Accuracy of channel resistance and current gain methods of Leff extraction

In this paper, we report a physically based interpretation of the MOSFET effective channel length, L_eff. L_eff of a MOSFET is usually extracted from measurements of the linear region current gain (1/K method) or the linear region resistance (resistance method) on transistors with different gate lengths. We find that the extracted L_eff of a submicron lightly doped drain/source (LDD) MOSFET is strongly influenced by diffusion of minority carriers (electrons for NMOSFETs) from the gate-controlled LDD overlap region into the inversion layer. This changes the resistance of the inversion layer at the source/drain ends and thus causes the extracted L_eff to deviate from the metallurgical channel length, L_met. The impact of minority carrier diffusion is stronger for shorter channel lengths; therefore, the extracted ΔL(=L_gate-L_eff where L_gate is the drawn gate length) increases with decreasing gate length. The choice of gate lengths used in L_eff extraction will, thus, influence the value for L_eff. MINIMOS 2D simulations have been used to study this effect. The results have been confirmed by actual measurements on submicron LDD MOSFETs. This analysis has been used to select the best choices of L_gate and bias condition to minimize errors in the measurement of L_eff