Improving Molecular-Dynamics Simulations for Solid-Liquid Interfaces with Machine-Learning Interatomic Potentials

Emerging developments in artificial intelligence have opened infinite possibilities for material simulation. Depending on the powerful fitting of machine learning algorithms to first-principles data, machine learning interatomic potentials (MLIPs) can effectively balance the accuracy and efficiency problems in molecular dynamics (MD) simulations, serving as powerful tools in various complex physicochemical systems. Consequently, this brings unprecedented enthusiasm for researchers to apply such novel technology in multiple fields to revisit the major scientific problems that have remained controversial owing to the limitations of previous computational methods. Herein, we introduce the evolution of MLIPs, provide valuable application examples for solid-liquid interfaces, and present current challenges. Driven by solving multitudinous difficulties in terms of the accuracy, efficiency, and versatility of MLIPs, this booming technique, combined with molecular simulation methods, will provide an underlying and valuable understanding of interdisciplinary scientific challenges, including materials, physics, and chemistry.

Two-dimensional bilayer ice in coexistence with three-dimensional ice without confinement

Icing plays an important role in various physical-chemical process. Although the formation of two-dimensional ice requires nanoscale confinement, two-dimensional bilayer ice in coexistence with three-dimensional ice without confinement remains poorly understood. Here, a critical value of a surface energy parameter is identified to characterize the liquid-solid interface interaction, above which two-dimensional and three-dimensional coexisting ice can surprisingly form on the surface. The two-dimensional ice growth mechanisms could be revealed by capturing the growth and merged of the metastable edge structures. The phase diagram about temperature and pressure vs energy parameters is predicted to distinguish liquid water, two-dimensional ice and three-dimensional ice. Furthermore, the deicing characteristics of coexisting ice demonstrate that the ice adhesion strength is linearly related to the ratio of ice-surface interaction energy to ice temperature. In addition, for gas-solid phase transition, the phase diagram about temperature and energy parameters is predicted to distinguish gas, liquid water, two-dimensional ice and three-dimensional ice. This work gives a perspective for studying the singular structure and dynamics of ice in nanoscale and provides a guide for future experimental realization of the coexisting ice.

Formation of a two-dimensional helical square tube ice in hydrophobic nanoslit using the TIP5P water model

In extreme and nanoconfinement conditions, the tetrahedral arrangement of water molecules is challenged, resulting in a rich and new phase behavior unseen in bulk phases. The unique phase behavior of water confined in hydrophobic nanoslits has been previously observed, such as the formation of a variety of two-dimensional (2D) ices below the freezing temperature. The primary identified 2D ice phase, termed square tube ice (STI), represents a unique arrangement of water molecules in 2D ice, which can be viewed as an array of 1D ice nanotubes stacked in the direction parallel to the confinement plane. In this study, we report the molecular dynamics (MD) simulations evidence of a novel 2D ice phase, namely, helical square tube ice (H-STI). H-STI is characterized by the stacking of helical ice nanotubes in the direction parallel to the confinement plane. Its structural specificity is evident in the presence of helical square ice nanotubes, a configuration unseen in both STI and single-walled ice nanotubes. A detailed analysis of the hydrogen bonding strength showed that H-STI is a 2D ice phase diverging from the Bernal-Fowler-Pauling ice rules by forming only two strong hydrogen bonds between adjacent molecules along its helical ice chain. This arrangement of strong hydrogen bonds along ice nanotube and weak bonds between the ice nanotube shows a similarity to quasi-one-dimensional van der Waals materials. Ab initio molecular dynamics simulations (over a 30 ps) were employed to further verify H-STI's stability at 1 GPa and temperature up to 200 K.

Terahertz electric field induced melting and transport of monolayer water confined in double-walled carbon nanotubes

Understanding the structures and dynamics of confined water in nanochannels holds great promise for various applications, ranging from membrane separation to blue energy collection. A setting of particular interest is the confined monolayer water within double-walled carbon nanotubes (DWCNTs), which demonstrates rich ice morphologies; however, the dynamics of this peculiar system are still unexplored. In this work, a series of molecular dynamics (MD) simulations reveal that the two-dimensional ice in DWCNTs can be effectively melted by terahertz electric fields but not by static electric fields, exhibiting an interesting ice to vapor-like transition along with extraordinary dynamical behaviors. Specifically, under appropriate field frequency, the water flow presents a sharp increase with the increase in field strength, indicating an excellent gating behavior. These remarkable findings are attributed to the resonance effect between the terahertz electric field and inherent vibration of water hydrogen bonds, causing the water molecules to change from the frozen to super permeation states. The amount of confined water exhibits a sudden reduction, confirming the breakdown of the hydrogen bond network. The distributions of density profiles, hydrogen bond number and dipole orientation demonstrate more details of the water structural change. Furthermore, under a certain field strength, the water flow shows a peculiar maximum behavior with the increase in field frequency, implying frequency optimization for water transport. These findings not only enhance our understanding of the phase transition behavior of water molecules confined within DWCNTs under the influence of a terahertz electric field but also provide a promising avenue for designing innovative nanofluidic devices.

Temperature-pressure phase diagram of confined monolayer water/ice at first-principles accuracy with a machine-learning force field

Understanding the phase behaviour of nanoconfined water films is of fundamental importance in broad fields of science and engineering. However, the phase behaviour of the thinnest water film - monolayer water - is still incompletely known. Here, we developed a machine-learning force field (MLFF) at first-principles accuracy to determine the phase diagram of monolayer water/ice in nanoconfinement with hydrophobic walls. We observed the spontaneous formation of two previously unreported high-density ices, namely, zigzag quasi-bilayer ice (ZZ-qBI) and branched-zigzag quasi-bilayer ice (bZZ-qBI). Unlike conventional bilayer ices, few inter-layer hydrogen bonds were observed in both quasi-bilayer ices. Notably, the bZZ-qBI entails a unique hydrogen-bonding network that consists of two distinctive types of hydrogen bonds. Moreover, we identified, for the first time, the stable region for the lowest-density [Formula: see text] monolayer ice (LD-48MI) at negative pressures (<-0.3 GPa). Overall, the MLFF enables large-scale first-principle-level molecular dynamics (MD) simulations of the spontaneous transition from the liquid water to a plethora of monolayer ices, including hexagonal, pentagonal, square, zigzag (ZZMI), and hexatic monolayer ices. These findings will enrich our understanding of the phase behaviour of the nanoconfined water/ices and provide a guide for future experimental realization of the 2D ices.

Evidence of Formation of 1-10 nm Diameter Ice Nanotubes in Double-Walled Carbon Nanotube Capillaries

Water exhibits rich phase behaviors in nanoscale confinement. Since the simulation evidence of the formation of single-walled ice nanotubes (INTs) in single-walled carbon nanotubes was confirmed experimentally, INTs have been recognized as a form of low-dimensional hydrogen-bonding network. However, the single-walled INTs reported in the literature all possess subnanometer diameters (<1 nm). Herein, based on systematic and large-scale molecular dynamics simulations, we demonstrate the spontaneous freezing transition of liquid water to single-walled INTs with diameters reaching ∼10 nm when confined to capillaries of double-walled carbon nanotubes (DW-CNTs). Three distinct classes of INTs are observed, namely, INTs with flat square walls (INTs-FSW), INTs with puckered rhombic walls (INTs-PRW), and INTs with bilayer hexagonal walls (INTs-BHW). Surprisingly, when water is confined in DW-CNT (3, 3)@(13, 13), an INT-FSW freezing temperature of 380 K can be reached, which is even higher than the boiling temperature of bulk water at atmospheric pressure. The freezing temperatures of INTs-FSW decrease as their caliber increases, approaching to the freezing temperature of two-dimensional flat square ice at the large-diameter limit. In contrast, the freezing temperature of INTs-PRW is insensitive to their diameter. Ab initio molecular dynamics simulations are performed to examine the stability of the INT-FSW and INT-PRW. The highly stable INTs with diameters beyond subnanometer scale can be exploited for potential applications in nanofluidic technologies and for mass transport as bioinspired nanochannels.

The pressure induced phase diagram of double-layer ice under confinement: a first-principles study

Here, we present double-layer ice confined within various carbon nanotubes (CNTs) using state-of-the-art pressure induced (-5 GPa to 5 GPa) dispersion corrected density functional theory (DFT) calculations. We find that the double-layer ice exhibits remarkably rich and diverse phase behaviors as a function of pressure with varying CNT diameters. The lattice cohesive energies for various pure double layer ice phases follow the order of hexagonal > pentagonal > square tube > hexagonal-close-packed (HCP) > square > buckled-rhombic (b-RH). The confinement width was found to play a crucial role in the square and square tube phases in the intermediate pressure range of about 0-1 GPa. Unlike the phase transition in pure bilayer ice structures, the relative enthalpies demonstrate that the pentagonal phase, rather than the hexagonal structure, is the most stable ice polymorph at ambient pressure as well as in a deep negative pressure region, whereas the b-RH phase dominates under high pressure. The relatively short O⋯O distance of b-RH demonstrates the presence of a strong hydrogen bonding network, which is responsible for stabilizing the system.

Self-assembled rhomboidal ammonia monolayer confined in two vertically stacked graphene oxide/graphene nanosheets

Confined water molecules have attracted widespread research interest due to their versatile phase behaviors. Ammonia (NH₃, isoelectronic with water) molecules are also expected to realize the delicate self-assembled hydrogen-bonded network like water in confinement. Here, the structures and phase behavior of NH₃ monolayers confined in two structurally symmetrical graphene oxide (GO) or graphene (G) nanosheets are investigated using first-principles calculations and ab initio molecular dynamics simulations. A highly ordered new rhomboidal phase with all NH₃ molecules adopting a Y-shaped configuration, in which one N-H bond is parallel to the confining planes and two other N-H bonds point to the top/bottom GO/G layers, respectively, was discovered at low temperature, resulting from the symmetrical confinement and subtle interlayer/intermolecular interactions. Remarkably, this new phase is so stable that a quite large strain is needed to destroy it. At room temperature, these NH₃ monolayers behave like a liquid. These rhomboidal NH₃ monolayers confined in GO/G nanosheets not only offer diverse hydrogen-bonded networks but also possess potential piezoelectricity for future device applications.

First-Principles Molecular Dynamics Simulations of the Spontaneous Freezing Transition of 2D Water in a Nanoslit

As with bulk ices, two-dimensional (2D) ices exhibit diverse crystalline structures, and the majority of these 2D structures have been predicted based on classical molecular dynamics (MD) simulations. Here, the spontaneous freezing transition of 2D liquid water within hydrophobic nanoslits is demonstrated for the first time using first-principles MD simulations. Various 2D ices are observed under different lateral pressure and temperature conditions. Notably, the liquid water confined to a 6.0 Å-wide nanoslit can spontaneously freeze into a monolayer ice consisting of an array of zigzag water chains at 2.5 GPa and 250 K. Moreover, within an 8.0 Å-wide nanoslit and at 4.0 GPa and 300 K, a previously unreported bilayer ice forms spontaneously that has a structure resembling that of the double surface layers of bulk ice-VII. Both 2D crystalline ices do not obey the ice rule, suggesting first-principles simulation can access a certain phase space that is not easily approached using classical simulations.

Computational Prediction of Novel Ice Phases: A Perspective

Although computational prediction of new ice phases is a niche field in water science, the scientific subject itself is representative of two important areas in physical chemistry, namely, statistical thermodynamics and molecular simulations. The prediction of a variety of novel ice phases has also attracted general public interest since the 1980s. In particular, the prediction of low-dimensional ice phases has gained momentum since the confirmation of a number of low-dimensional "computer ice" phases in the laboratory over the past decade. In this Perspective, the research advancements in computational prediction of novel ice phases over the past few years are reviewed. Particular attention is placed on new ice phases whose physical properties or dimensional structures are distinctly different from conventional bulk ices. Specific topics include the (i) formation of superionic ices, (ii) electrofreezing of water under high pressure and in a high external electric field, (iii) prediction of low-density porous ice at strongly negative pressure, (iv) ab initio computational study of two-dimensional (2D) ice under nanoscale confinement, and (v) 2D ices formed on a solid surface near ambient temperature without nanoscale confinement. Clearly, the formation of most of these novel ice phases demands certain extreme conditions. Ongoing challenges and new opportunities for predicting new ice phases from either classical molecular dynamics simulation or high-level ab initio computation are discussed.

Insights into the Emerging Networks of Voids in Simulated Supercooled Water

The structural evolution of supercooled liquid water as we approach the glass transition temperature continues to be an active area of research. Here, we use molecular dynamics simulations of TIP4P/ice water to study the changes in the connected regions of empty space within the liquid, which we investigate using the Voronoi-voids network. We observe two important features: supercooling enhances the fraction of nonspherical voids and different sizes of voids tend to cluster forming a percolating network. By examining order parameters such as the local structure index (LSI), tetrahedrality and topological defects, we show that water molecules near large void clusters tend to be slightly more tetrahedral than those near small voids, with a lower population of under- and overcoordinated defects. We show further that the distribution of closed rings of water molecules around small and large void clusters maintain a balance between 6 and 7 membered rings. Our results highlight the changes of the dual voids and water network as a structural hallmark of supercooling and provide insights into the molecular origins of cooperative effects underlying density fluctuations on the subnanometer and nanometer length scale. In addition, the percolation of the voids and the hydrogen bond network around the voids may serve as useful order parameters to investigate density fluctuations in supercooled water.

Direct observation of 2-dimensional ices on different surfaces near room temperature without confinement

Water-solid interfaces play important roles in a wide range of fields, including atmospheric science, geochemistry, electrochemistry, and food science. Herein, we report simulation evidence of 2-dimensional (2D) ice formation on various surfaces and the dependence of the 2D crystalline structure on the hydrophobicity and morphology of the underlying surface. Contrary to the prevailing view that nanoscale confinement is necessary for the 2D liquid-to-bilayer ice transition, we find that the liquid-to-bilayer hexagonal ice (BHI) transition can occur either on a model smooth surface or on model fcc-crystal surfaces with indices of (100), (110), and (111) near room temperature. We identify a critical parameter that characterizes the water-surface interaction, above which the BHI can form on the surface. This critical parameter increases as the temperature increases. Even at temperatures above the freezing temperature of bulk ice (Ih ), we find that BHI can also form on a superhydrophilic surface due to the strong water-surface interaction. The tendency toward the formation of BHI without confinement reflects a proper water-surface interaction that can compensate for the entropy loss during the freezing transition. Furthermore, phase diagrams of 2D ice formation are described on the plane of the adsorption energy versus the fcc lattice constant (Eads-afcc), where 4 monolayer square-like ices are also identified on the fcc model surfaces with distinct water-surface interactions.

Ice-Liquid Oscillations in Nanoconfined Water

Nanoscale confinement has a strong effect on the phase behavior of water. Studies in the last two decades have revealed a wealth of novel crystalline and quasicrystalline structures for water confined in nanoslits. Less is known, however, about the nature of ice-liquid coexistence in extremely nanoconfined systems. Here, we use molecular simulations to investigate the ice-liquid equilibrium for water confined between two nanoscopic disks. We find that the nature of ice-liquid phase coexistence in nanoconfined water is different from coexistence in both bulk water and extended nanoslits. In highly nanoconfined systems, liquid water and ice do not coexist in space because the two-phase states are unstable. The confined ice and liquid phases coexist in time, through oscillations between all-liquid and all-crystalline states. The avoidance of spatial coexistence of ice and liquid originates on the non-negligible cost of the interface between confined ice and liquid in a small system. It is the result of the small number of water molecules between the plates and has no analogue in bulk water.

Modelling the atomic arrangement of amorphous 2D silica: a network analysis

The recent experimental discovery of a semi two-dimensional silica glass has offered a realistic description of the random network theory of a silica glass structure, initially discussed by Zachariasen.

CO Separation from H2 via Hydrate Formation in Single-Walled Carbon Nanotubes

Hydrogen is an alternative fuel without generating greenhouse gas or other harmful emissions. Industrial hydrogen production, however, always contains a small fraction of carbon monoxide (CO) (∼0.5-2%) that must be removed for use in fuel cells. Here, we present molecular dynamics simulation evidence on facile separation of CO from H₂ at ambient pressure via the formation of quasi-one-dimensional (Q1D) clathrate hydrates within single-walled carbon nanotubes (SW-CNTs). At ambient pressure, Q1D CO (or H₂) clathrates in SW-CNTs are formed spontaneously when the SW-CNTs are immersed in CO (or H₂) aqueous solution. More interestingly, for the CO/H₂ aqueous solution, highly preferential adsorption of CO over H₂ occurs within the octagonal or nonagonal ice nanotubes inside of SW-CNTs. These results suggest that the formation of Q1D hydrates within SW-CNTs can be a viable and safe method for the separation of CO from H₂, which can be exploited for hydrogen purification in fuel cells.

AB-stacked square-like bilayer ice in graphene nanocapillaries

Water, when constrained between two graphene sheets and under ultrahigh pressure, can manifest dramatic differences from its bulk counterparts such as the van der Waals pressure induced water-to-ice transformation, known as the metastability limit of two-dimensional (2D) liquid.