Multi-criteria optimization is used for developing molecular models for ethylene carbonate (EC) and propylene carbonate (PC), organic solvents commonly used in Li-ion batteries. The molecular geometry and partial charges of the solvents are obtained from quantum mechanical calculations. Using a novel optimization strategy that combines systematic variations of the Lennard-Jones parameters with a reduced units approach, the models are fitted to experimental data on the liquid density, vapor pressure, relative permittivity, and self-diffusion coefficient. Since no experimental data for the self-diffusion coefficient of pure EC were available in the literature, they are measured in this work using a gradient-based nuclear magnetic resonance technique. For all pure component properties, excellent agreement between experiment and simulation is obtained. Moreover, the predictive capabilities of the new solvent models are assessed by comparison to experimental data for the liquid density and relative permittivity of mixtures of EC and PC. In addition, molecular models for the anions PF6−, BF4−, and ClO4− in solutions of their lithium electrolytes in PC are developed using experimental data on the solution densities. Finally, the self-diffusion coefficients of LiPF6 in PC and in aqueous solution are predicted and compared, showing that diffusion is much slower in the organic solution due to the formation of larger solvent shells around the ions. Furthermore, an analysis of the radial distribution functions in these solutions suggests that the ions have much less impact on the structure of the solvent PC than on water.
Nuclear magnetic resonance (NMR) spectroscopy is a powerful tool for process monitoring. However, the quantitative analysis of highly diluted components or of components in small sample volumes is difficult with this analytical method due to its inherent lack of sensitivity. The hyperpolarization technique Overhauser dynamic nuclear polarization (ODNP) holds promise to solve this problem because it can provide strong signal enhancements. Furthermore, ODNP operates on short time scales and can therefore be used on flowing samples. Despite these advantages, to our knowledge, ODNP has never been applied for quantitative analysis of mixtures–probably because NMR signal enhancements by ODNP can vary greatly for different molecules, making quantitative analysis of mixtures difficult. We demonstrated that this problem can be solved by a robust calibration: three binary mixtures were studied as test cases in a wide range of concentrations by ODNP‐enhanced H and C NMR spectroscopy in continuous‐flow experiments with a benchtop NMR spectrometer using a new tailored calibration procedure. We show that quantitative analysis with ODNP‐enhanced NMR spectroscopy is possible even under these challenging conditions.
3 September 2026 in Industrial & Engineering Chemistry Research