You are designing a standard cell for a 3nm process. A complex CMOS logic gate implements the Boolean function Y = ~(A + BC). Assuming all NMOS transistors are sized 'W' and PMOS are sized '2W' for an equivalent reference inverter (to balance rise/fall times), what should be the worst-case sizing of the PMOS transistors in the pull-up network to perfectly match the rise time of the reference inverter?
The function Y = ~(A + BC) is a NOR-like structure. By De Morgan's laws, the pull-up network (PUN) is A' * (B' + C'). This means PMOS A is in series with a parallel combination of PMOS B and C. The worst-case rise path goes through two PMOS transistors in series (either A and B, or A and C). To achieve the equivalent resistance of a single reference PMOS (sized 2W), series transistors must be doubled in width. Therefore, they must all be sized to 4W.