Probing Confinement Effects on the Infrared Spectra of Water with Deep Potential Molecular Dynamics Simulations

Abstract

The hydrogen-bond network of confined water is expected to deviate from that of the bulk liquid, yet probing these deviations remains a significant challenge. In this work, we combine large-scale molecular dynamics simulations with machine learning potential derived from firstprinciples calculations to examine the hydrogen bonding of water confined in carbon nanotubes (CNTs). We computed and compared the infrared spectrum (IR) of confined water to existing experiments to elucidate confinement effects. For CNTs with diameters >1.2 nm, we find that confinement imposes a monotonic effect on the hydrogen-bond network and on the IR spectrum of water. In contrast, confinement below 1.2 nm CNT diameter affects the water structure in a complex fashion, leading to a strong directional dependence of hydrogen bonding that varies nonlinearly with the CNT diameter. When integrated with existing IR measurements, our simulations provide a new interpretation for the IR spectrum of water confined in CNTs, pointing to previously unreported aspects of hydrogen bonding in this system. This work also offers a general platform for simulating water in CNTs with quantum accuracy on time and length scales beyond the reach of conventional first-principles approaches.

Accuracy of TIP4P/2005 and SPC/Fw Water Models

“…26−49 Recently, new models have also been parametrized from a deep neural network. 36,39 In general, each of these water models has been optimized to reproduce only a few of the thermodynamic properties of water, usually from experimental data. In this context, their success depends on being able to reproduce additional experimental properties in both bulk and confined water.…”

“…In the past few decades, much effort from the molecular dynamics (MD) simulation community has been devoted to developing force field models to simulate bulk and confined water. These include models with three, four, and five Coulomb interaction sites with rigid or flexible geometry and other models that incorporate the electronic polarization. − Recently, new models have also been parametrized from a deep neural network. , In general, each of these water models has been optimized to reproduce only a few of the thermodynamic properties of water, usually from experimental data. In this context, their success depends on being able to reproduce additional experimental properties in both bulk and confined water.…”

Integrating Machine Learning Potential and X-ray Absorption Spectroscopy for Predicting the Chemical Speciation of Disordered Carbon Nitrides

“…More recently, data science approaches provide a promising means to accelerate the theory-experiment feedback for interrogating the structure and chemistry of complex materials. Development in machine learning (ML) interatomic potentials enables efficient exploration of chemical space at time and length scales beyond the reach of conventional AIMD simulations. − Data science techniques also play a growing important role in elucidating structure–spectroscopy relationship and inferring structural information from spectroscopic measurements. − For example, Timoshenko et al utilized neural network (NN) models to extract geometric structures and morphology of Pt nanoparticles and different phases of bulk iron from experimental spectra. , Carbone et al applied a convolutional NN classifier to predict the coordination number of transition metal oxides from XANES spectra . The existing studies, however, have largely focused on elucidating the structure–spectrum relationship for either crystalline solids or small molecules, whereas disordered systems have received less attention.…”

Confinement Effects on Proton Transfer in TiO2 Nanopores from Machine Learning Potential Molecular Dynamics Simulations

“…Within porous materials, it is expected that the hydrogen-bond network, and therefore the energetics and kinetics for proton transfer, significantly deviate from that of bulk water, particularly for the narrowest nanopores with diameters smaller than 10 nm . In the past several decades, proton transfer in hydrophobic nanoporous materials has been extensively studied using both simulations and experiments. ,, Examples include carbon nanotubes (CNTs) that represent an ideal model system for studying confinement effects, thanks to their simple and highly controlled geometry. − Simulations and experiments both suggest that water undergoes an order–disorder transition for a CNT with a diameter smaller than 10 nm. , In addition, experimental measurements have shown that 0.8 nm diameter CNTs, which promote the formation of one-dimensional water wires, can support proton transport rates that exceed those of bulk water by an order of magnitude . Atomistic simulations based on first-principles and empirical valence bond methods support this view, suggesting that proton transport in narrow CNTs is faster than in the bulk …”