Navigation Content
University of Wisconsin Madison College of Engineering
You are here:
  1. Home > 
  2. News > 
  3. News archive > 
  4. 2013 > 
  5. New tools accelerate computational materials research

New tools accelerate computational materials research

Drawing on scripting languages that are used in applications ranging from video game development to enhancing functionality on a web page, a team of University of Wisconsin-Madison researchers is developing tools that automate and accelerate research for materials design.

Led by Dane Morgan, a UW-Madison associate professor of materials science and engineering, the group received a five-year, $2.05 million grant from the National Science Foundation to fund its efforts.

The grant will help the team create novel tools for high-throughput first-principles—or fundamental—material modeling. First-principles methods, which over the past decade have become increasingly more stable and robust, solve the fundamental quantum mechanical equations of matter to accurately calculate the properties of a given material so that a researcher doesn't need to physically measure that material. "Researchers have automated the core aspects of first-principles computation so humans don't have to solve all of the differential equations—but a lot is still left to the user," says Morgan.

Now, the next step is to automate the all of the steps researchers need to calculate the properties of many different systems, and that's where scripts come in. Scripts are a series of electronic instructions. On the web, for example, they create interactive, dynamic pages that respond to the ways in which a user interacts with the page.

Morgan's group is using scripts to develop tools that operate on "top" of the existing first-principles codes to help automate the entire computational materials research process. "Instead of having a graduate student set up the file, copy and paste it into a spreadsheet, compare the data to previous results, and do all the steps iteratively, we take every one of the steps and write them as computer scripts," he says. "If you can take the human being out of it, you can accelerate the number of calculations you can perform by orders of magnitude, opening up new opportunities for developing materials." 

That's not to say Morgan is removing humans entirely from materials research. Rather, the group is capitalizing on lightning-fast computation—the thing computers do best—to increase the speed at which people learn about materials. "It's like Star Trek," he says. "You walk up to the computer and ask it for something and it gives you the things that meet your criteria."

Morgan is applying the automation to study point defect chemistry, which plays a key role in semiconductors and energy technologies, and to diffusion in materials, or the ways in which atoms move around in materials, which he says is a critical area of data need. For example, the group can calculate an element's diffusivity in an alloy—information important in everything from understanding welding materials to materials corrosion and radiation effects to designing ionic conductors for batteries and fuel cells. "Previously, it was a big effort for a student to do this," he says. "Because we've automated it, we recently calculated the diffusion energetics of essentially all the pure elements in the Periodic Table in some of the most common structures. Now, we're doing every element 'A' in every possible element 'B.' With the tools we are developing, plus modern computers, high-throughput computations could in five years generate more diffusion data than has been measured in alloy systems over the last century."

Within a decade, Morgan hopes his and other groups can develop and share via the Internet an enormous amount of accurate information that can serve as a foundation for improved database-grounded materials design and analysis—kind of like a mega search engine for materials science. And while he is applying his tools to point defect chemistry and diffusion in materials, Morgan's goal is to make his tools easy to use and accessible so that other researchers can apply them to their own fields of study. In fact, he is collaborating with the Materials Project group at Lawrence-Berkeley National Laboratory and will provide his tools as part of its growing materials design environment.

His team includes undergraduates, graduate students and postdoctoral researchers from engineering and computer sciences at UW-Madison, as well as researchers from Lawrence Berkeley National Laboratory, Massachusetts Institute of Technology, and the University of Kentucky.

Renee Meiller