Multivariate analysis of noise-corrupted PECVD data
A prominent goal of plasma enhanced chemical vapor deposition (PECVD) is the description of solid film properties, such as refractive index, carbon content, etc, in terms of input variables such as gas pressure, substrate temperature and others. This goal is complicated because of the high dimensions of the input and response variable space. In this paper, data is analysed from C:H film deposition with proper account of measurement errors. It is shown that errors reduce correlations, and that the whole data set can be well described by a multivariate normal distribution.
Bibliographic Reference: Article: Thin Solid Films, Vol. 307 (1997) pp. 65-70
Record Number: 199810321 / Last updated on: 1998-03-09
Original language: en
Available languages: en