Melissa Hines is a researcher in search of perfection. Her goal is a mirror surface on which not even a single atom is protruding above the surface.

"There is no theoretical reason why you can't make things that are perfect," says Hines, an assistant professor of chemistry at Cornell University. "It was once thought there were no mechanisms for perfection." But within the next five years she expects researchers to be able to produce silicon surfaces that "are essentially totally flat."

Hines described her work in understanding perfection at the annual meeting of the American Physical Society in Los Angeles today (March 18). On April 1 she and her Cornell colleagues will address scientists at the annual national meeting of the American Chemical Society in Dallas.

Hines' research, which she began as a postdoctoral student at Bell Labs, is of great economic importance to the semiconductor industry because surface roughness, even on the atomic scale, can greatly decrease the performance of a transistor. "As we go down to smaller and smaller devices, roughness becomes a larger and larger problem," Hines says .

The possibility of surface perfection was serendipitously discovered about five years ago when Bell Labs researchers sought a new method of removing dust from the silicon wafers used to produce integrated circuits. The old method, developed in the 1960s, involves washing the silicon wafers in basic peroxide baths. But today's much smaller circuitry develops atomic-scale roughness from the chemical, significantly reducing the transistor's performance.

But by changing the acidity and composition of the chemical solution the researchers discovered they were able to produce small areas on the silicon surface that were totally flat, even at the atomic level. In fact the surface roughness was equal to only one protruding atom out of every 30,000 surface atoms.

However, this perfection is only reproducible on one type of silicon surface, called silicon (111), which is a different plane from the silicon (100) used for integrated circuits. Thus, says Hines, the goal of research is to find chemical solutions that will produce perfection on different surfaces. To do this, she says, it must first be understood how the chemicals used in her research, a basic hydrofluoric acid solution, etch away protruding atoms. "At this point we know what is going on," she says. "Next we have to change the chemistry to control the reactions. I'm completely convinced this is possible."

The most perfect surface Hines and her colleagues have achieved to date appears through the electron tunneling microscope as a series of steps, with every step only a single atom high. The steps are the result of almost imperceptible errors in cutting the silicon wafer. Because of the chemical action, each step is evenly spaced and almost straight.

Another dramatic example of surface chemistry is the production of equilateral triangles. In this case, the chemicals appear to burrow into small defects on the silicon surface, each a few atoms across, and then open the defects out into triangles about 1,000 atoms across. The bottom of each triangle is perfectly flat. "This had us confused for a very long time," says Hines. "It turns out there is an atomic defect in the crystal that is very reactive. When etched, the atomic structure becomes triangular."

Hines marvels at the chemical reactions that produce both the flat surfaces and the triangles. In both cases, the chemicals etch away surface atoms, one atom at a time, in a very precise order. She calls the process "unzipping," because neighboring atoms are etched in a sequential fashion in much the same way that teeth in a zipper are sequentially opened. It is this type of reaction that Hines is seeking to control in her quest for perfect surfaces.

The technique has many uses, she says. In addition to integrated circuit technology, the chemistry would be useful in micromachining of very small parts in which nanoscale control of manufacturing is essential. These chemistries, Hines says, could be applied not only to etching patterns in material but also to applying thin films.

"The nice thing about chemistry is that it does all this automatically," she says. "It's not as if you had to build a machine that removes one atom at a time. The chemistry has this built in."

The title of Hines' talk at the American Physical Society is "Towards chemical control of surface morphology: Aqueous etching of silicon." The title of her talk at the American Chemical Society is "The unexpected role of etchant diffusion in autocatalytic etching of Si(111)." Her collaborators in the Cornell chemistry department are Yi-Chiau Huang, Jaroslav Flidr and Theresa A. Newton.

This work was supported by the Beckman Young Investigator Program and by the National Science Foundation.