Trends in aluminum etch rate uniformity in a commercial inductively coupled plasma etch system

Abstract

Characterization of an aluminum etching process in an inductively coupled discharge using measurements of discharge impedance and current and voltage sensors The effects of process conditions and chamber geometry on the uniformity of Al etched by Cl 2 were measured in a Lam TCP™ 9600 SE etch reactor. A computer simulation accurately predicted etch uniformity and aided in the explanation of uniformity trends. Parameters used in the experimental matrix included pressures between 6 and 24 mT, flow between 25 and 100 sccm, power supplied to the plasma between 0 and 350 W, and chamber heights ranging from 6 to 12 cm. The distinctive features of this study include the large number of input parameters studied in a commercial reactor and the accurate predictions obtained from a self-consistent simulation without free parameters. Reducing residence time in the experiments by adjusting chamber height or flow rate produces a more center-fast etch, as expected. The flow simulations were useful in corroborating intuitive arguments and in explaining anomalous results such as the effect of pressure on etch uniformity. More specifically, comparison of simulations and measurements demonstrated the quantitative connection between the Peclet number, the residence time, and the edge uniformity over a large range of process conditions. In addition to explaining general trends with residence time, Peclet number considerations also clarify the differing effects of pressure, flow rate, and chamber height change on uniformity. No attempt was made to impose plasma effects in the flow simulation because measurements of the neutral temperature and dissociation fraction were not available. Plasma power was observed experimentally to slightly improve uniformity without changing the average etch rate.

Semiempirical profile simulation of aluminum etching in a Cl2/BCl3 plasma

“…This can be understood by considering that surface kinetics based regression model is only valid while Al etching is taking place and etch reactions have been omitted from our HPEM modeling thus far. The total chlorine flux over the wafer decreases from 2.55 For typical Al etch rates ͑ϳ0.75 m/min͒, Cl 2 flows ͑ϳ100 sccm͒, expected etch product ͓((AlCl 3 ) x )͔, 45°C electrode temperature, 10 mTorr chamber pressure, and 50% Al area exposed wafers having 200 mm diameter, we find that approximately half of the injected Cl 2 is utilized in the etch reaction ͑in agreement with Beale et al 45 ͒. Both of these effects lead to a reduction in total chlorine flux arriving at the wafer surface.…”

“…As we have previously discussed, the etch process provides a sink for chlorine and a source for etch product through which chlorine must diffuse in order to reach the wafer. Evidence for transport limited etching has been given by Beale et al 45 and Danner and Hess 49 who explain their observations of etch rate nonuniformity in terms of chlorine trans-port through an etch product boundary layer that forms over their substrates. An approximate etching mechanism has been added to our HPEM model in an attempt to gain qualitative verification of this effect.…”