According to the customer's various needs and the shrinking design rules for any device, it has been required to control process tools with too many input parameters, in order to cope with the more complicated formulas for high-mix manufacturing. In this paper, "Multiple SPC" is proposed as a new fault detection and classification (FDC) solution that can control the highly advanced tools in the nano-scale age. Multiple SPC is composed of three parts: (i) univariate SPC with control limits, robust enough to handle the frequent changes for tool maintenance and generic enough such that it is possible to normalize the input parameters, (ii) multivariate SPC with an integrated index correlating to the normalized parameters in part (i), and (iii) fault classification for classifying fault types with the patterns of the correlated parameters. The proposed method which covers the recipes (formulas) of various design rule from 130 to 65 nm, was tested by in an application to the plasma etcher.
展开▼