Comprehensive quality assurance methodology for BSIM4.5 corner parameter extraction

Quality assurance of shipped compact model parameters is one of the crucial problems for circuit designers as well as parameter providers in process foundries. However, very few reports have been published on this issue and it is virtually veiled so far. In this paper, we firstly propose a comprehensive methodology focused on corner parameter assessment for compact MOS models. Key technologies are: (1) rational target description of corner performances, (2) a new quantitative error definition of target performances, and (3) quality assurance strategy which can leads to a standardization scheme in parameter extraction framework. The new QA technology was verified with 90nm BSIM4.5 corner MOS model parameters. This activity has been supported by five companies who applied to STARC corner-parameter extraction contest using identical I-V and C-V experimental data and corner-performance specifications. The result shows significant features that the quality of the corner-parameters is widely spread in terms of accuracy. Our new definition of allover RMS error is found to range 1.32% to 11.18% depending on the corner-parameters from contest applicants. Note that this work will give a new quantitative flow & algorithm for comprehensive corner parameter assessment.