Basel – Scientists at the University of Basel have demonstrated the potential of artificial intelligence: it enabled them to identify crystals that can be regarded as new materials. This discovery would not have been made were it not for artificial intelligence.
The focus of the University of Basel research work is on the mineral elpasolite. According to a statement, it can exhibit four properties depending on its composition: it can be a metallic conductor, a semi-conductor or an isolator and may even emit light when exposed to radiation. Its versatility means that it can conceivably be used in a range of applications, and yet it is practically impossible, mathematically speaking, to calculate every theoretically viable combination of the four elements in its structure.
Felix Faber, a doctoral student at the Department of Chemistry, turned to modern artificial intelligence to solve this problem. First, he used quantum mechanics to generate predictions. He then used the results to train statistical machine learning (ML) models. Within just one day, the ML model was able to predict the formation energy of two million elpasolite crystals. In contrast, a calculation by quantum mechanical means would have taken a supercomputer more than 20 million hours, explains the university.
Thanks to this analysis, the researchers were able to identify 90 previously unknown crystals that should be thermodynamically stable. As a result of Basel discovery, elpasolite has now been entered in the Materials Project database. It plays a key role in the Materials Genome Initiative, which was launched to accelerate the discovery and experimental synthesis of new materials.
“The combination of artificial intelligence, big data, quantum mechanics and supercomputing opens up promising new avenues for deepening our understanding of materials and discovering new ones that we would not consider if we relied solely on human intuition,” said study director Anatole von Lilienfeld.