You've seen the movie scene: dilapidated skyscrapers, collapsed bridges, and empty, shell-like cars in a post-apocalyptic city. While Hollywood imagines fictional causes for this decay, in reality, the culprit is far more mundane: corrosion.
Corrosion costs trillions of dollars globally, with up to three percent of the U.S. GDP spent on failing materials. New research from Lawrence Livermore National Laboratory (LLNL) aims to tackle this issue by predicting failure and informing the design of better materials, up front.
"Our current knowledge of corrosion is based on historical data from well-known and well-characterized metal compositions and processing," said LLNL scientist and author Brandon Wood. "As soon as you alter the composition at all or alter the way the materials are processed, all bets are off."
Using a novel technique that employs advanced kinetic modeling, the team simulated corrosion processes with both speed and accuracy and identified the effects of operating conditions and material composition.
The researchers focused their simulation efforts on the natural protective oxide film that forms on metals. This film is crucial for keeping the metal intact. If it dissolves or fractures, or if it becomes permeable to attack, corrosion creeps in.
Former LLNL postdoctoral researcher Penghao Xiao, now at Dalhousie University, developed the multi-scale simulations that capture how the oxide grows, dissolves and changes composition over time in response to environmental factors like pH and voltage. Since this approach is too cumbersome to deploy for every material and environment, the team trained a machine learning-inspired model to predict when and why corrosion occurs.
With this framework, the authors examined three voltage regimes. While the environments with high and low voltages are well-studied and understood, the intermediate regime was a bit of a mystery.
"Until now, no one was really able to explain what exactly was going on in that regime," said LLNL scientist Chris Orme, who was the experimental lead on the project. "We showed there is competition between two processes: dissolution and reprecipitation. When molecules leave the surface, mix and redeposit, the oxide looks completely different."
While voltage may be applied directly in some systems, like batteries, the same phenomenon is surprisingly omnipresent in other contexts as well.
"Putting certain metals close to one another creates a sort of microbattery that can drive corrosion," Wood said. "This has been a problem in building ships and bridges, for instance. Our model can in principle account for such effects, while also being flexible enough to consider the interplay between the corrosive environment and the base metal composition."
That's just one example of a scenario where this model might be helpful. By advancing our understanding of corrosion and developing predictive tools, this research paves the way for designing materials that can withstand the test of time.
This work was part of an LLNL strategic initiative to predict materials failure. Funding was provided by the Laboratory Directed Research and Development program. Other LLNL authors include Christine A. Orme, S. Roger Qiu, Tuan Anh Pham, Seongkoo Cho and Michael Bagge-Hansen. Former LLNL postdoctoral researcher Penghao Xiao led the simulations.
