Simulation sheds light on the very small details of battery life, enhances understanding of fuel cells and catalysis

  • April 18, 2013
  • Science Highlights

Given the myriad uses we put lithium-ion batteries to — from the smartphones in our pockets to the new Boeing 787 Dreamliner passenger aircraft — how much we don’t know about the batteries' inner workings, especially at the molecular or atomic scales, is a little surprising.

Take the interfaces between different phases in a battery, like the interface between the solid electrodes and the liquid electrolyte that passes lithium ions between the electrodes, which is part of the process of generating electricity.

The electrode-electrolyte interface affects battery safety as well as performance, says Jeffrey Greeley, a Purdue associate professor of chemical engineering. Lithium-ion batteries can burn or explode, as the occasional news story about a fried laptop or Boeing’s 787 battery problem, which grounded the new airliner for a time, illustrate.

One reason such occurrences are news rather than routine is a protective layer that spontaneously forms where electrode and electrolyte meet. "That naturally occurring process works pretty well, but it's very poorly understood," Greeley says. "We think it could be made to work much better if you could find a more systematic way to first understand and then control it."

Greeley's lab is working to advance understanding of that interface and of other atomic-scale interfaces important in batteries, in fuel cells, and in heterogeneous catalysts, which are used for purposes ranging from filtering automobile pollutants to manufacturing materials like liquid fuels and plastics. Purdue’s community cluster supercomputers are an important tool for the research.

"What we love to do is calculate things that can't be measured experimentally, or at least where the experiments would be difficult and time consuming," Greeley says. "All of the interfaces in which we're interested have atomic-scale features. Characterizing what these interfaces look like is experimentally a significant challenge. But we can look, using various simulation techniques, at the behaviors of individual atoms or groups of atoms."

Among other things, such simulations could advance the use of silicon as a next-generation material for battery electrodes. Silicon can absorb a lot of lithium without much mass, which might make for a smaller, lighter battery that’s nonetheless more powerful and lasts longer. However, silicon also degrades considerably as a battery charges and recharges, one reason a working molecular-scale model of a silicon lithium-ion battery hasn't yet been developed.

"What we've been doing is trying to study some of the atomic-scale details of how lithium moves into the silicon surface, how it transforms the structure of the silicon, how it deforms, expands, pulverizes and then, ultimately, what the molecular properties are that we could tune to try to improve the performance," Greeley says.

Greeley's lab also works to simulate and optimize the properties of various classes of heterogeneous catalysts. Such catalysts are of critical importance for industrial processes ranging from petrochemicals processing to biomass reforming. A recent focus of this work has been the production of hydrogen from biomass; simulating this process, which involves large numbers of complex biomolecules, requires supercomputing.

Hydrogen is also important for many types of fuel cells, which are attractive as a cleaner energy source, emitting nothing but water in the case of the hydrogen-powered variety, for instance. Greeley’s work in this area involves the search for effective, but less expensive, alternatives to platinum, the catalyst commonly used in the devices now.

Greeley's simulations are based on mathematics that require lots of computational cycles to solve. "The more atoms you're simulating, the more expensive the calculations generally become. One of our goals is to move the simulations in the direction of larger systems, which in some cases can give us more realistic models of catalysts and batteries," he says.

In addition, time scales are a factor. They’re currently limited to hundreds of picoseconds for things like figuring the trajectories of molecules. The research also involves doing calculations on numerous targets in screening for new materials with desired properties — in essence a giant search problem.

The capability of Purdue’s Carter cluster supercomputer, a competitive price to buy into the cluster and the support of ITaP Research Computing (RCAC) staff sold Greeley on the Community Cluster Program.

"Knowing there was a group of experienced professionals I could rely on for support in establishing the computational infrastructure I needed, that was very comforting," says Greeley, who joined Purdue's faculty in January 2013. "You don't have to spend months trying to get your cluster running. You don't have to spend a lot of graduate student or post-doc time to maintain or expand it, either. We can focus more on the scientific problems, which are our primary interest."

More Information

Originally posted: July 1, 2014 4:23pm EDT