, Volume 66, Issue 2, pp 277–290

Dynamic Behavior of a Rare-Earth-Containing Mg Alloy, WE43B-T5, Plate with Comparison to Conventional Alloy, AM30-F


    • Materials Science and EngineeringUniversity of Virginia
  • Wilburn Whittington
    • Mechanical EngineeringMississippi State University
  • Andrew Oppedal
    • Mechanical EngineeringMississippi State University
  • Haitham El Kadiri
    • Mechanical EngineeringMississippi State University
  • Matthew Shaeffer
    • Mechanical EngineeringJohns Hopkins University
  • K. T. Ramesh
    • Mechanical EngineeringJohns Hopkins University
  • Jishnu Bhattacharyya
    • Materials Science and EngineeringUniversity of Virginia
  • Rick Delorme
    • Magnesium Elektron North America
  • Bruce Davis
    • Magnesium Elektron North America

DOI: 10.1007/s11837-013-0830-x

Cite this article as:
Agnew, S., Whittington, W., Oppedal, A. et al. JOM (2014) 66: 277. doi:10.1007/s11837-013-0830-x


The dynamic behavior of Mg alloys is an area of interest for applications such as crash-sensitive automotive components and armor. The rare-earth element-containing alloy WE43B-T5 has performed well in ballistic testing, so the quasi-static (~10−3 1/s) and dynamic (~600–5000 1/s) mechanical behaviors of two Mg alloys, rolled WE43B-T5 and extruded AM30-F, were investigated using servohydraulic and Kolsky bar testing in uniaxial tension and compression. The yield stress was surprisingly isotropic for WE43B-T5 relative to conventional Mg alloys (including extruded AM30-F). The WE43B plate was textured; however, it was not the typical basal texture of hot-rolled Mg-Al alloys. The effect of strain rate on the yield strength of WE43B-T5 is small and the strain-hardening behavior is only mildly rate sensitive (m = 0.008). The combination of high strength (~300 MPa), moderate ductility (0.07–0.20), and low density yield a material with good specific energy absorption capacity.


The Mg alloy designated WE43 [which contains additions of 3.7–4.3 wt.% yttrium (W), 2.4–4.4 wt.% neodymium-rich mischmetal (E), and at least 0.4 wt.% zirconium] was originally introduced as a casting alloy with similar high-temperature strength and improved ductility, relative to sister alloy, WE54.1,2 More recently, the alloy has been developed into a commercially produced direct-chill cast and hot-rolled plate product.3 This material has demonstrated potential as an armor material, given its good ballistic performance, relative to competing lightweight metallic materials, such as Al-Mg alloy 5083.4 The current study was initiated to determine whether there are any unique features regarding the constitutive behavior of this new commercially available plate alloy during high-strain-rate deformation. There have recently been quite a number of studies of the high-strain-rate behavior of Mg alloys,516 so there is a significant repository of data against which to compare the response of this new material.

The goal of the current study was to determine the high-strain-rate deformation behavior of a commercially produced thick plate of Mg alloy WE43B-T5. Tests were also performed on a more conventional Mg alloy AM30-F, which was recently proposed as a replacement for some potential applications of AZ31B, since it can be extruded at higher rates.17 One feature that has been emphasized in prior studies is the dominant role of crystallographic texture. Thus, a complete description of the texture in the materials was determined using x-ray and neutron diffraction prior to testing.


Effect of Grain Size Refinement on the Dynamic Response of Mg Alloys

Notably, one of the earliest high-strain-rate deformation studies was performed on the alloy presently of interest WE43.6 Mukai et al.6 concluded that a significant improvement in energy absorption could be achieved by strong grain refinement (1.4 μm) achieved by high-reduction (100:1) extrusion. While the Mg alloy was not as strong as the high-strength Al alloys (7075-T6 and IN905XL) to which it was compared, the dynamic tensile ductility of the grain-refined WE43 was shown to be more than twice that of the Al alloys. As a result of this high ductility and the low density characteristic of Mg alloys, the energy absorption per unit mass (q/ρ) was larger than that of Al-Mg alloy 5056-0 and more than double that of the high-strength Al alloys. The energy absorption per unit mass is determined by integrating the uniaxial stress–strain curve and dividing by the mass density ρ.
$$ \frac{q}{\rho } = \frac{1}{\rho }\int\limits_{0}^{{\varepsilon_{f} }} {\sigma \left( \varepsilon \right) \cdot d\varepsilon } $$

One wonders how much this result might have had to do with the weak dynamic recrystallization texture that rare-earth alloys can exhibit18 and that has achieved so much attention in recent years.19 The role of texture is one issue that will be explored in the current study.

The idea of grain refinement was also explored for alloy ZK60, first via conventional extrusion, which yielded a fine grain size of 3 μm,7 and more recently, even down to 1 μm.15 Alternatively, equal-channel angular pressing (ECAP) has been used to achieve a grain size of 0.8 μm.13 Mukai et al.7 again showed enhanced ductility and energy absorption capacity in the fine-grained material. Another interesting feature of some examples of fine-grained, extruded ZK60 is the absence of a tension compression yield strength asymmetry.15 Nevertheless, the strain-hardening behavior (and texture evolution) shows that much more twinning is occurring during compression along the extrusion axis than tension along the same direction. Interestingly, the ECAP-processed ZK60 did not experience the huge increase in ductility,13 which had been observed previously in alloy AZ31 at both quasi-static and dynamic strain rates.9 The effect in AZ31 was attributed to a change in crystallographic texture during ECAP, and indeed, one of the current authors has observed that different Mg alloys undergo different texture evolution during ECAP.20

Effect of Texture on the Dynamic Response of Mg Alloys

The studies of Livescu,10 Tucker et al.,12 and Ulacia et al.14 have made very clear the strong role that texture plays in the dynamic response of Mg alloy plates. First, they highlight the strong anisotropy that also is exhibited by textured, wrought Mg alloys at quasi-static strain rates. Comparisons of through-thickness compression with in-plane compression of plates with basal texture (〈00.1〉 || plate thickness direction) revealed parabolic hardening in the former and sigmoidal hardening in the latter. The S-shaped flow curve was characterized by an initially low-yield plateau followed by rapid hardening and finally slowed hardening and failure. This behavior is related to the tension–compression yield strength asymmetry that has been studied extensively in Mg alloys. The soft {10.2} extension-twinning mode is responsible for the lower yield strength in compression as well as the subsequent unusual hardening response.21 During through-thickness compression, other deformation mechanisms are required for generalized flow, such as 〈c + a〉 slip22 or compression twinning.23 Ulacia et al.14 employed both tension and compression testing and showed that the tension–compression yield strength asymmetry of textured Mg alloy, AZ31, plates persists at high strain rates.

Effect of Dynamic Strain Rates on the Flow Stress of Mg Alloys

The dependence of the flow stress on the strain rates is often approximated as obeying a power law, where the exponent is dubbed the strain-rate sensitivity m. This parameter may be determined by assessing the slope of the log–log plot of the flow stress at a fixed plastic strain versus the strain rate:
$$ m = \frac{\partial \ln \sigma \left( \varepsilon \right)}{{\partial \ln \dot{\varepsilon }}} $$

All three of the aforementioned studies of textured AZ31 sheet or plate10,12,14 highlighted the fact that the yield strength is essentially rate insensitive when the flow is dominated by extension twinning (e.g., in-plane compression of basal-textured plate.) In fact, Ulacia et al.14 observe that the stress to activate twinning appears to be constant regardless of rate or temperature, up to 400°C. This observation is consistent with the recent quasi-static analysis of the strain-rate sensitivity of twinning-dominated flow by Chun and Davies.24

Interestingly, the through-thickness compression test data also show only slight rate sensitivity, m = 0.007, which can be calculated from the values of the 0.5% flow stress reported by Ulacia et al.14 at ambient temperature. The current authors would associate such a low m-value, in this case, with the fact that the initial yield during through-thickness compression is controlled by basal slip,22 and basal slip in Mg is known to be essentially athermal at ambient temperatures and above,25 at least for quasi-static strain rates. On the other hand, when tested in tension within the plane of a sheet14; a higher value of m = 0.037 is observed. The flow stress during in-plane tension of such strongly textured plate materials is primarily controlled by nonbasal slip of 〈a〉 type dislocations.22 The thermally activated nature of nonbasal (i.e., prismatic) slip in Mg and its alloys is well established.2629

The strain-rate sensitivities of the ultimate stresses are rather independent of direction.10,12,14 For example, employing the data of Ulacia et al. shows the rate sensitivity of the maximum stress to be m = 0.016 ± 0.003, for tension along the sheet-rolling and transverse directions and compression along the sheet-rolling and normal directions.14 This relative isotropy in rate sensitivity is due to the fact that the flow stresses at larger strains are controlled by thermally activated mechanisms regardless of direction. For the in-plane tension direction, it remains largely controlled by nonbasal slip of 〈a〉 and 〈c + a〉 dislocations (with a potential influence of contraction twinning). For the through-thickness compression and in-plane compression, it is certainly the 〈c + a〉 slip (and/or contraction twinning) that controls the flow stress at these larger strains (~10–15%). The thermally activated nature of these latter mechanisms is established.3032 Experimentally, the roles of 〈c + a〉 slip and contraction twinning are difficult to separate,33,34 although atomistic simulation is beginning to provide insight.32

Yokoyama8 tested three conventionally extruded Mg alloys (AZ31B-F, AZ61F, and ZK60A-T5) in tension at quasi-static and dynamic strain rates. They observed parabolic hardening indicative of slip- (rather than {10.2} extension twinning-) accommodated deformation. Furthermore, they showed that all three alloys exhibit modest strain-rate sensitivity (at strain levels of 5% and 10%), with values in the range of m = 0.001 to 0.013, and a typical (average) value of mavg = 0.009 at 5% strain regardless of alloy. This value agrees reasonably well with the behavior observed for the aforementioned rolled products. Perhaps the small discrepancy may be attributed to distinctions in the crystallographic texture between rolled and extruded material.

Only a rough estimate of the strain-rate sensitivity can be developed for WE43. Based on the UTS data obtained from extruded samples by Mukai et al.,6 it is in the range of m = 0.004–0.016, with an average value of 0.010, depending upon thermomechanical processing history. This is similar to the values reported above for conventional wrought alloys AZ31, AZ61 and ZK60 (at larger strain levels of 5–15%). In the present study, estimates of the rate sensitivity are made as a function of straining direction and strain level.

Effect of Strain Rate on the Fracture Behavior of Mg Alloys

In addition to transitions in plastic flow stresses, ductility, and energy absorption capacity, which have been observed to occur with variations in grain size, texture, and strain rate, the fracture behavior has also been examined as a function of these variables. For instance, in their early study of alloy WE43, Mukai et al.6 commented on a transition in the fracture mode in alloy WE43, from one that was described as intergranular (for precipitation-hardened large grained material) to one that clearly exhibited ductile dimples. In the case of alloy ZK60, fracture was associated with cracks along twin boundaries in coarse-grained ZK60, while such strain localization-induced fracture seems to have been avoided in the fine-grained material.7 Thus, one wonders whether the assignment of intergranular failure to coarse-grained WE43 was correct; perhaps it only appeared so because twins are often observed to initiate at one grain boundary, cross the grain, and terminate on the opposite grain boundary. All three alloys that Yokoyama8 examined also exhibited a fracture surface transition from a single shearing surface to a more tortuous fracture surface (although still dominated by shear localization) under the conditions where higher ductility is observed.

Because correlations between ballistic performance and dynamic fracture mode are often observed, the microstructure evolution in the vicinity of ballistic penetrations of WE43B-T5 plate has already been investigated.35 The fracture mode of the plate samples tested under more controlled, uniaxial straining conditions will be briefly examined in light of these prior results, although it remains an area ripe for further study.

Experimental Methods


Ingots of Elektron WE43B alloy were direct-chill cast at the industrial casting facility of Magnesium Elektron, in Manchester, U.K. The slabs were cast with a nominal chemical composition (in wt.%) of 4 Y, 2 Nd, 1 rare-earth mischmetal, 0.5 Zr, balance Mg. They had dimensions of 870 × 315 mm cross section and 880 mm length. These slabs were scalped and hot rolled on a reversing mill at the facility of Magnesium Elektron North America, in Madison, IL down to 38-mm-thick plates at descending temperatures beginning in the range of 500°C. These as-deformed (F-temper) plates were artificially aged at 210°C for 48 h to obtain the peak-aged (T5-temper) condition examined in this study.

Alloy AM30 is the same material (chemical composition, wt.%: 2.54 Al, 0.4 Mn, 0.018 Zn, 0.03 Fe, 0.008 Si, 0.011 Cu, 0.025 Ce, 0.005 Ni, balance Mg) examined in a previous study of the quasi-static mechanical behavior.36 The material was direct-chill cast as a cylindrical ingot with ~457 mm (18 in.) nominal diameter and scalped to 450 mm (~17.5 in.). Subsequently, it was extruded to 180 mm (7 in.) diameter at ~570°C and 0.3 m/s ram speed, at the former Timminco Ltd. facility in Aurora, CO. This non-heat-treatable alloy was examined in the as-extruded (F-temper).


Optical and scanning electron microscope (SEM) imaging were employed to reveal the microstructure prior to testing. For optical-based metallographic examination, a sample was cut out from the rolling direction (RD)-normal direction (ND) plane of the WE43B-T5 material. It was cold-mounted in epoxy, mechanically ground down to 1200-grit SiC paper, polished with 3- and 1-μm oil-based diamond paste, and final polished with 0.06-μm colloidal silica. The specimens were etched in an acetal-nitric solution (10 mL nitric acid, 30 mL acetic acid, 120 mL ethanol, and 40 mL water) for 3–5 s. The microstructure of the WE43B-T5 was equiaxed, due to recrystallization during the hot-rolling process (Fig. 1). The mean lineal intercept grain size is 9 μm.
Fig. 1

Representative optical microscope image of the RD-ND plane of the WE43-T5 plate revealing an equiaxed microstructure which results from the hot-rolling. Black spots are due to pitting around second-phase particles

The AM30-F material was prepared for electron backscattered diffraction (EBSD) by grinding with SiC paper to 4000 grit in water and polishing with 5 and 0.3 μm Al2O3 suspended in ethylene glycol on Struers MD Mol and MD Chem cloths (Struers, Inc., Cleveland, OH). The samples received a light etch with 0.5% nital and ethanol rinse just before EBSD. EBSD data collection was performed with a Zeiss field-emission gun (FEG) Supra 40 SEM (Carl Zeiss, Oberkochen, Germany) equipped with an orientation image microscopy (OIM; EDAX TSL; EDAX Inc., Mahwah, NJ) at a step size of 100 nm on large regions of the samples. These EBSD analyses were completed using the commercially available TSL OIM Analysis software (EDAX Inc.). Microscopy reveals the AM30 material to consist of a partially recrystallized microstructure with large, elongated grains (hundreds of μm long) and smaller recrystallized grains in some regions (Fig. 2). Due to this duplex structure, no attempt to determine a mean lineal intercept grain size was made.
Fig. 2

Inverse pole figure (IPF) maps of preextruded AM30 in (a) section parallel to the extrusion direction and (b) section normal to the extrusion direction

The crystallographic texture of the WE43B-T5 plate was assessed using x-ray diffraction. The texture was assessed at the surface, one-sixth plane (6.3 mm depth), one-third plane (12.7 mm depth), and the mid-plane (19 mm depth), to explore the possibility of texture gradients. The measurements were made using a PANalytical X’pert Pro multi-purpose diffractometer (MPD; PANalytical, Almelo, The Netherlands) with Cu Kα radiation at 45 kV and 40 mA. The tube was used in point focus orientation, with cross-slit texture optics having 2 mm horizontal and 1 mm vertical dimensions, following the Schultz reflection method,37 to minimize defocusing at higher χ tilts. The {1 0. 0}, {0 0. 2}, and {1 1. 0} incomplete (χ = 0° to 80°) pole figures were obtained. The background was measured on each sample and an experimental defocusing curve was obtained from a random Ti sample, made by curing a mixture of Ti powder and epoxy. Background subtraction, defocusing correction, and generation of complete orientation distributions and full pole figures from the incomplete pole figures were carried out using the MATLAB (MathWorks, Natick, MA) toolbox, MTEX.38

Neutron diffraction texture analysis was conducted on the AM30 samples with the high-intensity pressure and preferred orientation (HIPPO) time-of-flight diffractometer at the Lujan Manual Jr. Neutron Scattering Center at Los Alamos National Laboratory. The pole figure data was determined via multiple whole-pattern Rietveld refinement using the MAUD software39 and imported into MTEX38 for further orientation distribution function (ODF) analysis and plotting.

Mechanical Testing

To explore the anisotropic behaviors of the two alloys, compression tests were performed on cylinders (or cuboids) with the compression axis oriented parallel to the RD, transverse direction (TD), and ND for the WE43B-T5 plate material and parallel to the extrusion direction (ED) and extrusion radial direction (ERD) for the AM30-F bar. The compression specimens of both alloys were right cylinders 10 mm in diameter with an aspect ratio of 1. The cuboidal specimen extracted from the WE43B-T5 plate had a square cross section 5 × 5 mm and a height to width ratio of 0.6 (Fig. 3). For low-strain-rate (0.001 1/s) compression testing, an Instron 5869 load frame (Instron Corporation, Norwood, MA) was used with a 50-kN load cell for gathering load data, and a 25-mm full-scale contact extensometer was for gathering strain data.
Fig. 3

Image of high-strain-rate tensile and cuboidal (5 × 5 × 3 mm) compression sample before testing. Larger compression samples were also employed (10 mm diameter by 10 mm height)

For high-strain-rate (~1000–5000 1/s) compression testing, the Kolsky Bar [also known as the Split-Hopkinson pressure bar (SHPB)] was used. The SHPB is a high-strain-rate testing apparatus comprised of a series of bars, namely the striker bar, incident bar, and transmitted bar. For compression, a gas gun is used to fire the striker bar at the incident bar, and a single stress wave is delivered to the specimen. Strain gauges on the incident and transmitted bars as well as a high-speed and high-resolution oscilloscope are used to resolve the reflected and transmitted stress waves. The stress–strain behavior of the specimen at a calculated strain rate can be determined from these waves. An illustration of the compression SHPB apparatus is shown in Fig. 4.
Fig. 4

Schematic of SHPB configuration showing bar arrangement, and strain gauge and specimen locations with an illustration of the bar velocities u1 and u2

For high-strain-rate tension, a tensile Kolsky bar apparatus was employed (Fig. 5). Small, threaded end tensile samples with gauge section diameter of 3 mm and length of 4 mm were extracted from the RD, TD, and ND of WE43B-T5 plate (Fig. 3). The stress wave in the tension bar is generated by clamping the bar at a point upstream from the sample and applying a tensile stress on the bar with a hydraulic actuator. When the clamp is released through the fracture of a holding pin, the stress wave is propagated to the sample. The advanced tensile Kolsky bar apparatus employs a novel laser occlusive radius detector (LORD) for a direct measurement of the strain during the test. This helps to eliminate strain effects from the threaded tensile sample ends seen at the strain gauges on the bars, ultimately permitting a more accurate assessment of the stress–strain response.
Fig. 5

Schematic illustration of the tension Kolsky bar employed in this study, showing the strain gauges (as in the conventional SHPB in Fig. 1) as well as the LORD

The theory of a SHPB is based on the elastic wave propagation in long cylindrical bars and on the principle of superposition of waves.40 The stress wave traveling through the bar is assumed to propagate longitudinally with elastic dispersion, due to our use of maraging steel as the bar material. The initial wave, called the incident wave, propagates down the incident bar and is recorded by a strain gauge attached to the incident bar, known as gauge 1. Because the specimen has different mechanical impedance than the bars, a portion of the incident wave is transmitted through the specimen, while the remainder of the wave is reflected off of the specimen-bar interface. The transmitted portion, known as the transmitted wave, is captured by a strain gauge located on the transmitted bar. The reflected wave is captured by gauge 1.

The velocities of the bars near the specimen-bar interfaces \( \dot{u} \) at some time t are shown to be:
$$ \dot{u}_{1} (t) = - c_{1} *(\varepsilon_{i} (t) - \varepsilon_{r} (t)) $$
$$ \dot{u}_{2} (t) = - c_{2} *\varepsilon_{t} (t) $$
where c is the longitudinal wave speed, ɛ is the strain gauge record, and the subscripts i, r, and t are the incident, reflected, and transmitted waves, respectively. The subscripts 1 and 2 represent the incident and transmitted bars and are illustrated in Fig. 1. By knowing the velocities of the bars at the specimen-bar interfaces, the strain rate εs and strain εs of the specimen can be found as:
$$ \dot{\varepsilon }_{\text{s}} (t) = \frac{{\dot{u}_{1} (t) - \dot{u}_{2} (t)}}{{L_{\text{s}} }} $$
$$ \varepsilon_{\text{s}} \left( t \right) = \mathop \int \nolimits \dot{\varepsilon }_{\text{s}} (t)dt $$
where L is the instantaneous length and the subscript s refers to the specimen. The forces F in the bars can then be found:
$$ F_{1} \left( t \right) = A_{1} E_{1} *(\varepsilon_{i} \left( t \right) + \varepsilon_{r} (t)) $$
$$ F_{2} \left( t \right) = A_{2} E_{2} \varepsilon_{t} \left( t \right) $$
where A and E are the cross-sectional area and elastic modulus, respectively. Knowledge of the forces on both sides of the specimen is important as this is the most widely used method of examining uniform loading in the specimen, which is critical for a valid compression test. If the forces at each specimen side agree well, then the specimen stress σ can then be found as:
$$ \sigma_{\text{s}} \left( t \right) = \frac{{F_{2} \left( t \right)}}{{A_{\text{s}} }} $$
using only the transmitted bar force due to the reduced ringing in the bar signal. These calculations allow for the determination of the effect of the true strain rate on the true stress–strain behavior of materials.


Crystallographic Texture

The WE43B-T5 plate material was shown to have a texture of moderate strength (Fig. 6). It is not the typical texture of a conventionally hot-rolled and annealed commercial magnesium alloy, such as AZ31B,22 which essentially has a 〈00.1〉 fiber texture aligned with the sheet normal direction. Aside from the surface, which does have a fairly intense basal texture, ~6 multiples of a random distribution (MRD), the two peaks (~4–5 MRD) in the basal pole intensity are tilted ~25° toward the rolling direction (Fig. 5). This is more typical of warm-rolled alloy AZ31B or other hot-rolled materials that contain rare-earth elements.41 Robson42 has recently made the argument that solute drag stalls dynamic recrystallization in the rare-earth containing alloys, which helps to explain why the texture does not look like that in hot-rolled conventional plate alloy AZ31. Alloy AM30-F was shown to have a strong (11 MRD) fiber texture, with 〈10.0〉 parallel to the extrusion direction, which is typical of Mg alloy extrusions (Fig. 7).
Fig. 6

Recalculated complete {10.0}, {00.2}, and {10.1} pole figures from hot-rolled and artificially aged WE43-T5 plate at the surface and through the thickness of the plate obtained by x-ray diffraction. Note the splitting of basal {00.2} pole intensity toward the RD
Fig. 7

{00.1} and {10.0} pole figures from the as-extruded AM30 obtained via time-of-flight neutron diffraction and multiple whole-pattern Rietveld refinement. Note the strong, essentially axis-symmetric 〈10.0〉 fiber texture

Mechanical Behavior

The low-strain-rate experiments allowed control of the strain rate via closed-loop operation using the extensometer placed near the specimen. For the high-strain-rate experiments, however, the SHPB does not allow such a function. For SHPB testing, a pulse shaper was used to control the initial stress wave shape, which has some determination on strain rate. As an example, several compression experiments were conducted with pulse shaper dimensions and it was found that AL-1100 cylindrical pulse shapers of 10 mm in diameter and 2 mm in length gave reasonably constant strain rates, close to 1000/s. Figure 8 shows a typical engineering stress–strain curve, as well as the associated strain rate, for the SHPB testing. In addition, the analysis described in Eqs. 39 can yield raw stress–strain data with considerable noise, as shown in Fig. 9. The tensile data presented in subsequent figures were smoothed using a 175-point moving average as shown in Fig. 9.
Fig. 8

Engineering stress–strain curves and strain rates, as function of the engineering strain under ND and RD compression of WE43-T5 using the split-Hopkinson pressure bar. Note that the strain rate is nearly constant throughout the test
Fig. 9

Representative high-strain-rate stress–strain data obtained from alloy WE43-T5 tested in tension along the rolling direction. Both the raw data and the averaged are shown

High-strain-rate and quasi-static compression test data for WE43B-T5 tested along the plate RD and ND reveal a number of interesting features (Fig. 10). First, the quasi-static yield strengths of WE43B-T5 are quite similar (~300 MPa), regardless of straining direction. For the data shown, 0.001 and ~1000 1/s, the strain-rate sensitivity of the yield strength is difficult to detect. In short, the yield strength is nearly isotropic and rate insensitive. On the other hand, the strain-hardening behavior is both anisotropic and rate sensitive. Notably, the stress–strain curve has a parabolic shape during compression along the plate ND (for both low and high strain rates), whereas the curve has a subtle sigmoidal (though nearly linear) hardening behavior for the RD test. This sigmoidal strain-hardening behavior is typical of in-plane compression of textured Mg alloy plates, and it is indicative of {10.2} extension-twinning activity.
Fig. 10

High-strain-rate (HSR ~ 1000/s) and quasi-static (QS = 0.001/s) compression test results for alloy WE43-T5 tested parallel to the RD and ND

While the compressive yield behavior is essentially isotropic, the material still exhibits a tension compression yield strength asymmetry in all three orthogonal directions examined (Fig. 11). For the in-plane directions (RD and TD), the tensile yield strength is higher than the compressive. For the through-thickness direction (ND), the opposite is true. This type of asymmetry has been observed for Mg alloy plate materials before,4346 but it is not a common observation because it requires tensile testing parallel to the plate normal direction. As observed in those prior studies, the tension–compression yield strength asymmetry is greatest for the ND. One reassuring result is the agreement between the RD compression curves obtained using small cuboidal samples and the larger cylindrical samples. Figure 12 presents representative high-strain-rate (~1000–5000 1/s) data obtained from the RD and TD showing a mild rate sensitivity over the tested rate range.
Fig. 11

High-strain-rate tension (~600–900 1/s) and compression (~1000 1/s) data along each of the orthogonal directions, (a) RD, (b) TD, and (c) ND for alloy WE43-T5 revealing a relatively low level of anisotropy and tension compression asymmetry
Fig. 12

Representative high-strain-rate (~1200, 3200, and 5100 1/s) compression test results for alloy WE43-T5 tested parallel to the (a) RD and (b) TD

Figure 13 presents the 1% offset yield strength and the flow stresses at a strain of 0.05 and 0.10, for the RD, TD, and ND. The dynamic tensile test data are included in this plot for reference even though the rate sensitivity of the tensile behavior has not yet been assessed. The plots emphasize that the yield stress (1% offset) is essentially rate insensitive up to ~1000 1/s. Above that rate, the yield stress does seem to increase, yielding an overall rate sensitivity of m = 0.009, if one fits the power law expression (Eq. 1) to all of the RD compression data collectively. A similar analysis for all three strain levels is presented for the RD and ND (Table I).
Fig. 13

Strain-rate sensitivity analysis for alloy WE43-T5 for (a) the 1% offset yield stress, (b) the flow stress at a strain of 0.05, and (c) the flow stress at a strain of 0.10 for the RD, TD, and ND. The dynamic tensile test data (open symbols) are included for reference

Table I

Strain-rate sensitivity (m) of the compressive flow stress of alloy WE43B-T5 as a function of strain for the RD and NDa

Strain Level




0.009 (0.000)



0.008 (0.008)



0.016 (0.010)


aCalculated based upon all data, up to 5000 1/s, or only the cylinder samples tested at 1 × 10−3 1/s and 1000 1/s and presented within brackets (#.###).

The range of rate sensitivities observed in this study is identical to that which can be derived from data published previously by Mukai et al.6

The compressive behavior of the extruded alloy, AM30-F, exhibits strong anisotropy (Fig. 14). The ED compression data show the characteristic sigmoidal hardening curve indicative of significant strain accommodation by the {10.2} extension-twinning mechanism. The ERD compression data show an initial linear hardening response that can be indicative of a combination of deformation mechanisms, including extension twinning.36 The rate sensitivity of the radial direction (ERD) compression strength is approximately equal at strains of 0.01 and 0.05, m = 0.013, suggesting that similar mechanism(s) are controlling the flow stress at these two strain levels. The apparent rate sensitivity of the 1% offset compressive yield strength along the ED is slightly negative (Table II), although more tests will need to be done to confirm this result.
Fig. 14

High-strain-rate (HSR ~ 1000/s) and quasi-static (QS = 0.001/s) compression test results for extruded alloy AM30 tested parallel to the ED and ERD

Table II

Strain-rate sensitivity (m) of the compressive flow stress of alloy AM30-F as a function of strain for the ED and ERD

Strain Level











At this point, it can be definitively stated that the rate sensitivity of the yield stress is low. The apparent rate sensitivity at a plastic strain of 0.05 is quite high, m = 0.042, but this is due to the fact that the twin-related rapid hardening takes place at an earlier strain level in the high-strain-rate test, relative to the quasi-static. Probably a better assessment is to compare the saturation flow stress after this rapid rise, which is essentially the fracture stress in this case. This yields a value very similar to that obtained for the ERD.

Most of the specimens were tested until failure, and detailed SEM fractography is underway. However, even preliminary optical and SEM images of the fractured specimens are revealing. Figures 15 and 16 show typical fractured samples after tension and compression testing, respectively. The tension samples fractured roughly perpendicular to the applied stress, whereas the compression samples underwent shear induced fracture with the fracture plane being highly inclined to the loading axis, indicative of shear failure. Optical fractography reveals that the RD and TD tensile samples have a similar rough fracture surface, whereas the fracture surface of the ND tensile surface was distinct (Fig. 15b–d). It has been widely observed that Mg alloys have little resistance to shear failure, especially at high strain rates, where adiabatic softening can play a role in localization. The SEM fractography presented in Fig. 16b shows that this shear localization is still accompanied by some evidence of ductile fracture behavior in the form of elongated dimples.
Fig. 15

(a) Typical failed tensile sample of alloy WE43-T5 showing macroscopic and optical images of the tensile fracture surfaces when tested parallel to the (b) RD, (c) TD, and (d) ND. The low-magnification optical images show the full section diameter
Fig. 16

Typical failed compression samples of alloy WE43-T5 showing (a) macroscopically flat, highly inclined (indicative of shear failure) fracture surface and (b) elongated dimples also indicative of ductile shear failure


The results obtained in this study serve to reiterate the rather ubiquitous nature of strength anisotropy and asymmetry in textured wrought Mg alloys, even at dynamic strain rates. In fact, the asymmetry can become even more dramatic at high strain rates, as observed by Tucker et al.12 and others. This is because different plastic deformation mechanisms, which control yield along different directions, can have different strain-rate sensitivities. For example, the {10.2} extension-twinning mechanism has been shown to be rather insensitive to deformation temperature14,47 and strain rate (perhaps even causing the material to exhibit a negative strain-rate sensitivity24). The current results lend support to the very low (perhaps even negative) strain-rate sensitivity of the yield stress, when twinning is controlling (see WE43B-T5 RD & AM30-F ED data in Tables I and II, respectively).

Along other directions and senses of straining, where slip mechanisms are known to dominate the yielding response at ambient conditions, the results vary. For ND compression of WE43B-T5, the rate sensitivity of yielding is also low, indicative of basal slip control as shown for quasi-static compression of AZ31B-O plate along ND.22 For other cases, such as ERD compression of AM30-F, the rate sensitivity is higher. A previous study by Oppedal et al.36 indicated that the strain was mainly accommodated by basal slip and {10.2} extension twinning in this case as well. Perhaps the interplay between the mechanisms is what yields the higher apparent rate sensitivity. Such a notion would be consistent with the observation that the strain-rate sensitivity of the uniaxial flow stress along all directions converges to a single value at larger strain levels. This was mentioned in the background section, with reference to the results of Ulacia et al.14 In fact, the rate sensitivity that they observed for the flow strength of alloy AZ31 at ambient temperature, m = 0.016, is quite similar to the value of m = 0.013 obtained for extruded alloy AM30-F in the current study. This suggests that the rate sensitivity of the flow stress at larger strain levels may be largely insensitive to the details of texture and deformation mechanisms for a given alloy. Furthermore, it suggests that the slight chemical difference between alloys AZ31 and AM30 may be insufficient to engender a significant change in rate sensitivity at ambient temperatures.

Although slightly lower values were obtained, mavg = 0.009, the similar results obtained by Yokoyama8 in the study of the dynamic tensile behavior of extruded AZ31B-F, AZ61A-F, and ZK60A-T5 suggest that the rate sensitivity of Mg alloys may even be relatively insensitive to alloy composition and heat treatment. As mentioned previously, the study of extruded WE43 by Mukai et al.6 permitted an estimate of m = 0.007, based on the ultimate tensile stresses reported. This is similar to the average value obtained for RD and ND compression of WE43B-T5, at strain levels of 0.05 to 0.10, m = 0.008, despite the very different thermomechanical processing history, microstructure, and texture. These results permit concluding that the good energy absorption capacity of WE43, which can be observed after specific thermomechanical treatment,6 is not because the material has a strain-rate sensitivity that is distinct from other Mg alloys. On the contrary, it seems very similar to other alloys.

The anisotropy and asymmetry exhibited by WE43B-T5 is lower than that frequently observed in Mg alloys, such as the presently observed conventional AM-series alloy, extruded AM30-F. While the texture strength is lower in the former than the latter, the WE43B-T5 plate material does not have a very weak texture, as was observed in extruded WE54.18 Rather, the reduced anisotropy seems to related to the relatively fine grain size (9 μm) and age hardening, which can strengthen some deformation mechanisms (e.g., basal slip) more than others.4850 Additionally, rare-earth element additions, especially Y, have been suggested to promote nonbasal slip of 〈c + a〉 dislocations,51 which would lead to more isotropic behavior, particularly when comparing the flow stresses at larger strains.

From a practical perspective, the specific energy absorption capacity of WE43B-T5 is high because it has a relatively high yield strength (~300 MPa), moderate ductility (fracture strains = 0.07–0.20) (in tension and compression) along all three orthogonal directions, and the low density of a Mg alloy. This characteristic of the alloy WE43 was highlighted in the early study of Mukai et al.6 The current results show that this observation is also relevant to commercially produced plate material, in addition to the ultrafine-grained extrusion examined in that early study (Table III).
Table III

Energy absorption, per Eq. 1, observed for the samples tested along various directions



Absorption Energy Density (MJ/m3)

Absorption Energy Per Mass (J/kg)

WE43B-T5 cuboidal samples











76.9 ± 4.6

41.8 ± 2.5

WE43B-T5 cylindrical samples










WE43B-T5 threaded samples











38 ± 16

20 ± 9

In-plane Average-T

47 ± 9

25.5 ± 4.8











Because there was no obvious trend with strain rate, over the range 1000–5000 1/s, all of the compression data from the cuboidal samples were analyzed collectively to improve the measurement statistics. At least four samples were averaged for the cuboidal sample compression data, and two samples were averaged for the tensile. The absorption energy in compression is apparently higher for the cuboidal samples than the cylindrical. It is not presently clear whether this is a direct effect of the sample geometry or the fact that the cuboidal samples were tested at a higher rate, which gave rise to slightly higher flow strengths (Fig. 13). The specific energy absorption of WE43B-T5 tested in tension is comparable with the high-strength Al alloys listed by Mukai et al.,6 except when pulled along the ND, in which case it is lower, on the level of pure Mg.6 In fact, the energy absorbed is better than that observed for aged WE43 in that prior study. It is likely that the current hot-rolled plate has a more uniform microstructure than the extruded and aged material examined previously. The AM30F material exhibits similarly unimpressive energy absorption along both tested directions, ED and ERD.

The worst-case scenario for WE43B-T5 is ND tension, where the yield strength is only ~250 MPa, lower than all other test directions, and the fracture strain is only 0.07. Unfortunately, this worst case is relevant to armor applications because spallation failures (see Ref. 35) are driven by tensile stress waves propagating through the plate thickness. Future work will focus on crystal plasticity modeling of the observed plastic response as well as further experimental characterization of the dynamic fracture behavior. Strategies to improve the poor dynamic tension response along the ND will be sought.


  1. 1.

    WE43B-T5 plate is shown to exhibit a relatively isotropic yield strength at quasi-static (0.001 1/s) and dynamic (~1000–5000 1/s) strain rates. This is despite the fact that the plate is revealed to have a crystallographic texture of moderate strength (six multiples of a random distribution).

  2. 2.

    On the other hand, WE43B-T5 plate does exhibit some tension–compression strength asymmetry and anisotropy in strain-hardening behavior, which is indicative of more {10.2} extension twinning during straining along some directions (RD and TD compression, and ND tension) than others (RD and TD tension, and ND compression). This behavior is consistent with the observed texture.

  3. 3.

    In comparison, extruded AM30-F exhibits stronger anisotropy in strain-hardening behavior, which is related to the strong 〈10.0〉 fiber texture and the significant level of {10.2} twinning that occurs during compression along the extrusion direction.

  4. 4.

    The strain-rate sensitivity of the yield strength is observed to vary with testing direction. This is related to the fact that different deformation mechanisms control yielding along the different directions, and these different mechanisms have different rate-controlling processes. For example, {10.2} extension twinning is known to be relatively rate insensitive, so the yield strength is observed to be rate insensitive when that mechanism is dominant.

  5. 5.

    At larger strain levels, the rate sensitivities along different directions converge. For AM30-F, the average strain-rate sensitivity at a plastic strain of 0.05 is m = 0.013 and that of WE43B-T5 is m = 0.008. These values compare well with those observed for a variety of conventional Mg alloys, processed in a variety of ways, suggesting that there is not a strong alloy, microstructure, or texture dependence.

  6. 6.

    The ductility of the two alloys is similar, although the average value (for different directions) is higher for the new plate alloy WE43B-T5. This cannot be associated with a higher resistance to strain localization due to higher rate sensitivity. Instead, it must be due to details of the failure mechanism. The details are currently under investigation.

  7. 7.

    As highlighted in an early study of the material,6 the energy absorption capacity of WE43B-T5 is generally good, due to the good combination of strength, ductility, and low density as compared with competing materials. The current results show that this good behavior can be observed in material produced on a commercial scale.



S.R.A. would like to thank McMaster University for sponsoring a visiting faculty appointment during which this article was written. The research at U.V.A. and M.S.U. was sponsored by the United States Army Research Office under contract number W911NF-12-1-0455 monitored by Dr. Suveen Mathaudhu. The research at MENA and JHU was sponsored by the Army Research Laboratory and was accomplished under Cooperative Agreement No. W911NF-07-2-0073 with technical monitor, Kyu Cho. The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the Army Research Laboratory or the U.S. Government. The U.S. government is authorized to reproduce and distribute reprints for government purposes notwithstanding any copyright notation hereon.

Copyright information

© The Minerals, Metals & Materials Society 2014