Deep Reactive Ion Etch (DRIE) has historically been regarded as a process possessing inherent variable response. These varying responses include etch rate, mask selectivity, etch depth uniformity across the wafer, and the overall profile of the features being etched. Several factors are thought to lend themselves to this observed variation. Among them are process temperature disparities and residual parasitic compounds within the reaction chamber itself.
A long term experiment was carried out to examine the statistical difference between DRIE runs with and without a specially defined pre-process conditioning recipe. This recipe was developed with the expectations of serving a twofold effect: the first serving as a “warm-up” of the process chamber to a steady state temperature, and the second being a stripping of residual organic compounds within the chamber that might otherwise add variance to the following DRIE process. The pre-process recipe has duration of ~30 minutes. The results of the experiment performed will clearly show that this conditioning recipe run prior to processing reduces the typical variance of DRIE processing.