When Gerbrand Ceder was studying metallurgy in his native Belgium, it struck him that materials science still had a hit-or-miss quality—like ancient smiths accidentally creating bronze by melding malleable copper and brittle tin. Wouldn’t it be possible, he wondered, to use known properties of the material world to virtually mix and match elements to predict useful new materials no one had yet envisioned?
Ceder, the R. P. Simmons Professor of Materials Science and Engineering, now describes his youthful notion as a bit naive. Yet his vision is almost exactly replicated in the Materials Project, originally called the materials genome because it aimed to do for materials what mapping the human genome did for medicine. Since 2011, Ceder’s freely available online database has allowed engineers and researchers to poke around like kids in a candy shop for as-yet-unimagined new materials to make, for example, a better solar cell, drug delivery system, or battery. “We have calculated the basic properties—crystal structure, strength, conductivity, density, energy, stability, corrosion, and so on—of nearly all of the approximately 35,000 inorganic materials in nature and another few thousand that exist only in theory,” Ceder says.
“It is really up to the creativity of scientists to figure out what new materials can be developed from it. Think of it as a LEGO kit,” he says, from which everything from fuel cells to computer chips can be fashioned.
Ceder’s fledgling idea for a computer-aided database emerged from a 2005 meeting with Procter & Gamble, which wanted to find a better cathode material for alkaline batteries. Would it be possible, executives asked, to computationally screen all known compounds? “In principle, you can compute almost anything,” says Ceder, who earned a PhD in computational materials science. With $1 million in seed funding and access to P&G supercomputers, the Alkaline Project screened 130,000 real and hypothetical compounds and generated a list of candidate materials. Ceder is energized by the fact that hundreds of research professionals consult the Materials Project’s web site every day for ways to bypass the frustrating and inefficient guesswork involved with designing a new material. With access to supercomputing clusters at the Lawrence Berkeley National Laboratory in California, MIT engineers use quantum mechanical models that simulate how materials behave in nature to virtually test thousands of materials at a time. “People have reported that a few minutes on the site saved them weeks or months of hands-on experiments,” Ceder says.
In today’s fast-paced manufacturing environment, timing is critical. Ceder hopes to not only design materials faster, but also to speed them from lab to marketplace. “The faster new materials can be commercialized,” he says, “the faster they can be applied to technologically relevant solutions to the urgent challenges of a warming, increasingly crowded planet.”