This thesis consists of three major parts; 'the effect of gas phase reaction on the roughness of silicon films in HF-CVD', 'morphological evolution of silicon films induced by charges in CVD', and 'new approach to SSD and SENTAXY in CVD'. All of them are closely related to 'charge-related deposition phenomena in silicon CVD'.
Chapter III dealt with 'the effect of gas phase reaction on the roughness of silicon films in HF-CVD', where the charges were introduced by the thermoemission effect of the hot-filament. For the $SiH_4-HCl-H_2$ system, combinations of the amount of the electron emission and the supersaturation ratio affected the deposition rate, the current, and finally the roughness of silicon film in a different way. Currents of 0.2 ~ 5 μwere measured in the hot filament CVD chamber for deposition of silicon. In such a highly ionizing atmosphere with the appreciable supersaturation for precipitation of silicon, it is highly probable that the charged clusters should be formed in the gas phase by ion-induced or photo-induced nucleation. The predicted nanometer silicon clusters were observed by TEM on a carbon membrane, which was placed in the hot-filament CVD chamber. The clusters for the filament temperatures of 1800℃ and 1600℃ were ~ 2 nm and 5 ~ 7 nm, respectively. Moreover, there was a linear relationship between the cluster sizes and the RMS roughness. These results, which could be explained by classical nucleation theory and Coulomb repulsion effect, leads to the following conclusion: higher electron emission enhances the gas phase nucleation with the help of the ionization process, but it prevents the charged clusters from growing to the large size, so smoother film can be deposited. Furthermore, at the constant electron emission, the gas phase with higher supersaturation ratio has lower nucleation energy barrier, and makes many small charged clusters by the ion-induced heterogeneous nucleation, thus producing smoother surface. From the TEM analysis of silicon clusters, they were demonstrated to be the magic sizes, and according to the theory of the magic size the clusters can form crystalline films even at low temperatures. Therefore, HF-CVD system will be appropriate for the study on the film formation from nano-sized clusters. The roughness of silicon film depends on the character of the deposition element, and both the electron emission and the supersaturation ratio in the gas phase are important factors in determining the deposition pattern of the HF-CVD. It seems that large clusters or grains leads to a rough surface of film, but more analytical research and observation must be carried out.
Chapter IV dealt with 'morphological evolution of silicon films induced by charges in CVD'. It is hard to devise a general law which can explain all the film morphologies found in CVD, because many intricate problems stem from various deposition parameters, chemical reaction paths and flow dynamics related to material transfers. Main factors affecting the morphology of CVD films consist of the substrate temperature, the supersaturation ratio, and electrochemical interactions between depositing species and growing films( or substrates). The various morphological evolution of CVD films mostly stems from the differences in chemical and physical properties between substrate and growing film, and the most active initial stage of surface chemical interactions takes place during the dynamic development of films. In this connection, the substrate must be a very important parameter, but many problems about it remain unsolved due to the reactivity with the depositing species and the unknown effects on the film formation. The special effect of the substrate on film morphologies such as a chemical reactivity was not considered in this study. Instead, whether there are some charges and charged particles, and, if there are, how they interact with the substrate were mainly examined. First of all, it was examined that in the conventional CVD there could be charges with both signs and the surface morphology of films could be smoother by a bias treatment, regardless of the charge sign. Moreover, the result obtained in HF-CVD indicated that the particle growth can prevail in a plasma ring. In silicon platelets on Ni, the morphological evolution of CVD platelet films could be explained by means of the relationship between adsorption amount of charges and increase of the film area. This result was not in accordance with an earlier report about the CVD platelet film. Cauliflower structure, which is a semi-spherical shape film, was considered to have formed by charged clusters as shown in the plasma CVD, but the morphological development should be studied continuously in CVD as well as in PVD. CTR of substrates or films became an important parameter in deciding the morphological evolution of silicon films, because it could be related to the attractive force between deposit species(or charged clusters) and growing films. This fact was revealed by the following experimental; deposition on various substrates, deposition on insulated-substrates, changes in microstructures with deposition time, and edge deposition pattern in HF-CVD.
Chapter V dealt with 'new approach to SSD and SENTAXY in CVD', where the earlier explanations of the selective silicon deposition had had one critical contradiction. From the thermodynamic point of view, the partially etching of silicon nuclei on SiO2 by HCl is impossible. A new mechanism to SSD and SENTAXY contains has three consecutive steps in the CVD process. That is to say: first, charged clusters(or species) are formed in the gas phase. Second, the state of the film surface becomes atomically an etching condition, which can be explained by CVD thermodynamics. Third, charged silicon clusters(species) and silicon radicals produced by electrochemical reaction in the gas phase are likely to be electronegative, which can cause the charging problem on SiO2. The theory of the bond strain energy, where Si-Si bond strain energy of silicon nuclei changes with substrates, can contribute to this mechanism. In SiH4-HCl-H2 system, the above model could be applied to SENTAXY. The formation and subsequent disappearance of many tiny silicon nuclei could be explained by the surface charging effect, and the continuous silicon deposition on metal substrates(Mo, Pt) by the attractive force between charged clusters(species, or polymers) and substrate.