2.1.1 Molecular Structure of the Silicon-Silicon Dioxide Interface

An in-depth analysis of the interface between silicon and silicon dioxide and how the molecular structure changes during the oxidation process is outside the scope here as this work mainly concerns itself with changing topographies and interfaces rather than atomistic-level simulations. However, it is worth mentioning that the interface between silicon and the grown oxide is not a perfect transition between a crystalline silicon and an amorphous SiO$ _2$. There are many suggestions regarding the molecular make-up of the interface and how the interface is built during the initial stages of oxidation [47], [87], [160], [186], [199], [208].

The initial proposed model for the atomic configuration of the Si(100)-SiO$ _2$ interface is shown in Figure 2.4a [47]. However, as noted a small amount of surface silicon atoms remain unbonded, resulting in unpaired electrons to localize on the defect silicon atom, forming a ``dangling bond'' [47]. Figure 2.4b and Figure 2.4c illustrate the modified interface, when a ``dangling bond'' is introduced to a (100) and a (111) oriented silicon, respectively. More recently, it has been proposed, and widely accepted, that the oxide layer, although amorphous, contains a crystalline structure close to the Si interface [186]. It has also been suggested that the bulk oxide itself is not a simple amorphous structure, but rather that, throughout the bulk of the oxide, an ordered bond structure exists, having an epitaxial relation with the silicon substrate [208].

Figure 2.4: Atomistic configurations of the Si-SiO$ _2$ interface.
(a) Atomic configuration model for the Si-SiO$ _2$ interface.
\includegraphics[width=0.7\linewidth]{chapter_oxidation/figures/atomicConfig100.eps} \includegraphics[width=0.48\linewidth]{chapter_oxidation/figures/atomicConfig111.eps}
(b) Dangling bond formation on a (100) Silicon. (c) Dangling bond formation on a (111) Silicon.

Much effort has also been spent in order to investigate the initial steps of silicon oxidation at the molecular level [20], [37], [75], [169], [214], [219]. Watanabe et al. [219] suggest a layer-by-layer oxidation, whereby nucleation of nanometer-scale two-dimensional oxide islands at the interface explains the initial stages of oxidation. It is suggested that atomically flat terraces, separated by single atomic steps, approximately 0.13nm in height, are formed as oxidation is initiated. Pasquarello et al. [169] suggest that the interface between Si and SiO$ _2$ can only be explained with the introduction of transition regions, which do not follow the properties of SiO$ _2$ or bulk Si, at the interface. Their suggested model gives a disordered Si layer (0.5nm - 1nm) containing a dense Si arrangement, which links the bulk silicon to the oxide layer. However, the oxide layer is also represented using an interface region ($ \sim $1nm) which acts as a sub-oxide transition region and contains silicon atoms in intermediate stages of oxidation. More recently, Hemeryck et al.[75] indicated that the penetration of oxygen atoms into the top layers of crystalline silicon depends on the starting and final surrounding environment, with activation energies ranging from 0.11eV to 2.59eV, and is not a simple case of the atom hopping from one Si-Si bond to another. The exact mechanism by which the oxidation process progresses is not yet fully understood at the atomistic level.

L. Filipovic: Topography Simulation of Novel Processing Techniques