Method for Predicting Dipole Moments of Complex Molecules for Use in Thermophysical Property Estimation

Abstract

We present a formalism for accurate estimation of dipole moments using quantum mechanics for complex molecules having conformational degrees of freedom. Dipole moments of complex molecules are often needed for use in correlations for estimating viscosities and other thermophysical properties. However, experimentally measured dipole moments are not available for many molecules, especially those used in proprietary industrial processes and products. Many complex molecules have dipole moments that change significantly in response to the conformation of the molecules. We show that proper accounting of the conformation-dependent dipole moment may be required to achieve an acceptably accurate estimate of the experimental dipole moment and provide recommendations on efficient estimation techniques. We also demonstrate that for molecules with dipole moments above about 1.3 D reasonably accurate estimation of the dipole moment is required for reliable prediction of vapor phase viscosity, whereas estimation of thermal conductivity is less sensitive to the dipole moment.

Assessing conformer energies using electronic structure and machine learning methods

“…Moreover, many computational predictions rely on Boltzmann‐weighted averages of multiple thermally accessible conformers, including NMR prediction, [ 1,2 ] reactions, and even understanding the effects of dipole moments on solvent viscosity. [ 68 ] Consequently, deriving accurate relative energies of molecular conformers is a crucial task, as discussed below.…”

“…One recent example comes from understanding the effects of dipole moments on solvent viscosity. [ 68 ] Finding all conformers with proper weighting is thus crucial to predicting the dipole moment of an ensemble of different conformers.…”

Assessing Conformer Energies using Electronic Structure and Machine Learning Methods

“…Moreover, many computational predictions rely on Boltzmann-weighted averages of multiple thermally accessible conformers, including NMR prediction, 1,2 reactions, and even understanding the effects of dipole moments on solvent viscosity. 67 Consequently, deriving accurate rankings as well as accurate relative energies of molecular conformers is a crucial task.…”

Assessing Conformer Energies using Electronic Structure and Machine Learning Methods

“…Moreover, many computational predictions rely on Boltzmann-weighted averages of multiple thermally accessible conformers, including NMR prediction, [2], [1] reactions, and even understanding the effects of dipole moments on solvent viscosity. [68] Consequently, deriving accurate relative energies of molecular conformers is a crucial task, as discussed below.…”

“…of multiple thermally accessible conformers, even small differences in conformer energies have large effects on populations involved in property prediction, as recently discussed with conformer and polarity effects on solvent viscosity. [68] Machine Learning Batch Evaluation An advantage for ML and force field predictions is the ability to batch evaluate by loading all conformers of a molecule at once and evaluating them as a batch opposed to evaluating one at a time, as with conventional quantum chemistry methods. Table 5 indicates the median sequential times from Table 1 and median time per single point in batch evaluation.…”