Characterisation of stress corrosion cracking (SCC) of Mg–Al alloys

Abstract

Stress corrosion cracking (SCC) of the Mg–Al alloys AZ91, AZ31 and AM30 in distilled water has been characterised using constant extension rate tests (CERTs) and linearly increasing stress tests (LISTs). AZ91 consists of an α-matrix with extensive β-particles, whereas AZ31 and AM30 consist only of an α-matrix with an Al-concentration similar to that in AZ91. The presence of β-particles in AZ91 was associated with: (i) a lower threshold stress, σ_SCC, for AZ91 (55–75 MPa) relative to AZ31 (105–170 MPa) and AM30 (130–140 MPa); and (ii) a different SCC initiation mechanism for AZ91 relative to AZ31 and AM30. The SCC velocity, V_c, for AM30 (3.6 × 10⁻¹⁰ to 9.3 × 10⁻¹⁰ m/s) was lower than that for AZ91 (1.6 × 10⁻⁹ to 1.2 × 10⁻⁸ m/s) and AZ31 (1.2 × 10⁻⁹ to 6.7 × 10⁻⁹ m/s). This was attributed to the influence of Zn and second phase particles, which are more concentrated in AZ31, on the diffusivity of H in the α-matrix.

Introduction

Our recent publications [1], [2], [3], [4], [5] have shown that there exists a considerable body of research outlining the phenomenology of transgranular stress corrosion cracking (TGSCC) of Mg alloys. It is generally accepted that the mechanism for TGSCC of Mg alloys is a form of hydrogen embrittlement (HE) with the hydrogen coming from the cathodic partial reaction (hydrogen generation) of the Mg corrosion reaction [6], [7], [8], [9]; however, the specific nature of the HE mechanism remains uncertain. The HE models that may be applicable for Mg alloys are: hydrogen enhanced decohesion (HEDE); hydrogen enhanced localised plasticity (HELP); adsorption-induced dislocation emission (AIDE); and delayed hydride cracking (DHC). AIDE [10] and DHC [11], [12], [13], [14], [15] have been proposed for TGSCC of Mg alloys; however, the evidence for both mechanisms is limited. HEDE and HELP also remain possible mechanisms. Detailed reviews of these mechanisms are provided in Birnbaum [16], Lynch [17] and Gangloff [18]. A brief review of factors relevant to this investigation is given below.

HEDE has been proposed as the dominant mechanism for high-strength alloys that do not form hydrides [18]. It involves reduction of the electron charge density between metal atoms in the region ahead of the crack tip, where H accumulates by stress-assisted diffusion. This causes weakening of the bonds between and eventually tensile separation of adjacent metal atoms.

DHC involves repeated stages of: (i) stress-assisted diffusion of H to the region ahead of the crack tip; (ii) hydride precipitation as the local H concentration exceeds the local solvus; and (iii) brittle fracture through the hydride. DHC has been proposed as the mechanism for TGSCC of Mg alloys, although evidence for this is limited [11], [12], [13], [14]. Our numerical model for DHC in Mg predicted crack propagation velocities in the lower range of those reported by previous workers; however, the results were based on speculative values for the H diffusion coefficient and solvus concentration [15].

HELP is attributed to enhanced dislocation mobility due to the interactions between H atmospheres at dislocations and obstacles to dislocation motion, resulting in microvoid coalescence more localised than that which occurs in inert environments. This necessitates a sufficiently high H diffusivity such that H atmospheres may move with their respective dislocations and reconfigure as they interact with stress fields and other H atmospheres [16], [17]. Kuramoto et al. [19] provided some evidence for HELP in the Mg alloy AZ31 by observing H evolution along slip lines during SCC. However, it should be noted that H transport by mobile dislocations might play a secondary role to other HE mechanisms. For example, dislocations emitted at the crack tip by adsorbed H atoms (as per AIDE) may also accumulate H atmospheres [17], which may explain the observations by Kuramoto et al.

Lynch [17], [20] proposed that H adsorbed at the crack tip and trapped within the first few atomic layers of metal may weaken metal–metal bonds causing dislocation emission. This would promote crack growth by alternating slip on specific planes, resulting in enhanced localised microvoid coalescence. Lynch and Trevena [10] proposed that TGSCC of pure Mg in NaCl + K₂CrO₄ solution occurred by AIDE based on the similarity of the fracture surface morphology with that produced by liquid metal embrittlement, which involves adsorption of metal atoms at the crack tip.

The influence of microstructural features on SCC characteristics is indicative of the predominant HE mechanism. Microstructural influences may be rationalised in terms of the rate of H transport within the matrix, which is partly dependent on the strength and distribution of traps and short-circuit diffusion paths. Short-circuit diffusion paths (e.g. grain and phase boundaries) may accelerate diffusion by two to four times the rate in the bulk matrix [21], [22]. The influence of traps (e.g. solute atoms, second phase particles, grain and phase boundaries, dislocations and twin boundaries) on H transport depends on their binding energy and whether they are reversible (acting as sinks or sources, depending on conditions) or irreversible (always acting as sinks). There is no data currently available for the binding energies of traps in Mg alloys.

Pressouyre [23] proposed that a uniform distribution of irreversible traps would increase resistance to HE by reducing the quantity of H arriving at crack initiation sites, provided that the primary mode of H transport is dislocation sweep or interstitial diffusion (not diffusion along short-circuit paths). Traps may themselves become crack initiation sites if their H concentration exceeds some critical level. The susceptibility of metals to HE is particularly dependent on the role of reversible traps (e.g. low angle grain boundaries and dislocations). The role of reversible traps depends on: (i) the source of H (internal or external); and (ii) the mode of H transport (dislocation sweep or interstitial diffusion) [23]. In the case of H transported with dislocations, reversible traps may act as sinks or as sources of H for mobile dislocations such that H may penetrate further into the specimen than in the absence of traps [23].

A comprehensive review of the phenomenology of SCC in Mg alloys is given in Winzer et al. [1]. Some important considerations, in the context of the present investigation, are given below.

It is difficult to isolate the specific role that alloying elements play in promoting or retarding SCC, since they may have competing influences on H activity, film integrity and dislocation mobility. Moreover, these influences may be specific to combinations of alloy, environment and mechanical loading. Mg alloys used in service typically contain Al and Zn as primary alloying elements. Mg alloys containing Al and Zn are particularly susceptible to SCC [24], [25], [26], [27], [28]. The SCC susceptibility of Mg alloys increases with increasing Al concentration [29] despite an increase in the repassivation rate [14]. Mg alloys with >2% Al concentration typically contain Mg₁₇Al₁₂ (β-phase) particles, which are highly cathodic relative to the α-phase. Consequently, the relationship between Al concentration and SCC susceptibility has previously been associated with the propensity for microgalvanic corrosion of the α-phase adjacent to Mg₁₇Al₁₂ precipitates [6], [7], [14], [27], [28], [30].

Many factors related to mechanical processing influence the SCC susceptibility of Mg alloys [1]. Extruded Mg–Al–Zn alloys may be more susceptible to SCC than cast alloys with similar overall composition due to the influences of microstructural defects and residual stresses [31]. For specimens loaded in tension, residual tensile stresses are detrimental to SCC resistance whereas compressive residual stresses are beneficial [32], [33]. The influence of grain size has been studied only with respect to the propensity for intergranular or transgranular SCC [11], [28], [34], [35]. Grain orientation is likely to have strong influence on TGSCC fracture morphology if the propagation mechanism is crystallographic [11], [36], [37], [12].

The influence of strain rate varies between stages of SCC (i.e. initiation and propagation). The influence of strain rate on SCC initiation is associated with the propensity for mechanical film rupture at the surface, which causes localised corrosion and H ingress. The influence of strain rate on SCC propagation is associated with: (i) the propensity for repassivation at the crack tip; (ii) the transition to the inert fracture mode (at high strain rates); and (iii) the time required for embrittlement and fracture of the region ahead of the crack tip [14], [16], [17], [18], [38], [39], [40], [41]. It should also be noted that the propensity for surface film breakdown and repassivation at the crack tip is also dependent on the interaction between the alloy and environment.

Our review [1] identified a need for a mechanistic understanding of the influences of environment, microstructure and mechanical loading on SCC of Mg–Al alloys to support the growing use of Mg alloys for stressed components in service such as for automotive applications. The principal issues in developing this understanding are: (i) the environmental and mechanical conditions causing film breakdown and H ingress; and (ii) the microstructural and mechanical conditions that promote accumulation of H in such concentrations as to cause HE. These issues may be resolved by comparing alloys under SCC conditions with respect to: (i) fracture surface morphology; (ii) crack propagation velocity; (iii) the sensitivity to microstructural modification; and (iv) the sensitivity to various environments. Thus, the present investigation focuses on evaluating, with respect to the possible HE mechanisms, the influence of microstructure on the SCC susceptibility of Mg–Al alloys by comparing the SCC characteristics of AZ91 (consisting of an α-matrix with extensive β-particles), AZ31 (consisting of an α-matrix with Al-concentration similar to that in AZ91) and AM30 (consisting of an α-matrix with similar concentration to AZ31, but with lower Zn-concentration).