The Interplay of Structure and Dynamics in the Raman Spectrum of Liquid Water over the Full Frequency and Temperature Range

Tensorial Properties via the Neuroevolution Potential Framework: Fast Simulation of Infrared and Raman Spectra

Infrared and Raman spectroscopy are widely used for the characterization of gases, liquids, and solids, as the spectra contain a wealth of information concerning, in particular, the dynamics of these systems. Atomic scale simulations can be used to predict such spectra but are often severely limited due to high computational cost or the need for strong approximations that limit the application range and reliability. Here, we introduce a machine learning (ML) accelerated approach that addresses these shortcomings and provides a significant performance boost in terms of data and computational efficiency compared with earlier ML schemes. To this end, we generalize the neuroevolution potential approach to enable the prediction of rank one and two tensors to obtain the tensorial neuroevolution potential (TNEP) scheme. We apply the resulting framework to construct models for the dipole moment, polarizability, and susceptibility of molecules, liquids, and solids and show that our approach compares favorably with several ML models from the literature with respect to accuracy and computational efficiency. Finally, we demonstrate the application of the TNEP approach to the prediction of infrared and Raman spectra of liquid water, a molecule (PTAF-), and a prototypical perovskite with strong anharmonicity (BaZrO3). The TNEP approach is implemented in the free and open source software package gpumd, which makes this methodology readily available to the scientific community.

Comparing machine learning potentials for water: Kernel-based regression and Behler-Parrinello neural networks

In this paper, we investigate the performance of different machine learning potentials (MLPs) in predicting key thermodynamic properties of water using RPBE + D3. Specifically, we scrutinize kernel-based regression and high-dimensional neural networks trained on a highly accurate dataset consisting of about 1500 structures, as well as a smaller dataset, about half the size, obtained using only on-the-fly learning. This study reveals that despite minor differences between the MLPs, their agreement on observables such as the diffusion constant and pair-correlation functions is excellent, especially for the large training dataset. Variations in the predicted density isobars, albeit somewhat larger, are also acceptable, particularly given the errors inherent to approximate density functional theory. Overall, this study emphasizes the relevance of the database over the fitting method. Finally, this study underscores the limitations of root mean square errors and the need for comprehensive testing, advocating the use of multiple MLPs for enhanced certainty, particularly when simulating complex thermodynamic properties that may not be fully captured by simpler tests.

Deciphering Density Fluctuations in the Hydration Water of Brownian Nanoparticles via Upconversion Thermometry

We investigate the intricate relationship among temperature, pH, and Brownian velocity in a range of differently sized upconversion nanoparticles (UCNPs) dispersed in water. These UCNPs, acting as nanorulers, offer insights into assessing the relative proportion of high-density and low-density liquid in the surrounding hydration water. The study reveals a size-dependent reduction in the onset temperature of liquid-water fluctuations, indicating an augmented presence of high-density liquid domains at the nanoparticle surfaces. The observed upper-temperature threshold is consistent with a hypothetical phase diagram of water, validating the two-state model. Moreover, an increase in pH disrupts the organization of water molecules, similar to external pressure effects, allowing simulation of the effects of temperature and pressure on hydrogen bonding networks. The findings underscore the significance of the surface of suspended nanoparticles for understanding high- to low-density liquid fluctuations and water behavior at charged interfaces.

First-principles spectroscopy of aqueous interfaces using machine-learned electronic and quantum nuclear effects

Vibrational spectroscopy is a powerful approach to visualising interfacial phenomena. However, extracting structural and dynamical information from vibrational spectra is a challenge that requires first-principles simulations, including non-Condon and quantum nuclear effects. We address this challenge by developing a machine-learning enhanced first-principles framework to speed up predictive modelling of infrared, Raman, and sum-frequency generation spectra. Our approach uses machine learning potentials that encode quantum nuclear effects to generate quantum trajectories using simple molecular dynamics efficiently. In addition, we reformulate bulk and interfacial selection rules to express them unambiguously in terms of the derivatives of polarisation and polarisabilities of the whole system and predict these derivatives efficiently using fully-differentiable machine learning models of dielectric response tensors. We demonstrate our framework's performance by predicting the IR, Raman, and sum-frequency generation spectra of liquid water, ice and the water-air interface by achieving near quantitative agreement with experiments at nearly the same computational efficiency as pure classical methods. Finally, to aid the experimental discovery of new phases of nanoconfined water, we predict the temperature-dependent vibrational spectra of monolayer water across the solid-hexatic-liquid phases transition.

Predicting the Raman Spectra of Liquid Water with a Monomer-Field Model

The Raman spectrum of liquid water is quite complex, reflecting its strong sensitivity to the local environment of the individual waters. The OH-stretch region of the spectrum, which captures the influence of hydrogen bonding, has only just begun to be unraveled. Here we develop a model for predicting the Raman spectra of the OH-stretch region by considering how local electric fields distort the energy surface of each water monomer. We find that our model is capable of reproducing the bimodal nature of the main peak, with the shoulder at 3250 cm-1 resulting almost entirely from Fermi resonance. Furthermore, we capture the temperature and polarization dependence of the shoulder, which has proven to be difficult to obtain with previous methods, and analyze the origin of this dependence. We expect our model to be generally useful for understanding and predicting how Raman spectra change under different conditions and with different probe reporters beyond water.

Structural evolution of water-in-propylene carbonate mixtures revealed by polarized Raman spectroscopy and molecular dynamics

The liquid structure of systems wherein water is limited in concentration or through geometry is of great interest in various fields such as biology, materials science, and electrochemistry. Here, we present a combined polarized Raman and molecular dynamics investigation of the structural changes that occur as water is added incrementally to propylene carbonate (PC), a polar, aprotic solvent that is important in lithium-ion batteries. Polarized Raman spectra of PC solutions were collected for water mole fractions 0.003 ≤ χwater ≤ 0.296, which encompasses the solubility range of water in PC. The novel approach taken herein provides additional hydrogen bond and solvation characterization of this system that has not been achievable in previous studies. Analysis of the polarized carbonyl Raman band in conjunction with simulations demonstrated that the bulk structure of the solvent remained unperturbed upon the addition of water. Experimental spectra in the O-H stretching region were decomposed through Gaussian fitting into sub-bands and comparison to studies of dilute HOD in D2O. With the aid of simulations, we identified these different bands as water arrangements having different degrees of hydrogen bonding. The observed water structure within PC indicates that water tends to self-aggregate, forming a hydrogen bond network that is distinctly different from the bulk and dependent on concentration. For example, at moderate concentrations, the most likely aggregate structures are chains of water molecules, each with two hydrogen bonds.

Developing machine-learned potentials to simultaneously capture the dynamics of excess protons and hydroxide ions in classical and path integral simulations

The transport of excess protons and hydroxide ions in water underlies numerous important chemical and biological processes. Accurately simulating the associated transport mechanisms ideally requires utilizing ab initio molecular dynamics simulations to model the bond breaking and formation involved in proton transfer and path-integral simulations to model the nuclear quantum effects relevant to light hydrogen atoms. These requirements result in a prohibitive computational cost, especially at the time and length scales needed to converge proton transport properties. Here, we present machine-learned potentials (MLPs) that can model both excess protons and hydroxide ions at the generalized gradient approximation and hybrid density functional theory levels of accuracy and use them to perform multiple nanoseconds of both classical and path-integral proton defect simulations at a fraction of the cost of the corresponding ab initio simulations. We show that the MLPs are able to reproduce ab initio trends and converge properties such as the diffusion coefficients of both excess protons and hydroxide ions. We use our multi-nanosecond simulations, which allow us to monitor large numbers of proton transfer events, to analyze the role of hypercoordination in the transport mechanism of the hydroxide ion and provide further evidence for the asymmetry in diffusion between excess protons and hydroxide ions.

Vibrational Coupling and Hydrogen-Bond Structure of Water in the Extended Hydration Shell of Metal Ions

Intra- and intermolecular vibrational coupling (VC) and hydrogen bonding (H-bonding) of water are sparsely understood in the hydration shell (HS) of a metal ion, though the corresponding knowledge for an anion is quite extensive. This is primarily due to the overwhelming effect of anions on water, which masks the subtle perturbing influence of most of the cations. Using Raman difference spectroscopy with simultaneous curve fitting (Raman-DS-SCF) in combination with isotopic dilution and polarized Raman spectroscopy, we have elucidated the VC and H-bonding of water in the HS of bi- and trivalent metal ions─Mg2+, Ca2+, La3+, Gd3+, Dy3+. Polarized Raman measurement of the HS water with VC "turned on" and "turned off" (using isotopically diluted water, HOD) reveals that water retains the intra- and intermolecular vibrational coupling in the HS of high-charge-density metal ions, which is in stark contrast to that of an anion. Hydration shell spectroscopy in HOD unambiguously shows that the average H-bonding of water becomes stronger in the HS than that of bulk water. The first HS water strongly donates two H-bonds to the second HS water (ν̅max ≈ 3200 cm-1) but weakly accepts a H-bond from the second HS water (ν̅max ≈ 3590 cm-1), which makes the HS water heterogeneous in terms of its H-bond structure. The weakly interacting OH (ν̅max 3585 cm-1 in HOD) red-shifts by ∼ 15 cm-1 while the VC is "turned on" (ν̅max ≈ 3600 cm-1 in H2O), revealing the intramolecular coupling of water in the HS of metal ions.

Data-Efficient Machine Learning Potentials from Transfer Learning of Periodic Correlated Electronic Structure Methods: Liquid Water at AFQMC, CCSD, and CCSD(T) Accuracy

Obtaining the atomistic structure and dynamics of disordered condensed-phase systems from first-principles remains one of the forefront challenges of chemical theory. Here we exploit recent advances in periodic electronic structure and provide a data-efficient approach to obtain machine-learned condensed-phase potential energy surfaces using AFQMC, CCSD, and CCSD(T) from a very small number (≤200) of energies by leveraging a transfer learning scheme starting from lower-tier electronic structure methods. We demonstrate the effectiveness of this approach for liquid water by performing both classical and path integral molecular dynamics simulations on these machine-learned potential energy surfaces. By doing this, we uncover the interplay of dynamical electron correlation and nuclear quantum effects across the entire liquid range of water while providing a general strategy for efficiently utilizing periodic correlated electronic structure methods to explore disordered condensed-phase systems.

Local solvation structures govern the mixing thermodynamics of glycerol-water solutions

Glycerol is a major cryoprotective agent and is widely used to promote protein stabilization. By a combined experimental and theoretical study, we show that global thermodynamic mixing properties of glycerol and water are dictated by local solvation motifs. We identify three hydration water populations, i.e., bulk water, bound water (water hydrogen bonded to the hydrophilic groups of glycerol) and cavity wrap water (water hydrating the hydrophobic moieties). Here, we show that for glycerol experimental observables in the THz regime allow quantification of the abundance of bound water and its partial contribution to the mixing thermodynamics. Specifically, we uncover a 1 : 1 connection between the population of bound waters and the mixing enthalpy, which is further corroborated by the simulation results. Therefore, the changes in global thermodynamic quantity - mixing enthalpy - are rationalized at the molecular level in terms of changes in the local hydrophilic hydration population as a function of glycerol mole fraction in the full miscibility range. This offers opportunities to rationally design polyol water, as well as other aqueous mixtures to optimize technological applications by tuning mixing enthalpy and entropy based on spectroscopic screening.

Realistic phase diagram of water from "first principles" data-driven quantum simulations

Since the experimental characterization of the low-pressure region of water's phase diagram in the early 1900s, scientists have been on a quest to understand the thermodynamic stability of ice polymorphs on the molecular level. In this study, we demonstrate that combining the MB-pol data-driven many-body potential for water, which was rigorously derived from "first principles" and exhibits chemical accuracy, with advanced enhanced-sampling algorithms, which correctly describe the quantum nature of molecular motion and thermodynamic equilibria, enables computer simulations of water's phase diagram with an unprecedented level of realism. Besides providing fundamental insights into how enthalpic, entropic, and nuclear quantum effects shape the free-energy landscape of water, we demonstrate that recent progress in "first principles" data-driven simulations, which rigorously encode many-body molecular interactions, has opened the door to realistic computational studies of complex molecular systems, bridging the gap between experiments and simulations.

Combinational Vibration Modes in H2O/HDO/D2O Mixtures Detected Thanks to the Superior Sensitivity of Femtosecond Stimulated Raman Scattering

Overtones and combinational modes frequently play essential roles in ultrafast vibrational energy relaxation in liquid water. However, these modes are very weak and often overlap with fundamental modes, particularly in isotopologues mixtures. We measured VV and HV Raman spectra of H₂O and D₂O mixtures with femtosecond stimulated Raman scattering (FSRS) and compared the results with calculated spectra. Specifically, we observed the mode at around 1850 cm^-1 and assigned it to H-O-D bend + rocking libration. Second, we found that the H-O-D bend overtone band and the OD stretch + rocking libration combination band contribute to the band located between 2850 and 3050 cm^-1. Furthermore, we assigned the broad band located between 4000 and 4200 cm^-1 to be composed of combinational modes of high-frequency OH stretching modes with predominantly twisting and rocking librations. These results should help in a proper interpretation of Raman spectra of aqueous systems as well as in the identification of vibrational relaxation pathways in isotopically diluted water.

Two-dimensional infrared-Raman spectroscopy as a probe of water's tetrahedrality

Two-dimensional spectroscopic techniques combining terahertz (THz), infrared (IR), and visible pulses offer a wealth of information about coupling among vibrational modes in molecular liquids, thus providing a promising probe of their local structure. However, the capabilities of these spectroscopies are still largely unexplored due to experimental limitations and inherently weak nonlinear signals. Here, through a combination of equilibrium-nonequilibrium molecular dynamics (MD) and a tailored spectrum decomposition scheme, we identify a relationship between the tetrahedral order of liquid water and its two-dimensional IR-IR-Raman (IIR) spectrum. The structure-spectrum relationship can explain the temperature dependence of the spectral features corresponding to the anharmonic coupling between low-frequency intermolecular and high-frequency intramolecular vibrational modes of water. In light of these results, we propose new experiments and discuss the implications for the study of tetrahedrality of liquid water.

Identification of Ion Pairs in Aqueous NaCl and KCl Solutions in Combination with Raman Spectroscopy, Molecular Dynamics, and Quantum Chemical Calculations

This work summarizes a theoretical analysis of the perturbation on Raman spectra in aqueous NaCl and KCl solutions with the aim to detect ion pairs. The experimental Raman spectra, both polarized and depolarized, are perturbed by these ions to a comparable extent or somewhat less by KCl than NaCl. This result appears to be contrary to the molecular dynamics (MD) simulation showing that the isolated and separated ions of KCl should have a larger perturbation than NaCl, as the solvation shell of K⁺ is larger than that of Na⁺. The apparent discrepancy signifies the ion pair formation which is more substantial for KCl than NaCl. The MD simulations and quantum chemical calculations revealed that KCl forms ion pairs more than NaCl and that the ion pair formation reduces the perturbation on the Raman spectra more for KCl. The present analysis shows that the perturbed Raman spectra provide a useful sign to evaluate the ion pair formation in aqueous solutions.

Spectroscopy from Machine Learning by Accurately Representing the Atomic Polar Tensor

Vibrational spectroscopy is a key technique to elucidate microscopic structure and dynamics. Without the aid of theoretical approaches, it is, however, often difficult to understand such spectra at a microscopic level. Ab initio molecular dynamics has repeatedly proved to be suitable for this purpose; however, the computational cost can be daunting. Here, the E(3)-equivariant neural network e3nn is used to fit the atomic polar tensor of liquid water a posteriori on top of existing molecular dynamics simulations. Notably, the introduced methodology is general and thus transferable to any other system as well. The target property is most fundamental and gives access to the IR spectrum, and more importantly, it is a highly powerful tool to directly assign IR spectral features to nuclear motion─a connection which has been pursued in the past but only using severe approximations due to the prohibitive computational cost. The herein introduced methodology overcomes this bottleneck. To benchmark the machine learning model, the IR spectrum of liquid water is calculated, indeed showing excellent agreement with the explicit reference calculation. In conclusion, the presented methodology gives a new route to calculate accurate IR spectra from molecular dynamics simulations and will facilitate the understanding of such spectra on a microscopic level.

Gigahertz elastic modulus and OH stretching frequency correlate with Jones-Dole's B-coefficient in aqueous solutions of the Hofmeister series

The ability of salts to change the macroscopic viscosity of their aqueous solutions is described by the Jones-Dole equation with B-coefficient for the linear concentration term. The sign and value of this coefficient are often considered as a measure of the salt's structure-making/breaking ability, while the validity of this assignment is still under discussion. Here, by applying Raman and Brillouin scattering spectroscopy to various salts from the Hofmeister series, we studied a possible relation between macroscopic Jones-Dole's B-coefficient and the microscopic dynamic response. Raman spectroscopy provides information about molecular vibrations and Brillouin spectroscopy about acoustic phonons with wavelengths of hundreds of nanometers. It has been found that Jones-Dole's B-coefficient correlates linearly with the coefficients, describing the concentration dependences of the average OH stretching frequency, real and imaginary parts of gigahertz elastic modulus. These relationships have been interpreted to mean that the OH stretching frequency is a measure of the ion-induced changes in the water network that cause changes in both viscosity and gigahertz relaxation. Depolarized inelastic light scattering revealed that the addition of structure-making ions not only changes the frequency of the relaxation peak but also increases the low-frequency part of the relaxation susceptibility. It was shown that the ion-induced increase in the gigahertz elastic modulus can be described by changes in the relaxational susceptibility without a noticeable change in the instantaneous elastic modulus. The isotropic Raman contribution associated with the tetrahedral-like environment of H₂O molecule does not correlate with Jones-Dole's B-coefficient, suggesting a minor influence of these tetrahedral-like configurations on viscosity.

Hydrogen bonding network dynamics of 1,2-propanediol-water binary solutions by Raman spectroscopy and stimulated Raman scattering

Raman spectroscopy and stimulated Raman scattering (SRS) were used to investigate the hydrogen bonding (HB) network in 1,2-propanediol (1,2-PD)-water binary solutions. Abnormal changes in hydrogen bonds (HBs) were detected when V_1,2-PD (volume fraction of the1,2-PD) was 0.4. In the case of Raman spectroscopy, the HB strength of water is weakened and then strengthened with the increase of 1,2-PD volume fraction. In the case of SRS, two new peaks at 3283 cm^-1 and 3319 cm^-1 were appeared, which demonstrated the appearance of ice-like structures near the methyl group and the weakening of HBs. Based on these phenomena, the HB structure of this binary system underwent a transition from H₂O-H₂O to H₂O-1,2-PD when the V_1,2-PD was 0.4 as V_1,2-PD increased. This work serves as a reference value for the study of HB networks in alcohol-water binary solutions.

High-dimensional neural network potentials for accurate vibrational frequencies: the formic acid dimer benchmark

In recent years, machine learning potentials (MLP) for atomistic simulations have attracted a lot of attention in chemistry and materials science. Many new approaches have been developed with the primary aim to transfer the accuracy of electronic structure calculations to large condensed systems containing thousands of atoms. In spite of these advances, the reliability of modern MLPs in reproducing the subtle details of the multi-dimensional potential-energy surface is still difficult to assess for such systems. On the other hand, moderately sized systems enabling the application of tools for thorough and systematic quality-control are nowadays rarely investigated. In this work we use benchmark-quality harmonic and anharmonic vibrational frequencies as a sensitive probe for the validation of high-dimensional neural network potentials. For the case of the formic acid dimer, a frequently studied model system for which stringent spectroscopic data became recently available, we show that high-quality frequencies can be obtained from state-of-the-art calculations in excellent agreement with coupled cluster theory and experimental data.

Coupled Cluster Molecular Dynamics of Condensed Phase Systems Enabled by Machine Learning Potentials: Liquid Water Benchmark

Coupled cluster theory is a general and systematic electronic structure method, but in particular the highly accurate "gold standard" coupled cluster singles, doubles and perturbative triples, CCSD(T), can only be applied to small systems. To overcome this limitation, we introduce a framework to transfer CCSD(T) accuracy of finite molecular clusters to extended condensed phase systems using a high-dimensional neural network potential. This approach, which is automated, allows one to perform high-quality coupled cluster molecular dynamics, CCMD, as we demonstrate for liquid water including nuclear quantum effects. The machine learning strategy is very efficient, generic, can be systematically improved, and is applicable to a variety of complex systems.

Charge Gradients around Dendritic Voids Cause Nanoscale Inhomogeneities in Liquid Water

Water is the matrix of life and serves as a solvent for numerous physical and chemical processes. The origins of the nature of inhomogeneities that exist in liquid water and the time scales over which they occur remains an open question. Here, we report femtosecond elastic second harmonic scattering (fs-ESHS) of liquid water in comparison to an isotropic liquid (CCl₄) and show that water is indeed a nonuniform liquid. The coherent fs-ESHS intensity was interpreted, using molecular dynamics simulations, as arising from charge density fluctuations with enhanced nanoscale polarizabilities around transient voids having an average lifetime of 300 fs. Although voids were also present in CCl₄, they were not characterized by hydrogen bond defects and did not show strong polarizability fluctuations, leading to fs-ESHS of an isotropic liquid. The voids increased in number at higher temperatures above room temperature, in agreement with the fs-ESHS results.

2D spectroscopies from condensed phase dynamics: Accessing third-order response properties from equilibrium multi-time correlation functions

The third-order response lies at the heart of simulating and interpreting nonlinear spectroscopies ranging from two dimensional infrared (2D-IR) to 2D electronic (2D-ES), and 2D sum frequency generation (2D-SFG). The extra time and frequency dimensions in these spectroscopies provides access to rich information on the electronic and vibrational states present, the coupling between them, and the resulting rates at which they exchange energy that are obscured in linear spectroscopy, particularly for condensed phase systems that usually contain many overlapping features. While the exact quantum expression for the third-order response is well established it is incompatible with the methods that are practical for calculating the atomistic dynamics of large condensed phase systems. These methods, which include both classical mechanics and quantum dynamics methods that retain quantum statistical properties while obeying the symmetries of classical dynamics, such as LSC-IVR, Centroid Molecular Dynamics (CMD) and Ring Polymer Molecular Dynamics (RPMD) naturally provide short-time approximations to the multi-time symmetrized Kubo transformed correlation function. Here, we show how the third-order response can be formulated in terms of equilibrium symmetrized Kubo transformed correlation functions. We demonstrate the utility and accuracy of our approach by showing how it can be used to obtain the third-order response of a series of model systems using both classical dynamics and RPMD. In particular, we show that this approach captures features such as anharmonically induced vertical splittings and peak shifts while providing a physically transparent framework for understanding multidimensional spectroscopies.

Rapid Water Dynamics Structures the OH-Stretching Spectra of Solitary Water in Ionic Liquids and Dipolar Solvents

In a recent study [ J. Phys. Chem. B 126, 4584 (2022)] we used infrared spectroscopy to investigate the solvation and dynamics of solitary water in ionic liquids and dipolar solvents. Complex shapes observed for water OH-stretching bands common to all high-polarity solvents were assigned to water in several solvation states. In the present study, classical molecular dynamics simulations of a single water molecule in four ionic liquids and three dipolar solvents were used to test and refine this interpretation. Consistent with past assignments, simulations show solitary water usually donates two hydrogen bonds to distinct solvent molecules. Such symmetrically solvated water produces the primary pair of peaks identified in the OH spectra of water in nearly all solvents. We had further proposed that additional features flanking this main peak are due to asymmetric solvation states, states in which only one OH group makes a hydrogen bond to solvent. Such states were found in significant concentrations in all of the systems simulated. Simulations of the OH stretching spectra using a semiclassical description and the vibrational map developed by Auer and Skinner [ J. Chem. Phys. 128, 224511 (2008)] provided semi-quantitative agreement with experiment. Analysis of species-specific spectra also confirmed assignment of the additional features in the experimental spectra to asymmetrically solvated water. The simulations also showed that rapid water motions cause a marked motional narrowing compared to the inhomogeneous limit, and that this narrowing is largely responsible for making the additional features due to minority solvation states manifest in the spectra.

Toward High-level Machine Learning Potential for Water Based on Quantum Fragmentation and Neural Networks

Accurate and efficient simulation of liquids, such as water and salt solutions, using high-level wave function theories is still a formidable task for computational chemists owing to the high computational costs. In this study, we develop a deep machine learning potential based on fragment-based second-order Møller-Plesset perturbation theory (DP-MP2) for water through neural networks. We show that the DP-MP2 potential predicts the structural, dynamical, and thermodynamic properties of liquid water in better agreement with the experimental data than previous studies based on density functional theory (DFT). The nuclear quantum effects (NQEs) on the properties of liquid water are also examined, which are noticeable in affecting the structural and dynamical properties of liquid water under ambient conditions. This work provides a general framework for quantitative predictions of the properties of condensed-phase systems with the accuracy of high-level wave function theory while achieving significant computational savings compared to ab initio simulations.

Water does not dance as ions sing: A new approach in elucidation of ion-invariant water fluctuations

Aqueous solutions of salts composed from monovalent ions are explored using temperature-dependent FT-IR spectroscopy in transmission. Water combination band, being extremely sensitive to the network of hydrogen bonds due to the contribution of water librations (ρ_LH₂O), is analyzed in uni- and multivariate fashion. Univariate analysis of the combination band maximum (ν_max) reveals that perturbation of water hydrogen bond network by ions is primary driven by electrostatic interactions between water and ions. Using multivariate curve resolution with alternating least squares and evolving factor analysis this band is separated into two components that represent low- and high-density water. The observed asymmetry in their behavior is interpreted in terms of fluctuations of a hydrogen bond network of two water components. The significance of the found phenomenon is unambiguously confirmed by performing analogous analysis in the spectral range that contains partial signature of water linear bending (δHOH) and is free from ρ_LH₂O, in which the asymmetry is absent. Additionally, we show that this phenomenon, namely ion-invariant behavior of water fluctuations, persists even in the regime of high ionic strengths. Although ions indeed participate in shaping of water hydrogen bond network, this straightforward approach shows that its temperature-dependent fluctuations are ion-independent.

Isotope effects on the structural transformation and relaxation of deeply supercooled water

We have examined the structure of supercooled liquid D2O as a function of temperature between 185 and 255 K using pulsed laser heating to rapidly heat and cool the sample on a nanosecond timescale. The liquid structure can be represented as a linear combination of two structural motifs, with a transition between them described by a logistic function centered at 218 K with a width of 10 K. The relaxation to a metastable state, which occurred prior to crystallization, exhibited nonexponential kinetics with a rate that was dependent on the initial structural configuration. When the temperature is scaled by the temperature of maximum density, which is an isostructural point of the isotopologues, the structural transition and the non-equilibrium relaxation kinetics of D2O agree remarkably well with those for H2O.

Weakly hydrated anions bind to polymers but not monomers in aqueous solutions

Weakly hydrated anions help to solubilize hydrophobic macromolecules in aqueous solutions, but small molecules comprising the same chemical constituents precipitate out when exposed to these ions. Here, this apparent contradiction is resolved by systematically investigating the interactions of NaSCN with polyethylene oxide oligomers and polymers of varying molecular weight. A combination of spectroscopic and computational results reveals that SCN^- accumulates near the surface of polymers, but is excluded from monomers. This occurs because SCN^- preferentially binds to the centre of macromolecular chains, where the local water hydrogen-bonding network is disrupted. These findings suggest a link between ion-specific effects and theories addressing how hydrophobic hydration is modulated by the size and shape of a hydrophobic entity.

Influence of Methylene Fluorination and Chain Length on the Hydration Shell Structure and Thermodynamics of Linear Diols

The interplay between the local hydration shell structure, the length of hydrophobic solutes, and their identity (perfluorinated or not) remains poorly understood. We address this issue by combining Raman-multivariate curve resolution (Raman-MCR) spectroscopy, simulation, and quantum-mechanical calculations to quantify the thermodynamics and the first principle interactions behind the formation of defects in the hydration shell of alkyl-diol and perfluoroalkyl-diol chains. The hydration shell of the fluorinated diols contains substantially more defects than that of the nonfluorinated diols; these defects are water hydroxy groups that do not donate hydrogen bonds and which either point to the solute (radial-dangling OH) or not (nonradial-dangling OH). The number of radial-dangling OH defects per carbon decreases for longer chains and toward the interior of the fluorinated diols, mainly due to less favorable electrostatics and exchange interactions; nonradial-dangling OH defects per carbon increase with chain length. In contrast, the hydration shell of the nonfluorinated diols only contains radial-dangling defects, which become more abundant toward the center of the chain and for larger chains, predominantly because of more favorable dispersion interactions. These results have implications for how the folding of macromolecules, ligand binding to biomacromolecules, and chemical reactions at water-oil interfaces could be modified through the introduction of fluorinated groups or solvents.

Water as a probe for pH measurement in individual particles using micro-Raman spectroscopy

Atmospheric aerosol acidity impacts numerous physicochemical processes, but the determination of particle pH remains a significant challenge due to the nonconservative nature of the H⁺ concentration ([H⁺]). Traditional measurements have difficulty in describing the practical state of an aerosol because they comprise chemical components or hypotheses that change the nature of the particles. In this work, we present a direct pH measurement that uses water as a general probe to detect [H⁺] in individual particles by micro-Raman spectroscopy. Containing the vibrational bands of ions and water influenced by ions, the spectra of hydrated ion were decomposed from the solution spectra as standard spectra by multivariate curve resolution analysis. Meanwhile, ratios of hydrated ions were calculated between the Raman spectra and standard spectra to evaluate concentration profiles of each ion. It demonstrated that good quantitative models between the ratio and concentration for all ions including H⁺ can be built with correlation coefficients (R²) higher than 0.95 for the solutions. The method was further applied to individual particle pH measurement. The pH value of sulfate aerosol particles was calculated, and the standard error was 0.09 using pH values calculated from the [HSO₄^-]/[SO₄^2-] as a reference. Furthermore, the applicability of the method was proven by detecting the pH value of chloride particles. Therefore, utilizing water, the most common substance, as the spectroscopic probe to measure [H⁺] without restriction of the ion system, this method has potential to measure the pH value of atmospheric particles with various compounds, although more work needs to be done to improve the sensitivity of the method.

Freezing of Dilute Aqueous-Alcohol Nanodroplets: The Effect of Molecular Structure

We investigate vapor-liquid nucleation and subsequent freezing of aqueous-alcohol nanodroplets containing 1-pentanol, 1-hexanol, and their 3-isomers. The aerosols are produced in a supersonic nozzle, where condensation and freezing are characterized by static pressure and Fourier transform Infrared (FTIR) spectroscopy measurements. At fixed water concentrations, the presence of alcohol enables particle formation at higher temperatures since both the equilibrium vapor pressure above the critical clusters and the cluster interfacial free energy are decreased relative to the pure water case. The disappearance of a small free OH peak, observed for pure water droplets, when alcohols are added and shifts in the CH peaks as a function of alcohol chain length reveal varying surface partitioning preferences of the alcohols. Changes in the FTIR spectra during freezing, as well as changes in the ice component derived from self-modeling curve resolution analysis, show that 1-hexanol and 1-pentanol perturb freezing less than their branched isomers do. This behavior may reflect the molecular footprints of the alcohols, the available surface area of the droplets, and not only alcohol solubility. The presence of alcohols also lowers the freezing temperature relative to that of pure water, but when there is clear evidence for the formation of ice, the ice nucleation rates change by less than a factor of ∼2-3 for all cases studied.

Study of hydrogen bonding interactions in ethylene glycol-water binary solutions by Raman spectroscopy

EG (ethylene glycol) is a good model system for the study of the fundamental hydrogen bonds in aqueous solutions. Using Raman spectroscopy, we have investigated the EG volume fraction induced variation in the hydrogen bonding interactions and conformations of EG-H₂O (water) binary solutions. New hydrogen bonding networks is evidenced by the appearance of remarkable changes in Raman spectra and the full width at half maximum (FWHM) when the mixing volume ratio is 0.5. The H-bond in water molecules firstly strengthened and then weakened with the increasing concentration of EG. Meanwhile, the dominant association structure also changed from H₂O-H₂O to EG-H₂O in binary solutions in this process. We provide a simple but effective method for studying EG-H₂O binary solutions. It also has exciting potential prospects and can be easily extended to other mixing situations.

Efficient Quantum Vibrational Spectroscopy of Water with High-Order Path Integrals: From Bulk to Interfaces

Vibrational spectroscopy is key in probing the interplay between the structure and dynamics of aqueous systems. To map different regions of experimental spectra to the microscopic structure of a system, it is important to combine them with first-principles atomistic simulations that incorporate the quantum nature of nuclei. Here we show that the large cost of calculating the quantum vibrational spectra of aqueous systems can be dramatically reduced compared with standard path integral methods by using approximate quantum dynamics based on high-order path integrals. Together with state-of-the-art machine-learned electronic properties, our approach gives an excellent description not only of the infrared and Raman spectra of bulk water but also of the 2D correlation and the more challenging sum-frequency generation spectra of the water-air interface. This paves the way for understanding complex interfaces such as water encapsulated between or in contact with hydrophobic and hydrophilic materials through robust and inexpensive surface-sensitive and multidimensional spectra with first-principles accuracy.

Combining Machine Learning and Computational Chemistry for Predictive Insights Into Chemical Systems

Machine learning models are poised to make a transformative impact on chemical sciences by dramatically accelerating computational algorithms and amplifying insights available from computational chemistry methods. However, achieving this requires a confluence and coaction of expertise in computer science and physical sciences. This Review is written for new and experienced researchers working at the intersection of both fields. We first provide concise tutorials of computational chemistry and machine learning methods, showing how insights involving both can be achieved. We follow with a critical review of noteworthy applications that demonstrate how computational chemistry and machine learning can be used together to provide insightful (and useful) predictions in molecular and materials modeling, retrosyntheses, catalysis, and drug design.

AENET-LAMMPS and AENET-TINKER: Interfaces for accurate and efficient molecular dynamics simulations with machine learning potentials

Machine-learning potentials (MLPs) trained on data from quantum-mechanics based first-principles methods can approach the accuracy of the reference method at a fraction of the computational cost. To facilitate efficient MLP-based molecular dynamics and Monte Carlo simulations, an integration of the MLPs with sampling software is needed. Here, we develop two interfaces that link the atomic energy network (ænet) MLP package with the popular sampling packages TINKER and LAMMPS. The three packages, ænet, TINKER, and LAMMPS, are free and open-source software that enable, in combination, accurate simulations of large and complex systems with low computational cost that scales linearly with the number of atoms. Scaling tests show that the parallel efficiency of the ænet-TINKER interface is nearly optimal but is limited to shared-memory systems. The ænet-LAMMPS interface achieves excellent parallel efficiency on highly parallel distributed-memory systems and benefits from the highly optimized neighbor list implemented in LAMMPS. We demonstrate the utility of the two MLP interfaces for two relevant example applications: the investigation of diffusion phenomena in liquid water and the equilibration of nanostructured amorphous battery materials.

Nuclear Quantum Effect and Its Temperature Dependence in Liquid Water from Random Phase Approximation via Artificial Neural Network

We report structural and dynamical properties of liquid water described by the random phase approximation (RPA) correlation together with the exact exchange energy (EXX) within density functional theory. By utilizing thermostated ring polymer molecular dynamics, we examine the nuclear quantum effects and their temperature dependence. We circumvent the computational limitation of performing direct first-principles molecular dynamics simulation at this high level of electronic structure theory by adapting an artificial neural network model. We show that the EXX+RPA level of theory accurately describes liquid water in terms of both dynamical and structural properties.

Strategies for the construction of machine-learning potentials for accurate and efficient atomic-scale simulations

Recent advances in machine-learning interatomic potentials have enabled the efficient modeling of complex atomistic systems with an accuracy that is comparable to that of conventional quantum-mechanics based methods. At the same time, the construction of new machine-learning potentials can seem a daunting task, as it involves data-science techniques that are not yet common in chemistry and materials science. Here, we provide a tutorial-style overview of strategies and best practices for the construction of artificial neural network (ANN) potentials. We illustrate the most important aspects of (a) data collection, (b) model selection, (c) training and validation, and (d) testing and refinement of ANN potentials on the basis of practical examples. Current research in the areas of active learning and delta learning are also discussed in the context of ANN potentials. This tutorial review aims at equipping computational chemists and materials scientists with the required background knowledge for ANN potential construction and application, with the intention to accelerate the adoption of the method, so that it can facilitate exciting research that would otherwise be challenging with conventional strategies.

Hydration and Seamless Integration of Hydrogen Peroxide in Water

Raman multivariate curve resolution is used to decompose the vibrational spectra of aqueous hydrogen peroxide (H2O2) into pure water, dilute H2O2, and concentrated H2O2 spectral components. The dilute spectra reveal four sub-bands in the OH stretch region, assigned to the OH stretch and Fermi resonant bend overtone of H2O2, and two nonequivalent OH groups on water molecules that donate a hydrogen bond to H2O2. At high concentrations, a spectral component resembling pure H2O2 emerges. Our results further demonstrate that H2O2 perturbs the structure of water significantly less than either methanol or sodium chloride of the same concentration, as evidenced by comparing the hydration-shell spectra of tert-butyl alcohol dissolved in the three aqueous solutions.

The freezing behavior of aqueous n-alcohol nanodroplets

We generate water-rich aerosols containing 1-propanol and 1-pentanol in a supersonic nozzle to study the effects of these solutes on the freezing behavior of water. Condensation and freezing are characterized by two complementary techniques, pressure trace measurements and Fourier Transform Infrared spectroscopy. When 1-pentanol and 1-propanol are present, condensation occurs at higher temperatures because particle formation from the vapor phase is enhanced by the decrease in interfacial free energy of mixed aqueous-alcohol critical clusters relative to those of pure water. FTIR results suggest that when ∼6 nm radius droplets freeze, the tetrahedral structure of the ice is well preserved up to an overall alcohol mole fraction of 0.031 for 1-propanol and 0.043 for 1-pentanol. In this concentration range, the ice nucleation temperature decreases continuously with increasing 1-propanol concentration, whereas the onset of freezing is not significantly perturbed by 1-pentanol up to a mole fraction of 0.03. Furthermore, once freezing starts the ice nucleation rates in the aqueous-alcohol droplets are very close to those for pure water. In contrast, at the highest mole fractions of either alcohol it is not clear whether droplets freeze to form crystalline ice since the final state of the particles cannot be adequately characterized with the available experimental techniques.

The key role of solvent in condensation: Mapping water in liquid-liquid phase-separated FUS

Formation of biomolecular condensates through liquid-liquid phase separation (LLPS) has emerged as a pervasive principle in cell biology, allowing compartmentalization and spatiotemporal regulation of dynamic cellular processes. Proteins that form condensates under physiological conditions often contain intrinsically disordered regions with low-complexity domains. Among them, the RNA-binding proteins FUS and TDP-43 have been a focus of intense investigation because aberrant condensation and aggregation of these proteins is linked to neurodegenerative diseases such as amyotrophic lateral sclerosis and frontotemporal dementia. LLPS occurs when protein-rich condensates form surrounded by a dilute aqueous solution. LLPS is per se entropically unfavorable. Energetically favorable multivalent protein-protein interactions are one important aspect to offset entropic costs. Another proposed aspect is the release of entropically unfavorable preordered hydration water into the bulk. We used attenuated total reflection spectroscopy in the terahertz frequency range to characterize the changes in the hydrogen bonding network accompanying the FUS enrichment in liquid-liquid phase-separated droplets to provide experimental evidence for the key role of the solvent as a thermodynamic driving force. The FUS concentration inside LLPS droplets was determined to be increased to 2.0 mM independent of the initial protein concentration (5 or 10 μM solutions) by fluorescence measurements. With terahertz spectroscopy, we revealed a dewetting of hydrophobic side chains in phase-separated FUS. Thus, the release of entropically unfavorable water populations into the bulk goes hand in hand with enthalpically favorable protein-protein interaction. Both changes are energetically favorable, and our study shows that both contribute to the thermodynamic driving force in phase separation.

Machine learning-accelerated quantum mechanics-based atomistic simulations for industrial applications

Atomistic simulations have become an invaluable tool for industrial applications ranging from the optimization of protein-ligand interactions for drug discovery to the design of new materials for energy applications. Here we review recent advances in the use of machine learning (ML) methods for accelerated simulations based on a quantum mechanical (QM) description of the system. We show how recent progress in ML methods has dramatically extended the applicability range of conventional QM-based simulations, allowing to calculate industrially relevant properties with enhanced accuracy, at reduced computational cost, and for length and time scales that would have otherwise not been accessible. We illustrate the benefits of ML-accelerated atomistic simulations for industrial R&D processes by showcasing relevant applications from two very different areas, drug discovery (pharmaceuticals) and energy materials. Writing from the perspective of both a molecular and a materials modeling scientist, this review aims to provide a unified picture of the impact of ML-accelerated atomistic simulations on the pharmaceutical, chemical, and materials industries and gives an outlook on the exciting opportunities that could emerge in the future.

On the existence of soliton-like collective modes in liquid water at the viscoelastic crossover

The problem of large-density variations in supercooled and ambient water has been widely discussed in the past years. Recent studies have indicated the possibility of nanometer-sized density variations on the subpicosecond and picosecond time scales. The nature of fluctuating density heterogeneities remains a highly debated issue. In the present work, we address the problem of possible association of such density variations with the dynamics of terahertz longitudinal acoustic-like modes in liquid water. Our study is based on the fact that the subpicosecond dynamics of liquid water are essentially governed by the structural relaxation. Using a mode coupling theory approach, we found that for typical values of parameters of liquid water, the dynamic mechanism coming from the combination of the structural relaxation process and the finiteness of the amplitude of terahertz longitudinal acoustic-like mode gives rise to a soliton-like collective mode on a temperature-dependent nanometer length scale. The characteristics of this mode are consistent with the estimates of the amplitudes and temperature-dependent correlation lengths of density fluctuations in liquid water obtained in experiments and simulations. Thus, the fully dynamic mechanism could contribute to the formation and dynamics of fluctuating density heterogeneities. The soliton-like collective excitations suggested by our analysis may be relevant to different phenomena connected with supercooled water and can be expected to be associated with some ultrafast biological processes.

Coupling of three intramolecular vibration modes of liquid H2O molecules in the framework of the fluctuation theory of hydrogen bonding

Vibrational spectra of liquid water contain a wealth of information about its structure and dynamics but there are no generally acknowledged interpretation of their band profiles, unfortunately. We have tried to calculate them taking into account only the coupling of three intramolecular vibrations in the set of H₂O molecules with different initial oscillator frequencies and intensities. The matter is that each water molecule forms hydrogen bonds of different strengths; thus the OH stretching band spans several hundreds of wave numbers (the fluctuation theory of hydrogen bonding). This distribution overlaps with the similar band of the first overtone of the HOH bending frequencies thus triggering a Fermi resonance between three vibrations. There were some problems causing some simplifications in previous theoretical modeling of vibrational transitions in condensed water. To solve them we extract the statistical distribution of OH frequencies of H₂O molecules directly from the experimental spectra of HOD molecules at the same conditions instead of defining it theoretically. Also the bending overtone is allowed to have non-zero intrinsic intensity when calculating the Fermi resonance. The test calculation of the isotropic component of the Raman spectrum which is most critical for interpretation shows that our algorithm can reproduces the characteristic peculiarities of the experiment. The spectrum consists of three non-Gaussian contours. The overtone of bending vibration, being strengthened by Fermi resonance, makes the greatest contribution to the dominating low-frequency spectrum component that was previously attributed to intermolecular coupling of adjacent OH oscillators by some authors. Further we plan to calculate similar band profiles in the IR and also Raman isotropic and anisotropic spectra of Н₂О and D₂О molecules within a wide temperature range for their quantitative comparison with experiment.

Temperature Dependence of the Water Infrared Spectrum: Driving Forces, Isosbestic Points, and Predictions

The temperature derivative of the infrared (IR) spectrum of HOD/D₂O is directly calculated from simulations at a single temperature using a fluctuation theory approach. It is demonstrated, on the basis of an energetic decomposition of the derivative, that the blue shift with increasing temperature is associated with the competition between electrostatic and Lennard-Jones interactions. The same competition gives rise, where their contributions cancel, to a near isosbestic point. The derivative is further used to define an effective internal energy (and entropy) associated with the IR spectrum, and it is shown how a van't Hoff relation can be used to accurately predict the spectrum over a wide range of temperatures. These predictions also explain why a precise isosbestic point is not observed.

Exploiting Machine Learning to Efficiently Predict Multidimensional Optical Spectra in Complex Environments

The excited-state dynamics of chromophores in complex environments determine a range of vital biological and energy capture processes. Time-resolved, multidimensional optical spectroscopies provide a key tool to investigate these processes. Although theory has the potential to decode these spectra in terms of the electronic and atomistic dynamics, the need for large numbers of excited-state electronic structure calculations severely limits first-principles predictions of multidimensional optical spectra for chromophores in the condensed phase. Here, we leverage the locality of chromophore excitations to develop machine learning models to predict the excited-state energy gap of chromophores in complex environments for efficiently constructing linear and multidimensional optical spectra. By analyzing the performance of these models, which span a hierarchy of physical approximations, across a range of chromophore-environment interaction strengths, we provide strategies for the construction of machine learning models that greatly accelerate the calculation of multidimensional optical spectra from first principles.