AVS 53rd International Symposium
    Electronic Materials and Processing Thursday Sessions
       Session EM2-ThA

Paper EM2-ThA7
Theoretical Analysis of the Interface between Zr(Hf)O@sub 2@ and Ge(100) for Ge-based MOSFET Devices

Thursday, November 16, 2006, 4:00 pm, Room 2003

Session: Electronic Properties of High-k Dielectrics, Ferroelectrics, and Their Interfaces
Presenter: T.J. Grassman, University of California, San Diego
Authors: T.J. Grassman, University of California, San Diego
S.R. Bishop, University of California, San Diego
A.C. Kummel, University of California, San Diego
Correspondent: Click to Email
In recent years there have been many attempts at the fabrication of high-quality Ge-based MOSFETs. One of the most successful dielectric materials used in these studies is ZrO@sub 2@, producing some of the best Ge-based MOS devices to date; HfO@sub 2@ has also found a fair amount of use in the field, with varying results. In order to better understand these MOS systems, and particularly the often-problematic SC/oxide interface, a systematic density functional theory study of the Zr(Hf)O@sub 2@/Ge(100) interfaces has been undertaken. Multiple initial first-layer bonding configurations of ZrO@sub 2@ and HfO@sub 2@ on Ge(100)-2x1/4x2 have been simulated in order to elucidate, through adsorption energy comparisons and comparison with experiments, what the interfacial configuration between the semiconductor and oxide actually is. These sites were also modeled for electronic structure in order to help explain or clarify the experimental results. It has been found that ZrO@sub 2@ bonds to the Ge(100) surface very strongly in both Zr- and O-end down configurations, with the Zr-end down geometry the strongest. It is also seen from the electronic structure calculations that the Ge-Zr bonds are covalent rather than metallic, and neither Zr- nor O-end down bonding configurations result in an increase in near-Fermi level DOS. Such results are consistent with the experimental findings that ZrO@sub 2@ is a good gate dielectric material for Ge(100). In addition, the calculations indicate that H-passivation of dangling bonds at the ZrO@sub 2@/Ge(100) interface, and potentially within the oxide itself, may be an effective method to improve MOS device properties. However, while H-passivation of dangling bonds on both Zr and O atoms produces a considerable reduction of near-Fermi level DOS, only the passivation of the O atoms is thermodynamically stable enough to be achievable in device processing conditions. All of these same calculations are currently being performed on HfO@sub 2@.