Powering the Dream_ The History and Promise of Green Technology - Alexis Madrigal [135]
Beyond photovoltaics, batteries and energy-scavenging devices to convert heat into electricity would both benefit from better materials. Wind turbines could get stronger and lighter. Building materials like cement could become less carbon intensive.23 To say materials science holds the potential for transformative change across the whole greentech spectrum is no exaggeration.
But there’s a problem.
Materials science remains substantially a trial-and-error bench science. If we want a material with certain properties—say, a much better battery—then we have to just keep trying different synthesis techniques to come up with new options. We test them to see what worked and then keep iterating on that process. Getting a material from the drawing board out into the field takes decades. Even once we have a material, creating the manufacturing process to make a lot of it involves years of trial and error.24 In 2008 Gerbrand Ceder, an MIT materials science professor, and other researchers wrote,
Searching for novel materials is typically a rather random and therefore somewhat unpredictable process. The initial focus on new materials is usually based on a single outstanding property. Hence, attempts towards scale-up and commercialization start without access to other key properties which affect whether a material can be commercialized: reliability, stability and degradation, processability, etc. Deficiencies of a material in any of these areas can throw a materials development program off track.25
In other words, materials scientists find a material that’s good at converting sunlight into electricity and start trying to make more of it. When they do, they discover all sorts of little quirks. As the years go by the once new, good-looking material starts to seem more like a drunk, unstable husband. If only scientists were able to predict which materials would be a good match over the long haul.
Ceder thinks he can do just that. The idea came to him in the waning days of the dot-com boom. He was doing some sidework with a colleague for a startup that used algorithms to predict consumer movie preferences—not unlike Amazon’s famous recommendation engine. After a few months he returned to his normal work of thinking about compounds and their various properties like strength, conductivity, and the ability to store electrons. When he did, suddenly all the meaningless work for the Internet startup took on new significance. “I remember sitting down and going, ‘Oh my god, this is just like movies: The people are the compound and the movies are the properties,’” Ceder recalled. “I really got that wacky idea there at the startup. It’s the only good thing that I got out of it. I think they sent me a hat, too.”
The idea, developed with Dane Morgan, now a materials scientist at the University of Wisconsin Madison, grew into what Ceder calls the Materials Genome Project. It is no less than an attempt to solve the quantum equations for nearly every inorganic substance that humans can produce. There are something like fifty thousand inorganics out there, Ceder estimates, and maybe another fifty-thousand left to be discovered. Over the last couple of years and just in Ceder’s lab with half a million dollars of computers, they have quantified the properties of thirty-one thousand materials. “We already have enormous coverage of the known universe of materials,” Ceder says. “We’re showing that this whole discovery process is about to be accelerated