Translucent Structural Steel (MoO₂): Genetic Algorithm + DFT + FEM — Corrected Dataset v3

Version 3 — Corrected electronic ground state (June 2026) This dataset accompanies the manuscript 'Translucent Structural Steel Based on MoO₂: Computational Design via Genetic Algorithms, Density Functional Theory, and Finite Element Structural Validation' (Journal of Constructional Steel Research, Elsevier). Correction in v3 (vs. v2): the diatomic Mo–O potential energy surface in v2 had two off-curve points caused by SCF convergence to an excited solution. Re-running with a robust, stability-checked SCF protocol revealed that the root cause was the spin state: v2 used the triplet, but the ground state (matching the experimental ⁵Δ) is the quintet. Confirmed at the same level (UB3LYP/def2-SVP+ECP28, PySCF 2.13): the quintet lies below the triplet throughout the well (≈ 4.9 kcal/mol near equilibrium), the PES is smooth, and the solution is essentially spin-pure (⟨S²⟩ = 6.03). Corrected DFT values (quintet ⁵Δ): r_eq = 1.6956 Å, E = −143.3043 Eh, HOMO–LUMO gap = 2.012 eV, dipole = 3.19 D, ν = 975.2 cm⁻¹ (experimental 897–1009 cm⁻¹). Unchanged from v2: structural/optical results — E = 288.6 GPa (Voigt–Reuss–Hill, Taib et al. 2019), n = 2.03, T ≈ 45%, and the full FEM analysis (anastruct; W14×211 columns / W12×190 beams; σmax = 251.9 MPa = 76.3% fy; all load combinations PASS). Version history: v1 (10.5281/zenodo.14345994) contained an ORCA H₂O simulation in error; v2 (10.5281/zenodo.20263263) corrected this to a PySCF Mo–O calculation but in the triplet state; v3 corrects the ground state to the quintet ⁵Δ. Patent: BR 10 2024 024314-5 (INPI, Brazil)Author: Luiz Ricardo Mantovani da Silva — UNISAGRADO, Bauru, SP, Brazil