Journal of Chemical Physics最新文献

The evolution of molecular dynamics (MD) simulations has been intimately linked to that of computing hardware. For decades following the creation of MD, simulations have improved with computing power along the three principal dimensions of accuracy, atom count (spatial scale), and duration (temporal scale). Since the mid-2000s, computer platforms have, however, failed to provide strong scaling for MD, as scale-out central processing unit (CPU) and graphics processing unit (GPU) platforms that provide substantial increases to spatial scale do not lead to proportional increases in temporal scale. Important scientific problems therefore remained inaccessible to direct simulation, prompting the development of increasingly sophisticated algorithms that present significant complexity, accuracy, and efficiency challenges. While bespoke MD-only hardware solutions have provided a path to longer timescales for specific physical systems, their impact on the broader community has been mitigated by their limited adaptability to new methods and potentials. In this work, we show that a novel computing architecture, the Cerebras wafer scale engine, completely alters the scaling path by delivering unprecedentedly high simulation rates up to 1.144 M steps/s for 200 000 atoms whose interactions are described by an embedded atom method potential. This enables direct simulations of the evolution of materials using general-purpose programmable hardware over millisecond timescales, dramatically increasing the space of direct MD simulations that can be carried out. In this paper, we provide an overview of advances in MD over the last 60 years and present our recent result in the context of historical MD performance trends.

Delocalized excitations, denoted excitons, play an important role in many systems in chemical physics. The characterization of their extent of delocalization is a crucial element in understanding these quasiparticles. In this paper, I will revisit the most common delocalization measures applied to Frenkel-type excitons. Based on this analysis, I propose to use a so-far ignored measure. The key advantage of this measure, which I will denote as the Manhattan exciton size, is that it directly connects with the oscillator strength of the excitons. It provides a strict upper bound on the oscillator strength of any given exciton for linear aggregates. Finally, I demonstrate that this exciton delocalization measure is more sensible for analyzing super-radiant states compared to, for example, the most commonly applied measure, i.e., the (inverse) participation ratio. However, these two measures together provide insight into the degree of exciton confinement.

In this paper, we examine the time-fractional Schrödinger equation from the perspective of non-Markovian dynamics in dissipative systems. First, we determine the range of the fractional derivative's order by examining the memory properties of the time-fractional Schrödinger equation. Next, we employ the Jaynes-Cummings model to identify the appropriate mathematical form of the imaginary unit. Finally, we use the refined equation to study quantum teleportation under amplitude damping noise. It was found that the time-fractional Schrödinger equation without fractional operations on the imaginary unit i might be more suitable for describing non-Markovian dynamics in dissipative systems. Our research may provide a new perspective on the time-fractional Schrödinger equation, contributing to a deeper understanding and further development of time-fractional quantum mechanics.

Reaction coordinates are a useful tool that allows the complex dynamics of a protein in high-dimensional phase space to be projected onto a much simpler model with only a few degrees of freedom, while preserving the essential aspects of that dynamics. In this way, reaction coordinates could provide an intuitive, albeit simplified, understanding of the complex dynamics of proteins. Together with molecular dynamics (MD) simulations, reaction coordinates can also be used to sample the phase space very efficiently and to calculate transition rates and paths between different metastable states. Unfortunately, ideal reaction coordinates for a system capable of these performances are not known a priori, and an efficient calculation in the course of an MD simulation is currently an active field of research. An alternative is to use geometric reaction coordinates, which, although generally unable to provide quantitative accuracy, are useful for simplified mechanistic models of protein dynamics and can thus help gain insights into the fundamental aspects of these dynamics. In this study, five such geometric reaction coordinates, such as the end-to-end distance, the radius of gyration, the solvent accessible surface area, the root-mean-square distance (RMSD), and the mean native hydrogen bond length, are compared. For this purpose, extensive molecular dynamics simulations were carried out for two peptides and a small protein in order to calculate and compare free energy profiles with the aid of the reaction coordinates mentioned. While none of the investigated geometrical reaction coordinates could be demonstrated to be an optimal reaction coordinate, the RMSD and the mean native hydrogen bond length appeared to perform more effectively than the other three reaction coordinates.

The Madrid-2019 intermolecular potential was developed for use in molecular simulations of salty aqueous solutions. The selection of the accurate TIP4P/2005 potential for water and the adoption of scaled charges for ions, ±0.85e for monovalent ions and ±1.70e for divalent ions, are the key features of the model. The use of scaled charges enhances the description of several properties, including solubility, transport properties, the density maximum, and the water activity in ionic solutions. In this study, we will investigate the performance of scaled charges in describing the properties of inorganic salts containing Cl-, Li+, Na+, and Ca+2 in another polar solvent, methanol. The ion charges and ion-ion interactions were taken from the Madrid-2019 potential, while the accurate OPLS/2016 model was selected for methanol. The protocol used in the development of the Madrid-2019 model, particularly regarding the selection of target properties in the fitting procedure, was applied to create this potential using LiCl, NaCl, and CaCl2 as inorganic salts. Its predictive ability was evaluated by calculating the density, dielectric constant, self-diffusion coefficients of methanol and ions, and viscosity for methanolic solutions of these three salts. As will be shown, the experimentally observed effects of salt addition are reproduced by the new model, not only qualitatively but also quantitatively. Furthermore, since the interaction potential is compatible with the Madrid-2019 model, we also demonstrated its accurate predictive ability in the ternary system methanol + water + NaCl.

Solvation plays a pivotal role in chemistry to effectively steer chemical reactions. While liquid water has been extensively studied, our molecular-level knowledge of other associated liquids capable of forming H-bond networks, such as liquid methanol, remains surprisingly scarce. We use large-scale ab initio molecular dynamics simulations to comprehensively study the structural, dynamical, and electronic properties of bulk methanol under ambient conditions. Methanol is an interesting species in the liquid state since it can only donate one H-bond while a significant fraction accepts two H-bonds, which imprints one-dimensional linear and cyclic H-bonding patterns subject to significant bifurcations. After validation of radial distribution functions and the self-diffusion coefficient with respect to experimental data, we carried out detailed analyses of the H-bond network topology in terms of chain-like, ring-like, and branched H-bonded aggregates, including lifetime assessment. The analysis revealed that nearly all methanol molecules are actively engaged in filamentary H-bonding, predominantly forming branched linear chains with a significant contribution arising from tetrameric to hexameric rings-in stark contrast to the compact three-dimensional H-bond network of water. Five-membered rings turned out to be the most long-lived cyclic structures with an intermittent lifetime of 4 ps, while rings consisting of only three methanol molecules as well as very large cyclic structures are merely transient motifs. Detailed analyses of the effective electric molecular dipoles disclose a pronounced sensitivity of non-additive polarization and charge transfer effects of the individual methanol molecules to the particular H-bond network structure they are a member of, including its topology, be it linear or cyclic.

Light-induced spin polarization can be produced in chromophore-radical conjugates by reversible transitions between the excited trip-doublet and trip-quartet states. The precise origin of this polarization is often difficult to elucidate because different transition pathways, promoted by different interactions, can occur depending on the nature of the conjugate. Moreover, the complexity of the expressions describing the evolution of the spin state populations and polarization generated by these transitions makes it difficult to estimate the dependence of the polarization on factors such as the exchange interaction and spin-orbit coupling. Here, we present a theoretical analysis and show that by making assumptions for specific cases, simplified expressions can be obtained that provide better insight into the physical origins of the polarization.

A full-dimensional potential energy surface (PES) represented by the neural network method for the first excited state S1(1A″) of HPCO is reported for the first time. The PES was constructed based on more than 51 000 ab initio points, which were calculated at the multi-reference configuration interaction level with Davidson correction using the augmented correlation consistent polarized valence triple zeta basis set. Based on the newly constructed PES, quasi-classical trajectory calculations were carried out to study the photodissociation dynamics of HPCO at the total energy ranging from 4.0 to 5.6 eV. At low total energies, the HP + CO product is dominant, while the product H + PCO becomes increasingly favored at higher energies. Furthermore, the translational energy distributions of two products are found to be energy-dependent. Owing to the strongly repulsive PES along the HP + CO dissociation pathway, the translational energy distributions of HP + CO are dominated by relatively higher energies in contrast to H + PCO. The diatomic products HP and CO are found to possess the vibrational distributions decaying monotonically with the vibrational quantum number and relatively cold rotational state distributions, consistent with the strongly repulsive potentials toward the HP + CO channel. In addition, the vibrational distributions of HP and CO are found to be quite similar due to their close frequencies, while the rotational distributions of CO have a much more highly excited rotational degree of freedom owing to its rotational constant approximately four times smaller than that of HP.

Liquid gallium exhibits a unique metallic-covalent coexistence. Leveraging the volume constant pressure molecular dynamics method and a well-trained neural network potential, we study the evolution of liquid Ga surface structures under varying temperatures and pressures. Our study presents a schematic P-T phase diagram of the liquid surface. We observe symmetric static structure factor main peaks in the outermost layers of the liquid Ga surface compared with asymmetric ones for inner layers, indicating a simple liquid behavior and a lack of Ga2 dimers at the surface. We calculate the surface energy and the surface tension, which reveal non-monotonic changes. All these results provide a further insight into understanding the physics of the strange metal gallium.

The development of experimental techniques at the nanoscale has enabled the performance of spectroscopic measurements on single-molecule current-carrying junctions. These experiments serve as a natural intersection for the research fields of optical spectroscopy and molecular electronics. We present a pedagogical comparison between the perturbation theory expansion of standard nonlinear optical spectroscopy and the (non-self-consistent) perturbative diagrammatic formulation of the nonequilibrium Green's functions method (which is widely used in molecular electronics), highlighting their similarities and differences. By comparing the two approaches, we argue that the optical spectroscopy of open quantum systems must be analyzed within the more general Green's function framework.