Direct extraction of small-signal model parameters for nanoscale MOSFETs

A small-signal model of a nanoscale MOSFET becomes very important for designing RFICs and characterizing processes and devices. To extract parasitic resistances and inductances, a direct method presented in D. Lovelace et al. (1994) based on S-parameter measurements at zero bias is developed, but the assumption of zero transconductance (g_gm) and output conductance (g_ds) at external Vgs=0V is not valid in real MOSFETs. Thus, Z-parameter extraction equations as stated in D. Lovelace et al. (1994) result in significant error according to the selected frequency range. Another extraction method according to S. Lee (2003) using a linear regression of high-frequency data has been proposed to determine resistances and inductances, but much higher frequency data are required to apply for nanoscale MOSFETs, because of high g_m and g_ds. In this paper, a direct method to extract resistances and inductances of nanoscale MOSFETs is proposed using a linear regression of high-frequency data at zero gate voltage.