Manufacturing Strategies to Mitigate Deformation Twinning in Magnesium

Magnesium (Mg) and its alloys have the potential to be promising engineering platform materials due to their high strength-to-weight and stiffness-to-weight ratios and their affordability compared to other lightweight materials such as carbon fiber. However, due to a low ductility at room temperature and predominantly brittle failure resulting from the hexagonal closed packed (HCP) crystalline structure, Mg alloys are difficult to incorporate into a commercial setting [1–3]. Thus, understanding the deformation mechanisms that drive microstructural evolution and influence ductility is critical to enable optimization of the Mg alloy composition and microstructure. Moreover, determining whether brittle failure in HCP lattice structures can be mitigated is crucial for producing a stronger, more ductile form of magnesium.

The low ductility of Mg alloys ultimately results from a lack of easy slip systems, which is innate to the HCP lattice. Like many HCP metals, Mg has an easy slip mode along the basal planes; however, the stress required to slip along the prismatic and pyramidal planes is orders of magnitude higher. The five independent slip systems required by Taylor criterion for arbitrary ductile shape change are not generally all operable at temperatures below 300 °C. At lower temperatures, while the basal slip is easy, extension and contraction of the ⟨c⟩ axis are difficult [4]. Due to this inability to initiate slip at room temperature, as well as the HCP lattices’ low symmetry, deformation twinning is promoted rather than ⟨c + a⟩ slip [5,6].

Deformation twinning is frequently observed in Mg and is a mechanism that accommodates plastic strain along the ⟨c⟩ axis, as well as ⟨c + a⟩ dislocation gliding [7,8]. It has long been known that the ${101¯2}⟨101¯1⟩$ tension twin, as well as the ${101¯1}⟨101¯2⟩$ compression twin, plays a very important role in deformation mechanisms in Mg as well as other HCP materials [9,10]. Twinning accommodates plastic strain; the reorientation of grains due to twinning has been suggested to promote dislocation gliding, and thus plastic straining [11–13]. Twin advancement is controlled by two processes: nucleation and growth. The nucleation of twins usually originates at the grain boundary and continues to propagate across the grain. Twin growth occurs by way of fast extension of a sharp twin tip and is then followed by a coordinated slower migration of the boundaries [14]. Many factors can influence twinning; among these factors is grain boundary misorientation, which is hypothesized to have a substantial influence on the twinning stress magnitude as well as the twin propagation stress [15–17]. Twins which collide with low-angle grain boundaries generally induce the same twin system to activate in the opposite grain because of the similarity of the twinning shears and the resolved shear stresses in a phenomenon known as accommodation. This results in twin bands that can quickly span the entire material [4,18,19].

Twin nucleation sites in Mg alloy are preferred at low misorientation angle grain boundaries, which tend to be the highest energy boundaries [20]. Free surfaces are also prevalent nucleation sites for twinning. The material conditions required to nucleate twins have been studied extensively and are still unclear. However, it is well known that the nucleation site must have a pre-existing defect structure which helps stabilize the twin embryo when it is small. Additionally, the nucleation site must accommodate the twinning shear without inducing too much strain energy due to the eigenstrain of the twin. This condition may be satisfied by a free surface to accommodate the shear or easy basal slip or twinning across a grain boundary.

There have been many proposed methods attempting to inhibit twinning in HCP materials, including grain refinement to promote randomized textures and alloying designed to promote easier ⟨c + a⟩ slip [21,22] and texture randomization [23]. Additional work has focused on attempting to restrict twin boundary migration by promoting segregation to twin boundaries. Unfortunately, the success of these results has been too limited to result in widespread applicability. Therefore, in this work, we propose a new strategy for the suppression of the twinning mechanism.

The conditions required for twin nucleation are, in many ways, similar to those required to nucleate a crack. It is known that crack initiation strongly depends on the surface condition of a specimen’s surface, and therefore crack initiation could be affected by a coating [24,25]. There are currently many applications for Mg coating treatments such as anodizing [26], conversion coating [27], as well as electroplating and electrolysis plating [28,29] and others. Mg alloys have previously been plated with chromium, copper, nickel, and niobium in attempts to inhibit corrosion as well as increase the strength and ductility [30–32]. Results from these previous studies showed that the electroplated Mg alloys produced improved surface hardness resulting in more plastic deformation resistance [30,31]; however the low-cycle fatigue life decreased for a Mg AZ61 alloy after alkaline Cu followed by acidic Cu, Ni/Cu, and Cr–C/Cu electroplating [31]. We hypothesize that this same process could be used to suppress twin nucleation on the surface. Twinning is accommodated at the free surface by kink formation [16,33]. Plating a sharply textured Mg alloy with other metals that can hinder such kink accommodation slows twin nucleation events, thus delaying the failure. A comparable study done in a highly textured sample could suppress the formation of twin bands, thus possibly resulting in higher overall ductility. Our effort in this work is to test this hypothesis by observing the coating materials’ effects on twin nucleation and propagation in a sharply textured Mg (MgAZ31).

Rectangular magnesium samples of size 0.635 mm (t) × 152.4 mm (l) × 25.4 mm (w) were machine polished and then sent to C.I.L Electroplating Inc to be plated by zinc (Zn), copper (Cu), and nickel (Ni). The samples are oriented such that the transverse direction is along the length of the sample, the normal direction is along the height of the sample, and the rolling direction is aligned parallel to the width of the sample as shown in Fig. 1(a). Due to magnesium’s propensity to chemically react, exceptions had to be made for certain types of plating to adhere to the matrix. Samples were dipped into electroplating baths, and some types of plating required a copper “prestrike” for adhesion. The HCP structured zinc plating adhered directly to Mg, but the face centered cubic (FCC) structured nickel and copper-plated samples required a prestrike. Other samples were sent to advanced surface technologies finishing, which underwent a chromium conversion process.

Separate plated samples were also analyzed using a digital imaging correlation (DIC) to determine the LaGrangian strain that each sample underwent locally. Samples were bent to 1% and 2% strains using the Universal Servo Hydraulic MTS Machine. The strain was measured using ASTM standard D6272, which agrees with Eq. (1).

Stress–strain curves were obtained from bending tests for magnesium AZ31 specimens with and without metal platings. Figure 2 shows stress–strain curves from unplated Mg specimens as well as each of the three types of plating (Zn, Cu, and Ni). While the experiments were performed in the same manner to replicate the results multiple times for most cases, for some cases only one test was performed. Figure 2 only shows a representative curve for different coated samples. As observed in Fig. 2, the stress flow curve differentiates between various plating of Mg specimen.

Table 1 shows the material behavior observed for different plating. Both yield and ultimate tensile strength (UTS) increase for Zn plating, whereas the strength reduces for Cu and Ni plating. The yield and UTS obtained for the unplated sample are 220 MPa and 265 MPa. The Zn plating performed significantly better than the unplated sample, demonstrating a yield strength of 240 MPa and UTS of 295 MPa. The Zn plated sample finally demonstrated a much later onset of softening than the unplated curve, with strength very near its UTS at the end of the test, unlike the unplated sample. On the other hand, the Cu plated and Ni samples actually demonstrated a lower yield strength and UTS than the unplated sample. The Cu plated sample was weaker than the Ni sample with a Cu prestrike.

Samples that were bent to 2% strain were examined under the optical microscope at the same magnification and location on the sample, as seen in Fig. 3. Twin bands are clearly visible in all of the samples. The separation distances between these twin bands and the twin band thickness for each specimen are tabulated in Table 2. The separation and thickness distances were measured along the horizontal directions of observed twin bands, 90 µm from the surface, as indicated by black arrows in Fig. 3. Table 2 clearly shows that twinning was less in the zinc plated sample. The twin bands in the Zn plated sample are thinner, with wider spaces between them. Meanwhile, Ni plated samples showed slightly lower decrease in twin thickness but greater than the Cu plated samples. The chromium conversion proved to have the thickest twin bands on average, followed by the copper and nickel.

Figure 4 shows the strain maps obtained from DIC analysis of four-point bending experiments. The DIC data for Ni samples were corrupted during acquisition, thus DIC analysis of strain field for Ni are not included in Fig. 4. These results showed good agreement with the strain fields from twin bands as observed in the optical microscopy results from Fig. 3 and stress–strain curves in Fig. 2. In particular, the unplated sample showed the most strain in the compression zone, including some banding, while the Zn plated sample showed the least strain. While the strain concentrations were evident at the surface of the Zn plated and the Cu plated samples, the strains in the unplated sample were more homogeneous. The overall plastic strain and the apparent curvature are much greater in the unplated sample than the others even though the samples were all subjected to the same deflection. This indicates that the plating process enhanced detwinning once the applied stress was released.

σ₀ appears to be affected by the modulus and crystal lattice of the plating material. While both Mg and Zn have HCP crystal structure, σ₀ increases for the Zn plated sample resulting from the difficulty to nucleate twin at the surface as Zn has a higher modulus than Mg. However, for a similar modulus for Cu and a higher modulus for Ni, σ₀ decreases significantly. On the other hand, the hardening, R_∞, increases considerably for Cu and Ni, but this effect is insufficient to increase their UTS to the levels seen for pure or Zn plated Mg.

We theorize that due to the highly textured nature of the substrate, the local interface between FCC and HCP will frequently take the form ${11¯1}FCC||{0001}HCP$⁠. This aligns other closepacked planes of the FCC plating near the second-order pyramidal plane of the substrate. Dislocations then nucleate from the surface on the {111} planes, move to the interface, and transmute through, combining into ⟨c + a⟩ dislocations in the Mg lattice with some defect content left behind at the interface. Since these dislocations nucleating at the surface have a much smaller Burgers vector than ⟨c + a⟩ dislocations, they nucleate more easily. ⟨c + a⟩ slip has been previously observed to harden extensively, which is compatible with our results showing more hardening with the Cu and Ni plating.

Alternatively, it is possible that enhanced twin nucleation at the phase boundary is responsible for the lower yield in Cu and Ni, followed by detwinning upon the release of stress. This process could be partially accommodated by slip in the plating. Further investigation will be needed to identify whether these processes occur.

Plating with Zn, which does not produce a FCC–HCP interface, appeared to have the anticipated effect. The HCP–HCP interface between Zn and Mg restricts plasticity rather than providing a nucleation site like the FCC–HCP interface. Twinning was suppressed, increasing both the yield and ultimate stresses. Overall, the effect of plating in all cases led to decreased twinning, increased detwinning, and increased hardening.

In summary, zinc plating seemed to discourage twin nucleation and thus delay yield and increase the strength of Mg AZ31. Not only was twin nucleation more infrequent, but twin thickening was also inhibited compared to an unplated sample with the same loading. It is theorized that this occurs because the Zn plating presents a hard surface barrier preventing the kinking of the surface necessary for the twin to nucleate. If this result is confirmed, it could lead to significant improvements in Mg alloy strength. Unlike slip, twinning produces highly localized deformation, which is specifically retarded by the stiff plate of Zn. Experiments with plating by Cu instead of Zn showed an opposite effect with earlier yield, even though twins still appear to be suppressed. The results show that Zn is specifically effective as a coating to strengthen Mg surfaces because Zn has excellent adherence to the Mg substrate and significantly higher stiffness than Mg.

The research described and the resulting data presented herein, unless otherwise noted, were funded under PE 0602784A, Project T53 “Military Engineering Applied Research,” Topic 3 under Contract W56HZV-17-C-0095, managed by the US Army Engineer Research and Development Center. The work described in this document was conducted in the Center for Advanced Vehicular Systems at Mississippi State University. OPSEC permission was granted to publish this information.

There are no conflicts of interest. This article does not include research in which human participants were involved. Informed consent was obtained for all individuals. Documentation provided upon request. This article does not include any research in which animal participants were involved.

The datasets generated and supporting the findings of this article are obtainable from the corresponding author upon reasonable request.