Using quantum atomistic computer simulations to solve industrial problems

PASADENA—In the world of engineering and applied science, ideas that look good on the drawing board often turn out to have annoying real-world problems, even though the finished products still look pretty good. An example is the aluminum car engine, which has the advantage of being lightweight, but tends to wear out more quickly than its heavier steel counterpart.

To solve such bedeviling problems, experts often find it necessary to go back to "first principles," which in the case of the aluminum engine may include a computer simulation of how the individual atoms slide around under wear and tear.

California Institute of Technology chemistry professor Bill Goddard had this type of problem in mind when he established a special center a decade ago within the campus's Beckman Institute. Christened the Materials and Process Simulation Center (MSC), Goddard's group set as their goals the development of computer simulation tools necessary to deal with materials and process issues, and the transfer of solutions to government and industry for the creation and improvement of products.

"We started the center to follow the dream of being able to predict chemical, biological, and materials processes with a computer," says Goddard. "The idea was to get a simulation that was close enough so that you wouldn't have to do the experiment."

Now that the MSC is celebrating its 10th anniversary, Goddard says the group has made some genuine progress on a number of real industrial problems—much to the satisfaction of corporate collaborators and sponsors, which at present are underwriting about 10 new projects each year.

In addition, the conference celebrates the 100th birthday of Arnold Beckman, the founder of Beckman Instruments and the benefactor of the Beckman Institute.

Since technology transfer and real-world results are a high priority, Goddard and his colleagues sponsor an annual meeting in which the collaborators showcase all their activities. This year's meeting, to be held March 23–24 at the Beckman Institute on campus, is also the 10th anniversary celebration of the center itself.

"There are several new accomplishments we'll discuss at this year's meeting," Goddard says. "We've had the first prediction of the structure of a membrane-bound protein, we've shown how to grow a new class of semiconductors to make real-world devices, and with our local collaborator Avery Dennison we've had success in predicting gas diffusion polymers.

"The bottom line is that it has worked out," he says. "In this center we have probably the most complete group of theorists in the world—about 40 people—and we've continued to have a flow of excellent grad students and postdocs who have gone on to be leaders in their fields."

A unique feature of the MSC is its emphasis in starting out with first principles, using quantum mechanics (the Schrödinger equation) to describe what is happening between atoms. For example, if the real-world problem is how best to lubricate a certain type of moving part (which is an actual industrially funded project the center has worked on), then the researchers would use the Schrodinger wave equation to build a simulation to show precisely how the electrons of a certain lubricant would interact with other electrons, how variable factors such as temperature and pressure would enter into the picture, and how a host of other interactions at the atomic level would play out.

But the quantum level is only the first in a hierarchy of regimes the center researchers might use in investigating complex problems. The quantum level with its Schrödinger equation is good for a system of about 100 atoms, but currently no computer can use quantum mechanics to predict the structure of hemoglobin, the protein that carries oxygen to our muscles.

Rather, for systems with up to about a million atoms, the center uses molecular dynamics techniques, essentially solving Newtonian equations.

For the billion or so atoms or particles that compose a "segment" of material, the MSC investigators employ the techniques of coarse-grain meso-scale modeling and tools such as phase diagrams. Beyond this point, for process simulation, materials applications, and engineering design involving the entire object, the center has developed yet another set of techniques.

This hierarchy of materials modeling is not describable merely by the number or size scale of particles. Time scales are also involved, with quantum mechanics operating at the femtosecond scale (a millionth of a billionth of a second), molecular dynamics at the nanosecond scale (a billionth of a second), coarse-grain meso-scale modeling at the millisecond scale, process simulation at the scale of minutes, and engineering design over periods ranging up to years.

Finally, the hierarchy has many crossover points, which particularly allow the center's research to be innovative and interdisciplinary.

"So you start with fundamentals of quantum mechanics, and imbed this in the next steps at all length scales and time scales," Goddard says. "The idea is to figure out why these things happen, and how looking at first principles can solve industrial problems."