Empirical-mechanistic Modeling for Monitoring Multiproduct Manufacturing Process Applied to the Fabrication of Integrated Circuits

• Activity Number: 428
• Type: Contributed
• Date/Time: Thursday, August 12, 2004 : 10:30 AM to 12:20 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): Spencer B. Graves*+
• Companies: PDF Solutions
• Keywords: modeling spatial variations ; statistical process control (SPC) ; Bayesian sequential updating ; Kalman filtering ; Monitoring low-volume, multiproduct production ; optimizing preventive maintenance and equipment replacement cycles

Abstract

Many manufacturing facilities produce multiple products (sometimes with low volumes or short life cycles) using the same or similar equipment. This presentation will discuss conceptual tools for decomposing variability into product and process specific components. This includes tracing variations to the structure and geometry of a product and to specific types of equipment malfunctions. Applications include optimization of preventive maintenance and equipment replacement cycles. These concepts will be illustrated in studying spatial variations across a wafer in the fabrication of integrated circuits (ICs/computer chips). Parameters estimated from appropriate models can be monitored over time, across products, and between early and mature production, decomposing variability between product and equipment. Simple models may not be available for other phenomena such as scratches, whose presence may be more easily detected in residuals. Parameter estimation can be built into Bayesian sequential analyses/Kalman filters. Multiple Kalman filters running simultaneously can provide sensitive data-mining tools tied to manufacturing physics and management.