Current Project Members: Kangzhan Zhang,
Ken
Nicholson
Past Project Members: Sungee Lee, John Bender, Neil Greeley, and Leah Meeuwenberg
We are also fortunate to have as collaborators on this project
F. R. McFeely of the IBM Thomas J. Watson Laboratory
Udo Pernisz of the Physics Expertise Center, Dow Corning
Brad
Orr of the Applied
Physics Program
Solid/solid interfaces play increasingly important roles in microelectronic devices and have been long recognized as critical for catalysis, yet the structure and reaction mechanisms of many of these interfaces remains largely a matter for conjecture. Included among this class, despite the apparent simplicity of containing only two types of atoms and three decades of intensive research because of relevance to the microelectronics industry, is the Si/SiO2 interface. In recent years, poorly understood reactions occurring at this interface have impeded the development of reliable devices with oxide thickness of under 100 Å.
The Banaszak Holl group has undertaken a study of the structure and reactivity of the Si/SiO2 interface. A central part of the research design has involved the creation of model interfaces derived from cluster molecules such as those shown above.
We have characterized three model interfaces synthesized
using the spherosiloxane clusters. These molecules, which simulate 3-4
layers of oxide, represent the first synthesis of silicon/silicon oxide
interfaces of known structure. The best model interface available previously
was water adsorbed on Si(100). The primary characterization tool for both
the model interfaces and the Si/SiO2
interface has been synchrotron soft X-ray photoemission of Si 2p core-levels.
Spectra for each of these molecules chemisorbed onto Si(100)-2x1 are shown
below. Our assignments have been rather controversial
as they challenge the orthodox methods for Si 2p core-level assignment
at silicon/silicon oxide interfaces. These data have been published in
Physical
Review Letters and the Journal of the American
Chemistry Society.
The monovertex binding modes shown above are also supported by Reflection Absorption Infrared Spectroscopy (RAIRS) and Scanning Tunneling Microscopy (STM). The STM image of a single H8Si8O12 cluster chemisorbed onto Si(100)-2x1 is shown below on the right. The bright blue lines are the Si dimers. The red regions roughly correspond to the electron density associated with the oxygen atoms. As can be seen by comparison to the electron density/electrostatic potential maps shown on the left, the experimental data is consistent with the monovertex model in FOUR important ways. 1) The electron density associated with the cluster is observed to be in between the dimer rows, not centered on the dimer row. 2) A square of four dots is observed. 3) The square of dots is oriented parallel and perpendicular to the dimer row directions. 4) Two of the dots are more intense than the other two and these pairs are oriented parallel to the dimer rows.
All four of these geometrical constraints are consistent with the predictions made by the electron density/electrostatic potential map of the monovertex model providing important experimental support for the RAIRS and XPS assignments we have previously made. Note that all four of these geometrical contraints are inconsistent with the predictions made by the electron density/electrostatic potential map of the cracked cluster model, effectively ruling out this binding mode as a reasonable proposal.
Our new approach to Si 2p core-level shift assignments has profound implications for work concerning the structure of the silicon/silicon oxide interface. We are also applying the new shift assignments to explore the reaction of hydrogen atoms in the interface region ( Appl. Phys. Lett. 1996, 68, 1081).