Evidence-graded coding data for AI and data-driven thin-film deposition in semiconductor manufacturing systems (DVL + T/O/L/A, codebook v1.1, assignment set r11)

Master coding workbook, codebooks, audit records, and historical-control search datasupporting the review article "AI and Data-Driven Thin-Film Deposition for SemiconductorManufacturing Systems: An Evidence-Graded Review from Molecular Screening to AuthorizedClosed-Loop Control". The workbook contains 132 coded rows drawn from 130 sources, of which 81 carry adeposition-relevant claim. Each claim is graded on a Deposition Validation Level (DVL)recording where its reported evidence touches the physical deposition process, togetherwith four independently coded descriptors: transfer (T), observability (O), latency (L),and authorization (A). Both codebook versions are included — v1.0, frozen on 2026-08-06and applying to DVL, T, L and A, and v1.1, issued 2026-08-13 and revising theobservability axis only — so that any assignment can be rechecked against the definitionin force when it was made. Also included are the boundary-case log, the full coding change log, the freeze recordwith its documented protocol deviation, a self-verifying count-check sheet, the 28-rowpost-freeze holdout with both codings side by side, the 14-row targeted audit of rarelevels, and a reference-number concordance between the freeze-date and currentmanuscript numbering. Every count reported in the article is regenerated from the Master_Coding sheet; README.mdgives the exact expressions and states the limits of the reproducibility evidence.