2016 Volume 57 Issue 11 Pages 1872-1879
Mechanical properties of magnesium alloys under dynamic loading are still unclear. To evaluate the impact fracture behavior of magnesium alloys, we constructed a novel impact three-point bending test apparatus using three elastic bars with Charpy standard-size specimen, and investigated the impact fracture properties of as-cast Mg-3Al-1Zn (hereafter denoted as AZ31) alloy. Finite element (FE) analysis were carried out to estimate the effect of inertial force of the specimen during the impact three-point bending. Based on the FE analysis, we successfully developed a small-scale apparatus for examining a quarter-size specimen, which was capable of carrying out the impact three-point bending test with minimized influence of the inertial force. Impact fracture behavior of Mg-6Al-1Zn-2Ca (hereafter denoted as AZX612) alloy was estimated and compared by using small-scale apparatus. The experimental results pointed out that the AZX612 had similar energy absorption capability to AZ31 against the dynamic loading, however, the crack propagation speed of the Ca bearing alloy was almost twice as fast as that of the AZ31 alloy.
This Paper was Originally Published in Japanese in J. Japan Inst. Light Metals 66 (2016) 258–265.
Recent years, in order to reduce the environmental load, magnesium alloys, which have less density than that of steels and aluminum alloys, are desired to application for the structural component of vehicles1,2). Several studies have reported that addition of rare earth elements, and grain refinement can improve the mechanical properties of Mg alloys3–6). Limited numbers of reports are, however, available on mechanical properties of Mg alloys under high-strain rate deformation. Because some Mg alloys have strain rate dependence of their mechanical properties7,8), it is important to evaluate the high-strain rate deformation behavior, which is significant aspect in collision safety for transportations, of the Mg alloys. In order to investigate the tensile – compressive properties at high-strain rate, split Hopkinson pressure bar (SHPB) method is commonly used9,10). The SHPB method can evaluate stress – strain relation at strain rates exceeding $1.0 \times 10^3\,{\rm s}^{-1}$ using elastic bars and strain gauges. However, the SHPB method can only estimate the dynamic “deformation” behavior, but can't directly evaluate dynamic “fracture” behavior. Charpy impact test is a standard testing method to estimate the resistance of materials against impact fracture11). The Charpy impact test has, however, two disadvantages derived from the structure of testing apparatus: the first one is limitation of maximum applied loading rate (5 m/s), and another is incapability of measurement for deformation behavior of specimen during the testing. Then, instrumented Charpy impact test was developed to measure load – displacement relation of specimen during the test12). However, the structure of instrumented Charpy impact test apparatus is too complicated to measure the stress waves with accuracy. Yokoyama et al. have developed novel impact three-point bending test procedure based on the SHPB method, in combination with finite element (FE) analysis, for determining the dynamic fracture initiation toughness $K_{\rm Id}$ of materials13). The impact three-point bending test using elastic bars is possible to evaluate the deformation and fracture behavior of materials with accuracy as well as achievement of higher loading rate than that of Charpy impact test. On the other hand, it is reported that there is great possibility of influence of inertial force on experimental results of the above-mentioned impact fracture tests14). For evaluation of the impact fracture properties of materials while taking effects of inertial force into consideration, application of several fracture mechanical parameters were proposed15,16). However, in practical use, it seems that the development of test method itself is desired with sufficiently suppression of influence of inertial force.
Taking these research backgrounds, in this study, we constructed impact three point bending apparatus using three elastic bars, which was developed by Yokoyama et al., with ultra-high speed video camera, and the effects of inertial force on experimental results was estimated by FE analysis. Then a small-scale impact three-point bending apparatus was successfully developed with minimized effects of the inertial force. Furthermore, the impact fracture behaviors of Mg-3Al-1Zn (AZ31) alloy, which was continuous casting material, and Mg-6Al-1Zn-2Ca (AZX612) alloy, which was extruded material, were evaluated using the novel small-scale testing apparatus.
Figure 1 shows schematic illustration of impact three-point bending apparatus, and Fig. 2 shows dimension and configuration of three point bending specimen, which is standard size specimen used in Charpy impact test11). This apparatus was consisted with one incident bar, two transmission bars, absorbers, striker bar, launching pipe, and air compressor. The length of incident bar, two transmission bars, and absorbers were 1295 mm, 1095 mm, and 400 mm, respectively. The incident bar, transmission bars, and absorbers were made of tool steel SKD11 with a diameter of 16 mm. The striker bar was made of brass with a diameter of 16 mm. The two transmission bars were set up with the distance between specimen support points of 40 mm. The contact end of incident bar and two transmission bars formed a diagonal hammer and anvil for Charpy impact test.
Schematic illustration of testing apparatus for impact three-point bending. The variables P and u indicate the applied force and displacement of each elastic bar, respectively.
Dimensions of V-notch specimen (standard: a = 1, quarter-size: a = 0.5).
In this apparatus, striker bar was accelerated by compressed air and impacted one end of the incident bar. Then, compressive stress waves were generated in contacted end, and propagating in the incident bar and the striker bar. When the compressive stress wave reached the opposite end of the striker bar, the stress wave was free-end-reflected and propagating in the striker bar as a tensile stress wave. The striker bar stopped and came off from the incident bar when the tensile stress wave approached the contacted end. The time until striker bar stopped and away from the contact with the incident bar is wavelength of the compressive stress wave (incident wave) propagating in the incident bar. The wavelength corresponded to duration of loading. Because the propagation speed of waves in elastic bars differs depending on materials of elastic bars, the duration of loading can be controlled by changing the materials, and/or the length of materials. By adjusting the magnitude of the compressive stress wave in the incident bar by changing air pressure, it is also possible to control the strain rate. On the other hand, when the incident wave reached contacted end with the specimen, a part of the incident wave propagated to the specimen depending on the energy required for deformation. Remains of the incident wave was free-end-reflected and propagating in the incident bar as a tensile stress wave (reflected wave). The stress wave transmitted through the specimen propagated in the two transmission bars as transmitted waves. The load – displacement curve during impact three-point bending was described by measurement of stress waves propagating in the incident bar and two transmission bars. In addition, the deformation and crack propagation behaviors of the specimen during the test were observed by using the ultra-high speed video camera with a sampling rate of 2.5 Mfps, and the captured image size of $312 \times 260$ pixels.
2.2 Evaluation of load – displacement relationThe applied load, and relative displacement of the specimen were calculated as followed procedure17,18).
In this study, the electrical resistance change of strain gauges was monitored as the voltage change with an oscilloscope. Figure 3 shows an example of detected the incident wave and the two transmitted waves. The strain $\varepsilon$ was calculated by
| \[\varepsilon = \frac{2 \times V}{V_{\rm g} \times K \times G}\] | (2-1) |
An example of detected stress waves.
The load and displacements of each elastic bars were defined as shown in Fig. 1. Based on the principles of one-dimensional elastic wave propagation, the displacement of the specimen $u$, and load P of three elastic bars are expressed as
| \[u = \int \frac{C_b}{2} \{ -4\varepsilon_i + 3(\varepsilon_{t_1} + \varepsilon_{t_2}) \} dt\] | (2-2) |
| \[\begin{array}{l} P_i = E_i A_i(\varepsilon_i + \varepsilon_r) \\ P_{t_1} = E_{t_1} A_{t_1} \varepsilon_{t_1} \\ P_{t_2} = E_{t_2} A_{t_2} \varepsilon_{t_2} \end{array}\] | (2-3) |
The crack propagation speed is an important evaluation criteria of investigation of the impact fracture behavior. The crack propagation length was measured from the images obtained by ultra-high speed video camera. The crack propagation speed was determined as the slope of the variations of crack propagation length with time.
2.4 Impact three-point bending test for as-cast AZ31 alloyThe impact three-point bending test was performed for as-cast AZ31 alloy using the above-mentioned apparatus. Figure 4(a) shows the optical micrograph of the initial microstructure of the as-cast AZ31 alloy. The observation plane was parallel to the casting direction. The chemical composition and quasi static tensile properties were summarized for Table 1 and Table 2, respectively. The shape of the cast material was a cylindrical ingot having a diameter of 155 mm. The specimen was machined from center of the ingot with longitudinal direction parallel to the casting direction. The V-notch was introduced perpendicular to the casting direction.
Microstructure of (a) as-cast AZ31, and (b) extruded AZX612 alloy.
| Al | Zn | Ca | Mn | Si | Cu | Ni | Fe | Mg | |
|---|---|---|---|---|---|---|---|---|---|
| AZ31 as-cast |
2.9 | 0.84 | — | 0.42 | 0.016 | <0.002 | <0.002 | 0.003 | Bal. |
| Extruded AZX612 |
5.9 | 0.53 | 2.04 | 0.27 | 0.020 | <0.002 | <0.002 | <0.002 | Bal. |
| AZ31 as-cast |
Extruded AZX612 |
|
|---|---|---|
| 0.2% proof stress [MPa] | 56 | 265 |
| Ultimate tensile strength [MPa] | 175 | 319 |
| Elongation to failure | 0.106 | 0.086 |
The sequence images of the Charpy standard size specimen during impact three-point bending were shown in Fig. 5. The right bottom number of each image indicated the elapsed time from start recording. The incident bar began to be displaced 48 $\mu$sec elapsed from start recording (Load start). After the 72 $\mu$sec elapsed from Load start, the initiation of the crack at the tip of the V-notch, which was indicated by white circle, was observed. Then, the crack propagated with the displacement of the incident bar. After the 272 $\mu$sec elapsed from Load start, the incident bar stopped being displaced by the first wave of incident stress wave. Figure 6 shows the variations of displacement with time obtained by data calculated from strain gauge output, and by measured using ImageJ from the images of ultra-high speed camera. Because, as shown in Fig. 6, the calculated displacements agreed with the measured displacements, it was confirmed that the stress waves were appropriately measured using strain gauges. The displacement rate of the incident bar was estimated from the slope of the variations of the displacement of the incident bar with time. In this study, the displacement rate of the incident bar was regarded as the impact loading speed of the impact three-point bending. From Fig. 6, the impact loading speed of the testing for the as-cast AZ31 alloy was estimated as 8.8 m/s.
Sequence images of the standard size specimen during the impact three-point bending test.
Comparison of displacement between the data calculated from strain gauge output and measured in ultra-high speed camera images.
Figure 7 shows the variations of the crack propagation length with time. The crack length was measured from the images of ultra-high speed video camera using ImageJ. The crack propagation speed of as-cast AZ31 alloy was evaluated to be 9.2 m/s. Figure 8 shows the normalized load – displacement curve of the as-cast AZ31 alloy. In this study, the length of stress waves measured without interference was limited by the scale of the test apparatus, namely, the length of the elastic bars. Therefore, it is unable to obtain the load – displacement curve until the specimen completely to be fractured. The design of the apparatus which enables to measure load – displacement relation until the specimen completely to be fractured is a topic for future study. The white outline mark indicates the displacement for the crack initiation point observed by ultra-high speed video camera. From these results, the developed test apparatus in this study achieved that estimation for crack propagation speed, evaluation of the load required to crack initiation, and observation of decrease of the load due to the crack propagation. The conventional test methods such as the Charpy impact test are difficult to investigate these impact fracture behaviors. Therefore, the novel impact three-point bending test apparatus constructed in this study has usefulness for quantitative evaluation of impact fracture properties.
Variations of crack length with time in impact three-point bending tests.
Normalized Load – Displacement curve of as-cast AZ31 and extruded AZX612 alloys. The white outline and hatched marks indicate the displacements for the crack initiation point and the crack propagation to X mm (X = 0.8 (Standard specimen), 0.4 (Quarter-size specimen)), respectively.
However, due to the size of the specimen used in this impact three-point bending was relatively large, it is possible that the inertial force, which is generated by steep velocity change of the specimen, influences on the experimental results such as load – displacement relation. Then, the FE analysis of impact three-point bending test for as-cast AZ31 alloy was performed to estimate the effects of the inertial force.
Kawa et al. conducted the impact three-point bending simulation for as-cast AZ31 alloy by means of the FE analysis in previous study17). Figure 9 shows the mesh model which was constructed in this previous study. The FE analysis software ANSYS/LS-DYNA was used for numerical simulation of impact three-point bending. In this simulation, the length of incident bar and two transmission bars were set to 3000 mm and 1500 mm, respectively, in order not to cause interference with incident wave length, which can completely fracture the specimen, within the measurement time. Other configuration and dimensions were same as that of the actual test apparatus. The element type used in this simulation was three-dimensional eight nodes element. The elements of the incident bar, the two transmission bars and the striker bar were treated as elastic body, and the elements of the specimen were treated as piecewise linear elasto-plastic body. The material properties of the incident bar, the transmission bars, and the specimen for FE analysis were summarized in Table 3. Kawa et al. reported that load – displacement relation, and crack propagation behavior of the specimen in this impact three-point bending simulation agreed with the experimental results, therefore, this simulation is assumed to have the validity17). Then, in this study, we estimate the effects of the inertial force by means of this impact three-point bending simulation.
FE model of impact three-point bending set-up17).
| SKD11 | Brass | AZ31 as-cast |
|
|---|---|---|---|
| Young's modulus [GPa] | 205 | 105 | 45 |
| Poisson's ratio | 0.35 | 0.35 | 0.35 |
| Density [kg/m3] | 8,000 | 8,500 | 1,740 |
| Yield stress [MPa] | — | — | 63 |
| plastic strain at failure | — | — | 0.159 |
The inertial force generated in the specimen during the impact three-point bending was estimated from FE analysis. The speed of each element at each time can be obtained in ANSYS/LS-DYNA. Furthermore, we can also obtain the volume of each element. The inertial force of the specimen $F_{t_n}$ at the time $t_n$ were calculated by
| \[F_{t_n} = \sum_{i=1}^N \left( \rho V_i \frac{\dot{u}_i^{t_n} - \dot{u}_i^{t_{n-1}}}{t_n - t_{n-1}} \right)\] | (3-1) |
Figure 10(a) and Fig. 10(b) show the variations of elastic strain wave propagated in the incident bar with time, and the variations of inertial force of the specimen with time calculated from eq. (3-1), respectively. The time of $t = 0$ indicates the time, which is the incident bar begins to be displaced. As shown in Fig. 10(b), when the stress wave, which steeply rose at 20 $\mu {\rm sec}$ (Steep shaped wave in Fig. 10(a)) was set, the very high inertial force was immediately generated after the start of the load to the specimen. By this high inertial force, it is assumed that the load required to deform the specimen is overestimated in the two transmission bars. In this simulation, the stress wave was generated by impacting the striker bar to the incident bar with controlling the speed of the striker bar. Then, by controlling the speed of the striker bar so as to sinusoidally change, the start of the incident wave gradually rose (Gradual shaped wave in Fig. 10(a)). As shown in Fig. 10(b), the inertial force of the specimen with gradual shaped wave was effectively lowered than that of steep shaped wave. Figure 10(c) shows load – displacement relation obtained by FE analysis. The vibration of the load occurred at the initial states of the load – displacement curve with steep shaped wave. However, the vibration was efficiently reduced in the load – displacement curve with gradual shaped wave. On the other hand, the aluminum small piece of $2\,{\rm mm} \times 2\,{\rm mm} \times 1\,{\rm mm}$ was attached the contacted end of the incident bar as the pulse shaper in the impact three-point bending test described in the previous chapter, in order to apply the relatively gradual shaped incident wave as can be seen in Fig. 3. Therefore, it is indicated that moderating the rising of the incident wave by using the pulse shaper is effective in order to reduce the inertial force of the specimen.
Effect of loading rate in FE analysis; (a) two variations of incident wave as the input data, (b) variations of inertial force, and (c) variations of load-displacement curve as the output data.
The FE analysis revealed that the influence of the inertial force can be reduced by applying the gradual shaped incident wave. However, the influence of the inertial force is yet to be perfectly ignored. It is believed that reducing the mass of the specimen is effective to minimize the effects of inertial force. Thus, the impact three-point bending simulation with small size specimen model, which have quartered cross-sectional area compared to the Charpy standard size specimen model, were carried out to estimate the effects of the inertial force. Hereafter, we called the FE model with the Charpy size specimen as the standard specimen model, and the FE model with the small specimen having quartered cross-sectional area as the quarter-size specimen model. The configuration and dimensions of the quarter-size specimen were similar to the Charpy standard specimen as shown in Fig. 2. The diameter of the elastic bars were 8 mm with quarter-size specimen. The dimensions of the contact end of incident bar and two transmission bars were half similar shape with the dimensional ratio in accordance with the quarter-size specimen. In addition, the gradual shaped incident wave was set in this simulation. Figure 11(a) shows the load – displacement curve, and Fig. 11(b) shows the variations of the inertial force with time. As shown in Fig. 11(b), the inertial force of the quarter-size specimen model drastically decreases compared to the standard specimen model. The values of the inertial force are sufficiently lower than that of the load measured from the stress waves in the transmission bars. Therefore, it is suggested that the impact three-point bending test is executable using the small-scale apparatus and the quarter-size specimen with minimize the influence of the inertial force of the specimen.
Result of FE analysis for quarter-size specimen; (a) load-displacement curve and (b) inertial force.
Based on the FE analysis, the novel small-scale impact three-point bending apparatus with the quarter-size specimen were actually constructed to evaluate the impact fracture behaviors of the materials. The impact three-point bending test for as-cast AZ31 alloy were performed using the small-scale test apparatus. In this testing, the impact loading speed was 7.1 m/s. Figure 7 and Fig. 8 show the variations of the crack propagation length with time, and normalized load – displacement curve, respectively. From Fig. 7, the crack propagation speed of as-cast AZ31 alloy, which was estimated using quarter-size specimen, was determined as 8.3 m/s. This value was almost equal to the value of the crack propagation speed obtained using the standard specimen. While this test apparatus has slight deviations of the impact load speed, the crack propagation speed was divided by the impact load speed of each testing in order to normalize. The normalized crack propagation speed of as-cast AZ31 alloy with the standard specimen was 1.04, and that with the quarter-size specimen was 1.18. These values are almost equal.
From Fig. 8, though the normalized load – displacement curve obtained using the quarter size specimen mostly agreed with that of the quarter-size specimen, in the initial state of the normalized load displacement curve, the load per unit area of the standard specimen was higher than that of quarter-size specimen. It is implied that due to the influence of the inertial force of the standard specimen is greater than that of the quarter-size specimen, the measured load for the standard specimen is larger than the load necessary for actual deformation of the specimen. In Fig. 8, the white outline and hatched marks indicate the displacements for the crack initiation point and the crack propagation to X mm (the standard specimen: X = 0.8, the quarter-size specimen: X = 0.4), respectively. The crack initiation point and the crack propagates to X mm point of quarter-size specimen were in accord with those of the standard specimen.
In this study, in order to evaluate the impact toughness, the absorbed energy per unit volume E was calculated. E was calculated as the area under the normalized load – displacement curve. The absorbed energy per unit volume by crack initiation, and from crack initiation until crack propagates to X mm (above-mentioned) were calculated. Figure 12 shows the results of amount of absorbed energy per unit volume E of each testing. From Fig. 12, while the results of amount of absorbed energy per unit volume of the quarter-size specimen agreed with those of the standard specimen, the absorbed energy per unit volume by crack initiation of the standard specimen was higher than that of the quarter-size specimen due to the influence on the inertial force.
Absorbed energy per unit volume of as-cast AZ31 (Standard and Quarter-size specimen) and extruded AZX612 (Quarter-size specimen).
Taking these comparison of the crack propagation speed, normalized load – displacement curve, and amount of absorbed energy per unit volume, it is suggested that the experimental results of the quarter-size specimen have no extremely large differences from those of the standard specimen. Therefore, the novel small-scale test apparatus with the quarter-size specimen enables to investigate the impact fracture behaviors of materials with minimizing the influence of the inertial force. The downsizing of the test apparatus and the specimen is not only effective for the suppression of the inertial force but also useful for evaluation of impact fracture properties of small test piece such as welded joint. It is believed that the developed novel small-scale impact three-point bending test apparatus is useful for the evaluation of the impact fracture behaviors of a wide variety of materials.
4.2 Impact fracture properties of extruded AZX612 alloyThe impact fracture behaviors of the extruded AZX612 alloy was evaluated using the small-scale apparatus. The AZX612 alloy is the flame-resistant Mg alloy by addition of 2 mass% calcium19,20). Recent years, these frame-resistant Mg alloys bearing Ca are desired to apply to the components of bullet trains2,21). Figure 4(b) shows the optical micrograph of the initial microstructure of the extruded AZX612 alloy. The observation plane was parallel to the extrusion direction. The chemical composition and quasi static tensile properties were summarized for Table 1 and Table 2, respectively. The shape of the extruded material was a sheet having a width of 65 mm and a thickness of 6 mm. The specimen was machined from center of the sheet with longitudinal direction parallel to the extrusion direction. The V-notch was introduced perpendicular to the extrusion direction. Figure 7 and Fig. 8 show the variations of crack propagation length with time and normalized load – displacement curve of the extruded AZX612 alloy, respectively. In this testing, the impact loading speed was 7.9 m/s. The normalized load – displacement curve indicates that the load per unit area at crack initiation of the extruded AZX612 alloy was about 1.9 times higher than that of the as-cast AZ31 alloy. Furthermore, from Fig. 7, the crack propagation speed and normalized crack propagation speed of the extruded AZX612 alloy were determined as 16.4 m/s and 2.08, respectively. These values were twice as fast as those of the as-cast AZ31 alloy. This trend corresponded to the decrease of the load after the crack initiation of the extruded AZX612 alloy is more rapidly than that of the as-cast AZ31 alloy. The absorbed energy per unit volumes by crack initiation of the extruded AZX612 alloy was almost equal to that of the as-cast AZ31 alloy (Fig. 12). In addition, the amount of absorbed energy per unit volume from crack initiation until crack propagates to 0.4 mm of the extruded AZX612 alloy was slightly higher than that of the as-cast AZ31 alloy.
Therefore, it is suggested that the extruded AZX612 alloy has high strength than the as-cast AZ31 alloy under dynamic loading, however, the extruded AZX612 alloy is easier to propagate the crack than the as-cast AZ31 alloy.
4.3 Fracture mode by impact three-point bendingThe fractography by scanning electron microscopy (SEM) for the as-cast AZ31 alloy and the extruded AZX612 alloy after the impact three-point bending were carried out. Figure 13 shows the SEM images of fracture sufaces. Figure 13(a) and (b) typically shows parallel streaks on the fracture surfaces of the as-cast AZ31 alloy. It means that the crack propagates along the some sort of the interfaces. Somekawa et al. reported that the cracks propagate in parallel to the $\{10\bar{1}2\}$ mechanical twin boundary through the static fracture toughness test for the AZ31 alloy22). Thus, it is believed that the cracks propagate along the interfaces between the twin boundary and $\alpha$-magnesium of the as-cast AZ31 alloy under dynamic loading. The fracture surface of the extruded AZX612 alloy, on the other hand, was imitate intergranular fracture surface such like uneven surface with contrasted brightness and shade (Fig. 13(c)). As shown in Fig. 4(b), the extruded AZX612 alloy has the black precipitates due to the addition of Ca more than the solubility. It is suggested that due to the cracks propagates along the interface between $\alpha$-magnesium and the precipitates clumped together around the grain boundary, the fracture surface of the extruded AZX612 alloy resembles the intergranular fracture surface. In addition, the micro-cracks assumed to be cracked precipitates, which were indicated white circles in Fig. 13(d), were typically found on the fracture surface of the extruded AZX612 alloy. The cracked precipitates promote stress concentration and the crack propagation of materials. Therefore, it is suggested that the precipitates have baneful influences on impact toughness of the extruded AZX612 alloy. At the same time, it is believed that controlling dispersion of the precipitates is important material design criteria in order to improve the impact fracture toughness of Ca bearing flame-resistant Mg alloys.
Fracture surfaces of (a) as-cast AZ31 (Standard specimen), (b) as-cast AZ31 (Quarter-size specimen), and (c) extruded AZX612 alloy (Quarter-size specimen) after impact three-point bending test and (d) Micro cracks on fracture surface of AZX612 alloy.
In this study, the impact three-point bending apparatus for quantitative evaluation of the deformation and fracture behavior of the materials under dynamic loading was constructed. The influence of the inertial force was estimated using FE analysis, and the novel small-scale test apparatus, which can more precisely evaluate the impact fracture properties of materials with minimizing the influence of the inertial force, was successfully developed. The results obtained in this study are summarized as follws.
(i) The impact three-point bending test using the Charpy standard size specimen for as-cast AZ31 alloy were carried out. Through the testing, it is indicated that the crack propagation speed and load – displacement relation were able to be measured with accuracy.
(ii) The FE analysis for the constructed impact three-point bending apparatus were performed to investigate the effects of the inertial force of the specimen. From the FE analysis, it is found that the inertial force decrease by the moderating the start of the incident wave. In addition, it is suggested that use of the quarter-size specimen compared to the Charpy standard size specimen and the small-scale apparatus can sufficiently suppress the influence of the inertial force.
(iii) We actually developed the small-scale impact three-point bending apparatus. The experimental results obtained using the quarter-size specimen and the small-scale apparatus were almost agreed with that of the standard size specimen. Therefore, it is indicated that the impact fracture behavior of the materials was able to be evaluated using the novel small-scale apparatus.
(iv) The impact three-point bending test were conducted for the extruded AZX612 alloy using the small-scale apparatus. Direct comparison to the as-cast AZ31 alloy, the load required to the crack initiation of AZX612 alloy was about twice as high as that of AZ31 alloy, however, the absorbed energy per unit volume of AZX612 by crack initiation was same level of that of the AZ31 alloy. Furthermore, the absorbed energy per unit volume of AZX 612 during the crack propagation was slightly higher than that of AZ31 alloy. The crack propagation speed of AZX612 was more than twice as fast as AZ31 alloy. Therefore, it is suggested that the extruded AZX612 alloy is easy to propagate the crack while have higher strength than as-cast AZ31 alloy.
(v) The fractography were performed for the as-cast AZ31 alloy and the extruded AZX612 alloy after the impact three point bending. From the fractography for the AZX612 alloy, the cracks of the precipitates were typically observed on the fracture surface. Thus, it is implied that the cracks along the interface between $\alpha$-magnesium and the precipitates, and the cracks of precipitates itself cause stress concentration, which promote the crack propagation. It is believed that controlling the dispersion of the precipitates is important for improvement of the impact fracture toughness of Mg alloys bearing calcium.
Contribution of the first and second authors are equal. This work was supported in part by JSPS KAKENHI Grant No. 25246012 and the Light Metals Education Foundation, Japan. The results of AZX612 alloy were obtained from a project commissioned by the New Energy and Industrial Technology Development Organization (NEDO).
