This dataset is created as part of the article using phase-field modeling a physics-based computational thermodynamics approach: - Khatamsaz, Danial, Vahid Attari, and Raymundo Arroyave. "Microstructure-Aware Bayesian Materials Design." arXiv preprint arXiv:2502.03727 (2025). The design space, defined by four model parameters, is sampled 10,000 times using Latin Hypercube Sampling (LHS). For each sampled condition, a forward phase-field simulation is performed to generate time-series microstructures under an isothermal heat treatment at 600°C. To reduce model complexity, the heat treatment temperature and duration is fixed across all simulations during Bayesian Optimization, as the variations in microstructure resulting from changes in composition and elastic parameters provide sufficient diversity. Therefore, the processing condition is represented by a constant thermal treatment. A representative snapshot of the resulting dataset under these conditions is included. Included files: CSV File: Tabulated chemistry-microstructure-property columns. ZIP file containting static images corresponding to 10,000 simulation. If you need time-series, please contact Dr. Vahid Attari. There is no sufficient space in Zenodo to upload time-series data. A jpg file showcasing a pair-plot for exploratory analysis for the tabulated dataset.