The first thing you notice about Gerbrand Ceder’s materials science lab at MIT is that there are no crucibles, no furnaces, no crystal-growing instruments. Instead, you find a row of high-resolution computer displays with grad students and postdocs tweaking code and constructing colorful 3-D images. It’s in this room, quiet except for the hum of fans cooling the computer power, where new high-tech ceramics and electronic materials that have never been seen or made before are being forged. They are taking form “in virtuo”-designed from scratch on the computer, distilled out of the basic laws of physics.
The next thing you’re likely to notice is how young Ceder is. Quick to laugh but intensely passionate in explaining his work, the 33-year-old associate professor is one of a new breed of materials researchers, trained in traditional processing techniques, who have turned to discovering materials using computers. The dream is simple: Replace the age-old practice of finding new substances by trial and error, with calculations based on the laws of quantum mechanics that predict the properties of materials before you make them.
You can, in theory at least, design metals, semiconductors and ceramics atom by atom, adjusting the structure as you go to achieve desired effects. That should make it possible to come up with, say, a new composition for an electronic material much faster. Even more important, tinkering with atomic structure on a computer makes it possible to invent classes of materials that defy the instincts of the trial-and-error traditionalists.
It’s an idea that has been kicking around for at least a decade. But with the explosion in accessible computer power, as well as the development of better software and theories, it’s becoming a reality. Last year, Ceder and his collaborators at MIT synthesized one of the first materials that had actually been predicted on a computer before it existed. This new aluminum oxide is a cheap and efficient electrode for batteries. And while it may or may not lead to a better, lighter rechargeable battery, the success of Ceder’s group-and related work at a handful of other labs-is proving that useful materials can be designed from the basic laws of physics.
Designing from first principles represents a whole new way of doing materials science, a discipline that Ceder describes as “a collection of facts with some brilliant insights thrown in.” It’s a transformation he’s been aiming at since his undergraduate days in the late 1980s at Universit Catholique de Louvain in Belgium. “My background is heat and beat metallurgy,” he explains. “But I always thought there should be more to it, some way to calculate things using all the great physics of quantum mechanics.”
Getting there, however, won’t be easy. Scientists have known for decades that, according to the rules of quantum mechanics, if you could detail the position of the electrons swarming around atoms, you could then calculate physical properties of the material. Yet the sheer difficulty of carrying out these calculations has made the task seem hopeless. The computations are hard for even one molecule, but for the huge numbers of atoms that make up even the smallest chunk of a solid material, the chore is truly intimidating.
In the search for new vaccines and drugs, where computer-aided design has taken off, progress has been achieved precisely because the designers have been able to skip the influence of particular electrons, along with the rigorous quantum calculations. But inorganic materials are tougher. The properties of metals, alloys, semiconductors and oxides result from a vast sea of interconnected atoms and electrons. “With metals and ceramics, we really need to include the electrons as active players,” explains Erich Wimmer of Molecular Simulation, a software company that markets molecular and materials computer modeling programs. “And that means we need quantum mechanical methods.”
The linchpin of quantum mechanics is the Schrdinger Equation, which describes how electrons arrange themselves around atoms and how atoms share electrons to form chemical bonds. The Schrdinger Equation generates a “wave function” giving the probability that an electron will be at a given location at a given time. What makes it so powerful is that the wave function can reveal physical properties of the system: energy, optical absorption, conductivity. If done right, you insert the atomic masses and crystal structure, and out pops the physical properties of the material. The calculations are called “first principles” or “ab initio” because you start with the most fundamental information about the atoms and use the most basic rules of physics. The price researchers pay for this ability is that the Schrdinger Equation requires immense computer power to solve, even for simple atomic structures.
The Right Stuff
Eight years ago, ceder showed up at MIT (he received his PhD from the University of California, Berkeley, in 1991) as a newly minted professor ready to give the Schrdinger Equation a try on one of materials science’s most pressing problems: better batteries. Today’s frenzy for cell phones and laptops has driven the quest for lighter and more powerful storage materials. Lithium cobalt oxide is the electrode of choice for lightweight high-power applications, particularly products like cell phones. But lithium is expensive-and cobalt is even pricier. The costs can be passed on in high-tech gadgets like cell phones and laptops but for other uses, such as powering electric cars, it’s prohibitive.
Several of Ceder’s MIT colleagues began working on an improved battery. But it was obvious they needed better, cheaper oxide materials to serve as the electrodes. The traditional materials science strategy would have called for mixing up a batch of an oxide and then adding and subtracting components a little at a time. But there are almost unlimited combinations of ingredients you can put in-and every ingredient affects every other ingredient. “Every time you make a chemical change, lots of other things get altered too,” explains Ceder. “There may be structural changes, and who knows what else.”
Ceder attacked the problem by creating the samples and testing them not on the bench, but on the computer. “The advantage is that you have full control over what you do,” he says. “If I take a crystal, and add a little bit of some element, I can see exactly how the electrical conductivity will change.” Systematically changing the electrode composition, Ceder used his software code to calculate the effects on battery voltage. By replacing the cobalt with titanium, or vanadium, could they get a peppier energy cell?
What they found was surprising. The voltage didn’t depend strongly on the cobalt. In fact, the voltage was highest if all the cobalt were replaced with aluminum. This was entirely unexpected, Ceder says, because aluminum had been thought to be a nonplayer in battery oxides. But there was a hitch. Lithium aluminum oxide is an insulator. So even though the numbers showed the voltage would be at a peak, there would be no way to run a current through it. Calculations indicated, however, that a mixture of cobalt and aluminum might just do the job: enough cobalt to keep the electrode conductive, and aluminum to replace the rest.
That, at least, is what the computer showed. Someone still had to make the stuff. Enlisting the help of MIT ceramists, Ceder and his colleagues developed methods to synthesize the predicted cobalt-aluminum electrodes-materials that had never been made before. It turns out that, in fact, the mixture of cobalt/aluminum gives a higher battery voltage than cobalt alone. At the same time, the aluminum lowered the overall weight of the material, so that the energy density-another important figure for a good battery-went up, and the projected cost of the material went down.
Don’t expect to find cobalt-aluminum electrodes in a battery anytime soon, however. Although the new battery material is superior in voltage and costs to conventional electrodes, the conservatism of the power-storage business means that it’s not likely to be commercialized in the near future. Still, the success of Ceder’s group marks a significant advance because it means a material has been designed on the computer and found actually to have the predicted qualities.
Scientists are far from being able to plunk themselves down at a keyboard, tap in some desired properties, and have a new substance pop out. Ceder points out that electronic and optical properties are relatively amenable to calculation, whereas other important characteristics, like hardness or corrosion resistance, are more difficult to compute. Those properties, for one thing, depend on events occurring over a wide range of size and time scales (see sidebar: “Cracking a Tough Problem”). “There are still a lot of problems you can’t address by these methods,” he says.
The dark secret behind those cute AI-generated animal images
Google Brain has revealed its own image-making AI, called Imagen. But don't expect to see anything that isn't wholesome.
The hype around DeepMind’s new AI model misses what’s actually cool about it
Some worry that the chatter about these tools is doing the whole field a disservice.
The walls are closing in on Clearview AI
The controversial face recognition company was just fined $10 million for scraping UK faces from the web. That might not be the end of it.
This horse-riding astronaut is a milestone in AI’s journey to make sense of the world
OpenAI’s latest picture-making AI is amazing—but raises questions about what we mean by intelligence.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.