High-Dimensional Atomistic Neural Network Potentials for Molecule–Surface Interactions: HCl Scattering from Au(111)

Kolb, Brian; Luo, Xuan; Zhou, Xueyao; Jiang, Bin; Guo, Hua

doi:10.1021/acs.jpclett.6b02994

Cited by 101 publications

(95 citation statements)

References 68 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The applicability of artificial neural networks to describing chemical reactions has already been proven for, for example, bulk materials, 22,23 organic reactions, 24,25 simulations of solid–liquid interfaces, 26 molecules interacting with static surfaces, 27−29 and, very recently, energy transfer to metal surfaces in scattering of HCl from Au(111). 30 However, in the latter molecule–surface scattering study, reaction probabilities and scattering probabilities were not yet calculated, although final rotational state distributions in scattering were presented. In this work, for the first time, we apply the HD-NNP approach to reactive molecule–surface scattering with inclusion of surface atom motion.…”

mentioning

confidence: 99%

Accurate Neural Network Description of Surface Phonons in Reactive Gas–Surface Dynamics: N₂ + Ru(0001)

Shakouri

Behler

Meyer

et al. 2017

J. Phys. Chem. Lett.

137

167

View full text Add to dashboard Cite

Ab initio molecular dynamics (AIMD) simulations enable the accurate description of reactive molecule–surface scattering especially if energy transfer involving surface phonons is important. However, presently, the computational expense of AIMD rules out its application to systems where reaction probabilities are smaller than about 1%. Here we show that this problem can be overcome by a high-dimensional neural network fit of the molecule–surface interaction potential, which also incorporates the dependence on phonons by taking into account all degrees of freedom of the surface explicitly. As shown for N2 + Ru(0001), which is a prototypical case for highly activated dissociative chemisorption, the method allows an accurate description of the coupling of molecular and surface atom motion and accurately accounts for vibrational properties of the employed slab model of Ru(0001). The neural network potential allows reaction probabilities as low as 10–5 to be computed, showing good agreement with experimental results.

show abstract

mentioning

confidence: 99%

Accurate Neural Network Description of Surface Phonons in Reactive Gas–Surface Dynamics: N₂ + Ru(0001)

Shakouri

Behler

Meyer

et al. 2017

J. Phys. Chem. Lett.

137

167

View full text Add to dashboard Cite

show abstract

“…1-6 Effective local environment descriptors must be invariant under translation, rotation, and permutation of homonuclear atoms, and have the properties of uniqueness and differentiability. 7 Examples of such descriptors include symmetry functions 1,8 , smooth overlap of atomic positions (SOAP) 4,9 , bispectrum 2,5 , Coulomb matrix 3,10,11 , among others.A typical approach is to fit the PES as a function of these descriptors by machine learning on ab initio data sets, using techniques ranging from simple linear regression 5,12 to kernel ridge regression 6,7 to neural networks [13][14][15][16] .Thus far, the development of ML potentials based on local environment descriptors have largely been limited to elements and oxides. The Gaussian approximation potential (GAP) using the SOAP descriptor has been applied on Si 4 , C 17,18 , W 9 , P 19 , and Fe 20 , and neural network models based on symmetry functions have been fitted for Si 21 , C 22 , Na 23 , ZnO 24 , TiO 2 25 , GeTe 26 , and Li 3 PO 4 27 .…”

mentioning

confidence: 99%

Quantum-accurate spectral neighbor analysis potential models for Ni-Mo binary alloys and fcc metals

Chen

et al. 2018

Phys. Rev. B

View full text Add to dashboard Cite

In recent years, efficient inter-atomic potentials approaching the accuracy of density functional theory (DFT) calculations have been developed using rigorous atomic descriptors satisfying strict invariances, for example, to translation, rotation, permutation of homonuclear atoms, among others. In this work, we generalize the spectral neighbor analysis potential (SNAP) model to bcc-fcc binary alloy systems. We demonstrate that machine-learned SNAP models can yield significant improvements even over well-established, high-performing embedded atom method (EAM) and modified EAM (MEAM) potentials for fcc Cu and Ni. We also report on the development of a SNAP model for the fcc Ni-bcc Mo binary system by machine learning a carefully-constructed large computed data set of elemental and intermetallic compounds. We demonstrate that this binary Ni-Mo SNAP model can achieve excellent agreement with experiments in the prediction of Ni-Mo phase diagram as well as near-DFT accuracy in the prediction of many key properties such as elastic constants, formation energies, melting points, etc., across the entire binary composition range. In contrast, the existing Ni-Mo EAM has significant errors in the prediction of the phase diagram and completely fails in binary compounds. This work provides a systematic model development process for multicomponent alloy systems, including an efficient procedure to optimize the hyper-parameters in the model fitting, and paves the way to long-time, large-scale simulations of such systems. arXiv:1806.04777v2 [cond-mat.mtrl-sci] 16 Aug 2018 Machine learning (ML) models based on robust local environment descriptors have recently emerged as an approach to describe the potential energy surface (PES) of systems of atoms with near-quantum accuracy at several of orders magnitude lower cost than ab initio methods. 1-6 Effective local environment descriptors must be invariant under translation, rotation, and permutation of homonuclear atoms, and have the properties of uniqueness and differentiability. 7 Examples of such descriptors include symmetry functions 1,8 , smooth overlap of atomic positions (SOAP) 4,9 , bispectrum 2,5 , Coulomb matrix 3,10,11 , among others.A typical approach is to fit the PES as a function of these descriptors by machine learning on ab initio data sets, using techniques ranging from simple linear regression 5,12 to kernel ridge regression 6,7 to neural networks [13][14][15][16] .Thus far, the development of ML potentials based on local environment descriptors have largely been limited to elements and oxides. The Gaussian approximation potential (GAP) using the SOAP descriptor has been applied on Si 4 , C 17,18 , W 9 , P 19 , and Fe 20 , and neural network models based on symmetry functions have been fitted for Si 21 , C 22 , Na 23 , ZnO 24 , TiO 2 25 , GeTe 26 , and Li 3 PO 4 27 . Thompson and Wood 5,28 have developed linear and quadratic models based on the SO(4) bispectrum -the Spectral Analysis Neighbor Potential or SNAP -for bcc Ta and W. Chen et al. 12 later showed that a line...

show abstract

“…Important steps towards an accurate, yet efficient firstprinciples-based modeling of the energy uptake into phononic d.o.f. have recently been taken [3][4][5][6][7][8][9][10]. In contrast, the explicit description of e-h pair excitations and corresponding nonadiabatic couplings directly from first principles still poses a formidable challenge.…”

mentioning

confidence: 99%

Nonadiabatic Vibrational Damping of Molecular Adsorbates: Insights into Electronic Friction and the Role of Electronic Coherence

2017

View full text Add to dashboard Cite

We present a perturbation approach rooted in time-dependent density-functional theory to calculate electron-hole (e-h) pair excitation spectra during the nonadiabatic vibrational damping of adsorbates on metal surfaces. Our analysis for the benchmark systems CO on Cu(100) and Pt(111) elucidates the surprisingly strong influence of rather short electronic coherence times. We demonstrate how in the limit of short electronic coherence times, as implicitly assumed in prevalent quantum nuclear theories for the vibrational lifetimes as well as electronic friction, band structure effects are washed out. Our results suggest that more accurate lifetime or chemicurrentlike experimental measurements could characterize the electronic coherence. DOI: 10.1103/PhysRevLett.119.176808 The tortuous ways in which kinetic and chemical energy is transferred between adsorbates and substrate atoms fundamentally govern the dynamics of surface chemical reactions, for instance, in the context of heterogeneous catalysis or advanced deposition techniques. For metal substrates, the two main energy dissipation mechanisms in this regard are the adsorbate interaction with lattice vibrations, i.e., substrate phonons, and the excitation of electronhole (e-h) pairs. The latter are attributable to the nonadiabatic coupling of nuclear motion to the substrate electronic degrees of freedom (d.o.f.) and seem to be substantial in order to rationalize an increasing number of experimental findings [1,2]. Important steps towards an accurate, yet efficient firstprinciples-based modeling of the energy uptake into phononic d.o.f. have recently been taken [3][4][5][6][7][8][9][10]. In contrast, the explicit description of e-h pair excitations and corresponding nonadiabatic couplings directly from first principles still poses a formidable challenge.In this regard, electronic friction theory (EFT) [11,12] has become a popular workhorse to effectively capture the effects of such nonadiabatic energy loss on the adsorbate dynamics in a computationally convenient way [13][14][15][16][17][18][19]. Inspired by vibrational lifetimes obtained via response theory [20] or Fermi's golden rule in the nuclear system [21], a Langevin equation for the nuclei emerges from a semiclassical picture implying complete electronic decoherence in terms of the Markov approximation [12]. This approach thus avoids an explicit propagation of the electron dynamics and concomitant ultrafast time scales by coarse-graining the effects into electronic friction forces linear in nuclear velocities. This enables an efficient combination even with density-functional theory (DFT) based ab initio molecular dynamics (AIMD) simulations on high-dimensional potential energy surfaces as required for surface dynamical studies [15,22,23].Independent of the particular recipe employed to obtain the electronic friction coefficients [12,14,17,20,24,25], however, the downside of the coarse-graining of the electron dynamics is that it precludes a more fundamental understanding of the underlying e-h pair excit...

show abstract

High-Dimensional Atomistic Neural Network Potentials for Molecule–Surface Interactions: HCl Scattering from Au(111)

Cited by 101 publications

References 68 publications

Accurate Neural Network Description of Surface Phonons in Reactive Gas–Surface Dynamics: N₂ + Ru(0001)

Accurate Neural Network Description of Surface Phonons in Reactive Gas–Surface Dynamics: N₂ + Ru(0001)

Quantum-accurate spectral neighbor analysis potential models for Ni-Mo binary alloys and fcc metals

Nonadiabatic Vibrational Damping of Molecular Adsorbates: Insights into Electronic Friction and the Role of Electronic Coherence

Contact Info

Product

Resources

About

High-Dimensional Atomistic Neural Network Potentials for Molecule–Surface Interactions: HCl Scattering from Au(111)

Cited by 101 publications

References 68 publications

Accurate Neural Network Description of Surface Phonons in Reactive Gas–Surface Dynamics: N2 + Ru(0001)

Accurate Neural Network Description of Surface Phonons in Reactive Gas–Surface Dynamics: N2 + Ru(0001)

Quantum-accurate spectral neighbor analysis potential models for Ni-Mo binary alloys and fcc metals

Nonadiabatic Vibrational Damping of Molecular Adsorbates: Insights into Electronic Friction and the Role of Electronic Coherence

Contact Info

Product

Resources

About

Accurate Neural Network Description of Surface Phonons in Reactive Gas–Surface Dynamics: N₂ + Ru(0001)

Accurate Neural Network Description of Surface Phonons in Reactive Gas–Surface Dynamics: N₂ + Ru(0001)