Guest Post by Ken Kingery, Pratt School of Engineering

If you think ordering a drink from Starbucks can be a tall order, try picking out the right material for a new product or experiment. An iced, half-caff, four-pump, sugar-free, venti cinnamon dolce soy skinny latte may be a mouthful, but there are hundreds of thousands of known—and unknown—compounds to choose from, each with their own set of characteristics.

Screenshot of the AFLOW library search interface. The software combs through four giant databases of chemical compounds.

Screenshot of the AFLOW library search interface. The software combs through four giant databases of chemical compounds.

For example, take the family of materials Bi2Sr2Can-1CunO2n+4+x, or bismuth strontium calcium copper oxide for short, if you can call it that. These compounds have the potential to change the world because of their ability to become superconducting at relatively high temperatures. But each sibling, cousin or distant relative has its own physical variations…so how to choose which to pursue?

Thanks to Stefano Curtarolo and his research group, now there’s an app for that.
Curtarolo leads a collection of research groups from seven universities specializing in something they call materials genomics. The group—called the AFLOW consortium—uses supercomputers to comb databases for similar structures and builds theoretical models atom-by-atom to predict how they might behave.

With help from several members of the consortium, Pratt School of Engineering postdocs Cormac Toher, Jose Javier Plata Ramos and Frisco Rose, along with Duke student Harvey Shi, have spent the past few months building a system that combs through four materials databases. Users can choose the elements and characteristics they want a material to have—or not to have—and the website will play matchmaker.

Want a two-element compound containing either silicon or germanium—but not gallium—that is stable enough to withstand high temperatures? Not a problem. How about an electric insulator made from transition metals with a certain crystal structure? AFLOW has you covered.

Stefano Curtarolo

Stefano Curtarolo is a professor of mechanical engineering and materials science at Duke

One of the four databases searched by the program draws from an international collection of compounds with structures known from experimentation. The other three contain single- double- or triple-element compounds, and are not limited to previously explored materials. Through their molecule-building algorithms, the AFLOW consortium is constantly adding prospective materials to these four libraries.

The search engine currently can sort through more than 622,000 known and unknown compounds, and more than 1,000 new ones are added each week. Curtarolo, a professor of materials science and physics, hopes that the open-source program will continue to grow in materials and searchable characteristics to help scientists connect to their ideal material. To see how it works for yourself, take it for a test drive here.

As for an equivalent database of coffee drinks, that has yet to be built. So if you’re looking through AFLOW for a hot, lactose-free drink featuring 150 to 200 mg of caffeine and less than 200 calories made with beans from South America, you’re out of luck.