Theoretical and Computational Advances in Solvent Effect on NMR Spectroscopy

Accurate computational prediction of NMR chemical shifts requires proper treatment of solvent effects, which influence nuclear shielding through electrostatic polarization, hydrogen bonding, and dispersion interactions. This review examines computational methodologies from continuum solvation models (PCM, COSMO) to explicit approaches incorporating ab initio molecular dynamics and QM/MM schemes. We evaluate method performance and provide practical guidelines for methodology selection based on solute-solvent interaction types and system size. Emerging machine learning approaches are discussed. While significant progress has been achieved, quantitative accuracy requires careful integration of quantum mechanical theory, statistical mechanics, and validation against experimental data.