Probing Defects and Correlations in the Hydrogen-Bond Network of ab Initio Water

Piero Gasparotto, Ali Hassanali, Michele CeriottiJournal of Chemical Theory and Computation 2016 · DOI · arXiv

Abstract

The hydrogen-bond network of water is characterized by the presence of coordination defects relative to the ideal tetrahedral network of ice, whose fluctuations determine the static and time-dependent properties of the liquid. Because of topological constraints, such defects do not come alone but are highly correlated coming in a plethora of different pairs. Here we discuss in detail such correlations in the case of ab initio water models and show that they have interesting similarities to regular and defective solid phases of water. Although defect correlations involve deviations from idealized tetrahedrality, they can still be regarded as weaker hydrogen bonds that retain a high degree of directionality. We also investigate how the structure and population of coordination defects is affected by approximations to the interatomic potential, finding that, in most cases, the qualitative features of the hydrogen-bond network are remarkably robust.

Introduction

While liquid water is frequently conceptualized as a distorted tetrahedral network, its true complexity resides in its "coordination defects"—molecules that deviate from the ideal two-donor, two-acceptor ($2_D2_A$) pattern found in ice. These defects are not merely random glitches; their fluctuations drive the anomalous thermodynamic and dynamic properties of the liquid. Understanding these defects is challenging because traditional geometric definitions of hydrogen bonds (HBs) are often too rigid to capture the "fuzzy" nature of bonds in transition. Furthermore, these defects are relatively rare, requiring advanced sampling techniques to observe their interactions and clustering.

Methods

To probe the hydrogen-bond network, the researchers utilized Ab Initio Molecular Dynamics (AIMD) based on Density Functional Theory (DFT). To overcome the sampling limitations of rare defect states, they employed Parallel Tempering with the Well-Tempered Ensemble (PTWTE).

A key methodological innovation is the use of Probabilistic Analysis of Molecular Motifs (PAMM). Unlike binary "bond or no bond" rules, PAMM provides a continuous, probabilistic fingerprint for H-bonds. This allows for a robust classification of coordination states (e.g., $1_D2_A$ or $2_D3_A$) that is less sensitive to specific simulation parameters. The study compared various electronic structure approximations, including BLYP and B3LYP functionals, van der Waals (vdW) dispersion corrections (D3), and Nuclear Quantum Effects (NQE) via the PIGLET method, alongside state-of-the-art force fields like MB-pol.

Free energy surface $F(s_A, s_D)$ and population analysis of water coordination states using the PAMM descriptor.

Results and Discussion

Defects as "Weak Bonds"

A central finding is that under-coordinated defects—molecules missing a bond—still exhibit strong directional correlations. Rather than being "broken" bonds, these interactions behave like weaker, more flexible versions of standard hydrogen bonds. The spatial distribution functions (SDFs) for various defect states reveal that even highly distorted configurations retain a high degree of directionality.

Topological Clustering and the Ice VIII Connection

Defects do not appear in isolation. Due to the topological constraints of the network, a defect at one site strongly influences the likelihood of a defect at a neighboring site. Interestingly, the local environment around under-coordinated defects in liquid water bears a striking resemblance to the structure of Ice VIII, a high-pressure solid phase. This suggests that the liquid phase explores structural motifs found in various solid phases during its natural fluctuations.

Spatial distribution functions and radial distribution functions decomposed by specific coordination defect types, highlighting the directional nature of defect correlations.

Robustness Across Models

The standard oxygen-oxygen Radial Distribution Function (RDF) often masks this underlying topological complexity. Different models can produce nearly identical RDFs while possessing significantly different defect populations. The inclusion of vdW corrections was found to be the most critical factor for improving the BLYP model, preventing an "over-structured" network. Despite these sensitivities, the qualitative features of the defect network—how they cluster and their directional nature—remain remarkably consistent across high-quality models like MB-pol and B3LYP.

Conclusion

This work demonstrates that coordination defects are fundamental, highly correlated components of the liquid water network. While this study focuses on static thermodynamic properties, these defects are likely crucial for understanding water's reorientational dynamics and flow. Future research will extend this framework to more complex environments, such as water near ions, interfaces, or in confinement, where defect populations are expected to play an even more prominent role.

Summary updated

Abstract

The hydrogen-bond network of water is characterized by the presence of coordination defects relative to the ideal tetrahedral network of ice, whose fluctuations determine the static and time-dependent properties of the liquid. Because of topological constraints, such defects do not come alone but are highly correlated coming in a plethora of different pairs. Here we discuss in detail such correlations in the case of ab initio water models and show that they have interesting similarities to regular and defective solid phases of water. Although defect correlations involve deviations from idealized tetrahedrality, they can still be regarded as weaker hydrogen bonds that retain a high degree of directionality. We also investigate how the structure and population of coordination defects is affected by approximations to the interatomic potential, finding that, in most cases, the qualitative features of the hydrogen-bond network are remarkably robust.
BibTeX
@article{gasparotto2016probing,
  author = {Piero Gasparotto and Ali Hassanali and Michele Ceriotti},
  title = {{Probing Defects and Correlations in the Hydrogen-Bond Network of ab Initio Water}},
  journal = {Journal of Chemical Theory and Computation},
  year = {2016},
  doi = {10.1021/acs.jctc.5b01138},
}
Figure