Fast modeling method of MOS transistor electricity statistical model

Abstract

A rapid modeling method of a MOS transistor electricity statistic model is provided, which comprises a step S1 of collecting lots of process parameters in the production line and acquiring process standard deviation related to the process parameters, a step S2 of selecting 6 model parameters and analyzing the model parameters and numerical difference method sensitivity of the process standard deviation, a step S3 of pushing out standard deviation of the model parameter reversely by the process standard deviation collected in the production line on the basis of given sensitivity and a step S4 of writing the standard deviation of the model parameter in model documents and simulating and adjusting the standard deviation value of the model parameter until the results of the standard deviationof simulated process and standard deviation data of measured process correspond to each other. The method can complete the modeling of the MOS transistor electricity statistic model more simply and rapidly and improve the efficiency and accuracy of integrated circuit design work.

Description

Technical field The invention relates to a fast modeling method for an electrical statistical model of a MOS transistor. Background technique In the process of manufacturing integrated circuit products, generally hundreds of process links are required. Due to the statistical uncertainty of each process, even for products of the same design, the circuit performance will be due to different manufacturing workshops, different batches of processes, different wafers and different chip positions. And corresponding changes occur. Therefore, the device model established during integrated circuit design should fully consider the influence of these uncertain statistical factors, and establish the corresponding device electrical statistical model. When using the established model to do Monte Carlo simulation, the statistical distribution of circuit performance obtained by simulation should be basically consistent with the statistical distribution of circuit performance manufactured by t...

AI Technical Summary

Problems solved by technology

However, these methods, especially the sensitivity analysis techniques, often involve complex mathematical operations and tedious numerical optimization

Images