From solar panels to batteries, algorithms are becoming key to designing new materials

MidSummer solar panels

In the future, materials that could make a super efficient solar panel or a breakthrough battery probably won’t be discovered by a smart human scientist. Like everything else in this world, computers and software are increasingly identifying the best combination of materials to deliver a desired result, and then human researchers are testing out those computers’ choices in the lab.

For University of Colorado professor Alex Zunger, that idea is a fundamental change in materials research. Zunger is the chief theorist at the Center for Inverse Design, and at the SunShot Summit last week he spoke about how “inverse design” — identifying specific properties that are desired in a material, then determining that material’s required atomic structure — could transform sectors like solar.

Silicon wafers (solar)

For decades, materials for new applications have been selected to be tested “rather casually,” said Zunger, based on “simple ideas,” or even “availability in the lab.” But now, thanks to sophisticated algorithms, scientists can use computer intelligence to make these choices.

Zunger is particularly interested in using inverse design and computer intelligence to figure out the optimal materials to use quantum dots for solar materials. Quantum dots are little pieces of semiconductor crystals — less than 10 nanometers — that are so small they have different properties and characteristics than larger semiconductor pieces. But so far, Zunger says, there hasn’t been an obvious winning combination for solar quantum dots.

Zunger isn’t the only one doing this. It’s actually a hot trend for some of the most cutting-edge materials startups out there.


For example, a startup called Pellion Technologies, which was spun out of MIT, developed advanced algorithms and computer modeling that enabled it to test out 10,000 potential cathode materials to fit with a magnesium anode for a battery. Now the startup is developing a magnesium battery, which could have a very high energy density, and if it works could be important for electric vehicles and grid storage.

A founder of Pellion, MIT professor Gerbrand Ceder, helped develop the Materials Genome Project at MIT, which is a program that uses computer modeling and virtual simulations to deliver innovation in materials. The Economist once described Ceder’s work with the Materials Genome Project as “a short cut” for discovering electrodes and the interactions of inorganic chemical compounds.


Other smart people are also working on this idea. Columbia University’s Institute for Data Sciences and Engineering spearheaded important work in the area, and professors Venkat Venkatasubramanian and Sanat Kumar recently published research on their work designing nanostructured materials with an inverse design framework and genetic algorithms.

While this trend might seem like yet another way that computers are replacing humans, it’s actually an example of ways that computers can leverage massive data sets (that humans can’t) to advance society and make life better — for humans. It’s similar to the way that automated vehicles will make driving more efficient, safer and more productive. Odds are that the material breakthroughs of the future will come from this combination of artificial intelligence and human intelligence.


Comments have been disabled for this post