Harnessing Halogen Bonds: Iodine and N -oxide Interactions in Cocrystals, Low-Transition Temperature Mixtures, and Fractional Deep Eutectic Solvents

Abstract

Deep eutectic solvents (DES) and low-transition temperature mixtures (LTTMs) formed through halogen bonding (XB) are a promising class of tunable media, and despite their unique properties, systems containing a neutral iodine (I2) component have been scarcely reported. Here, crystalline and liquid halogen-bonded systems of iodine and various heteroaromatic N-oxide acceptors are examined to determine the factors that govern phase formation. Four novel N-oxide.I2 cocrystals, three polyiodide salts, two fractional DES mixtures, and one LTTM were structurally and/or thermally characterized. Several liquid-like mixtures displayed glass transitions, melting points, or the combination thereof, and were analyzed via rheometry to support liquid-like behavior. Five machine learning (ML) models for viscosity predictions, trained off a dataset of 4004 established DES systems, were benchmarked, revealing XGBoost as the most accurate architecture, and enabling the prediction of viscosity values with a mean absolute error of 6.0 cP above 340 K. Parallel computational analysis using molecular dynamics, noncovalent interaction (NCI) mapping, and periodic energy decomposition analysis (pEDA) revealed the intermolecular interactions that differentiate crystalline and liquid phases. Our results establish a defined framework for classifying thermally ambiguous systems and for understanding the intermolecular factors that drive crystalline or liquid formation. Additionally, rheometry-supported machine learning can provide a powerful framework and pre-experiment screen for identifying phase formation in underexplored classes of LTTM/DESs.