Integrated circuit performance modeling using MOSFET in-line parameters
Current methods of predicting the final-test performance of complex MOSFET devices are time consuming and can be unreliable. Also, large quantities of empirical data, required to develop a correlation to package-test speed, can only be collected weeks into the fabrication process. It would be useful early in the process to predict the final test IC performance before unnecessary commitments are made to a customer. To help solve this problem, statistical analysis of measurable, in-line parameters of a MOSFET transistor are used to develop process and circuit models of the devices. The parameters determined to have the greatest influence on speed performance, based on simulating IC speed, are determined. The known variation in these parameters can aid in better predictions of IC performance spreads.