Research: Engineering News
CIT Materials Scientist Solves a 100-Year-Old Mystery
Old problems are usually hard problems, and those are the type of challenges Katayun Barmak likes.
Barmak is one of a handful of materials scientists in the world to map nanoscale-size polycrystalline structures. Using a technique that took her ten years to implement, she can analyze crystalline materials a thousand times smaller than the diameter of a hair. The applications of this research loom large and affect engineered systems, ranging from silicon chips to fuel cells to medical devices.
"If you want to create nanoscale technology, you have to have tools for measurement. Engineering is quantitative. You need numbers and models." The mapping process is opening avenues for securing much-needed data. "We all say materials behave differently at the nanoscale. But we have to show the nature of the differences and quantify them. This is not easy."
Executing the mapping program is an achievement, but Barmak says, "I am not interested in developing tools. I want to do science." And here is where Barmak has quietly yet soundly shaken up the field of materials science and engineering. She and colleagues from the University of Central Florida have solved a classic problem that has challenged scientists for more than 100 years.
A professor in Carnegie Mellon's MSE department since 1999, Barmak begins her tale of science and determination with a history lesson. In 1897 the British physicist J. J. Thomson discovered the electron. A few years later, he was measuring electrical resistivity, or how materials resist the flow of electric current, when he observed that the thinner his samples, the greater their resistivity. He couldn't explain why resistivity was subject to a size effect. Quantum mechanics hadn't come along yet.
By the 1930s, explanations for Thomson's discovery began to "surface." Scientists learned that electrons scatter when they bounce off the surfaces of materials. This surface effect was deemed responsible for materials resisting electric charges and that was explanation enough. Then in the 1960s, silicon semiconductors emerged. Researchers were wiring the semiconductors with aluminum when they discovered that metals were comprised of polycrystals that had grain boundaries or surfaces between the crystals. A new theory formed: higher resistivity was the result of the electrons getting scattered by grain boundaries.
Interest in resistivity fizzled out again, that is until around 1998. Researchers at Intel reported that the dimensions of the copper wires used in their chips were getting so small that the size effect that Thomson reported was coming into play and it was affecting performance. A great debate soon ensued: Were surfaces or grain boundaries responsible for obstructing the flow of electrical charges?
Answering this question has been a tremendous struggle because identifying and measuring enough grains to provide reliable data is a Herculean task. Yet, amid an uproar of controversy, Barmak and colleagues have definitely proven that grain boundaries are the dominant scattering mechanism that causes higher resistivity.
"We conducted the very first, most careful and quantitative experiments, where we measured thicknesses and roughness of surfaces and grain sizes," says Barmak. Measuring grain sizes is very difficult. Working with pure copper, it took her team many months to measure more than 19,000 grains to obtain 22 samples. "We had to do it by hand. The images are very complex and they are not amenable to automated image analysis. We have tried."
Their painstaking work paid off. They published a major paper, "Surface and Grain-Boundary Scattering in Nanometric Cu Films," in Physical Review B. "This is exciting," says Barmak. "When we started this work, we couldn't get funding. Initially, all of our papers were rejected. For a long time, people have believed that surfaces were the culprit," says Barmak.
Getting experts to let go of long-held beliefs is not easy. A couple of years ago, Barmak was invited to speak at a conference where other researchers said that surfaces were the key to electrical resistivity. They presented three data points as proof of their theory. Barmak's presentation followed. She decreed that grain boundaries were the culprit. She showed the 44 data points in their model that substantiated their theory. "People in the audience were very vocal in telling me I was wrong. But people are now listening because it is hard to argue against very good data," laughs Barmak.
Unraveling Thomson's mystery is important for science and obviously for the semiconductor industry because it will help researchers decide if they should work to improve surfaces or grain boundaries. "Implementing solutions can be very costly, so you better solve the right problem," says Barmak.
What frustrates Barmak is that currently "we can not easily extend this study to others. There is no way that we are going to measure the grains by hand again." But hopefully, technology like the mapping process she has implemented will lead to more advanced metrology (measuring) and characterization tools. Barmak believes now that they've identified the nature and the extent of the nanoscale differences in copper, in theory they can do it with other metals.
"People have to engineer things at the nanoscale, but they don't have a data bank of properties or easy methods for getting the data." This confines the selection of materials that could be used in new technologies. Developing automated ways of obtaining accurate nanoscale information is vital, and Barmak is continuing work in this area. "I develop tools for projects. There is an excitement in revisiting old, difficult problems with new tools. This allows us to do better science," concludes Barmak.
By Sherry Stokes
Share this story: