35.1 TECHNOLOGY OVERVIEW

Advanced Process Control (APC) within semiconductors has essentially come to mean three separate efforts, each engaging software tools and mathematical techniques best suited to the scope and complexity of the task at hand.

The first of these is, effectively, line level data analysis and overall process optimization, typically conducted by a Product Engineering (sometimes known as Characterization) organization acting under the mantle of Yield Management to essentially tune the manufacturing line for production of an optimum parts distribution at high yield. This typically involves limited experimentation combined with data mining and modeling techniques polynomial models are popular, with more comprehensive multidimensional methods (Geometric Process Modeling) and neural nets occasionally in use as well but seldom approaches what would be recognized as control in other industries. Detection of problems visible at the level of in-line parametric measurements is also a focus and, with rare exception, all fabs have standing efforts of this nature in place.

Second in popularity is R2R Control. Originating at Semiconductor Research Corporation s MIT SEMATECH Center of Excellence (SRC s MIT SCOE) as Run-by-Run Control in the late 1980s, this became Run-to-Run Control once effort was focused at SRC s University of Michigan SCOE and has since been abbreviated by industry as R2R control. This relies on a model to adjust variable settings in a process recipe. That model is usually linear, as limited time series effects associated with a previous wafer or wafers...