High-resolutioncore-level and Auger-electron spectroscopy, polarization-dependent near-edgex-ray absorption, and angle-resolved photoemission are used to study the electronic structure and the bonding at the CaF2/Si(111) interface. Si core-level shifts of + 0.4 and -0.8 eV show that both Ca and F bond to Si and that the interface is atomically sharp. Interface-derived Ca and F core-level and Auger-electron shifts are found indicating layer-by-layer growth. The interface Ca 2p, 3p, 3score-level shifts are about 2 eV and the Ca 2p Auger energy shift is =4.S eV. The F 1s,2s core levels show no interface shift but a shift of 1.7 -2 eV in the initial adsorption regime indicating a rearrangement of F after the completion of the first layer. The F 1s Auger electrons show an interface shift of 2.0 eV. Initial-state and relaxation contributions to the shifts are considered. In the Ca 2p and F 1s near-edge x-ray-absorption fine-structure (NEXAFS) spectra several unoccupied Caand F-derived interface states are found. The orientation of the corresponding orbitals is revealed by the polarization dependence. The oxidation state of the Ca atoms at the interface is found to be changed to 1 + . The CaFq valence bands start to form at 2 layers with an overall bandwidth of 3.3 eV. An occupied interface state is found at 1.2 eV. The Fermi level shifts by 0.6 -0.6S eV when 2 CaF2 layers are deposited and a new pinning position is established at the Si valence-band maximum. A bonding model for the interface is proposed.

No abstract

No abstract

The rate of silicon etching by XeF2 is dependent on the type of dopant and the doping level.Soft-x-ray photoemission was used in an effort to elucidate the mechanism responsible for this phenomenon. Si(111)samples were subjected to sufficient XeF2 to reach the steady-state etching regime, and spectra were collected of the fluorosilyl reaction intermediate species that form on the surface. Samples that were lightly doped (10' cm ) showed virtually no difference between p and n doping, while heavily doped samples (10 cm ) showed a marked difference. The heavily doped n-type sample had a slightly thinner reaction layer than did the lightly doped samples, while the heavily doped p-type sample had a much thicker layer, in which the composition of the layer was also changed. These results are discussed in terms of possible reaction mechanisms.