Hydrogen-induced intergranular failure in nickel revisited

Abstract

Using a combination of high-resolution scanning and transmission electron microscopy, the basic mechanisms of hydrogen-induced intergranular fracture in nickel have been revisited. Focused-ion beam machining was employed to extract samples from the fracture surface to enable the examination of the microstructure immediately beneath it. Evidence for slip on multiple slip systems was evident on the fracture surface; immediately beneath it, an extensive dislocation substructure exists. These observations raise interesting questions about the role of plasticity in establishing the conditions for hydrogen-induced crack initiation and propagation along a grain boundary. The mechanisms of hydrogen embrittlement are re-examined in light of these new results.

Introduction

The degradation in the mechanical properties of metals due to the presence of hydrogen is often accompanied by a change in fracture mode from ductile transgranular to intergranular [1], [2], [3], [4], [5], [6], [7], [8], [9], [10], [11]. The former has been described as strain-controlled with plasticity concentrated at the macroscale, and the latter as stress-induced decohesion in which plasticity is blocked at the microscale and is incidental to the propagation of intergranular cracks [8]. The role of hydrogen was assumed solely to be to reduce the cohesive strength of the grain boundary such that crack initiation and propagation along the boundary was favored over slip transmission and ductile failure processes. The transition in failure mode in steels has also been attributed commonly to the presence of other segregants in addition to hydrogen at the grain boundary [8], although more recent studies using high-purity steels have suggested hydrogen alone can cause intergranular embrittlement [12]. This result is consistent with studies using relatively pure polycrystalline metals.

For example, Lassila and Birnbaum showed that the degree of intergranular failure in nickel was dependent on the concentration of sulfur and of hydrogen, with the combination of the two being more potent than either individually [6], [13], i.e. 100% intergranular fracture was possible if sufficient time was allowed to segregate a critical, but unknown and undetermined, concentration of hydrogen to the grain boundaries. Similarly, with sufficient sulfur segregated to the grain boundary, the fracture was completely intergranular, whereas with low sulfur coverage on the grain boundary, complete intergranular failure occurred only in the presence of hydrogen; again, the combined effect of the two segregants was more damaging than the individual contributions of each species.

To examine this further, Heubaum conducted in situ deformation experiments in a scanning electron microscope (SEM) of a hydrogen-charged nickel sample which, to retain the hydrogen, was stored at 77 K before testing [14]. Prior to intergranular failure, extensive surface rumpling and slip band formation was evident on the free surface. Additionally, slip traces were in evidence on the intergranular fracture surfaces. More recently Bechtle et al. demonstrated that the susceptibility of Ni-201 to hydrogen embrittlement could be reduced by thermomechanical processing to generate a high fraction of special grain boundaries, which were predominantly twins [15]. The reduced hydrogen susceptibility induced by the high special boundary fraction was rationalized by invoking one of two arguments. The first required more hydrogen segregating to, and accumulating at, random grain boundaries than special boundaries such that the magnitude of the reduction in cohesive strength would be less at the special boundaries. The second argument assumed the segregated concentration was independent of grain boundary character and as special grain boundaries are stronger than random ones, higher hydrogen concentrations would be needed to achieve a sufficient and necessary reduction in the cohesive strength to cause failure on special grain boundaries.

Although there is abundant evidence supporting transport of hydrogen by dislocations [16], [17], [18], there is a lack of information regarding the enhanced diffusion of hydrogen (or deuterium) along grain boundaries as well as its dependence on grain boundary type. Tsuru and Latanision measured the diffusivities of hydrogen in polycrystalline nickel and reported diffusion to be faster along grain boundaries (D_latt = 3.52 ± 1.02 × 10⁻¹⁴ m² s⁻¹, D_gb = 2.05 ± 1.50 × 10⁻¹² m² s⁻¹) [19]. One of the few studies to consider the dependence of diffusion on grain boundary character is that of Ladna and Birnbaum, who compared diffusion along Σ = 9 and Σ = 11; only along the Σ = 9 boundaries was the diffusion enhanced [20]. In addition to the lack of data on diffusion in the grain boundary, there is still a lack of data on the concentration of hydrogen that can exist in grain boundaries, as well as the effect that this hydrogen concentration will have on the cohesive strength of the boundary.

Electronic structure calculations have shown that the presence of impurities at grain boundaries changes the local bonding, such that the segregants decrease the cohesive strength of the boundary [21], [22], [23]. However, the magnitude of the reduction in the cohesive energy as a function of hydrogen on a particular type of grain boundary is not well documented. For example, using density functional theory, Dadfarnia et al. calculated that the reduction on a Σ3 grain boundary in iron was not significant and did not vary linearly with concentration [24].

Robertson and Birnbaum reported that cracks in nickel deformed in situ in a hydrogen environment in a transmission electron microscope (TEM) followed the contour of the grain boundary but did not necessarily propagate along the grain boundary [25]; Tabata and Birnbaum reported similar observations in iron [26], [27]. Specifically, the crack path followed the slip planes, which paralleled, rather than propagated in, the grain boundary. These observations raised the question about the actual crack path for intergranular fracture in these instances. However, based on results since these studies [28], [29], including those presented herein, it is likely that the crack path observed in some, but not all, in situ TEM deformation studies may be special as it occurs when the dislocations approaching the grain boundary are not accommodated in it, but rather cross-slip to follow a path that parallels the boundary. Furthermore, intergranular fracture will not occur if the hydrogen content segregated and accumulated at the grain boundary is below some critical concentration. The failure to reach the critical hydrogen concentration at the grain boundaries is reasonable given these straining experiments were performed in a gaseous hydrogen environment using electron-transparent samples. That is not to say that intergranular failure has never been observed in this type of experiment as intergranular fracture was reported to occur in Ni–S [30] and in Ni₃Al [31] deformed in the TEM in a hydrogen environment.

In an effort to relate the intergranular failure observed in macroscopically tested hydrogen-charged nickel [15] to the microstructure produced during the deformation that proceeded failure, the focused-ion beam (FIB) lift-out technique has been used to produce electron-transparent samples from the volume immediately beneath the fracture surface. The results from this investigation are reported and discussed in terms of known hydrogen embrittlement mechanisms. A new mechanism for hydrogen-induced intergranular failure of nickel emerges from this discussion.