We provide an exhaustive characterization of structural properties and nuclear dynamics in tungstic acid (WO3·H2O). To this end, we employ neutron and X-ray diffraction (ND and XRD) combined with inelastic neutron scattering (INS) and neutron Compton scattering (NCS) experiments, and we corroborate the analysis with extensive ab initio modeling. The first step in our analysis is the elucidation of the crystal structure based on the refinement of low-temperature powder ND data, extending the knowledge gained from XRD analysis of a mineral specimen of tungstite. These results are confronted with low-temperature INS experiments and zero-temperature phonon calculations. The analysis reveals an inconsistency in the definition of the structure of confined water with respect to crystallographic data, also showing a concomitant failure of the phonon calculations due to a strongly anharmonic confining potential. Extending the computational route toward ab initio MD (AIMD) simulations allows us to probe different structural configurations and provides an improved description of the vibrational dynamics as compared to high-resolution INS experiments, nevertheless, requiring the use of effective classical temperatures. The analysis of both INS and the NCS data reveals a remarkable similarity to the nuclear dynamics earlier reported for water confined in single-wall carbon nanotubes (SWNT), which has been qualitatively described as a new phase of ice. Our analysis reveals a strong two-dimensional hydrogen-bonding network, similar to the shell model for water in SWNT. The reported NCS data show narrowing of the hydrogen momentum distribution with respect to the reference ab initio calculations, indicating a great deal of conformational freedom due to spatial delocalization of protons in the ground state of the system, a clear signature of the quantum character of the nuclei.

The authors gratefully acknowledge the financial support from the U.K. Science and Technologies Facilities Council (STFC), granting the beamtime on WISH, TOSCA, and VESUVIO. The authors also acknowledge the computing resources provided by the STFC Scientific Computing Department’s SCARF cluster. K.D. gratefully acknowledges financial support from the Gipuzkoako Foru Aldundia under Grant Number 2020-CIEN-000009-01. This work has been gratefully supported by PL-Grid Infrastructure and the PROMETHEUS facility.

Peer reviewed