Achieving contact angles on graphene surfaces with the potential of machine learning

Researchers are working to precisely predict and control how graphene and water interact, a key element in the development of nanofluidic devices, sensors, and energy technologies. Darren Wayne Lim and Xavier R Advincula from the University of Cambridge, along with William C Witt, Fabian L Thiemann and Christoph Schran, presented a detailed atomistic understanding of water wetting on graphene using high-precision machine learning potentials. Their study resolved the long-standing debate about the water contact angle of free-standing graphene, finding a value of 89.1 degrees after accounting for finite size effects. Importantly, the researchers demonstrate that graphene’s dynamic morphology and mechanical strain significantly influence its wetting behavior, providing a convincing explanation for the discrepancy in experimental results and opening new avenues for controlling wettability in future technologies.

Water wetting of graphene by molecular dynamics simulation

Scientists have long sought to understand how water wets surfaces, a key factor in predicting and controlling water behavior in applications ranging from nanofluidics to energy storage. A key metric for quantifying wetting is the contact angle formed by a droplet on a surface, but experimental measurements on free-standing graphene have shown a wide range, hindering the development of a definitive understanding. This study revealed that the contact angle of water on free-standing graphene after finite size correction is 72.1 ±1.5°.
The research team achieved this breakthrough by developing a new methodology to precisely account for the inherent waviness of free-standing graphene to define the contact angle of spherical nanodroplets on non-flat surfaces. The simulations employ machine learning capabilities trained to reproduce density functional theory calculations at the revPBE-D3 level, enabling simulations at a scale and accuracy previously unattainable. This enabled the modeling of droplets containing 9,540 to 22,680 atoms. This is significantly larger than previous ab initio studies and is important to reduce the effects of finite size that impede nanoscale contact angle measurements. By focusing on free-standing graphene, the researchers isolated its intrinsic wetting behavior and eliminated the influence of supporting substrates and surface contaminants.

Experiments showed that the three-phase contact line of nanoscale water droplets strongly couples with the intrinsic thermal ripples of free-standing graphene, demonstrating that the wetting properties are highly sensitive to mechanical strain. Application of tensile strain significantly increases the hydrophobicity of graphene, whereas compressive strain induces coherent ripples in which the droplets effectively “surf”, resulting in pronounced anisotropic wetting and contact angle hysteresis. This study demonstrates that the wetting properties of graphene are governed not only by its chemical composition but also by its dynamic morphology and provides a convincing explanation for the variability observed in experimental measurements. Furthermore, this study reveals that mechanical strain presents a practical method to control wetting in graphene-based technologies, with promising implications for nanofluidic devices and nanofiltration applications. This study demonstrates the bidirectional coupling between surface waviness dynamics and three-phase contact lines and highlights the outsized influence of strain on graphene wettability. These findings suggest that mechanical strain manipulation may be a powerful means to tailor graphene surfaces for specific applications that require precise control of water behavior at the nanoscale.

Graphene wetting revealed by machine-learned molecular dynamics

Scientists used molecular dynamics simulations to investigate the wetting behavior of water on free-standing graphene, addressing a long-standing discrepancy in experimental measurements. This work pioneered a new methodology for defining contact angles on non-flat surfaces, which is important for accurately calculating the thermal ripple inherent in free-standing graphene. The researchers developed the potential of trained machine learning for density functional theory calculations at the revPBE-D3 level, enabling simulations with both first-principles accuracy and achievable computational cost. This approach facilitated the modeling of significantly larger droplets containing 9,540 to 22,680 atoms than was previously possible with ab initio methods and enabled robust finite size corrections.

The experiment simulated spherical water droplets of various sizes placed on a large, dynamically evolving graphene sheet. The contact angle was measured using a geometry-based method, analyzing the intersection of the time-averaged interface of the water droplet and the time-averaged graphene height map. To correct for the effects of finite size, the research team plotted the microscopic contact angle against the radius of the three-phase contact line and extrapolated to obtain a macroscopic contact angle of 72.1 ±1.5°. This extrapolation relies on the assumption of a linear relationship between the cosine of the contact angle and the inverse of the contact line radius and incorporates line tension effects.

Furthermore, in this study, we investigated the interaction between ripples on the graphene surface and wetting of nanoscale droplets by applying mechanical strain to the graphene sheet. The scientists found a strong correlation between surface ripple mechanics and three-phase contact lines, revealing that tensile strain increases hydrophobicity, while compressive strain induces coherent ripples that promote droplet movement. This finding demonstrates that the wetting properties are determined not only by the chemical properties of graphene but also by its dynamic morphology, providing a potential explanation for the variation in experimental results and suggesting mechanical strain as a viable control mechanism for nanofluidic and nanofiltration techniques.

Benchmarked graphene wettability revealed by molecular dynamics simulations

Scientists have achieved an ab initio benchmark for water droplet contact angles on free-standing graphene using molecular dynamics simulations. The research team measured a finite-size-corrected contact angle of 72.1 ±1.5°, resolving a long-standing discrepancy in experimental measurements. These simulations employed machine learning potentials trained to reproduce density functional theory calculations, allowing simulations with significantly larger droplets than previously possible and reducing the effects of finite size. Researchers have constructed a new methodology to define the contact angle of spherical nanodroplets on non-flat surfaces by precisely taking into account the surface ripples of free-standing graphene.

Experiments reveal a strong coupling between the three-phase contact line of nanoscale water droplets and the inherent thermal ripples of free-standing graphene. The data show that the wetting properties of graphene are very sensitive to mechanical strain, and tensile strain significantly increases the hydrophobicity of graphene. Conversely, compressive strain induces coherent ripples, allowing the droplet to “surf” along the surface, resulting in significant anisotropic wetting and contact angle hysteresis. The measurements confirm that not only the chemical properties of graphene but also the dynamic morphology of graphene govern its wetting properties, providing a new explanation for the variability observed in the experimental results.

The results show that the interaction between surface ripples and nanoscale water droplet wetting is substantial. The researchers found a bidirectional link between surface waviness mechanics and three-phase contact lines, showing that mechanical strain has a significant effect on graphene’s wettability. Specifically, applying tensile strain increases the hydrophobicity of the surface, while applying compressive strain creates ripples that promote droplet movement. These findings suggest that mechanical strain may serve as a practical method to control wetting in graphene-based technologies, providing potential benefits for nanofluidic and nanofiltration applications.

Tests demonstrate that the simulation accurately captures graphene’s inherent wettability and isolates graphene behavior from substrate effects and contamination. This study utilized droplets several orders of magnitude larger than previous ab initio studies, allowing us to more accurately resolve finite size effects. Additionally, this study highlights the importance of capturing the film-like flexibility of free-standing graphene, where thermal ripples reduce vibrational free energy and influence water diffusion and transport.