We have measured the absolute fluorine atom concentrations in electron cyc!o!ron resonance (ECR! and reactive ion etching !ME) plasmas by optimizing the actinometry technique. The major difference between this work and conventional actinometry is that the Ar concentration measurements were pelformed by a residual gas analyser (RGA). The emission intensities of F (7037 A) and Ar (7504 A) were simultaneously measured by an optical multichannel analyser (OM) and the Ar concentration by a RGA. The F atom concentration at the wafer stage in the CF4 ECR plasma was measured to be (0.4 to 4)x1Ol2 ~m -~, the microwave power from 500 to 900 W, pressure from 0.5 (3.7 sccm) to 3.5 mTorr (58 sccm), and a fixed RF bias voltage of -50 V. The F atom concentration of the CF, ECR plasma was four times larger in the source region than in the downstream region. The F atom concentration of the CF4 RIE plasma for pressures from 13 to 62 mTorr was measured as (0.8 to 4 . 2 ) ~l O ' ~ ~m -~, for a CF4 flow rate of 20 sccm, and power inputs from 250 to 1500 W. The F atom concentration was larger in the RIE etcher than in the ECR etcher, but the F atom production efficiency was eight times larger in the ECR etcher than in the RE etcher for the same power level. In spite of the possibility of a factor two discrepancy in the measurements from absolute values because of the uncertainty in the absolute values of the cross sections, this technique provides a relatively simple and consistent reproducible measurement of F atom concentration to compare operation of the two types of etchers, which is reproducible to 110%.We further explored the actinometry for vacuum ultraviolet (vuv) emission region by using F (955 A) and Ar (1048 A). A similar trend of F atom concentration was found as for the visible actinometry, but the absolute value of fluorine atom concentration was typically 15% larger for the vuv actinometry.

SiO2 and Si etching in a CF4/O2/Ar plasma has been carried out in an electron cyclotron resonance etcher over a wide range of conditions. The etch rate has been compared with the ion energy flux to the wafer surface, JiEi, and the atomic fluorine density in the gas phase, nF. It is found that the etch rate can be divided into two regimes by a critical value of nF/(JiEi), the ratio of the atomic fluorine density to the ion energy flux. The critical value can be determined from a contour plot of the etch rate as a function of the ion energy flux and the atomic fluorine density. The critical value of nF/(JiEi) for Si is higher than that for SiO2. For nF/(JiEi) higher than the critical value, the SiO2 etch rate linearly increases with the ion energy flux, and the Si etch rate shows a nonlinear increase with the ion energy flux. For nF/(JiEi) lower than the critical value, both SiO2 and Si etch rates linearly increase with the atomic fluorine density.

Wafer temperature, etch rate, and etch uniformity measurements of SiO2 wafers were made to characterize the use of back side helium cooling with an electrostatic wafer holder in an electron cyclotron resonance etching tool. The etch rate was found to be independent of the wafer temperature in the range between 20 and 110 °C. A 7% increase in etch nonuniformity (3σ) at higher backside pressures was attributed to helium, which leaked around the edge of the wafer, displacing the etchant gas. A back side pressure of 2–3 Torr provides a balance between wafer temperature control and helium leak rates.

In this article, we examine the influence of contaminants on an electron cyclotron resonance discharge. For our discharge, the measured level of contaminant was highest just after startup, decreasing to a stable level after approximately 20 min.This is consistent with a limited source of gas trapped in the chamber walls. It is shown that the presence of these contaminants cause both the plasma and floating potentials vary by several volts. These potential variations are largest when the contamination is largest. Such variations were found in N2, CF4, and CHF3 discharges. The observed variation in the plasma potential should be of great concern in plasma-aided manufacturing environments. The larger variations represent approximately 10% of the potential that one would typically apply between a device and the plasma. Such large changes in the potential might result in substantial changes in both etch rate and anisotropy. This can have particularly adverse effects on those devices having fine structures. Thus, one should monitor these contaminants and not process devices while the contaminant level varies.

No abstract