Single-Molecule Water Motion on h-BN and Graphene: A Paradigm Shift in Understanding the Behaviour of Water on 2D Material Interfaces

TL;DR

The paper investigates single-molecule water diffusion on two-dimensional polar interfaces, comparing h-BN/Ni(111) with graphene/Ni(111). It combines high-resolution helium spin-echo experiments with ab initio calculations (DFT with vdW corrections and AIMD) to map adsorption energetics, diffusion pathways, and friction. Key findings include quasi-continuous rotational-translational diffusion of H2O on h-BN/Ni with a much lower activation barrier than on graphene/Ni, where diffusion occurs via discrete hops; the water’s behavior is governed by a multidimensional PES and rapid molecular reorientation around the center of mass. Inclusion of the metal substrate reverses the relative friction between the two materials, revealing substrate-controlled vibrational coupling and PES corrugation as essential determinants of molecular friction; the results challenge classical jump-diffusion models and have implications for designing 2D-material interfaces for microfluidics and lubrication.

Abstract

Understanding water behaviour on 2D materials is crucial for sensing, microfluidics, and tribology. While water/graphene interactions are well studied, water on hexagonal boron nitride (h-BN) remains largely unexplored. Despite structural similarity to graphene, h-BN's slightly polar B-N bonds impart a large band gap, high thermal conductivity, and chemical stability, making it promising for electronics, lubricants, and coatings. Moreover, existing water studies often focus on multilayer water dynamics, overlooking single-molecular details. We bridge this gap by studying single-molecular water friction and diffusion on h-BN, comparing it with graphene using helium spin-echo experiments and ab initio calculations. Our findings show that water diffusion on h-BN/Ni follows a complex rotational-translational dynamic, unlike graphene. While conventional views treat water motion as discrete jumps between equivalent adsorption sites, we demonstrate that on h-BN, water molecules rotate freely around their centre of mass. Although the binding energies of water on h-BN and graphene are similar, the activation energy for water dynamics on h-BN is 2.5 times lower than on graphene, implying a much lower barrier for molecular mobility. The fundamentally different diffusion characteristics which classical models cannot capture, underscores the need to rethink how we model water on polar 2D materials. Moreover, our analysis reveals that the metal substrate strongly influences water friction, with h-BN/Ni showing a markedly lower friction than graphene/Ni, in stark contrast to the free-standing materials. These findings challenge assumptions about 2D material-water interactions, highlighting the crucial role of substrate effects in chemistry and material science and offer insights for designing next-generation microfluidic devices that require precise water mobility control.

Figures (6)

• Figure 1: Measurement of single-molecule water diffusion on h-BN. (a) Illustration of the HeSE method: Two wavepackets scatter from the surface with a time difference $t_{\mathrm{SE}}$, allowing the motion of molecules on the surface to be determined by the loss of correlation, which is measured through the polarisation of the beam. The inset shows a typical measurement of the diffusion of water on h-BN ($T = 120K$, $\Delta K = 0.2\per\angstrom$). The reduction in surface correlation with increasing spin-echo time follows a single exponential decay (solid line), characterised by the dephasing rate, $\alpha$. (b) A one-dimensional diffraction scan illustrates the epitaxial growth with the same symmetry as that of the pristine Ni substrate. The greater intensity of the h-BN peak compared to that of the substrate peak indicates the stronger corrugation of h-BN. The bottom panel shows the elastic component after exposing the surface to water ($C$ in \ref{['eq:fit']}), illustrating that no ordered superstructure due to the adsorbed water is present.
• Figure 2: Adsorption energy landscape from ab initio methods. An investigation of the adsorption energy landscape by DFT reveals a weakly corrugated potential energy surface (PES). The adsorption geometries shown in (a) illustrate that water favours adsorption near a boron atom (pastel pink), with the hydrogen atoms pointing towards a nitrogen atom (blue) due to the weak intermolecular bond between the oxygen and the partially positively charged boron atom (see text). The PES as a function of $x$ and $y$ in (b) is characterised by small energy differences between the sites, with the only exception being the nitrogen atom site. The PES further presents a rather weak $z$-dependence of the adsorption sites, as shown in (c), with the rightmost panel in (a) illustrating the geometry for a $z$ distance of $0.2\,\hbox{\AA}$ above the minimum energy site.
• Figure 3: Small activation energies for water dynamics on h-BN/Ni(111) (a) Temperature-dependent measurements at a constant momentum transfer of $\Delta K = 0.6\per\angstrom$ along both high-symmetry orientations $\overline{\Gamma \text{M}}$ (blue data points) and $\overline{\Gamma \text{K}}$ (green data points) show an extremely low activation energy of $E_a = 24meV$. (b)-(c) Pathways showing the migration of H$_2$O between sites on h-BN/Ni(111) along with the associated DFT-calculated transition state energy barriers, which are in excellent agreement with the experimentally determined barrier. (b) illustrates translation coupled with rotation, while (c) demonstrates translation without rotation. The consistently low energy barriers along each pathway suggest ready accessibility, facilitating migration of water across the surface.
• Figure 4: Diffusion of water on hexagonal boron nitride. (a) Momentum transfer dependence of the dephasing rate $\alpha ( \Delta K )$ (blue dots) at $T=120K$, from which the diffusion mechanism of H$_2$O on h-BN/Ni follows. An analytical model (red dash-dotted curve) shows that the motion contains a jump component for the translation of the molecules that follows the periodicity of the substrate. In addition, the motion is dominated by a strong normal component, which cannot be reproduced by the analytic model but is confirmed by ab initio calculations. The error bars correspond to the confidence bounds (1$\sigma$) in the determination of $\alpha$ from the measurements. (b) Single-molecule motion of H$_2$O on graphene/Ni ($T=125\,$ K) according to tamtogl2021 for comparison, where the water dipole remains perpendicular to the substrate (see inset) and H$_2$O moves through a series of discrete jumps.
• Figure 5: Details of the molecular motion from ab initio methods (a) Typical trajectory for H$_2$O on h-BN/Ni from AIMD simulations, illustrating that the hydrogen atoms precess around the oxygen atom along the trajectory, leading to a spinning motion perpendicular to the molecular direction of travel, similar to a corkscrew. The position of the O-H bonds with respect to the oxygen atom can thus easily flip, as shown in the $z$-variation versus time plot in the top-left inset. (b) A stark contrast is observed when comparing the probabilities of water being located at a specific surface site during motion on h-BN/Ni (left panel) and graphene/Ni (right panel). For h-BN/Ni, H$_2$O is hardly found at the N site, while the likelihoods for all the other sites are quite similar, leading to continuous motion. In contrast, on graphene/Ni, H$_2$O motion most likely proceeds via jumps to adjacent equivalent sites, thus resulting in a more disconnected hopping motion.