Using AI to find out what happens in chemical reactions could provide enormous benefit to the aerospace industry.
A team of scientists from the Science Foundation Ireland centre for Advanced Materials and Bioengineering Research (AMBER) and the CRANN Institute is hoping to make the work of many industries in materials science significantly easier.
In a paper published to Science Advances, the team revealed it has developed a new method to model the atomic world using artificial intelligence (AI). This aims to enable fast and efficient ways for scientists to find out what happens within chemical and biochemical reactions.
Not only benefiting those working in the latest research, the new method could also prove very useful to model experiments for the aerospace industry. Currently, it is difficult and costly to identify and test prototype materials that maintain their properties under very high pressure and temperature. A more accurate AI model could act as a precursor to identify better, more robust materials before physical construction and testing.
Existing models can predict what will happen when molecules form covalent bonds, requiring scientists to use the fundamental equation of quantum mechanics: the Schrödinger equation. This generally requires significant computing power and can take a considerable amount of time to complete.
However, this new method uses AI to understand the underlying physics and chemistry associated with a covalent bond. Explaining further, Dr Alessandro Lunghi of the CRANN Institute said: “In a sense, our models learnt the chemistry of the chemical bond just by looking at the reference molecular configurations we provided.”
Lead investigator of the study, Prof Stefano Sanvito, said this breakthrough – and machine learning in general – will lead to further great accomplishments in materials science.
“Using machine learning, which is a branch of AI research, it allows us to simulate any material at the atomic level in a shorter amount of time than traditional methods,” he said.
“We have invented a novel way to systematically construct atomistic models for materials, which are as accurate as the computationally expensive first-principles approach.”