# Scanner/Track Lithography Model Card v1

## Purpose

Provide a deterministic, normalized teaching replay that keeps scanner imaging, resist coating, focus-dose balance, PEB/development, overlay, stochastic defects, metrology, and RIE handoff in one evidence chain.

## Mechanism Boundary

The model has no physical units and no calibrated physics. Its coefficients and thresholds are original category-shaping choices. It does not solve optics, exposure, reaction-diffusion, dissolution, stochastic chemistry, overlay, stage dynamics, resist profile, or etch transfer.

## Validation Boundary

Independent validation proves complete Cartesian order, equation replay, exact ledgers, closed-label reachability, prohibited zeros, evidence-pair identity, CSV/JSON cell identity, two byte-identical isolated reruns, source boundary, public-v109 hold, and hygiene. It does not prove semiconductor-process accuracy.

## Required Replacement

Before any practical use, replace every normalized state with owned wafer, tool, material, scanner, track, metrology, integration, yield, reliability, and disposition evidence. Keep DUV, EUV, coat, PEB, develop, overlay, stochastic, and etch claims separately accountable.

## Reachability Amendment

The first deterministic sweep reached all eight regimes but masked four downstream labels after precedence filtering. RA-001 changes only categorical label rules: yield RIE-handoff review from <0.62 to <0.64, yield metrology review from <0.62 to <0.66 after the RIE branch, and the two downstream RIE limited bands from <0.54 to <0.60. All 20 proxy equations, 16 integer ledger equations, sources, axes, evidence pairs, prohibited zeros, and public-v109 boundaries remain unchanged.