Description
This study investigates the structural properties of the diiododioxouranium(VI) dihydrate complex UO2I2(OH2)2 using a multi-scale computational framework.
Initial molecular coordinates were derived from published literature (https://doi.org/10.1021/ja030260r, Figure2) utilizing deep-learning-assisted coordinate generation.
The structural refinement workflow began by deploying MACE machine-learned interatomic potentials, specifically extended via custom model files to handle high-atomic-number actinide elements like uranium (U). A dedicated Python pipeline was developed to automate the geometry analysis and execute the initial MACE-based structural optimization.
Following the machine-learning phase, high-level quantum-chemical calculations were designed and executed within the NWChem framework. Density functional theory (DFT) simulations were performed, initially through single-point energy evaluations and subsequently via full geometry optimization, utilizing effective core potentials (ECPs) to account for scalar relativistic effects.
While the current ECP framework accurately captures scalar relativity, resolving the more accurate U–I bond distance requires accounting for more rigorous electronic structure treatment. Future work will expand this multi-scale workflow to incorporate spin-orbit relativistic interactions, isolating the specific impact of spin-orbit coupling on actinide-halogen bonding.
Working files are available at https://github.com/miroi/Molecular-and-Materials-Modeling-2026-August/tree/main/molecular-modeling/abstract_Ucompound_mjc .