An improved statistical modeling method based on silicon plant back-end ET data

This paper presents an efficient, accurate method for statistical MOSFET modeling to include effects of manufacturing process fluctuations. The variations of manufacturing process can be represented by the back-end electric test data routinely which were used to monitor manufacturing performance. Determined casing limits with model casing definitions helps to set up several types of statistical process control limits to improve IC process and yield and provide easy use manner for circuit optimization. The results described in the paper are not only benefits to the circuit designers, and also are benefits to manufacturing process control. The relations of statistical models and statistical process control are also discussed.