Chemical challenge of submicron oxide etching

Abstract

In order to achieve the critical dimension control and selectivity requirements of submicron etching, industrial oxide etching tools are utilizing low pressure, high density plasma etching processes. At low pressures, the mean free path (1 cm at 5 mTorr) approaches the distance between the wafer and the chamber interior. This permits chemical interaction between the chamber and the wafer, since some of the gas species leaving the chamber interior can reach the wafer before undergoing further gas/gas collisions. The contribution from gas species leaving the wafer clamp is even greater, since the distance between the wafer edge and the clamp is smaller than the mean free path under all pressure conditions. This leads to the manufacturing problem that processing results become dependent on the previous thermal and chemical history of the chamber interior and the clamp. Results obtained using an Applied Materials high density plasma contact etcher will illustrate the generic nature of this history problem. The dependence of the etching performance on the chamber temperature and previous chamber oxygen cleans will be discussed. The manufacturing challenge is to reproducibly achieve acceptable processing results in systems where the contribution to the processing from the chamber is continuously varying. Not only must the current state of the etching tool be known, but the processes required to recover a baseline operating condition must be understood. It will be argued that this will require real time diagnostic control of the process. It is also suggested that the trend to low pressure systems may have been taken too far, and pressures in the 20–40 mTorr range might result in better process control in terms of reducing the chamber “history” effect. (This article summarizes an invited presentation given by McNevin at the 42nd Annual American Vacuum Society national meeting in the Plasma Science Division.)