An ideal crystalline surface is perfectly flat, but actual crystal surfaces are usually tilted slightly, with atomically flat terraces separated by atom-high steps. These steps can move when the material is heated, as surface atoms evaporate. In the 29 March PRL, two researchers in Japan shed new light on the movements of atomic steps on a silicon sample. The researchers also made moving steps collide to learn how the steps interact with one another. Physicists want to understand these surface features because silicon surfaces are the basis for microchip and other high-tech devices.
Although the theory behind atomic steps and their role in crystal growth has been around for almost 50 years, in the past decade researchers have advanced the techniques for studying steps. Paul Finnie and Yoshikazu Homma, both of NTT Basic Research Laboratories in Japan, fabricated silicon samples with several atomic steps in the form of a crater–a series of concentric circular terraces leading down to a flat bottom. Heating such a sample above 1000 degrees Celsius caused the surface atoms to evaporate and made the craters expand as each step moved outward from the center. According to Finnie, a step moves by ejecting atoms onto the terraces above and below, where the atoms diffuse around in the “adatom sea” before leaving the surface. Finnie and Homma made movies of the propagating steps by running cycles of heating, rapidly cooling, and imaging of the samples with a scanning electron microscope. The team found that with wide terraces separating them, the steps moved at a constant speed, but as they approached the boundary of the ultraflat region, the steps slowed down because of interactions between steps.
To further study the interactions, the researchers learned how to set up collisions between moving atomic steps by creating two craters near one another. “If two steps are close together, both ejecting atoms into the same sea, the sea starts filling up,” says Finnie, “and it is more and more likely that atoms will stick to the steps. So the steps slow down.” The team explains this apparent “repulsion” in terms of a theory from 1951, long before such effects could be observed directly. They also arranged for “constructive” collisions, where an inner step moved faster than the step just beyond it, causing the two to fuse into a double atomic step. Finnie notes: “For particle physicists, it is a time-honored technique to learn about the interaction between particles by crashing them together. Why should they have all the fun?”
No one has previously quantified the repulsion between colliding atomic steps, according to Ellen Williams of the University of Maryland. Kit Umbach of Cornell University is impressed with Finnie and Homma’s ability to repeat and control the geometry of the steps. Although the researchers used a different type of crystal surface from the one used for today’s microdevices, Umbach predicted, “The principles that they’re demonstrating will likely be transferrable” to the surfaces used in industry.