When researchers want to create better batteries, solar cells, and medical devices, they often look for answers in new materials. Materials with optimal properties can improve existing technologies and spark ideas for new ones. But finding materials that have just the right properties can take many years of trial and error.
“Suppose you want to find a material that would make a good solar cell, but you don’t have a design strategy,” said Christopher Wolverton, professor of materials science and engineering. “You would have to explore in the dark.”
Wolverton’s group in Northwestern’s McCormick School of Engineering and Applied Science has created a database that takes some of the guesswork out of designing new materials. The team performed systematic analyses of both known and imagined chemical compounds to find their key properties and established a database of the results. Called the Open Quantum Materials Database (OQMD), it launched in November and is the largest database in the world of its kind.
So far, the OQMD contains analyses of 285,780 compounds and continues to grow. Northwestern’s high-performance computer cluster, Quest, was used to construct most of the database, which is available and can be downloaded online. The original paper about the project, “Materials Design and Discovery with High-Throughput Density Functional Theory: The Open Quantum Materials Database,” was featured in the November 2013 issue of the journal JOM.
The purpose of the OQMD is to identify candidate materials for specific applications by screening them for various properties before they are tested in the lab. This dramatically accelerates the search, narrowing down candidates for possible materials to a mere handful that require further experimentation.
“The calculations are faster and easier with less cost than conducting experiments,” Wolverton said. “And it’s all on computers, so users can explore things—like toxic elements and radioactive elements—that they probably wouldn’t want to do in their labs.
The OQMD allows users to search for materials by composition, create phase diagrams, determine ground state compositions, and visualize crystal structures. Wolverton said his group has also implemented machine-learning models, trained on the database, that can learn chemistry and predict the possible existence of new compounds that have not yet been synthesized.
“Using sophisticated data mining, we could turn materials science into a big data problem,” he said. “We could use algorithms to make recommendations for materials the same way Netflix recommends movies you might like.”
Unlike other similar databases, the OQMD is completely open to the public. Wolverton said closed databases can only be used the way its creators intended. By keeping the database open, more people can use it, adding their own compounds and growing its potential.
“People will use the database in ways that we couldn’t possibly imagine right now,” he said. “People will improve it, change it, and use it in different ways. They will search for applications of materials that my group isn’t interested in, and that’s great. They will get value out of it that we never would have.”
Source: Northwestern Univ.
