Structure and Dynamics of the Magnetite(001)/Water Interface from Molecular Dynamics Simulations Based on a Neural Network Potential

Journal of Chemical Theory and Computation, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 13, 2025

The magnetite/water interface is commonly found in nature and plays a crucial role various technological applications. However, our understanding of its structural dynamical properties at the molecular scale remains still limited. In this study, we developed an efficient Behler-Parrinello neural network potential (NNP) for system, paying particular attention to accurate generation reference data with density functional theory. Using NNP, performed extensive dynamics simulations magnetite (001) surface across wide range water coverages, from single molecules bulk water. Our revealed several new ground states low coverage on Subsurface Cation Vacancy (SCV) model yielded profile that exhibits marked layering. By calculating mean square displacements, obtained quantitative information diffusion SCV different revealing significant anisotropy. Additionally, provided qualitative insights into dissociation mechanisms surface.

Language: Английский

---

Atomistic simulation of batteries via machine learning force fields: from bulk to interface

Journal of Energy Chemistry, Journal Year: 2025, Volume and Issue: unknown

Published: March 1, 2025

Language: Английский

---

Protons Accumulate at the Graphene–Water Interface

ACS Nano, Journal Year: 2025, Volume and Issue: unknown

Published: April 28, 2025

Water's ability to autoionize into hydroxide and hydronium ions profoundly influences surface properties, rendering interfaces either basic or acidic. While it is well-established that protons show an affinity the air-water interface, a critical knowledge gap exists in technologically relevant surfaces like graphene-water interface. Here we use machine learning-based simulations with first-principles accuracy unravel behavior of at Our findings reveal accumulate ion predominantly residing first contact layer water. In contrast, exhibits bimodal distribution, found both near further away from it. Analysis underlying electronic structure reveals local polarization effects, resulting counterintuitive charge rearrangement. Proton propensity interface challenges interpretation experiments expected have far-reaching consequences for conductivity, interfacial reactivity, proton-mediated processes.

Language: Английский

---

Free energy profiles for chemical reactions in solution from high-dimensional neural network potentials: The case of the Strecker synthesis

The Journal of Chemical Physics, Journal Year: 2025, Volume and Issue: 162(17)

Published: May 6, 2025

Machine learning potentials (MLPs) have become a popular tool in chemistry and materials science as they combine the accuracy of electronic structure calculations with high computational efficiency analytic potentials. MLPs are particularly useful for computationally demanding simulations such determination free energy profiles governing chemical reactions solution, but to date, applications still rare. In this work, we show how umbrella sampling can be combined active high-dimensional neural network (HDNNPs) construct systematic way. For example first step Strecker synthesis glycine aqueous provide detailed analysis improving quality HDNNPs datasets increasing size. We find that, addition typical quantification force errors respect underlying density functional theory data, long-term stability convergence physical properties should rigorously monitored obtain reliable converged solution.

Language: Английский

---

Terahertz calorimetry spotlights the role of water in biological processes

Nature Reviews Chemistry, Journal Year: 2025, Volume and Issue: unknown

Published: May 9, 2025

Language: Английский

---

Density Isobar of Water and Melting Temperature of Ice: Assessing Common Density Functionals

Published: June 6, 2024

We investigate the density isobars of water and melting temperature ice using six different functionals. Machine-learning potentials are employed to ensure computational affordability. Our findings reveal significant discrepancies between various base Notably, even choice damping can result in substantial differences. Overall, outcomes obtained through functional theory not entirely satisfactory across most utilized All functionals exhibit deviations either or equilibrium volume, with them predicting an incorrect volume difference water. heuristic analysis indicates that a hybrid 25% exact exchange van der Waals averaged zero Becke-Johnson dampings yields closest agreement experimental data. This study underscores necessity for further enhancements treatment interactions and, more broadly, theory, enable accurate quantitative predictions molecular liquids.

Language: Английский

---

Structure and thermodynamics of defects in Na-feldspar from a neural network potential

Physical Review Materials, Journal Year: 2024, Volume and Issue: 8(7)

Published: July 18, 2024

The diffusive phase transformations occurring in feldspar, a common mineral the crust of Earth, are essential for reconstructing thermal histories magmatic and metamorphic rocks. Due to long timescales over which these proceed, mechanism responsible sodium diffusion its possible anisotropy has remained topic debate. To elucidate this defect-controlled process, we have developed neural network potential (NNP) trained on first-principle calculations Na-feldspar (albite) charged defects. This force field reproduces various experimentally known properties including lattice parameters elastic constants as well heat capacity DFT-calculated defect formation energies. A new type dumbbell interstitial is found be most favorable, free energy at finite temperature calculated using thermodynamic integration. necessity electrostatic corrections before training an NNP demonstrated by predicting more consistent Published American Physical Society 2024

Language: Английский

---

The wetting of H2O by CO2

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 161(8)

Published: Aug. 28, 2024

Biphasic interfaces are complex but fascinating regimes that display a number of properties distinct from those the bulk. The CO2–H2O interface, in particular, has been subject studies on account its importance for carbon life cycle as well capture and sequestration schemes. Despite this attention, there remain open questions nature particularly concerning interfacial tension phase behavior CO2 at interface. In paper, we seek to address these ambiguities using ab initio-quality simulations. Harnessing benefits machine-learned potentials enhanced statistical sampling methods, present an initio-level description Interfacial tensions predicted 1 500 bars found be close agreement with experiment pressures which experimental data available. Structural analyses indicate buildup adsorbed, saturated film forming low pressure (20 bars) similar bulk liquid, preferential perpendicular alignment respect monolayer coincides reduced structuring water molecules This study highlights predictive macroscopic biphasic interfaces, mechanistic insight obtained into dioxide aggregation interface is high relevance geoscience, climate research, materials science.

Language: Английский

---

Transfer learning for accurate description of atomic transport in Al–Cu melts

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 161(17)

Published: Nov. 1, 2024

Machine learning interatomic potentials (MLIPs) provide an optimal balance between accuracy and computational efficiency allow studying problems that are hardly solvable by traditional methods. For metallic alloys, MLIPs typically developed based on density functional theory with generalized gradient approximation (GGA) for the exchange-correlation functional. However, recent studies have shown this standard protocol can be inaccurate calculating transport properties or phase diagrams of some alloys. Thus, optimization choice specific calculation parameters is needed. In study, we address issue Al-Cu in which Perdew-Burke-Ernzerhof (PBE)-based cannot accurately calculate viscosity melting temperatures at Cu-rich compositions. We built different functionals, including meta-GGA, using a transfer strategy, allows us to reduce amount training data order magnitude compared approach. show r2SCAN- PBEsol-based much better describing thermodynamic particular, r2SCAN-based deep machine potential quantitatively reproduce concentration dependence dynamic viscosity. Our findings contribute development quantum chemical accuracy, one most challenging modern materials science.

Language: Английский

---

Self-learning path integral hybrid Monte Carlo with mixed ab initio and machine learning potentials for modeling nuclear quantum effects in water

The Journal of Chemical Physics, Journal Year: 2024, Volume and Issue: 161(20)

Published: Nov. 27, 2024

The introduction of machine learned potentials (MLPs) has greatly expanded the space available for studying Nuclear Quantum Effects computationally with ab initio path integral (PI) accuracy, MLPs' promise an accuracy comparable to that at a fraction cost. One challenges in development MLPs is need large and diverse training set calculated by methods. This dataset should ideally cover entire phase space, while not searching this using methods, as would be counterproductive generally intractable respect computational time. In paper, we present self-learning PI hybrid Monte Carlo Method mixed ML potential (SL-PIHMC-MIX), where allows study larger systems extension original SL-HMC method [Nagai et al., Phys. Rev. B 102, 041124 (2020)] methods systems. While generated can directly applied run long-time ML-PIMD simulations, demonstrate PIHMC-MIX trained exact reproduction structure obtained from PIMD. Specifically, find simulations require only 5000 evaluations 32-bead structure, compared 100 000 needed PIMD result.

Language: Английский

---