Reducing Wafer-to-Wafer Mean Thickness Variability on CVD Processes Using Novellus Vector Tools
Publication Date: 2004-Oct-26
Publishing Venue
The IP.com Prior Art Database
Abstract
Disclosed is a method that uses a multi-pedestal chamber to improve the wafer-to-wafer mean film thickness. The disclosed method accomplishes this by scaling a deposition parameter (i.e. pressure, RF power, gas flow, dep time, etc.) as a function of the number of wafers present in the process chamber. Benefits include improved control over process conditions.
Reducing Wafer-to-Wafer Mean Thickness Variability on CVD Processes Using Novellus Vector Tools
Disclosed is a method that uses a multi-pedestal chamber to
improve the wafer-to-wafer mean film thickness. The disclosed method
accomplishes this by scaling a deposition parameter
(i.e. pressure, RF power, gas flow, dep time, etc.) as a function of the number
of wafers present in the process chamber. Benefits include improved control
over process conditions.
Background
Currently, in a multi-pedestal process tool, such as the
Novellus Vector tool, the RF power is not evenly distributed when the chamber
is partially loaded with wafers. The result is an increase in the film
deposition rate for the wafers in a partially (i.e. < 4 wafers) loaded
chamber
(see Figure 1).
General Description
The disclosed method removes any systematic shifts in the deposition rate due to a partially loaded chamber.
Figure 4 shows the chamber impedance and current flow for four pedestal chambers without any wafers in the process chamber. Each shower head-pedestal system is represented from an electrical standpoint as an impedance. There is a perfect impedance balance among the shower head-pedestal systems in an empty chamber, which is mathematically expressed by the equality of the currents that flow through the shower heads.
Figure 5 shows the chamber impedance and current flow for four pedestal chambers with one wafer in the process chamber. In this scenario, the showerhead-pedestal one system has a lower RF impedance and a higher RF current flow than the other three showerhead-pedestal systems that are unoccupied. The disclosed method uses a user defined scaling factor = C1 (C1 < 1). The RF power delivered to the occupied pedestal matches the RF power delivered to each pedestal when all pedestals are occupied. Figures 2 and 3 show that scaling down the RF power linearly reduces the deposition rate. This is a method to reduce the higher deposition experi...