Optical emission spectral analysis using FastICA for diagnosing wafer edge uniformity in plasma etching

Spatially integrated signals in optical emission spectroscopy (OES) inherently mask localized reaction dynamics at the wafer edge. This study applies independent component analysis (ICA) to standard plasma emission spectra to resolve this spatial limitation without hardware modification. The FastICA algorithm decouples edge-specific transient dynamics from the bulk plasma background. Subsequently, second-order temporal derivative analytics isolate the distinct oscillatory signature induced by the edge-localized impedance mismatch. Analysis reveals that a characteristic intensity modulation of argon emission lines correlates with transient shifts in the electron energy distribution function (EEDF). The predictive model, validated via leave-one-out cross-validation (LOOCV), achieves a cross-validated coefficient of determination Q 2 of 0.913. These results demonstrate that the proposed method provides a highly reliable and robust computational alternative to dedicated spatial sensors for diagnosing wafer edge uniformity. This computational framework transforms a standard 1D optical sensor into a spatially resolvable diagnostic tool.