Jim Chelikowsky keeps a dinner plate-sized wafer of silicon--his favorite material--in his ICES office as a tangible reminder of his 50-year career end-goal.
"More than a quarter of a million scientific papers have been written about silicon," Chelikowsky said, referring to a number produced by an online database when silicon is searched as a keyword. "For almost any theory in condensed matter physics…you can try it on silicon."
But his slice of work on that silicon wafer involves the nearly imperceptible nanomaterials within it, and how they behave. He is currently funded for
more than $4 million in research in the area.
Silicon wafers are the starting point for transistors, the semiconductors that make computer processing possible, and for more than 50 years, computing demands have been driving transistors to smaller and smaller sizes.
As transistors approach nanoscale dimensions, hardware developers have to account for changes in physical properties present in the tiny samples that aren't a factor in the microscopic devices of today. That's where Chelikowsky's work comes in. If that nanomaterial happens to contain "dopants" --impurities introduced into materials to optimize their device performance--a host of new physical properties could come with it.
Chelikowsky, the director of the Center for Computational Materials, has spent a large portion of his career working out just what these properties are at the quantum level.
Chelikowsky doesn't study transistors or other applied uses for silicon. Instead, his work focuses on understanding more about the behavior of silicon at the quantum and atomic level--what its electrons are doing and the properties of the crystalline structures its atoms form. This basic research has helped predict materials and properties before they were discovered in the lab, and help guide engineers toward what forms of silicon might be best for a particular application. It's also an area that relies on high-performance computing to process the complex mathematics used to describe quantum mechanics.
Chelikowsky credits Marvin Cohen, a professor of physics at the University of California Berkeley, with introducing him to how computational methods could be used to investigate material science questions in a realistic way. Cohen advised Chelikowsky's Ph.D. work at UC Berkeley in the 1970s.
"Going to Berkeley and working with Marvin Cohen is probably one of the best decisions I ever made," Chelikowsky said, who earned his bachelor's degree at Kansas State University. "He pioneered work in applying computational approaches to materials, and that's how I got in the field."
When Chelikowsky was in graduate school, he said that the materials science community thought that sound research happened by studying physical materials in a real lab, not by running scenarios on a computer. But Cohen and his research group felt differently and developed computational methods for studying in great detail how electrons interact at the interface of a material, using a technique known as the "empirical pseudopositional method." Cohen said that Chelikowsky's thesis work applying this technique to silicon semiconductors was a starting point for a rich array of research.
"He ended up publishing 25 papers based on his thesis work," Cohen wrote in an email. "This is highly unusual and may be a Berkeley record."
Cohen also noted Chelikowsky's manner of providing clear concise answers to problems, whether in lab discussions or in print.
"His office was piled high with computer paper and messy, but on top of the mess was a neat summary and the most important results," Cohen said.
After earning his Ph.D., Chelikowsky applied his skills at research positions in industry, working at both Bell labs and Exxon Mobil conducting basic research on semiconductors. After seven years at Exxon, he returned to academia in 1987, accepting a position as a professor in the Chemical Engineering and Materials Science Department at the University of Minnesota.
At that time, the National Science Foundation was very interested in funding computational science research and was soliciting research proposals for computational work. Chelikowsky answered the call by teaming up with Yousef Saad, a professor of computer science at the University of Minnesota, to develop software for running materials science research on parallel computers. One of their research successes was PARSEC--now a well-known software package for intensive quantum calculations.
"I enjoyed interacting with Yousef," Chelikowsky said. "I gave him lectures on quantum mechanics and the types of equations we needed to solve, and he gave me lectures on hardware and software, and taught me a lot about the types of algorithms we need to use."
At ICES, Chelikowsky is continuing his work on materials science, a field that has grown in size and scope as advances in silicon transistors have enabled faster and more efficient processing. Chelikowsky has built a career leveraging these advances while also conducting research that could help build the materials that could drive the high performance computers of tomorrow. His multi-disciplinary work has earned him positions as professor in three departments: physics, chemical engineering, chemistry and biochemistry.
"The community is calculating hundreds, if not thousands, of material properties each day," Chelikowsky said, commenting on the state of materials research today. "And these computations are just a drop in the bucket compared to the numerous possibilities."
To honor his professional achievements, ICES and his international colleagues celebrated Chelikowsky's 70th birthday with a conference June 2 entitled "Practical Quantum Mechanics for Electronic Materials."
--By Monica Kortsha