How do I use EM simulation to optimize an impedance matching network at millimeter wave frequencies?

Using EM simulation to optimize an impedance matching network at millimeter-wave frequencies (30-100+ GHz) is essential because circuit-level (schematic) simulation with ideal or simple parasitic models is insufficient at these frequencies; the physical layout geometry dominates the circuit behavior. The EM simulation workflow is: design the initial matching network in a circuit simulator (ADS, AWR) using ideal elements to establish target component values, create a physical layout that implements these values using transmission line sections (open stubs for capacitors, short stubs or high-impedance lines for inductors) and remaining lumped components (if any, only 0201/01005 footprints), simulate the layout in a 3D EM solver (Ansys HFSS for highest accuracy, or 2.5D solvers like Keysight Momentum or Sonnet for faster results) with accurate material models (substrate Er and tan_d at mmW, copper conductivity and surface roughness, component S-parameter models from manufacturers), extract the resulting S-parameters from the EM simulation and import them into the circuit simulator to compare against the target response, adjust the layout dimensions (stub lengths, line widths, component values) to compensate for parasitic effects revealed by the EM simulation, and iterate until the EM-simulated response meets the specification. Typically 3-5 iterations are needed. The final EM simulation serves as a virtual prototype, providing confidence that the manufactured circuit will perform as designed.