A Correlation-Considered Variation-Aware Interconnect Parasitic Profile Extraction Method

This paper presents a novel technique of performing correlation-considered statistical interconnect parasitic parameter extraction on 40-nm process technology. Five test structures are designed for the extraction. Then, 3-D simulations are performed, and typical interconnect technology profile (ITP) file is extracted. Based on the full mapping measurements of test structures, their distribution analysis, and particularly the correlations between parasitic coupling capacitance of each metal layer, statistical parameters are extracted using principal-component-analysis method. Then, a complete flow of correlation-considered variation-aware ITP file extraction is concluded. According to the comparison between distributions of Monte Carlo simulations and measurements of output signal periods of ring oscillators, the extracted statistical ITP is proved to be accurate.