A breakthrough in utilization maximization via real-time tool performance feedback

In semi-conductor manufacturing, it is a well-known fact that capital equipment depreciation is the largest Fab cost Pareto item. Therefore, it is critical to keep these expensive machines (or tools) highly utilized. Nevertheless, a level of idleness of 10% and even 20% on machines, in many High Volume Manufacturing (HVM) facilities, is still not uncommon. There are two approaches to maximizing tool utilization by minimizing idleness. The first deals with it at the capacity planning level, attempting to minimize tool purchases, thus enforcing lower gaps and consequently higher utilization on any set of tools. The second approach attempts to deal with it at the tactical-operational level, by providing effective monitoring and prompt feedback on opportunities to improve utilization. In this paper, we present a methodology that supports the latter, denoted as ´No-Loss´. ´No-Loss´ stands for NOrmalized LOading per Specific Shift. It is a methodology which is supported by a set of indicators at the tool, area, and factory level. The indicators provide online real-time feedback to the manufacturing technicians and managers that assist in keeping the tools loaded to the maximum. This methodology was successfully implemented in an HVM facility at Intel, and has resulted, since its implementation during 2002, in significant improvements to tools´ utilization and consequently cost savings. For example, for Diffusion tools, utilization improved by 5.8% on the average per tool, and work-in-process at the Diffusion area decreased by 17% post the implementation. Additionally, thanks to avoidance of partial batches, weekly cost savings were estimated at about $3.4 K per tool, or the equivalent of approximately $1.5 M annually for an HVM Diffusion area alone. The ´No-Loss´ methodology is currently in process of proliferation to other functional areas and across other production facilities at Intel. In this paper, we shall review the methodology in detail, provide the fundamental formulae for the indicators, illustrate usage via screen shots from the clean-room, present the actions required for successful implementation, and discuss the results of the implementation to date in terms of utilization improvements and operational cost savings.