Atomic-Scale Structure of the Si-SiO 2 and SiC-SiO 2 Interfaces and the Origin of Their Contrasting Properties
- PDF / 2,453,808 Bytes
- 6 Pages / 414 x 635.4 pts Page_size
- 17 Downloads / 218 Views
2 INTERFACES
AND THE ORIGIN OF THEIR CONTRASTING PROPERTIES Ryszard Buczko*, Stephen J. Pennycook and Sokrates T. Pantelides Solid State Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831 Department of Physics and Astronomy, Vanderbilt University, Nashville, TN 37235 ABSTRACT One of the reasons for the dominance of Si in microelectronics is the quality of the Si-SiO 2 interface. In contrast, development of SiC-based MOSFETs for power applications is hampered primarily by poor carrier mobility at the SiC-SiO 2 interface. Here we review recent calculations that elucidate the reasons of the contrasting properties of the two interfaces. In the case of Si, the interface energy is in fact lower when the interface is abrupt and smooth because of the intrinisic geometry of the Si (001) surface and the softness of the Si-O-Si angle. However, two energe ically degenerate phases are possible, leading to domain boundaries, that are the cause of subcxide bonds, steps, and dangling bonds. In principle, these effects may be avoidable by lowtemperature deposition. In contrast, the geometry and bond lengths of SiC surfaces are not suitable for abrupt and smooth interfaces, requiring the existence of a nonstoichiometric interlayer that may be the cause of the reduced mobility. INTRODUCTION A key component of metal-oxide-semiconductor field-effect transistors (MOSFETs) is the semiconductor-oxide interface. The predominance of Si in microelectronics is due primarily to the properties of the Si-SiO 2 interface, which can easily be made very abrupt (minimal suboxide layer) and smooth (minimal steps) with a minimal density of point defects.[1] As scaling laws are pushing the technology to ultrathin SiQ layers, understanding and control of the interface on the atomic scale remain open challenges. For power MOSFETs, a semiconductor with a wider band gap is preferred. SiC, whose native oxide is also Si 2, is one of the options, but efforts to develop SiC-based MOSFETs have been thwarted by poor-quality SiC-Si0 2 interfaces. Ryszard Buczko The abruptness and smoothness of the Si-SiQ interface are puzzling because SiO 2 is amorphous. In addition, 0 has a high solubility and diffusion constant in Si. In contrast, 0 is not known to be an abundant impurity in SiC and a recent calculation finds that 0 has very low solubility in SiC[2] Yet, SiC interfaces are generally found to be rough for MO SFETs. Photoemission [3] has provided information on bonding arrangements at the Si-SiOQ interface while microscopy and theory led to several interface models [4-10]. Total-energy calculations have shown that the oxidation process, though accompanied by emission of Si interstitials, does not lead to formation of dangling bonds [ 11 ]. Other calculations found that, during thermal oxidation, lateral growth is preferred [10]. Yet, the mechanisms that control abruptness and smoothness remain an open issue that is critical for the ultimate control of ultrathin gate oxides. In this paper we review the results of a recent systematic study [12] of bondin
Data Loading...