Physics-Informed Compact Model for SF₆/O₂ Plasma Etching

We propose a process simulation/emulation flow which incorporates the generation and application of compact models for fast geometry generation, suitable for design-technology co-optimization (DTCO). The methodology is applied to develop a compact model for etching in a SF6/O-2 plasma using 35 data points from experimental measurements and results from a calibrated physical model. The compact model is tested against physical simulations using 80 random points in the input parameter domain consisting of varying the oxygen content and pressure in the plasma chamber. The model is shown to produce highly accurate geometries with an average error for the depth and width at half depth (WAHD) of 2.0% and 1.0%, respectively. The simulation speedup is more than three orders of magnitude, when compared to the physical model; it should also be noted that the parallelized physical model was executed on 40 cores, while the compact model was run on a single core.