When good metals go bad

Air travel may become safer as a result of research at USC into corrosion-induced failure in high-performance metals used in aerospace and other demanding applications.

Air travel may become safer as a result of research the University of Southern California (USC) will carry out as its $2 million part of a $3.8 million investigation into corrosion-induced failure in high-performance metals used in aerospace and other demanding applications.

Supercomputing specialist Priya Vashishta and his colleagues will model the behaviour of hundreds, thousands and millions of individual atoms to gain greater understanding of how and why alloys of titanium and other metals suffer stress corrosion cracking, potentially catastrophic damage resulting from mechanical strain in chemically unfriendly environments.

Vashishta and longtime collaborators Aichiro Nakano and Rajiv K. Kalia will be carrying out their investigations as part of an Information Technology Research project funded by the National Science Foundation. Like Vashishta, Nakano and Kalia hold joint appointments in the Viterbi School departments of computer science, materials science and (for Kalia) biomedical engineering; and in the USC College department of physics and astronomy.

The trio will partner with Caltech and Purdue on the effort, one of 120 ITR projects funded by the NSF.

Vashishta and co-investigators will use new techniques of nanoscience to supplement the traditional structural engineering approach, known as “continuum mechanics.” This technique involves extensive testing of pieces of material to establish parameters of performance, which are then be expressed as predictive equations engineers use to design structures.

This works well, Vashista said, in providing reliable forecasts of how the material will behave when new. But it offers little guidance into how and when materials may fail because of stress corrosion cracking (SCC), damage from corrosion that starts when ordinary strain on the metal produces tiny cracks that allow the entrance of moisture and oxygen.

Nanoscientific analysis can supply such guidance, Vashishsta said. The idea is to go down to the basic atomic structure of the material and simulate the behaviour of individual atoms at the point where cracks appear in the surface.

“We start by accurately modelling the behaviour of collections of a few hundred atoms at one point; proceed from there to modelling thousands of atoms along the surface, going to millions of atoms over a larger area,” he explained.

The results of the nanoanalysis have to produce the same predictions for behaviour as the traditional continuum approach, Vashishta said. “But by understanding exactly what is going on, in detail, at the point where the material is failing, we can find better ways to prevent damage, and create more corrosion resistant materials.”

“Corrosion is an enormously complex technological and economic problem with an annual cost of about 3% of the US gross domestic product,” according to the proposal for the study. “Most critical here is premature and catastrophic failure of materials resulting from chemically influenced corrosion. Safety and reliability [of structures] is endangered by the uncertainties in SCC. To prevent SCC requires that we understand the atomistic mechanisms underlying SCC.”

Such understanding demands huge computational resources. Vashishta’s group has its own 480-CPU supercomputer, and also uses the 2048-CPU supercomputer at the USC centre for High Performance Computing and Communications (HPCC).