Number 366 (Story #2), April 9, 1998 by Phillip F. Schewe and Ben Stein|
THE SILICON-SILICON DIOXIDE INTERFACE in ultrasmall silicon-based transistors must be smooth on the atomic level or else their performance is degraded. If the boundary is too rough, electrons moving through the semiconducting silicon layer can scatter from the insulating SiO2 boundary, increasing electrical resistance to undesirable levels. Addressing this problem at the recent APS March Meeting in Los Angeles, Melissa Hines of Cornell (607-255-3040) showed that an ammonium fluoride solution could etch away surface roughness on Si(111) and produce surfaces of near-atomic smoothness over a large area. Hines hopes to find similar chemical methods for the Si(100) surfaces used in integrated circuits. Marcus Weldon of Lucent Technologies (908-582-5645) presented studies of how H2O reacts with silicon at elevated temperatures during the beginning stages of forming a silicon dioxide layer. Marrying infrared spectroscopy and quantum chemistry calculations, Weldon and colleagues discovered for the first time a silicon "epoxide," a triangular arrangement of silicon-oxygen-silicon that apparently dominates the surface at the intermediate stages of these reactions. Controlling the quality of SiO2 layers is increasingly important in state-of-the-art silicon devices; one recently fabricated SiO2 layer has a thickness of just three SiO2 molecular units in an ultrasmall silicon transistor announced by Lucent last year and envisioned for mass production by 2010. Built by Greg Timp and colleagues at Lucent, this 60-nm transistor is four times smaller, five times faster, and needs 60 to 160 times less power than present transistors.