THEORETICAL CALCULATIONS OF DEFECT PROCESSES ON THE SURFACE AND BULK OF SEMICONDUCTORS

Abstract

In conjunction with theoretical calculations to simulate excitation of semiconductor surfaces from a DIET (LID) point of view, calculations are also performed on defects in the bulk of semiconductors to investigate defect processes and in the process, the negative-U (Anderson 1975) characters of the O impurity, the Ga antisite defect and the P vacancy in GaP are theoretically established for the first time. Bulk defect processes, e.g. a defect can adopt a different geometry in different charge states (Bourgoin-Corbett 1972) and negative-U behaviour where electron- or hole-pairing is aided by lattice relaxation, are shown to have implications for surface defects under excitation and can explain the experimental observations of laser-induced desorption (LID) of atoms from defect sites on the semiconductor surface. The ability of the defects to exist in different charge states with different atomic configurations suggests that excitation should be multiple in nature to induce desorption as the initial excitation event (one-hole (lh) state) may merely lead to a change in geometry (due to a change in charge state to a more positive one) or hole-pairing to induce multi-hole localisation at the defect sites, instead of bond breaking, and hence, the existence of a threshold laser fluence observed experimentally. The change in geometry (lattice relaxation) between excitation events corresponding to the change in charge state also implies the appearance of intermediate states after lattice relaxation, which require further excitation before a final antibonding state is attained and can explain the non-linear dependence of the desorption yield on the laser fluence. Thus, the desorption process can be described in terms of a series of cascade/multiple excitation-relaxation processes (Hattori et al 1992), the nature of which can be elucidated in terms of well-established defect processes. The nature of the intermediate states formed, i.e., whether they are metastable or stable can also be inferred from a negative-U consideration. If the defect involved is a negative-U centre and is directly on the surface, then the equilibrium configuration of the excited state after lattice relaxation is most likely metastable and higher in energy. However, if the negative-U centre lies just immediately below the surface (this negative-u centre can still be regarded as a bulk defect/atom with a surface atom as one of its nearest neighbours) , the excited state after lattice relaxation can still be stable, but only if excitation of the two-hole state of the negative-U centre results in each hole occupying separate orbitals with one of the orbitals belonging to the surface atom just above the negative-U centre. In the bulk, this would be unstable for the two-hole state of a negative-U system whose nearest-neighbours are also bulk atoms as lattice relaxation would make hole-pairing on the same orbital exothermic. However, the appearance of this stable excited state for the negative-U system just below the surface is possible because the lack of symmetry of the surface atoms compared to those in the bulk renders them more conducive to undergo a higher degree of lattice distortion resulting in bond-weakening and requiring further excitation to finally break the bonds of the surface atom (see chapter 6). Thus, the role which the lack of symmetry of surface atoms plays in aiding hole localisation should be considered only when the holes are distributed over the surface and bulk atoms leading to unequal degrees of lattice distortion between them.We believe our calculations represent the first theoretical demonstration of the role played by the lack of symmetry of surface atoms as well as the modification in the stability of negative-U systems near the surface. With regards to the phenomena of hydrogen-induced passivation of doped semiconductors, our calculations for the H-P system in silicon (Khoo and Ong 1991) demonstrate clearly that passivation inn-type in Si can be due to pairing (although weak) between Hand the donor, providing the theoretical support to the suggestion (Pantelides 1987) that H has a deep donor level in the gap. This is because our model for the global equilibrium configuration of the H-P system in Si has the H atom in the interstitial space between the P atom and a Si atom (site Lp in Fig. 8.10, see chapter 8) with no intervening Si atoms between H and P atoms, unlike the other models (Johnson et al 1986, Denteneer et al 1990) proposed so far for the H-P system in Si. On the other hand, passivation in p-type Si is due to direct compensation and H-acceptor pairing (Ong and Khoo 1991) is a consequence, rather than the cause of passivation.