Step into a high-end restaurant or bar in the city and you might see a mixologist in action, whipping up colourful concoctions of masterfully blended liqueurs. Cocktails and alloys have a lot in common—they’re both mixtures of different components in set proportions which produce a specific flavour (or function).
Among alloys, high entropy alloys (HEAs) are particularly complex cocktails. Where traditional alloys might just blend two metals, HEAs can involve five or more, blended in roughly equal proportions. This complexity can give them exceptional strength, durability and corrosion resistance.
Naturally, a challenge materials scientists face is figuring out which metal combinations create HEAs with favourable physical properties. “HEA design spaces are incredibly vast. Once you’ve chosen the metals involved, there can be over 100,000 feasible compositions to investigate for the best options,” explained Viacheslav Sorkin, a Senior Scientist at A*STAR’s Institute of High-Performance Computing (IHPC).
For instance, the same five metals in a quinary HEA might be combined in over 7,000 possible patterns. Vary their concentrations and this number skyrockets, dragging out HEA design and development timelines. “Existing computational methods to screen HEAs are either lengthy and costly, or trade accuracy for efficiency,” Sorkin added.
One such method is the small set of ordered structures (SSOS). Imagine floor tiles with the same geometric pattern: by studying one tile, you can surmise the larger patterns a roomful of them might create. Likewise, SSOS works out a HEA’s physical properties—such as stability, formation energy and mass density—by simulating small, repeating three-dimensional sections within the alloy, instead of modelling every atom inside.
“While SSOS is currently considered the most promising way to accurately screen many HEA candidates at once, it still involves a relatively huge design space and expensive calculations, making it lose efficiency,” said Sorkin.
To improve on SSOS, Sorkin and colleagues developed the Preordered SSOS (PSSOS) method which selects feasible options from a ‘pre-optimised’ library of HEA structures identified through extensive screening and Density Functional Theory (DFT) structural simulations.
Using a quinary AlCoCrFeNi HEA as a test case, the team reported that PSSOS was more efficient and as accurate in predicting stable HEA compositions as the special quasi-random structures (SQS) method, commonly used in HEA design today. The method also shrunk a design space of around 50,000 feasible SSOS structures into just 1,000.
“PSSOS significantly outperforms SSOS in efficiency, bringing high-fidelity, high-throughput HEA screening closer to reality,” explained Sorkin.
The method can pave the way for developing advanced HEAs in challenging industrial applications such as biomedical materials and extreme-temperature coatings. Sorkin’s own team is currently working on applying PSSOS to streamline the production of lightweight HEAs.
“It’s one of the hottest research topics in metallurgy,” said Sorkin, adding that such next-generation HEAs might one day outperform traditional aluminium and titanium-based lightweight alloys.
The A*STAR-affiliated researchers contributing to this research are from the Institute of High Performance Computing (IHPC).