Sulfur-impurity segregation at grain boundaries in nickel, resulting in embrittlement of this otherwise ductile material, is a long-standing materials problem whose atomic level origins have remained elusive. Publishing in Nature Communications, a research group at the University of California, San Diego (UCSD) headed by Jian Luo has now identified the key players. “Combining modeling and experiments, we found that the universal formation of two types of bipolar, yet largely disordered, interfacial structures, namely amorphous-like and bilayer-like complexions, at faceted general grain boundaries in sulfur-doped nickel is the root cause for grain-boundary embrittlement in this material,” Luo says.

Complexions are phase-like entities with their own structures, compositions, and properties that can form at interfaces such as grain boundaries. Since they cannot exist independently apart from the interface, they are not traditional phases but they otherwise behave like them. For example, they can exhibit two-dimensional phase-like transitions at well-defined temperatures.

Until recently, to make complexions accessible to available experimental and theoretical techniques, most studies of grain boundaries, as well as other interfaces, have been limited to nearly ideal configurations that are inherently symmetric and periodic, such as special tilt or twist boundaries in bi-crystals. But “general” grain boundaries in real-world polycrystals can have orientations lacking symmetry and long-range order, thereby making them difficult to visualize experimentally at the atomic level and resistant to theoretical treatment. In 2011, a group headed by Martin Harmer of Lehigh University, along with Luo, successfully used aberration-corrected scanning transmission electron microscopy (AC-STEM) to demonstrate the formation of bismuth-rich complexions at general grain boundaries in bismuth-doped nickel, another well-known embrittlement system. Then in 2017 they further reported that interfacial reconstructions at general grain boundaries in bismuth-doped nickel led to the formation of superstructures, an extremely ordered complexion, which they verified by density functional theory (DFT) calculations. Characterizing disordered complexions, particularly those marked by partial orders, such as the hidden bipolar orders revealed in the current nickel-sulfur study, presents a stiffer challenge.

At UCSD, development of grain boundary “phase” (complexion) diagrams has been a goal of much of the Luo group research. The group chose to investigate the more disordered complexions in sulfur-doped nickel because, Luo says, “sulfur embrittlement of nickel is not only a classic problem in materials science, it is also of practical importance because many engineered alloys are used in sulfur-bearing environments.” For their study, the UCSD group used AC STEM combined with quantitative energy-dispersive x-ray spectroscopy to examine a large number of randomly selected general boundaries in samples that were annealed to obtain thermal-equilibrium structures and then quenched to lock in these structures.

“Bismuth and sulfur represent two extreme cases of interfacial structures—ordered versus disordered, respectively. Yet, the more disordered interfacial structures in nickel-sulfur are more challenging to characterize; e.g., DFT was no longer effective and we had to rely on the hybrid Monte Carlo/Molecular Dynamics simulations with a DFT-derived reactive force-field potential,” Luo says. Simulations in semi-grand canonical ensembles with a DFT-derived reactive force-field potential were used to obtain the equilibrium interfacial structures. Simulated AC STEM images from these atomic structures were compared with experimental images.

“The UCSD group applied a combination of high-resolution transmission electron microscopy and atomistic modeling techniques very creatively to the embrittlement problem in sulfur-doped nickel, while also utilizing a deep knowledge of the thermodynamics of complexions,” says Tim Rupert of the University of California, Irvine, who was not involved in this study. “There is a beautiful synergy of the techniques used in this study.”

What the research group found was the unexpected predominance at the same general grain boundary of two types of facets hosting complexions with excess sulfur. The complexions were primarily either amorphous-like layers (Type A facet) with nanoscale thickness fixed by thermal equilibrium or disordered bilayer-like layers (Type B facet). There also were complexions without sulfur (Type C), which were only observed at low-energy boundaries. No other type of facet was observed. Planes with low Miller indices played a decisive role in the type of facet that could be observed.

Type A facets only formed when the facet surfaces on one side of the grain boundary were (100) planes, and then the matching facet surfaces on the other side of the boundary always had high Miller indices. Type B facets formed with low-index planes other than (100) on one side of the interface and high-index planes on the other side. While it had been thought that it was the misorientation between the grains across the grain boundary that determined the interfacial structure, studies of the facets showed that it is the orientation of the low-index plane that plays that role.

Atomistic simulations further revealed the existence of polar sulfur-nickel entities in both amorphous-like Type A and bilayer-like Type B complexions. In the latter case, one can visualize the bilayer as two planes of polar sulfur-nickel structures oriented head-to-head in opposite directions approximately normal to the planes, to form a bipolar interfacial structure. The Type A case is not so easily visualized but conceptually similar.

The key point is that tensile simulations showed that the grain boundary can break between the planes of polar sulfur-nickel structures; they are the weak point that leads to sulfur embrittlement. That the computed tensile toughness is inversely correlated with a bipolar index testifies to this mechanism.

“Based on our findings, potential strategies to mitigate grain-boundary embrittlement include designing heat treatments or co-doping strategies to destabilize the bipolar interfacial complexions that are the root cause of grain boundary embrittlement, guided by grain boundary ‘phase’ diagrams,” Luo says.

Tim Rupert emphasizes that the UCSD findings are generalizable to other materials systems and problems. “The general framework established by the UCSD group should be very powerful and allow a number of other problems to be treated. An important extension could be to study the hydrogen embrittlement of engineering alloys, but many other problems where interfacial structure is important can also be addressed,” Rupert says.

Read the article in Nature Communications.