This guide may be used when changing or adding specifications to the order form aspect of silicon wafer specifications, such as SEMI M1, SEMI M57, SEMI M61, or SEMI M62.

Statistical specifications apply to all processes that have been statistically characterized. Solutions are given for two product distribution shapes (normal & lognormal), and the advantages of these solutions are explained. Appendix 1 shows the statistically characterized shape for many common silicon wafer processes.

This guide applies to processes related to the production and use of silicon wafers. It may also be applied to the production and use of other materials.

This approach implies that the quality level shipped is the same as the quality level produced.

This methodology can be an effective tool for driving quality improvement.

This procedure can be coupled with other techniques for centering the mean and reducing the variation within the process distribution as deemed necessary by the customer and supplier.

Appendix 2 further explains the rationale for this approach.

NOTICE: This standard does not purport to address safety issues, if any, associated with its use. It is the responsibility of the users of this standard to establish appropriate safety and health practices and determine the applicability of regulatory or other limitations prior to use.

Purpose

Specifications are based on requirements negotiated between trading partners. This document describes an explicit specification form that defines the risk level as a part of parametric specifications. This approach uses process capability information to focus quality improvement efforts, reduce sampling, and maintain low risks. It is based on the fundamental belief that specifications should facilitate the movement toward processed-in quality instead of inspected-in quality.

It is important for suppliers and their customers to acknowledge and mutually agree on quality levels so the methods employed will satisfy their expectations. Statistical specifications provide a convenient way to do this.

Statistical specifications are designed to facilitate the movement toward processed-in quality. They are most appropriate for processes that have been statistically characterized. This means the shape of the statistical distribution that created the product is known, or can be approximated to the satisfaction of the customer and the supplier. It also implies that the statistical control of the process and the measurement systems are defined to the level that is necessary to meet the current needs.