Caso 1: Modalidad de Intervención de Apoyo Educativo

A picture of the main mechanisms which, in our view, regulate the oxide growth has been presented. The emphasis was on space-charge effects as our experiments indicate that the increased pool of electrons available for reaction have a role to play i n the initial stages of the Si oxidation (at least). A role for stress-related effects or the presence of micropores or other structural characteristics cannot be excluded a priori, but we do not find any direct correlation with our experimental evidence, although there is some evidence that the density of fixed positive charge might be related to stress in the oxide (see [IRENE, 1987; WOLTERS et a l . , 1989] for references).

As with thermal oxidation, it is not possible to determine the precise role of electrons in the photonic oxidation. Even in the case of low temperature oxidation, the fact that the data could be fitted by a Cabrera-Mott type of growth rate doesn't in itself justify the existence of a particular mechanism. Furthermore, the kinetic data do not give enough information to uniquely pin-point a reason for the apparent reduction in the activation energy for the ionic movement into (and within) the oxide network.

enhances the adsorption (with creation of negative ions) and/or the dissociation of oxygen at the oxide surface in the initial stages. 0, 02" or 0' could generate a flux of oxidant (migrating to the interface either interstitially or by some exchange mechanism, with or without the assumption of an electric field) in parallel with the main interstitial 02 (in the assumption that oxygen moves mainly interstitially in our experimental c o n diti ons, which is by no means certain for thin oxides). This possibility would diminish with increasing oxide thickness as less electrons would be able to reach the oxide surface. If, instead, we assume that electrons interact with oxygen mainly at the boundary of the 'blocking' or 'reactive' layer, it is more difficult to explain why the enhancement should decrease exponentially with oxide thickness, as such a layer is likely to be always present during all stages of oxidation. However, it has been recently suggested [MOTT et a l ., 1989] that this reactive layer might increase in thickness with time (and total oxide thickness) until a dynamic equilibrium is reached, at each temperature.

On the other hand, the extra flux of electrons likely to be trapped into the oxide could, initially, compensate for the high density of fixed positive charge, thereby allowing a greater flux of ionic oxid.ant to reach the Si/Si02 interface. Accounting for Coulomb repulsion, the trapping rate is proportional to exp(-N/No) [WOLTERS et a l ., 1989 and references therein], where N is the number of trapped charges and No the maximum number of places where charges can be accommodated (generation of traps with visible light is not very likely as the photon energy is too low for breaking Si-0 bonds). This neutralizing effect could therefore diminish down to an equilibrium situation as the oxidation progresses. The ionic current would decrease accordingly [and so would the compensating electronic current, as required by Wagner's theory (see Wolters et a l . model in p a r .1.6)].

In this last section IV.4, all the comments applied equally to both <100> and <111> orientations. We have seen (IV.1.1) that <100> Si exhibits a slightly higher percent rate enhancement than <111> Si, with a similar thickness and temperature dependence. This confirms that the oxidation rate enhancement is not uniquely determined by the

number of photo-generated carriers, but is also moderated by the interfacial chemical/physical structure and/or charge and trap distribution at, or near, the interface (which are all orientation dependent, as briefly discussed in Appendix I). It must be remarked that in the Cabrera-Mott model no great sensitivity to crystal orientation is expected [MOTT, 1981, 1982].

Young [YOUNG, 1988] tentatively explained the difference between <100> and <111> enhancements by referring to recent studies on dipole moments at the Si/Si02 interface [MASSOUD, 1988] according to which the <111> orientation has the largest electron retarding potential by « 0.47eV. Therefore, under equivalent irradiation conditions, the <111> Si would emit a somewhat smaller flux of electrons than the <100> Si. A confirmation of this suggestion comes also from an older article by Weinberg and Hartstein [WEINBERG et a l ., 1983] on tunnelling from Si to S i 0 2 . The reduced tunnelling found for <111> was explained by a higher barrier (~ 0.5eV) . It was suggested that the barrier increase could be due to a bonding arrangement at the <111> Si/SiC>2 interface yhich resulted in a dipole layer of higher density.

Though it is a quite convincing explanation, it opens up some doubts on how to explain the greater thermal rate for <111> than for <100> if one assumes that the hot electron flux "moderating" role in 02 dissociation is the rate limiting step. Besides, if the densities of bonds in sub-oxides at the interface are assumed to be those reported by Grunthaner et a l . [GRUNTHANER et a l ., 1987] for <100> and <111> Si, then this difference in dipole moments could decrease to « 0.05V [MASSOUD, 1988].