Silicon dioxide and the chalcogenide semiconductors; similarities and differences

The purpose of this article is to examine current hypotheses about the optical and electrical properties of the amorphous chalcogenide semiconductors, and to examine whether they can be extended to explain some of the properties of amorphous silicon dioxide. For the selenides and tellurides we assume the validity of the model of ‘charged dangling bonds’ introduced by Street and Mott. Absorption of light by these defects leads to photoluminescence with a large Stokes shift; formation of an electron-hole pair far from a defect, on the other hand, leads to radiationless recombination. In enquiring why this is so, we examine the nature of excitons in non-crystalline systems, and the mechanism of self-trapping both of excitons and of free carriers in crystalline and non-crystalline materials. Self-trapping by the V _K-type mechanism, if it occurs, takes place after a certain delay for a free hole or exciton, but not for one bound at a defect, so that free-exciton absorption bands are narrower than those from bound excitons at defects. The radiationless recombination of a free or bound exciton normally occurs through the mechanism first considered by Dexter et al., if more energy than ～1 eV is emitted.

In SiO₂ electrons are exceptionally mobile, but on the basis of work by Hughes, we think that holes are self-trapped after a delay, and can then move by polaron-type hopping. We propose that the first absorption peak (at 10.2 eV) in the optical spectrum is due to an optically allowed exciton, which recombines by the Dexter mechanism, but only after a delay of a type proposed here. The delay allows a comparatively narrow absorption peak, of width given by that of the exciton band. The absorption peak due to a Na⁺ ion linked to a non-bridging oxygen atom is considerably broader because it is due to a bound exciton. We estimate the band-gap in SiO₂ to be about 10.6 eV and also the position of the centre of gravity of the level due to a Na⁺ ion with a non-bridging oxygen atom to be ～1.5 eV above the valence band of SiO₂, less than usually supposed.

Finally, we examine the nature of the positive and negative charges which occur in SiO₂ grown thermally on silicon; these are responsible for Anderson localization of electrons in the inversion layer, and for the ‘slow states’ of silicon technology. We tentatively conclude that they are non-bridging oxygen atoms and that these, like dangling bonds in chalcogenides, are normally positively and negatively charged. We examine their positions in the gap and the activation energies of charging and discharging.