Statistical simulation for process design and optimization in IC manufacture

An approach to multidimensional statistical simulation for process design and optimization in IC manufacture is proposed. It essentially takes account of the sensitivity of circuit parameters to the random variability of process parameters. The approach is implemented in an algorithm and software for process statistical analysis and optimization. The response-surface methodology and pattern-recognition techniques are used for the approximation of relations between process and circuit parameters. The capabilities of the approach are evaluated from simulated and measured data on the fabrication of transistors by routine bipolar and CMOS technologies.