Grant: Discovery Grants Program - Individual

The discovery and deployment of advanced materials are essential to Canada's technological innovation and economic growth. However, traditional approaches to material discovery involve time-consuming synthesis-test-repeat cycles and often take many years to yield new materials. With the fast rise of computing power, computer simulations can now accurately predict materials' behaviours and are being used to lead material discovery at unprecedented speed. Over the next five years, we will develop high-throughput machine-learning schemes for new materials prediction. Building on our strong expertise in high-pressure science, we will develop the new methodology around hydrogen-rich superconductors that can be synthesized under pressure. Superconductors are materials that conduct electricity without resistance below a certain temperature Tc (-135°C or lower). The requirement of low temperature is a major bottleneck towards widespread application of superconductors. Recent experimental breakthroughs have shown that certain binary (two elements) hydrides can superconduct at near room temperature (-23°C) under enormous pressure. Thus, finding near-room-temperature superconductors at lower pressure and, ultimately, atmospheric pressure is a major scientific pursuit. Our research will focus on mapping out the compositional space of complex (three elements or more) hydrides and uncover the most promising superconductors that are stable at low (close to ambient) pressures and have a high chance of being synthesized. We will develop machine-learning models to facilitate accelerated prediction of new materials from scratch (without the need for experimental input). The prediction of crystal structures for targeted materials will be carried out using our in-house software based on evolutionary algorithm interfaced with machine learning potentials. A set of `descriptors', i.e., the fingerprint representation of a superconductor, will be formulated and used to detect superconductors from a large library of predicted structures via a machine learning model. We will also develop machine-learning models to improve the function form of Tc commonly used for conventional superconductors. Once promising materials are predicted, we will pass the results to our experimental collaborators for validation. This research is expected to lead to the ground-breaking discovery of hydride superconductors. The machine learning methods for material prediction and other enabling tools developed in this program can be used to guide the experimental search for new materials. This program will provide excellent research opportunities for highly qualified personnel (HQP) recruited through a strong EDI focus to enhance the participation and achievement of women, Indigenous, and other underrepresented groups. HQP will be trained with top-quality technical and professional skills that prepare them to be future leaders in their specialized fields in Canada and other parts of the world.