Isotope effects in 2D correlation infrared spectra of water: HEOM analysis of molecular dynamics-based machine learning models

We model, simulate, and analyze the intramolecular modes of liquid H2O and D2O to elucidate how energy excitation, relaxation, and vibrational dephasing interplay through anharmonic mode-mode coupling. Our analysis employs two-dimensional (2D) correlation spectra, a representative observable in nonlinear infrared vibrational spectroscopy. Accurate reproduction of these 2D spectral profiles requires not only a precise dynamical description of intramolecular vibrations but also an appropriate treatment of thermal environmental effects arising from strong interactions with surrounding molecules, which act as thermal baths. Capturing the essential features of the 2D spectra further demands a non-Markovian, non-perturbative, and nonlinear description of the interactions between intramolecular modes and their baths. To this end, we adopt a hierarchical equations of motion framework to compute the 2D spectra. By comparing the resulting spectra of H2O and D2O, we explore the underlying mechanisms governing their complex energy and phase relaxation dynamics.