Mapping the World at a Molecular Level

“Chemistry is a bottom-up discipline. We explore what things are made of and how reality works. Materials and tools that people use every day—a sheet of steel or a nylon rope, for example—result from a complex arrangement of electrons and protons,” says Troy Van Voorhis, head of the Department of Chemistry, the Robert T. Haslam and Bradley Dewey Professor of Chemistry, and principal investigator in the Van Voorhis Group at MIT. Chemists are interested in the building process—how atoms and molecules come together to determine the properties of that sheet of steel or nylon rope—because understanding that process is essential to developing hypotheses about how new materials might be designed.

However, human brains don’t easily grasp this building process. These systems are not linear; it’s a time-consuming and tedious endeavor for humans to test and map these complex arrangements, and once the research gets down to an extremely small scale, experiments cannot access or manipulate the particles without modifying their interactions. “But these are ideal tasks for computers,” Van Voorhis says. “They are perfectly suited to automation.”

This is the promise of computational chemistry. “If we start with the idea that everything begins with atoms and molecules,” Van Voorhis says, “then we can teach very powerful computers to map the multiple reactions that define the building process, down to the most minute level. We can then access that information and translate it into predictions about how new materials might behave.” Expediting this process of discovery would not only allow scientists to efficiently improve existing materials and develop new technologies, but it would also help to better predict probable outcomes, like the efficacy of a new drug.

The potential of artificial intelligence as it relates to chemistry is vast, according to Van Voorhis, impacting virtually every aspect of our lives—the tools we use, the way we power our homes and businesses, the speed and breadth of our communication systems, human and climate health, and the quality of our food and water supply. Van Voorhis eagerly anticipates the scientific advancements that may be realized through the use of computational chemistry enhanced by artificial intelligence. “The best way to improve something is to first figure out how it works.”

This story was originally published in February 2020.

Computing School of Science