Critical-dimension escapes after lithography get argued as dose, focus, or resist age. Those matter. Under many stacks sits a thinner film that rarely makes the war room slide: the bottom anti-reflective coating (BARC). When BARC thickness walks, reflectivity swings, standing-wave effects move, and the resist process window shrinks while the scanner recipe still claims yesterday’s center point.
Photoresist edge-bead removal and spin-chuck vacuum hold are coater-track neighbors. Wafer bow fights the chuck flatness budget. This page is BARC thickness and coat/bake honesty as a lithography input—not scanner stitching, not EUV pellicle supply.
BARC exists to damp substrate reflectivity so the resist sees a more stable optical environment. Thickness targets are tight because small deltas change the interference condition. Causes of drift are ordinary and expensive: cup airflow imbalance, dispense volume or viscosity lot shift, post-apply bake hot-plate uniformity, and chuck vacuum that lets the wafer skate during spin. A “resist-only” disposition that never pulls BARC metrology will keep rewriting exposure dose forever.
Metrology that belongs before the dose knob
Trend BARC thickness (and within-wafer range) on the same cadence as resist thickness for critical layers. After a track PM, bake-plate swap, or BARC chemistry lot change, run monitors before production wafers. When CD maps show radial or crescent signatures that match coat hardware—not scanner slit—hold the lot against BARC genealogy.
Lithography owns the exposure. The track owns whether the optical stack underneath still matches the model the process window assumed. Sites that only trend resist thickness fund quiet focus and CD fights that a BARC map would have ended in one shift.