Skip to main content Accessibility help

Corrosion and stress play separate roles in Ag–Au alloy failure

By Arthur L. Robinson October 5, 2018
Electron micrograph of stringy nanoporous gold-enriched structures resulting from dissolution of silver at the surface of a silver–gold alloy exposed to a corrosive environment. The individual ligaments in the image are about 5 nm in diameter and are perforated with 2-nm-diameter holes (dark areas). So long as the nanoporous layer adheres well to the uncorroded alloy, it can mechanically transmit cracks into and through the alloy through grain boundaries, but the layer de-adheres as the holes quickly coarsen, degrading the interface and halting the transmission of intergranular cracks. The fast time scale is responsible for the prior understanding that corrosion and stress had to exist simultaneously to cause intergranular stress-corrosion cracking. Credit: Karl Sieradzki, Arizona State University.

Intergranular stress–corrosion cracking (IGSCC) is an important failure mode for many metal alloys exposed to corrosive environments. Materials researchers have long believed that along with a susceptible material, this kind of cracking requires the simultaneous presence of a corrosive environment and stress. Building on clues accumulating in the literature, a research group at Arizona State University and the Pacific Northwest National Laboratory headed by ASU’s Karl Sieradzki has now conclusively refuted this belief. The researchers reported their results in Nature Materials.

The research group applied a suite of state-of-the-art imaging tools, including aberration-corrected scanning transmission electron microscopy (STEM) and atom-probe tomography, in combination with statistical analysis and a novel experimental protocol to decouple the roles of corrosive environments and stress and prove they do not always have to be present simultaneously for IGSCC to occur. “The possibility of decoupling corrosion from stress is not a new idea,” says Roger Newman of the University of Toronto, “but the ASU–PNNL work provides a new level of rigor in quantifying how it can occur.”

Cracking leading to fracture in metal alloys is a multifaceted phenomenon. The cracking can propagate along grain boundaries (intergranular) if they are weaker than the grains or through the grains along lattice planes (transgranular). In IGSCC, different failure mechanisms include hydrogen embrittlement, which can occur if environmental hydrogen diffuses into the alloy, and selective metal dissolution, in which one alloy component is dissolved. The latter can be further sub-divided into impurity segregation to grain boundaries and de-alloying resulting in a nanoporous structure.

To eliminate the possibility of hydrogen embrittlement and isolate the de-alloying process, the research group focused on a silver–gold alloy, neither of whose components absorbs hydrogen. In addition, this alloy has long served as a model system for IGSCC and thus has been the subject of much prior work, giving the researchers a foundation from which to frame their experiments.

It was already known that de-alloying occurs faster at grain boundaries than in the grains, owing to the disordered structure of the grain boundaries. Previous comparisons of stress–corrosion cracking rates with corrosion rates along grain boundaries for silver–gold alloys revealed that crack propagation occurred far more rapidly than any reasonable silver dissolution rate, suggesting that a mechanical component accompanying each crack advance was also operative. Similar observations have been made for stainless steels.

The researchers developed a set of experiments aimed at separating the roles of corrosion and stress in the IGSCC of their silver–gold alloys. They chose two alloy compositions (Ag0.70Au0.30 and Ag0.72Au0.28) similar to those studied previously. The first step was to anneal polycrystalline samples of each composition to eliminate any residual stress before immersing them in perchloric acid. This acid has the advantage that it does not corrode silver–gold alloys without the addition of an applied voltage so the researchers could control the thicknesses of the de-alloyed, nanoporous layers—they chose 500 nm and 2.2 µm in these experiments.

Imaging, elemental analysis, and statistical analysis measurements were carried out on some of the de-alloyed samples, while others were held in air or in deionized water to halt corrosion before being stressed by bending and measured. An important ingredient in the protocol was a variable delay time of 3–180 seconds in air between being removed from the corrosive perchloric acid and stressing. The delay time turned out to be a key factor in demonstrating the decoupling between corrosion and stress.

The unstressed samples provided a baseline for interpreting the overall results. De-alloying was indeed observed to penetrate deeper along grain boundaries than at the surface. Porous structures were found as far down as de-alloying proceeded but no further. The researchers stressed the samples in what they called crack injection experiments by bending them by hand to a radius of curvature of 500 µm. None of the samples held in air for longer than 10 seconds or in water fractured all the way through when bent, whereas for shorter times, all the samples fractured.

For a closer look at what happens at long delay times, the researchers prepared a different set of samples, held them in air for 180 seconds before subjecting them to tensile stress. Selecting 10,000 grain boundaries that were wide enough for analysis, they measured the penetration depth of the de-alloyed portion below the de-alloyed surface layer. For the intergranular cracks, statistical analysis showed that the cracks continued on average a few times farther than the penetration depth of the de-alloyed portions of the grain boundaries. Examination of the porous layer by STEM revealed the formation of elliptical voids near and parallel to the interface between corroded and uncorroded layers. The voids led to de-cohesion of the corroded layer, preventing it from transmitting cracks much farther than grain-boundary corrosion had penetrated.”

In sum, these experiments demonstrated that stress–corrosion cracking leading to fracture can occur when corrosion and stress are not simultaneous, but only if the stress is applied before the rapid degradation of the mechanical integrity of the interface between the corroded and uncorroded material interrupts the crack transmission.

How do these results apply to engineering alloys subjected to corrosive environments? Sieradzki warns against uncritically generalizing the findings to all engineering alloys. However, he says, “Our findings do provide a path forward for obtaining a more complete understanding of IGSCC in important engineering alloys such as stainless steels and nickel-based alloys in critical applications.” Recent electron microscopy experiments have demonstrated the presence of grain-boundary nanoporosity due to selective chromium oxidation in stainless steels in water at high temperatures and of nickel-based alloys at the high temperatures and pressures found in nuclear reactors, yielding a direct link between the model silver–gold system and these engineering alloys.

Read the abstract in Nature Materials.