Predictive Model Extraction for Polysilicon Low-Pressure Chemical Vapor Deposition in a Commercial Scale Reactor

Theoretical optimization of a polysilicon low-pressure chemical vapor deposition process in a commercial scale reactor for 6 in. wafers was achieved by extracting a predictive reaction model from wafer-radial and reactor-axial deposition rate distributions. The extraction involved the radial deposition rate profile on the wafer, axial deposition rate profile in the reactor, and wafer-spacing dependence of the deposition rate at the wafer center. The deduced reaction model contains two film precursors as well as Si H 4 . Based on the radial deposition rate profile, the respective sticking probabilities of these two precursors were 5 × 10 − 2 and 7 × 10 − 4 at 550 ° C . Because the two precursors could not reach the wafer center due to their high sticking probabilities, as evidenced by the low deposition rate at the wafer center, the sticking probability of Si H 4 was 1.7 × 10 − 6 at 550 ° C with an activation energy of 161 kJ ∕ mol . The elementary reaction simulation done here included the mass transport and the gas-phase and surface elementary reactions in the reactor. Comparison between this simulation and the experimental analysis revealed that the two film precursors probably were Si H 2 and Si 2 H 6 .