Aluminum–Water Hydrolysis Hydrogen-Yield and Kinetics Dataset (provenance-tracked, leakage-aware), v1.0

A provenance-tracked, quality-tiered dataset for hydrogen production by aluminum–water hydrolysis, assembled from 31 independent published studies (315 yield rows; 76 kinetic rows). Each row traces to a study DOI and a table/figure source reference. Methodology covariates (temperature-control method, hydrogen measurement apparatus, vessel type, activation method) and quality tier are recorded per row and verified by blind independent-agent inter-rater reliability (post-adjudication Cohen's κ = 1.00 for temperature_control and measurement_method; κ = 0.826 for system_class). The dataset accompanies the paper "Apparent Contradictions in the Aluminum–Water Hydrolysis Literature are Predominantly Methodological: a Provenance-Tracked, Leakage-Controlled Analysis" (D. Unal, Energy and AI). Files:- aih2_v1.csv — 315 experimental yield rows × 30 columns (target: h2_yield_pct; features: alloy composition, temperature, alkali, activation method, methodology covariates; provenance).- rate_extraction.csv — 76 kinetic rows (activation energy, peak rate, rate constant, t₈₀).- data_dictionary.md — full column definitions for both tables.- CITATION.cff — citation metadata.- LICENSE_CC-BY-4.0.txt — license text. Scope: aqueous aluminum–water/alkaline hydrolysis for H₂ production only. Out of scope: electrolysis, photocatalysis, non-aqueous routes. Dataset is observational; all reported analyses are associational.