A curated dataset of thermal stability measurements for atomic layer deposition precursor compounds

Atomic layer deposition (ALD) is a critical thin film technology, yet thermal stability data for ALD precursors are scattered across peer-reviewed literature, safety data sheets, and public databases with inconsistent formats. We present a curated dataset compiled by screening approximately 2,000 publications, comprising 1,117 records for ALD precursor compounds, of which 719 (64.4%) include thermal stability measurements, covering 939 unique molecular structures and over 50 elements. Each record is annotated with a canonical SMILES string, source provenance, temperature type, application status classification, ligand class, and data quality flags. A systematic curation workflow identified self-accelerating decomposition temperature contamination (75 entries), extreme outliers (6 entries), and conflicting duplicate measurements (21 groups). The cleaned subset contains 566 validated entries. All data files are deposited in a public repository under CC BY 4.0 license. Descriptive statistics characterize the dataset composition across metal groups and ligand classes, while a baseline predictive model confirms that measurement heterogeneity dominates over structure–property relationships.