If you live on the equator, you may rely on the ubiquitous air conditioner to get relief from the oppressive heat and humidity. These devices can perform their important functions of heat harvesting and refrigeration thanks to a family of thermoelectric materials called bismuth telluride (Bi2Te3) alloys.
For decades, materials scientists have been trying to recreate their blockbuster success with bismuth telluride by searching for novel thermoelectric materials from over half a million inorganic compounds. They have made very little traction, despite the potential upside for industries as wide-ranging as cold storage, computing, automotive and aerospace.
“Finding new thermoelectric materials is time consuming, inefficient and expensive. Thus, limited progress has been made in the discovery of high-performance candidates, which is a bottleneck for thermoelectric technology,” explained Shuo-Wang Yang, a Senior Scientist at A*STAR’s Institute of High Performance Computing (IHPC) and a co-corresponding author on the study.
A new computational method developed by Yang and colleagues—called Energy-dependent Phonon- and Impurity-limited Carrier Scattering Time AppRoximation, or EPIC STAR—can identify promising high-performance thermoelectric materials in less than one-tenth of the time taken by other methods.
“Our benchmark shows that, compared with the widely used electron-phonon Wannier (EPW) interpolation method, EPIC STAR can provide results with the same accuracy in less than one-tenth of the computational time,” Yang said.
Instead of a trial-and-error, brute-force approach to the discovery of thermoelectric materials, EPIC STAR bypasses the challenges posed from the calculations of electron-phonon coupling, a time-consuming but essential calculation that is necessary to predict the thermoelectric performance of a candidate material.
To speed up their calculations, the researchers simplified the numerical integrations that are conducted in the Brillouin zones, which explain the behavior of an electron in a perfect crystal but become increasingly complex at every stage.
Yang and his team have used EPIC STAR to identify a low-cost, high-performance thermoelectric candidate, NaInSe2, which could be experimentally tested by industry partners for thermoelectric and electronic applications.
“The composing elements of NaInSe2 are more abundant than bismuth- and tellurium-based materials. Therefore, NaInSe2 and its analogs may have useful applications as a new low-cost, high-performance thermoelectric material,” Yang said.
To facilitate the screening of thousands of inorganic materials in a high-throughput manner, the team is leveraging on computational resources at the A*STAR Computational Resource Centre (ACRC) and the National Supercomputing Centre Singapore (NSCC).
“We expect to find new high-performance candidates and design principles that could enable breakthroughs in the field,” Yang shared. “The data we collect will be used to build an open-source database for machine learning-aided materials design.”
The A*STAR-affiliated researchers contributing to this research are from the Institute of High Performance Computing (IHPC) and the Institute of Materials Research and Engineering (IMRE).