Transgranular corrosion fatigue crack growth in age-hardened Al-Zn-Mg (-Cu) alloys

2.1 Materials and specimens

Al-6.27Zn-2.94Mg (wt.%) “single crystals” (i.e. material with an ~10 mm grain size, produced by strain annealing, which were notched such that crack growth occurred within a single grain) were used for most of the work (Figure 1). The specimens were tested in a variety of ageing conditions after solution treating at 450°C for 1 h, quenching in boiling water, and then ageing at 100°C for 1 h and then at 180°C for various times to produce underaged, peak-aged, and overaged conditions. The specimens were notched and then fatigued in cantilever bending at various cycle frequencies, generally using a sinusoidal waveform. A variety of grain orientations were tested, and the environment (and ageing condition) was generally changed during testing, so that their effects could be determined for the same crystal orientation and nominal ΔK value. The estimates of ΔK in single crystals were made by comparing striation spacings with literature da/dN-ΔK data for commercial 7xxx alloys.

Figure 1:

Cantilever bend specimen (40×10×3 mm) used for fatigue crack-growth studies in Al-Zn-Mg “single crystals.”

A few tests were also carried out using a 7475-T761 alloy (nominal composition Al-5.2–6.2Zn-1.9–2.6Mg-1.2–1.9Cu-0.18–0.25Cr) and a 7050-T7451 alloy (nominal composition Al-5.7–6.7Zn-1.9–2.6Mg-2.0–2.6Cu-0.08–0.15Zr). For the 7050 alloy, standard compact-tension specimens cut from 35-mm-thick plate in the L-T crack-plane orientation were tested at a constant ΔK value of 15 MPa√m.

2.2 Environments

Streams of dry hydrogen (99.999% purity), dry oxygen, and dry inert gases (at ~100 kPa), from commercial compressed gas cylinders, were passed through a desiccant and then over the specimens for some tests. Dry air (where specimens were sealed into a tube and surrounded by a desiccant), laboratory air (~50% relative humidity), distilled water, aqueous chloride and iodide solutions, and a liquid Bi-Pb-In-Sn-Cd eutectic (melting point 47°C) were also used. The tests were performed at ~20°C, except when temperature effects were investigated, and for those in the bismuth alloy, which were carried out at ~60°C. The Bi alloy was removed from fracture surfaces/cracks by dissolution in concentrated nitric acid, which does not significantly affect the underlying Al alloy fracture surface.

Some Al-Zn-Mg specimens were charged with hydrogen before testing in dry air or an inert alcohol (dodecanol). The specimens (~1 mm thick) were cathodically charged for up to 3 days at 20°C in (i) 3.5% NaCl solution, with 250 mg/l NaAsO₂ added to increase the rate of atomic hydrogen absorption, at -1.6 V relative to the saturated calomel electrode (SCE) (Klimowicz & Latanision, 1978), and (ii) HCl (pH 1) at -1.5 V (SCE). Some specimens (~2.5 mm thick) were also charged in the above solutions and at the above potentials while being strained to promote hydrogen diffusion by dislocation transport (Albrecht, Bernstein, & Thompson, 1982). These specimens were rapidly preloaded to ~70% of the yield stress, and then simultaneous charging and straining was carried out at a strain rate of ~2×10^-6 s^-1 until strains of ~2% had occurred. The hydrogen levels and distributions were not measured, but it was assumed that solute hydrogen concentrations were equal or greater than those that would be produced ahead of crack tips during fatigue in aqueous environments (which are difficult to measure and are not well established).

4.1 Precharged specimens

Fatigue crack growth in underaged and peak-aged Al-Zn-Mg single crystals in dry air or an inert alcohol at 20°C after cathodically charging with hydrogen (then immediately testing) resulted in ductile striations with the same appearance and spacing as those produced in specimens that were not precharged, even near specimen sides where hydrogen concentrations (not measured) would have been highest (Figure 10). Various charging solutions and conditions reported in the literature were used (see Section 2.2), but ductile behaviour was observed in all cases.

Figure 10:

SEM of fatigue fracture surface of peak-aged Al-Zn-Mg single crystal tested in an inert alcohol solution (dodecanol) at 20°C and 1 Hz (ΔK ~40 MPa√m), showing ductile striations with the same spacing after hydrogen charging [HCl, pH 1 at -1.5 V (SCE) at 20°C for 20 h] as for the preceding area before charging. Note also that there is no difference in striation spacing in the centre of specimens and near the side surfaces after hydrogen charging.

4.2 Effect of ΔK

Striation spacings obviously increase with increasing ΔK, but it is noteworthy that the spacings of brittle striation on cleavage-like {100} <110> fracture surfaces were up to ~0.2 mm at very high ΔK at cycle frequencies of 24 Hz in peak-aged (HV ~165) Al-Zn-Mg single crystals tested in distilled water at 20°C (Figure 11). In such circumstances, crack growth (which occurs only during loading) would reach velocities of ~10 mm/s-probably too high for diffusion of hydrogen ahead of cracks. The confirmation that cleavage-like cracking in fatigue precracked peak-aged Al-Zn-Mg single crystals in aqueous environments could occur at such high crack velocities was provided by fractographic observations after cantilever bend tests under high-rate “monotonically increasing” displacements, where crack growth on specimen sides was monitored by a high-speed cine camera (Figure 12) (Lynch, 1982, 1984a). Rapid cleavage-like cracking (approximately several millimetres per second) under monotonically increasing displacements has also been observed in commercial 7475-T761 and 7050-T7451 Al-Zn-Mg-Cu alloy specimens (L-T orientation) when corrosion fatigue cracks were broken open. However, cleavage-like cracking only occurred in favourably oriented grains just ahead of fatigue cracks with surrounding areas exhibiting large, deep dimples (Figure 13).

Figure 11:

(A) Optical micrograph and (B) SEM of fracture surface of peak-aged Al-Zn-Mg single crystal tested in distilled water at 20°C at a high ΔK (50±5 MPa√m; following precracking at a lower ΔK) using a cycle frequency of 24 Hz, showing brittle striations with a spacing of up to ~0.2 mm, such that crack growth during loading would be occurring at ~10 mm/s.

Figure 12:

SEM of fracture surface of peak-aged Al-Zn-Mg single crystal (orientation in inset) tested at very high displacement rates in distilled water showing cleavage-like cracking (from fatigue precrack) at velocities of ~10 mm/s, as measured from a high-speed cine film of the specimen sides, with 0, 1/8, and 1/4 s frames shown.

Note the significant lateral contraction of specimen side surfaces (indicative of high strains) and dimpled area (D) in the centre of the specimen near the back face.

Figure 13:

SEM of fracture surface of 7050-T7451 alloy showing cleavage-like area surrounded by dimpled areas produced by rapid cracking (several millimetres per second) in deionised water after fatigue crack growth.

4.3 Effect of ageing conditions

The characteristics of corrosion fatigue in Al-Zn-Mg single crystals in aqueous environments, viz. cleavage-like {100} <110> fracture surfaces with slip occurring especially on {111} planes intersecting crack tips, were similar for underaged, peak-aged, and overaged conditions, provided that the hardness exceeded a minimum value (HV 100–125). The minimum hardness was somewhat higher for underaged conditions and for crystal orientations where {100} planes were more steeply inclined to the stress axis. For a given ΔK (in the range 25–40 MPa√m) and crystal orientation, brittle striation spacings were somewhat larger (~50%) for overaged conditions compared with underaged conditions with the same hardness (Figure 14). For liquid-metal environments, brittle striations were observed in the solution-treated condition as well as aged conditions, although striation spacing did increase with increasing hardness (as they did for aqueous environments).

Figure 14:

Optical micrograph of fracture surface of Al-Zn-Mg single crystal after fatigue crack growth in distilled water at 1 Hz first for an underaged (UA) condition (HV 137) and an overaged (OA) condition (HV 136; ΔK ~35 MPa√m), showing brittle striations parallel to two <110> directions on a near-{100} plane, with a somewhat larger spacing for the OA condition.

The detailed appearance (at the nanoscale) of cleavage-like fractures differed for underaged, peak-aged, and overaged materials in that nanoscale dimples were clearly apparent for overaged but not for underaged material (Figure 15). The size and spacings of the nanodimples increased with increasing ageing, consistent with nanovoids being nucleated at age-hardening precipitates (at least for peak-aged to overaged conditions). The appearance of river lines also depended on the ageing condition: serrated steps (mating and interlocking on opposite fracture surfaces) were observed for peak-aged material, whereas tear ridges (with peak matching peak on opposite fracture surfaces) were observed for underaged and overaged materials. Numerous slip lines (produced by strains just behind crack tips) were observed for underaged and peak-aged materials but not for overaged material, which is not surprising because underaged material exhibits planar slip characteristics (due to the shearing of coherent age-hardening precipitates), whereas overaged material with incoherent precipitates deforms more homogeneously (Figure 16).

Figure 15:

TEM of replicas of (macroscopic) cleavage-like fracture surfaces of Al-Zn-Mg single crystals cracked rapidly (~10 mm/s) in distilled water (then immediately washed in alcohol and dried, so that no significant corrosion occurred), showing (A) nanoscale dimples for an overaged condition (HV 102) and (B) relatively smooth areas, except for slip lines and steps for an underaged condition (HV 126).

Figure 16:

Optical micrograph of deformation surface produced by fatigue crack growth on side surface of single crystal in distilled water (1 Hz), showing marked change in slip characteristics (from planar slip to more homogeneous slip) associated with a change in ageing condition from underaged (UA) to overaged (OA).

4.4 Effect of solution composition

The most dramatic effect of solution composition was observed for peak-aged Al-Zn-Mg single crystals tested first in distilled water (or NaCl solution) and then in a 2 m nitrate solution, which resulted in an abrupt change from brittle striations to ductile striations with a 10- to 20-fold reduction in striation spacing (Figure 17). Chromate solutions, on the other hand, resulted in a relatively small decrease (50%) in striation spacing compared with those produced in water, and striations still had “brittle” characteristics (S.P. Lynch, unpublished research). The changes in pH (from 2 to 12) for 3.5% NaCl solutions did not significantly affect the spacing of brittle striations, at least for the high crack-growth regime studied (Figure 18).

Figure 17:

SEM of fracture surface of peak-aged Al-Zn-Mg single crystal showing widely spaced brittle striations (arrowed) produced by fatigue crack growth at 20°C in distilled water and ductile striations produced when a 2 m nitrate solution was used (for the same ΔK=~30 MPa√m and cycle frequency of 1 Hz).

Figure 18:

Optical micrograph of fatigue fracture surface of peak-aged Al-Zn-Mg single crystal tested in 3.5% NaCl solution with pH values of 2, 7, and 12 (ΔK=~30 MPa√m, 20°C, and 0.3 Hz), showing little or no effect of pH on the spacing of brittle striations.

4.5 Effect of electrode potential

There was a small increase in the spacing of brittle striations (and slight change in roughness) when the potential was changed from -0.5 to -2 V (Ag/AgCl) for peak-aged Al-Zn-Mg single crystals tested in 3.5% NaCl solution at ~20°C (Figure 19; with some other specimens showing very little, if any, increase). In contrast, changing the potential from anodic (-0.8 V) to cathodic (-1.6 V; Ag/AgCl) for a commercial 7475-T761 Al-Zn-Mg-Cu alloy (under the same experimental conditions) resulted in an abrupt change from brittle striations to ductile striations (and an eightfold decrease in spacing) for tests in 3.5% NaCl solution at ~20°C (Figure 20).

Figure 19:

Optical micrograph of fatigue fracture surface of peak-aged Al-Zn-Mg single crystals tested in 3.5% NaCl solution at a ΔK=~25 MPa√m, 20°C, and 0.3 Hz at the (Ag/AgCl) potentials indicated showing that there was little or no effect of potential on the spacing of brittle striations.

Figure 20:

SEM of fracture surface of a 7475-T761 alloy tested in NaCl solution (ΔK=~30 MPa√m, 20°C, and 1 Hz), showing abrupt changes from brittle striations (arrowed) to ductile striations, as the potential was cycled from -0.8 to -1.6 V) (Ag/AgCl).

4.6 Effects of temperature

For peak-aged Al-Zn-Mg single crystals tested in distilled water, increasing temperature decreased the spacing of brittle striations (and also increased the density of steps) (Figure 21). In 3.5% NaCl solution, on the other hand, increasing temperature from -5°C to 65°C increased the spacing of brittle striations by about a factor of 2 (Figure 22).

Figure 21:

Optical micrograph of fatigue fracture surface of a peak-aged Al-Zn-Mg single crystal tested in distilled water at 0.3 Hz at 65°C and 4°C sequentially, showing brittle striations with a smaller spacing at the higher temperature.

Figure 22:

Optical micrograph of fatigue fracture surface of a peak-aged Al-Zn-Mg single crystal tested in 3.5% NaCl solution (ΔK=~25 MPa√m, 0.3 Hz) at 20°C, -5°C, and 65°C, showing brittle striations with a larger spacing at the higher temperature.

4.7 Effect of cycle frequency and waveform

For peak-aged Al-Zn-Mg single crystals tested in laboratory air (~50% relative humidity), increasing the cycle frequency from 1 to 24 Hz resulted in abrupt changes from brittle to ductile striations (S.P. Lynch, unpublished research). In aqueous environments, the change in cycle frequency from 1 to 24 Hz decreased the spacing of striations, but crack growth still tended to occur on {100} planes in <110> directions (Figure 23). Short hold times (<1 min) at peak load had no significant effect on striation spacing for tests in aqueous environments, but longer times increased striation spacing due to contributions from (sustained load) stress corrosion cracking (S.P. Lynch, unpublished research).

Figure 23:

SEM of fatigue fracture surface of peak-aged Al-Zn-Mg single crystal tested in an aqueous potassium iodide solution (ΔK=~25–30 MPa√m, 20°C) at 1 and 24 Hz sequentially, showing an increase in the spacing of brittle striations for the lower cycle frequency.

A particularly interesting effect of cycle frequency (triangular waveform) on the corrosion fatigue of a commercial 7050-T7451 alloy (L-T orientation) tested in deionised water at a constant ΔK (15 MPa√m) has recently been observed by the present authors (Byrnes et al., 2014). Abruptly changing the rise time from 1 to 25 s (with fall times of 1 and 25 s) resulted in an abrupt change from brittle striations on cleavage-like facets to ductile striations (with no effect of fall time on striation spacings). Changing the rise time from 25 s back to 1 s resulted in a transition back to brittle striations on cleavage-like areas but initially only at steps along the crack front. Cleavage-like areas then fanned out and coalesced. The areas produced by the “short” (1 s) rise-time cycles between the fan-shaped cleavage-like areas (before their coalescence) exhibited ductile striations with a spacing that was initially the same as the preceding ductile striations produced by the “long” (25 s) rise-time cycles. The ductile striation spacing then gradually increased as the size of the ligaments between the cleavage-like areas decreased, thereby increasing the “local” ΔK. The above observations are shown in Figure 24 and schematically in Figure 25.

Figure 24:

SEM of fracture surface of 7050-T7451 alloy tested in deionised water at 20°C (ΔK=15 MPa√m, R=0.3), with the various cycle frequencies/waveforms indicated, showing abrupt change from brittle to ductile striations along the crack front (A-A) when the rise time was increased and then a “staggered” change from ductile to brittle behaviour when the rise time was decreased (dotted line) as a result of brittle cracking initiating only at some steps (arrowed) along the crack front.

Figure 25:

Schematic diagram illustrating the complex behaviour shown in Figure 24 for the effects of changes in rise time on the spacing and appearance of fatigue striations.

5.1 Mechanisms of corrosion fatigue crack growth

The reason for the general assumption that hydrogen is the embrittling species for moist air and aqueous environments is partly because atomic hydrogen is readily produced by the chemical dissociation of water molecules or by electrochemical reactions on “fresh” aluminium surfaces at strained crack tips and partly because hydrogen is a ubiquitous embrittler (for numerous metals). For aluminium alloys, however, hydrogen gas does not dissociate on clean surfaces and, hence, no embrittlement is observed in dry hydrogen. For metals where hydrogen gas does dissociate (e.g. nickel), cleavage-like cracking with the same characteristics as for Al alloys in aqueous environments is observed, viz., {100} <110> crystallography, extensive slip on planes intersecting crack fronts, high localised strains, and nanoscale/microscale dimples on fracture surfaces (Lynch, 1984b, 2011).

The characteristics of transgranular corrosion fatigue crack growth in Al-Zn-Mg single crystals and commercial Al-Zn-Mg-Cu alloys described in the previous sections can be best explained by an adsorbed-hydrogen localised plasticity process, and specifically by the adsorption-induced dislocation emission (AIDE) mechanism, first proposed by Lynch (2011, 2012a). Mechanisms based on hydrogen-enhanced localised plasticity (HELP) associated with solute hydrogen in the plastic zone ahead of cracks, or by hydrogen-enhanced decohesion at or ahead of cracks, are not supported by the experimental observations. The effects of variables, such as rise time, solution composition, and temperature, can all be rationalised in terms of the AIDE model, although there is still much that is unknown regarding the effects of these variables on surface reaction kinetics, hydrogen generation, and adsorption.

Many of the reasons for concluding that the AIDE mechanism is responsible (or at least predominates) for cleavage-like cracking in Al alloys (and some other materials, such as Ni, Fe, Ti, and Mg) have been discussed in detail in previous publications (Lynch, 1988, 2011, 2012a,b) but are summarised below. It could be argued that some reasons are not completely persuasive when considered in isolation, but a convincing case for the AIDE mechanism can be made if the evidence is considered as a whole.

Observations that lend support to a localised-slip crack-growth process rather than an atomically brittle decohesion process for Al alloys include the following:

1. Very high, localised strains around {100} <110> cleavage-like cracks with extensive slip on planes intersecting crack fronts (which would tend to relax local stresses, such that the high stresses required for decohesion at crack tips are not attained);

2. Dimpled fracture surfaces (generally on the nanoscale but with occasional microscale dimples); and

3. Crack planes that can be up to 10° away from the low-index {100} plane, whereas decohesion would be expected to occur exactly on low-index plane (as for low-temperature cleavage in bcc materials).

Key observations that lend support to an adsorbed-hydrogen mechanism rather than one based on solute hydrogen for cleavage-like cracking in Al alloys are as follows:

1. Crack velocities that can be as high as 10 mm/s (under special conditions), such that hydrogen diffusivity to crack velocity ratios are <10^-8 cm, so that hydrogen should not diffuse more than one or two atomic distances (at most) ahead of cracks;

2. Ductile behaviour of hydrogen-charged specimens tested in inert environments (although hydrogen concentrations in charged specimens should be equal or greater than those ahead of crack tips for specimens tested in aqueous environments);

3. Similarities such as the {100} <110> crystallography of fracture surfaces between cracking in aqueous (and moist air) environments and cracking in liquid-metal environments, where only adsorption is likely to occur due to the high rates of cracking and low mutual solubilities associated with the large size difference between embrittling metal atoms and substrate atoms; and

4. Abrupt transitions from brittle to ductile striations (or abrupt changes in the spacing of brittle striations) on changing variables, such as temperature, rise time, and electrode potential.

Accepting that adsorbed hydrogen (i.e. hydrogen on crack-tip surfaces and between the first and second atomic layers) is responsible for cleavage-like cracking and that cracking occurs by localised plastic deformation, the only possible explanation (given the range of influence of adsorbed species is only one or two atomic distances) is that adsorption facilitates the nucleation of dislocations at crack tips. Moreover, the AIDE mechanism is supported by some quantum mechanical modelling, the defactant concept, and surface science observations that show that adsorption can have dramatic effects on bonding at surfaces (Lynch, 1988, 2011, 2012a,b). The AIDE mechanism can also account for all the characteristics of cleavage-like cracking, such as the {100} <110> crystallography, as discussed in detail elsewhere (Lynch, 1988, 2011, 2012a,b).

Despite the above arguments, the HELP mechanism continues to receive widespread support. However, this appears to be just on the basis that (i) hydrogen can diffuse ahead of cracks in many circumstances, (ii) hydrogen can enhance dislocation activity, and (iii) crack growth occurs by localised plasticity rather than on any detailed consideration of the cleavage-like {100} <110> characteristics of cracking and consideration of alternative explanations. Moreover, the extent of softening by solute hydrogen in bulk specimens generally appears to be small and would not be expected to have much effect on fracture characteristics. Apparently large effects of solute hydrogen on dislocation activity observed by TEM in strained thin foils should be treated with scepticism because other effects such as stresses arising from nonuniform hydrogen distributions and surface effects are probably involved in addition to HELP (Lynch, 1988, 2011, 2012a,b).

The solute hydrogen-based HELP phenomenon could make a contribution to the AIDE process by facilitating the movement of dislocations away from crack tips and reducing the back-stress on subsequent dislocation emission, but any contribution is likely to be a minor one. The HELP mechanism by itself (whereby solute hydrogen facilitates and localises “general” dislocation activity ahead of cracks) cannot account for cleavage-like cracking on {100} planes in <110> directions: Just facilitating/localising dislocation activity ahead of cracks would not result in a change in fracture plane from one normal to the applied stress in inert environments to a {100} plane steeply inclined to the stress axis in a hydrogen-bearing environment. Crack growth in <110> directions at large angles to the overall direction of cracking would also not be expected.

5.2 Explanations for the effects of variables on crack growth