Unified system for manufacturing process control and data collection

In the Endicott manufacturing lines, there were many varied methods used for controlling process parameters, ranging from paper SPC charts to automated closed loop systems. There were 'islands of automation', many processes being controlled or monitored using unrelated and incompatible: tools. It was not possible to share data between processes, or even to merge data for statistical analyses. In many cases, it was not even possible to gain access to the data from an individual process, It was also pointed out that many processes did not have any controls in place at all, and some that were using paper SPC charts had no easy means of acquiring the data for further analysis. These disparities were recognized as problems which impeded the ability to effectively analyze defects and yield detractors and develop solutions for them. By either not having the right data collected, or not being able to access necessary data, engineers were constrained in their efforts to improve their processes. Recognizing the need for a central focus, two groups got together and formed a larger, cross functional team to, once and for all, do it right. Meetings were held in the Endicott Decision Support Center with representatives of all of the involved organizations (Manufacturing Engineering, with detailed process knowledge, especially of critical process and product parameters; Quality Engineering, with statistical analysis and control techniques; Manufacturing, with implementation support; Information Systems, with hardware/software infrastructure and databases; and Tool Control Engineering, with machine and tool control expertise) to brainstorm together to identify current tools and techniques being used, current problems being encountered and potential solutions for the future. The primary objectives of this team were to consolidate the tools used for controlling manufacturing tools and processes; to minimize the variability in the manufacturing process parameters; to develop a method of sharing data among the processes, enabling a proactive feed forward information exchange; and to align the process data with product data (e.g. yields) for correlation analyses. The Team designed a solution based on a suite of tools that are capable of working together and sharing data. This architecture was built around a statistical process control application that had been in use in other areas of the Endicott site. Since there was already local expertise in using this tool, deployment was made much easier. The SPC tool utilizes a relational database, so the process data would be able to be merged with product data. Another component of the architecture was to provide a mechanism to collect data at various locations and have the data automatically input into the SPC database. Individual processes are already realizing benefits from increased attention and control, with smaller and fewer yield hits, and shorter failure analysis times when upsets do occur. We anticipate similar results from other processes as our roll out continues throughout the plant.