AVS 46th International Symposium
    Plasma Science and Technology Division Friday Sessions
       Session PS-FrM

Paper PS-FrM8
Real Time Control of Plasma Tools During Recipe Changes and Transients@footnote 1@

Friday, October 29, 1999, 10:40 am, Room 609

Session: Emerging Plasma Applications
Presenter: M.J. Kushner, University of Illinois, Urbana
Authors: M.J. Kushner, University of Illinois, Urbana
S. Rauf, Motorola Inc.
Correspondent: Click to Email
Successful development of real time control (RTC) of plasma tools should enable more rapid process development. A full etch process often includes multiple recipe changes to optimize, for example, break-through, main etch and over-etch. If the recipes are markedly different in gas composition or pressure, process parameters such as uniformity may significantly change during the transient. In the absence of RTC, maintaining desired process parameters requires changes in actuators based on interpolation between the beginning and end conditions. In this paper, the Virtual Plasma Equipment Model (VPEM) is used to investigate RTC strategies during recipe changes and transients. The VPEM is a "wrapper" for the Hybrid Plasma Equipment Model which contains simulated sensors, controllers and actuators. Recent improvements in the VPEM allow investigation of "real-time" (as opposed to run-to-run) control. Results from the VPEM will be discussed using response-surface based controllers to maintain process uniformity and rate during recipe changes for Cl@sub 2@ and C@sub 2@F@sub 6@ chemistries in ICP reactors. It was found that during recipe changes which, for example, significantly change mole fractions, the linearized response surfaces based on conditions at any given mole fraction are not adequate. Control is maintained for a portion of the transient but is eventually lost. To address this problem, real time mass spectrometer sensor data is used to interpolate between response surfaces which are the basis of (2 x 2) controllers. The response surfaces were generated using results from steady-state experiments. This strategy was able extend the dynamic range of control throughout the transient. @FootnoteText@ @footnote 1@This work was supported by AFOSR/DARPA, SRC and LAM Research.