Beyond cost-of-ownership: a causal methodology for costing wafer processing

Abstract

Classical cost-of-ownership data provides detailed cost data of equipment assets but does not provide wafer processing costs. Starting with a cost-of-ownership model, a wafer processing cost model was developed and validated. This cost-of-processing model provides wafer processing cost data from raw wafer through final passivation and parametric testing. This new model goes beyond classical cost-of-ownership data and captures more than just equipment costs-process, product, and fabricator costs are also captured. These costs are then causally spread to wafers via various algorithmic methodologies. In order to do this, some historical cost problems had to be addressed, such as how to properly weight equipment usage and account for dedicated equipment requirements, deal with measurement sampling, incorporate idle time and contingency, and account for different photolithographic field sizes. Output from the model was fully validated against actual spending and tied to accounting data in order to assure a full dollar capture. The model is currently being used for product costing, decision making, and cost reduction activities at the IBM Microelectronics Division Manufacturing Facility in Essex Junction, Vermont.