Numerically exact VRT levels of H2O and D2O reveal accurate predictions for infrared tunnelling, suggesting improved potential energy surfaces.
Numerically exact vibrational-rotational-tunnelling (VRT) levels of (H2O)2 and (D2O)2 have been computed in full dimensionality on a new highly accurate two-body potential energy surface (PES). Inter-molecular levels are computed with a basis of products of contracted intramolecular basis functions and inter-molecular functions that are products of three Wigner functions. Intra-molecular levels are computed with a product of contracted intra-molecular basis functions and contracted inter-molecular basis functions. We use a two-body PES that is fitted using the fundamental-invariant neural network (FI-NN) method using 740,000 ab initio points. Energies for points near the bottom of the well are computed without the frozen-core approximation. The PES has a root-mean-square fitting error of only 0.70 cm−1. All the experimental VRT fork origins and tunnelling splittings in the terahertz region (up to 150 cm−1) are in excellent agreement with our calculated levels. The largest error is 0.44 cm−1 for (H2O)2 and 0.85 cm−1 for (D2O)2. The calculated levels also agree very well with the 22, out of a possible 24, observed OD stretch vibration-tunnelling levels of (D2O)2 [J. Chem. Phys. 150, 164307 (2019);160, 114314 (2024)], the largest error being 0.35 cm−1. Coupling, which causes predissociation, makes OH stretch states of (H2O)2 difficult to observe. For the only rotationally resolved experimental OH stretch state, the as[A] “2s” state [J. Chem. Phys 91, 6613 (1989)], near 3738 cm−1, the 3 observed vibrational-tunnelling levels agree with the calculated levels to within 0.35 cm−1.
No takes yet. Share an insight, caveat, or question.
Wang et al. (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: