Maintenance Scheduling of Plasma Etching Chamber in Wafer Fabrication for High-Yield Etching Process

Plasma etching is commonly employed in the chemical etching process of wafer fabrication because of its low cost and good control of etching profile and uniformity. In the plamas etching process, particle per wafer pass (PWP) test will be carried out periodically to reduce chamber contamination. When the number of particles exceeds a predefined threshold, the chamber must be wet-cleaned. Unfortunately, in actual manufacturing, the gaseous recipes used during etching vary with the etched materials, producing unexpected by-products, and render the particle count from PWP highly stochastic. Consequently, the high particle counts could exceed the threshold before the scheduled wet-clean maintenance, resulting in an unexpected high-yield loss. In this paper, we analyze the daily PWP results from an inductively coupled plasma etching (ICP) chamber. The behavior of the particle counts is modeled as a stochastic function of the accumulated gaseous flowing though the chamber. The particle count is found to follow a negative binomial (NB) distribution. The parameters of the NB distribution are determined from the industrial data, and with these parameters, we determine the optimal schedule for the wet-clean process with significant reduction in the failure occurrence.