Tension–compression creep asymmetry in strong-textural Mg-RE alloy due to <c + a> dislocation reconfiguration

Abstract

Strong-textural Mg-RE alloy exhibits creep asymmetry at 200°C, with a much higher creep resistance in compression than in tension, albeit a higher yield strength in tension than in compression. Creep asymmetry is ascribed to the non-Schmid effects of pyramidal <c + a> slip, which enables a higher friction in compression and thus induces a lower mobility of pyramidal <c + a> slip in compression than in tension. Additionally, the dissociation of <c + a> dislocations into sessile structures in compression is more frequent than in tension, which contributes to the hardening response and thus lowers the compressive creep rate.

GRAPHICAL ABSTRACT

IMPACT STATEMENT

A new mechanism of the dissociation behavior difference of pyramidal <c + a> slip and its non-Schmid effect accounts for the creep asymmetry in highly textural Mg alloys.

KEYWORDS:

• Creep asymmetry
• dislocation
• texture
• Mg-RE alloy

1. Introduction

Magnesium (Mg) alloys have garnered a lot of interest in the automotive and aerospace industries due to their low density. In contrast to cubic metals, the yield strength of most wrought Mg alloys differs markedly between tension and compression modes. After Mg alloys are thermo-mechanically processed, such as rolling and extrusion, their grains would be endowed with the common crystallographic texture where the basal plane is oriented predominantly parallel to the rolling direction (RD) or the extrusion direction (ED) [1,2]. In such wrought alloys, their basal slip would be significantly inhibited when the load along the ED or RD because of its low Schmid factor [3]. It is known that the {10-12} twinning can be activated by a tensile stress perpendicular to basal plane or a compressive stress parallel to basal plane [4]. Therefore, {10-12} twinning is easily activated under compression but not under tension along the ED or RD, which gives rise to tensile yield strength (TYS) being much higher than compressive yield strength (CYS). Expectedly, wrought Mg products usually exhibit tension–compression yield asymmetry [1–3].

Except for the yield asymmetry, tension–compression asymmetry in terms of creep and fatigue behaviors is also observed in metallic materials, such as Al-, Zr-, and Ni-based alloys [5,6]. For instance, Zhang et al. [5] reported that in the creep of ZL109 Al alloy at high temperatures and stresses, the tensile creep rate is larger than the compressive creep rate because of more cavity nucleation developed during tensile creep than compressive creep. Moreover, Sondh et al. [6] found that the initial creep rate in compression is much larger than in tension and this asymmetric creep response is caused by the presence of an internal compressive stress field in the Ni-base superalloy matrix. Unfortunately, fewer efforts have been devoted to focusing on tension–compression creep asymmetry in Mg alloys. Wang et al. [7] reported that an extruded Mg–10Gd–3Y–0.5Zr alloys exhibits tension–compression creep asymmetry, with the minimum creep rate is higher in tension than in compression, which is caused by different deformation mechanisms, such as grain-boundary sliding dominated in compression while dislocation creep in tension. Li et al. [8] observed tension–compression creep asymmetry in hot-rolled Mg-3Gd alloy, and they confirmed that dislocation cross-slip is operated during tension creep whereas both cross-slip and twinning are activated during compressive creep. In this work, we observed the reversed creep asymmetry in a highly textural Mg-rare earth alloy, with compressive creep resistance being stronger than tensile creep resistance at 200°C, while its CYS is yet smaller than the TYS at this temperature. The aim of this paper is thus to explore the origin of the creep asymmetry in this extruded Mg-rare earth alloy under specific creep conditions.

2. Materials and methods

Mg–5Sm–0.6Zn–0.4Zr (S5, wt.%) alloys were fabricated by casting, solid solution treatment at 520°C for 12 h. Then the alloy was processed by backward extrusion at 300°C with an extrusion speed of ∼0.3 mm/s to produce bars of 30 mm in diameter, which corresponds to an extrusion ratio of ∼6:1. Finally, as-extruded bars were annealed at 200°C for 30 h. Tensile samples have a cylindrical geometry with a gauge section of 5 mm in diameter and 30 mm in length. Compressive samples have a size of 8 mm in diameter and 12 mm in height. The axis direction of both samples is parallel to the ED. Tension and compression tests are performed at 200°C, with an initial strain rate of 10⁻³ s⁻¹. Tensile and compressive creep tests are made at 200°C under various stress levels, and their sample’ gauge size are the same as the mechanical test samples. Creep strain was directly measured by two mechanical extensometers (RD50 with the accuracy of ±0.1 mm) connected to the gauge section of the creep samples. Microstructure examinations were performed by optical microscopy (OM, Olympus GX71), transmission electron microscopy (TEM, Philips CM20) at an accelerating voltage of 200 kV and electron backscattered diffraction (EBSD, FEI Helios NanoLab 600i) system operated at 15 kV. For creep microstructure examinations, tests were interrupted at strains of about 0.3% which corresponds to the minimum creep rate. EBSD analyzations were completed using Aztec Crystal software. TEM samples were polished to a thickness of 30∼50 µm and then further thinned by a low-energy ion beam with a cooling system with liquid nitrogen (milling parameters: Ar, 4 kV, 90 min).

3. Results

True stress–true strain curves of the S5 alloy in tension and compression modes along the ED at 200°C are displayed in Figure 1(a). The TYS of 288 MPa is much higher than the CYS of 237 MPa, thus creating yield asymmetry at 200°C. A concave-up shape occurs in compressive curve but not in tensile curve, revealing twinning activation after compression yielding [9]. Figures 1(b and c) show the microstructure of hot-worked grains in the S5 alloy after tension and compression strained to ∼0.03 (just over yielding point). As seen, the expected {10-12} twin is observed in compressive sample (Figure 1b), while profuse dislocations were developed in tensile sample (Figure 1c), which accounts for yield asymmetry in this alloy [10].

Figure 1. (a) Ture stress–true strain curves under tension and compression modes of the S5 alloy at 200°C. Bright-field TEM of hot-worked grains in (b) compressive sample and (c) tensile sample with a true strain of ∼0.03 (just over yielding point).

Figure 2(a) shows creep strain versus creep time curves under tension and compression modes at 200°C under 130 and 150 MPa. The plot of creep rates versus strain is presented in Figure 2(b), where creep rates are calculated by differentiating the creep strain with respect to the creep time. Apparently, the alloy exhibits the different creep response under tension and compression modes, with tensile creep proceeding much faster than compressive creep. In the primary creep stage, the instantaneous creep strain is almost the same in both tension and compression modes under an equal stress level (see the inset in Figure 2a). After primary creep stage, tensile and compressive creep reached the minimum creep rate ($\dot{\epsilon }$_min) at the stain of ∼0.3%. Unexpectedly, the $\dot{\epsilon }$_min value in compression is about six times smaller than in tension, namely the creep resistance in compression much larger than in tension, although the TYS is far higher than the CYS at 200°C. Therefore, the S5 alloy exerts creep asymmetry in terms of the minimum creep rate in tensile and compressive loads, which is just opposite to the yield asymmetry. The minimum creep rate as a function of applied stress at 200°C was plotted in Figure 2(c). Stress exponent values in tension and compression creep were thus obtained to be 5.5 and 5.9, respectively, implying that dislocation creep is the dominated creep mechanism in this alloy, without respect to loading modes [11].

Figure 2. (a) Creep strain versus creep time curves, (b) creep strain versus creep rate of S5 samples under tension and compression creep at 200°C and (c) minimum creep rate as a function of applied stress at 200°C.

To clarify the creep asymmetry of the S5 alloy, microstructures before and after creep were investigated. As-extruded S5 alloy possessed a bimodal grain structure consisting of low-proportion recrystallized grains and high-proportion hot-worked grains (Figure 3a). The area fraction of hot-worked grains is measured as ∼70%. Bright-field TEM image shows a lot of dynamical precipitates with an average size of ∼80 nm in the alloy (Figure 3b). These precipitates are identified as Mg₃Sm particles by HR-TEM analysis (Figure 3c) [4], which are randomly distributed in recrystallized grains interiors and boundaries, and the relatively uniform distribution in hot-worked grains. Inverse pole figure map (Figure 3d) shows that the mean grain size of recrystallized grains is ∼1 µm. Typical basal fiber texture of [10-10]//ED was developed in the S5 alloy (Figure 3e). Hot-worked grains have a strong fiber texture with [10-10]//ED, whereas recrystallized grains have random orientations. Further Schmid factor (SF) analysis for basal slip system (Figure 3f) shows that SF values for hot-worked grains and recrystallized grains are 0.09 and 0.32, indicating basal slip is in hard orientation in hot-worked grains and soft orientation in recrystallized grains [2].

Figure 3. Microstructure of the S5 alloy before creep: (a) OM image, (b) bright-field TEM image, (c) HR-TEM image of one precipitate, (d) inverse pole figure map, (e) inverse pole figures of various regions and (f) SF distribution histograms of basal slip system.

After the tensile or compressive creep strained to ∼0.3%, there is almost no detectable microstructural variation including recrystallized grain size, texture/orientation and recrystallized ratio (Figure 4a and b), compared to the initial microstructure, indicating a good thermostability of microstructure. In addition, the number fractions of sub-grain boundary with a misorientation angle <15° in initial, tensile and compressive samples are about 20.2%, 27.3% and 25.8%, respectively (Figure 4c). Note that sub-grain boundary was configured by dislocation rearrangement during creep, thus dislocation creep is operative in this alloy [12], in accord with the above result (Figure 2c). The geometrically necessary dislocations (GNDs) density of initial, tensile and compressive samples is calculated to be 1.74 × 10¹⁵, 1.74 × 10¹⁵, 1.87× 10¹⁵ and 1.82 × 10¹⁵ m⁻² by EBSD data, respectively. As mentioned before, {10-12} twinning is responsible for the yield asymmetry in highly textural Mg alloys [3]. However, twins are not found in compressive creep sample (Figure 4b), where the twin boundary misorientation should correspond to a misorientation angle between 80° and 90° in Figure 4(c). This demonstrates that creep asymmetry, differing from yield asymmetry, is caused by others, rather than the widely-accepted twinning. It is known that dynamic precipitation usually occurs during creep [12]. As compared to the initial sample without precipitates (Figure 4d), many precipitates were developed in hot-worked grains after tension/compression creep (Figures 4e and f). These precipitates align with basal planes are less than 1 nm in thickness and 30–60 nm in length, which are previously reported to be γ’’ phase [13]. The fraction of γ’’ phase is nearly the same in both crept samples, thereby their hardening effect is also identical during tension/compression creep. Furthermore, we observed that these precipitates are surrounded by profuse dislocations since creep-induced dislocations can act as sites for heterogeneous nucleation of precipitates [12].

Figure 4. Inverse pole figure maps and inverse pole figure of (a) tensile creep and (b) compressive creep samples. TD and CD is the tensile and compressive creep direction, respectively, both of which are parallel to the ED And (c) boundary misorientation angle histograms. TEM images showing precipitates in hot-worked grains of (d) initial, (e) tensile creep and (f) compressive creep samples.

Further TEM observations were executed for exploring the creep asymmetry. Figure 5 shows bright-/dark-field TEM image of dislocation configurations developed in hot-worked grains of tensile creep and compressive creep sample, and includes the initial sample for comparison. Note that these TEM images were captured under two-beam diffraction of g = [0002] to contrast <c + a> dislocations based on the g·b = 0 invisible criterion [14]. In initial sample (Figure 5a and b), a very limited number of <c + a> dislocations were retained possibly because most residual dislocations in hot-worked grains have been eliminated during long-term annealing [15]. Almost all <c + a> segments marked by red arrows in initial sample are parallel to the basal plane trace. Basal-oriented <c + a> dislocations were reported to be always sessile [16]. After creep, more dislocations were generated in compressive creep and tensile creep samples. Apart from most sessile <c + a> dislocations on the basal plane, a few <c + a> dislocation segments (indicated by blue arrows) deviated from the basal plane trace, which is termed as glissile <c + a> dislocations [17], are also observed in compressive creep sample. When the trace of <c + a> dislocations is not parallel to basal plane, they are deemed as glissile <c+a> dislocations [16,17]. Therefore, dislocation configurations in compressive creep sample are characterized by most of sessile <c + a> dislocations and a few glissile <c + a> dislocations. However, significantly different dislocation configurations were developed in tensile creep sample, in which the majority of <c + a> segments were inclined away from the basal plane and thus they are pyramidal-oriented <c + a> dislocations. As a result, there are more glissile <c + a> dislocations and fewer sessile <c + a> dislocations formed in tensile creep sample. It is known that <c + a> slip occurs predominantly on the pyramidal plane, but <c + a> core is kinetically metastable and during straining always transforms into new basal-oriented dislocations that are thermodynamically stable and immobile [17,18]. Figure 5(f) evidences the pyramidal to basal transformation of <c + a> dislocations, where <c + a> dislocation segments are lying on basal and pyramidal plane highlighted by blue circles. These results show that the transformation of easy-glide <c + a> dislocations is potentially affected by stress states upon creep loading, and this transformation progress seems to be accelerated under compressive creep stress. Molecular dynamics simulations showed that pyramidal <c + a> dislocation slip behavior is highly anisotropic in Mg [19]. Accordingly, based on TEM observation under g = [0002] for five hot-worked grains, the number fraction of sessile <c + a> dislocations in compressive and tensile creep samples is statistically estimated to be about 85% and 17%, respectively, which further confirmed that the transition of easy-glide <c + a> dislocations into basal-oriented sessile structure is higher frequent in compressive creep.

Figure 5. Bright- and dark-field TEM images of <c + a> dislocation configurations in hot-worked grains, with B near to ∼ [11–20]: (a, b) a limited number of sessile <c + a> dislocations in initial sample before creep, (c, d) an increased number of sessile <c + a> dislocations and several glissile <c + a> dislocations in compressive creep sample, (e, f) a limited number of sessile <c + a> dislocations and amounts of glissile <c + a> dislocations in tensile creep sample. The ∼0.3% strain just corresponds to the reaching minimum creep rate for both tension and compression creeps.

4. Discussion

The aforementioned results indicate that tension/compression creep deformation are dominated by pyramidal <c + a> slip; however, the minimum creep rate is unexpectedly higher in tension than that in compression. It was documented that pyramidal <c + a> slip is highly asymmetric, and its activity strongly depends on stress states [19,20]. We therefore proposed that the minimum creep rate asymmetry of S5 alloy could be principally attributed to the asymmetrical nature of the pyramidal <c + a> slip under both loads. It should be noticed that the asymmetric <c + a> slip behavior in hcp metals has been mentioned in early investigations [20–23]. For example, strongly textured Ti alloys present the tension–compression yield asymmetry even in the absence of twinning, in which pyramidal <c + a> slip is functional in both tension and compression [20,21]. Asymmetrical <c + a> dislocation causes a distinct anisotropy of Peierls stress that alters on the sign of the shear stress, which results in the different critical resolved shear stress (CRSS) of pyramidal <c + a> slip in tension and compression modes [19,22]. In general, the effective CRSS of pyramidal <c + a> slip in compression is higher than that in tension, thereby the activity of pyramidal <c + a> slip in compression is largely reduced [24]. Importantly, such response in the present alloy could be more significant due to strong texture. Further, when considering the influence of the hydrostatic stress condition on the CRSS values, the asymmetrical <c + a> slip behavior renders logical. The activation of <c+a> slip is also very sensitive to the hydrostatic stress, because the hydrostatic stress state affects the CRSS for <c+a> slip [25]. In compressive mode, the hydrostatic stress is higher, which is ascribed to increase in the friction stress for <c+a> dislocation glide as a result of hydrostatic pressure as the main mechanism. Thus the CRSS for <c+a> slip in compression is higher and deformation is harder and the converse is true for tension. According to the crystal plasticity finite element simulation results [24,26], the CRSS of pyramidal <c + a> slip can be influenced by not only the shear stress state but also the hydrostatic stress state, and thus the <c + a> slip behavior deviates from the Schmid law. Accordingly, the non-Schmid behavior of pyramidal <c + a> slip would give rise to a larger lattice friction in compression than in tension, thus leading to a lower mobility of <c + a> dislocations in compression than in tension. Evidentially, the weakly-textural Mg–Mn–Nd alloy shows the reversed yield asymmetry, which is mainly attributed to the impact of the hydrostatic stress states on the mobility of pyramidal <c + a> slip [24]. As a result, the asymmetrical minimum creep rate of the S5 alloy in tension and compression modes could be related to the non-Schmid behavior of pyramidal <c + a> slip.

Furthermore, it has been established that the pyramidal <c + a> dislocation core is metastable and that it will dissociate into one of three new basal-oriented dislocation structures having low energy in response to thermal activation and stress [17,18]. Note that such basal-oriented dislocation structure is sessile, which performs as a ‘forest’ to hinder the movements of any additional dislocations that would cause c-axis strain, thus contributing to the high hardening response [16]. Atomic scale simulation studies show that non-Schmid stresses might have an impact on the pyramidal <c + a> slip associated with dynamic dissociation [14]. Thus the slip and dissociation of pyramidal <c + a> system are directional because of the asymmetry of the dislocation core. At the minimum creep rate, in this work, the final configuration of pyramidal <c + a> dislocation is various in compression and tension modes, and most dislocation structures are mobile in tension but immobile in compression (Figure 5). The above results indicate that the dissociation behavior of <c + a> dislocation could depend on the stress states, which is consistent with the non-Schmid behavior of pyramidal <c + a> slip. As evidenced by Hidalgo–Manrique et al. work that the higher CYS than TYS in extruded Mg alloys was attributed to a higher frequency of dissociation of <c + a> dislocations in compression [24]. It is therefore expected that basal-oriented <c + a> sessile dislocations in compression creep can block subsequent easy-glide <c + a> dislocations, so that the hardening rate is dramatically enhanced, which resists creep deformation and thus reduces the creep rate.

5. Conclusion

In summary, the highly textural S5 alloy exhibited tension–compression creep asymmetry at 200°C, with the minimum creep rate in compression being about six times smaller than in tension, which is just opposite to yield stress asymmetry. The origin of the creep asymmetry is primarily attributed to the different <c + a> dislocation behavior in compression and tension creep. The larger level of dissociation of <c + a> dislocations in compression creep into basal-oriented sessile dislocations, which acts as the barriers for other easy-glide dislocations and thus decreases the creep rate. The lower activity of pyramidal <c + a> slip in compression than in tension due to the non-Schmid effects also reduces the creep rate in compression.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work is supported by National Natural Science Foundation of China (Nos. 52201111 and 52275389) and Taiyuan University of Science and Technology Scientific Research Initial Funding (No. 20232102).