%, 87% and 96%, respectively. The corresponding strengthening mechanisms included grain refinement strengthening, Orowan strengthening and load-bearing effects.