Modified PCA algorithm for the end point monitoring of the small open area plasma etching process using the whole optical emission spectra

An end point detection algorithm for small area etching was developed using the modified principal component analysis. Because the traditional end point detection techniques used a few manually selected wavelength, noise render them ineffective easily. And it is hard to select the important wavelength when the open area gets small. Modified principal component analysis including the concept of ‘product’ with the whole optical emission spectra was developed for the effective end point detection. The ‘product’ means the multiplication of the raw data and loading vector of itself whereas the score vector uses the normalized data and loading vector. And this algorithm was applied for the small open area of SiO
2
etching. In conclusions, the single wavelength signals of SiF (440.2nm), CO (483.5nm), and Si (505.6nm) was compared with the third product monitoring by the definition of ‘signal change to noise ratio’. As the results, end point of 0.6% open area could be monitored using this algorithm, whereas the traditional single wavelength method could hardly detect.