A feature-to-wafer-scale model of etch-rate non-uniformity in deep reactive ion etching/

Abstract

Deep Reactive Ion Etching (DRIE) is an inherently complex dry etching process commonly used in the semiconductor manufacturing industry. This work presents a new modeling approach to capture global etch rate variation in DRIE by integrating wafer- and feature-scale nonuniformity models that are grounded on an ion-neutral synergy model for etch rate. Our method focuses on diffusive transport and local depletion of Fluorine radicals above the wafer surface to facilitate integration of wafer- and feature-scale models. Our results show that the wafer-level model achieves a success comparable to that of other wafer-level models previously developed with an etch rate RMS error percentage between 2.1% and 8.2%. The coupled wafer- and feature-level model shows that the feature-level etch evolution substantially impacts the waferlevel Fluorine concentration and thereby modifies the wafer etch rate uniformity. Similarly, the wafer-level etch rate directly impacts the rate of feature-level etch evolution. The coupled model is observed to over-predict the feature etch depth by an amount that increases with time and decreases for larger features, thus suggesting that the over-prediction arises from our assumption of negligible Fluorine consumption at the feature sidewall. Within-wafer etch depth variation of high aspect ratio features is also over-predicted, likely due in part to the negligible sidewall Fluorine consumption assumed. Suggestions to improve all levels of the model are examined.