Simulation multi-échelles des phénomènes de nucléation/croissance de nanocristaux de silicium sur SiO2 par LPCVD
Résumé
The microelectronic industry is in permanent evolution, due to the need of integration in many systems of the everyday life (PC, car, MP3, mobile, …). A typical example concerns non volatile memories for which the poly-silicon floating gate of the Flash memories could be replaced by a floating gate made up of silicon nanocrystals. The deposition of nanocrystals by LPCVD (Low Pressure Chemical Vapor Deposition) from silane SiH4 on SiO2 surfaces remains one of the most promising ways of synthesis. In particular, it is mandatory to reach an area density of 1012 dots/cm2 with a radius lower than 5 nm so as reliable Flash memories could be industrialized. To overcome these problems, we have decided to model physical and chemical phenomena from the scale of the precursor molecules and surface bonds to that of the industrial reactor. Despite a huge experimental effort, fundamental understanding of the key mechanisms for nucleation and growth remains elusive.
Experiments presented in the literature have shown that surface pre-treatment forming Si-OH groups to the detriment of Si-O-Si bonds increased the nucleation rate. Previous modelings of LPCVD reactors showed that the silane pyrolysis leads to the formation of unsaturated species which are poorly concentrated but extremely reactive with surfaces.
In this paper, we propose a multiscale approach combining several models in hierarchical order to extend progressively the experimental requirements while keeping the predictive nature of the results by transferring parameters from a model to another.
Firstly, we study the fundamental chemical reaction taking place at the initial stage of deposition, i.e. nucleation and growth, via first principles calculations (density functional theory or DFT). Chemical pathways and associated activation barriers for SiH2/SiH4 reactions onto oxide surface species are detailed. Parameters are thus defined (kinetics of the reactions, sticking coefficients) for further chemistry modeling at the continuum level coupled with reactor scale modeling features (Computational Fluid Dynamics code Fluent).
Origine : Fichiers produits par l'(les) auteur(s)