Thermal transport in phase-change materials from atomistic simulations

Sosso, Gabriele C.; Donadio, Davide; Caravati, S.; Behler, Jörg; Bernasconi, Marco

doi:10.1103/physrevb.86.104301

Cited by 97 publications

(99 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Structures were visualized using AtomEye. 9 deed, the only reported ML potential dealing with amorphous matter is a neural-network potential for the phasechange data-storage material GeTe 46 that enabled largescale simulations of thermal transport 47 and atomistic processes during crystallization. 48 Amorphous materials are structurally much more diverse than their crystalline counterparts, and despite the lack of long-range translational symmetry, their properties depend crucially on structural order on the local and intermediate length scales.…”

Section: -6mentioning

confidence: 99%

Machine learning based interatomic potential for amorphous carbon

2017

View full text Add to dashboard Cite

We introduce a Gaussian approximation potential (GAP) for atomistic simulations of liquid and amorphous elemental carbon. Based on a machine-learning representation of the density-functional theory (DFT) potential-energy surface, such interatomic potentials enable materials simulations with close-to DFT accuracy but at much lower computational cost. We first determine the maximum accuracy that any finite-range potential can achieve in carbon structures; then, using a novel hierarchical set of two-, three-, and many-body structural descriptors, we construct a GAP model that can indeed reach the target accuracy. The potential yields accurate energetic and structural properties over a wide range of densities; it also correctly captures the structure of the liquid phases, at variance with state-of-the-art empirical potentials. Exemplary applications of the GAP model to surfaces of "diamond-like" tetrahedral amorphous carbon (ta-C) are presented, including an estimate of the amorphous material's surface energy, and simulations of high-temperature surface reconstructions ("graphitization"). The new interatomic potential appears to be promising for realistic and accurate simulations of nanoscale amorphous carbon structures.

show abstract

Section: -6mentioning

confidence: 99%

Machine learning based interatomic potential for amorphous carbon

2017

View full text Add to dashboard Cite

show abstract

“…12,[18][19][20][21][22][23][24][25][26] Computational costs limit DFT calculations to less than 100 atoms, however, making it challenging to explicitly incorporate the effects of disorder. 12,20,22,25,[27][28][29] Disorder is typically included in the ALD framework using Abeles' virtual crystal (VC) approximation, whereby the disordered solid is replaced with a perfect VC with properties equivalent to an averaging over the disorder (e.g., atomic mass and/or bond strength). 14 The ALD calculations are performed on a small unit cell with the averaged properties (i.e., all vibrational modes are phonons) and phononphonon and phonon-disorder scattering are included as perturbations.…”

Section: Introductionmentioning

confidence: 99%

Predicting alloy vibrational mode properties using lattice dynamics calculations, molecular dynamics simulations, and the virtual crystal approximation

Larkin

McGaughey

2013

Journal of Applied Physics

View full text Add to dashboard Cite

The virtual crystal (VC) approximation for mass disorder is evaluated by examining two model alloy systems: Lennard-Jones argon and Stillinger-Weber silicon. In both material systems, the perfect crystal is alloyed with a heavier mass species up to equal concentration. The analysis is performed using molecular dynamics simulations and lattice dynamics calculations. Mode frequencies and lifetimes are first calculated by treating the disorder explicitly and under the VC approximation, with differences found in the high-concentration alloys at high frequencies. Notably, the lifetimes of high-frequency modes are underpredicted using the VC approximation, a result we attribute to the neglect of higher-order terms in the model used to include point-defect scattering. The mode properties are then used to predict thermal conductivity under the VC approximation. For the Lennard-Jones alloys, where high-frequency modes make a significant contribution to thermal conductivity, the high-frequency lifetime underprediction leads to an underprediction of thermal conductivity compared to predictions from the Green-Kubo method, where no assumptions about the thermal transport are required. Based on observations of a minimum mode diffusivity, we propose a correction that brings the VC approximation thermal conductivities into better agreement with the Green-Kubo values. For the Stillinger-Weber alloys, where the thermal conductivity is dominated by low-frequency modes, the high-frequency lifetime underprediction does not affect the thermal conductivity prediction and reasonable agreement is found with the Green-Kubo values.

show abstract

“…For each atom, a vector of symmetry function values is used as input for a separate atomic feed-forward NN, which then yields the corresponding atomic energy contribution to the total energy if trained properly to first principles reference data. Potentials of this type have been constructed very successfully for a number of materials like silicon [40,50,52], sodium [53,54], carbon [55,56], copper [57,58], the phase change material GeTe [59][60][61], large copper clusters supported at zinc oxide [62], and the methanol molecule [58]. The accuracy of the Behler-Parrinello method depends on the chosen cutoff radius of the symmetry functions, which defines the atomic environments.…”

Section: High-dimensional Neural Networkmentioning

confidence: 99%

A Full-Dimensional Neural Network Potential-Energy Surface for Water Clusters up to the Hexamer

Morawietz

Behler

2013

Zeitschrift Für Physikalische Chemie

Self Cite

View full text Add to dashboard Cite

Water clusters have attracted a lot of attention as prototype systems to study hydrogen bonded molecular aggregates but also to gain deeper insights into the properties of liquid water, the solvent of life. All these studies depend on an accurate description of the atomic interactions and countless potentials have been proposed in the literature in the past decades to represent the potential-energy surface (PES) of water. Many of these potentials employ drastic approximations like rigid water monomers and fixed point charges, while on the other hand also several attempts have been made to derive very accurate PESs by fitting data obtained in high-level electronic structure calculations. In recent years artificial neural networks (NNs) have been established as a powerful tool to construct high-dimensional PESs of a variety of systems, but to date no full-dimensional NN PES for water has been reported. Here, we present NN potentials for water clusters containing two to six water molecules trained to density functional theory (DFT) data employing two different exchange-correlation functionals, PBE and RPBE. In contrast to other potentials fitted to first principles data, these NN potentials are not based on a truncated many-body expansion of the energy but consider the interactions between all water molecules explicitly. For both functionals an excellent agreement with the underlying DFT calculations has been found with binding energy errors of only about 1%.

show abstract

Thermal transport in phase-change materials from atomistic simulations

Cited by 97 publications

References 40 publications

Machine learning based interatomic potential for amorphous carbon

Machine learning based interatomic potential for amorphous carbon

Predicting alloy vibrational mode properties using lattice dynamics calculations, molecular dynamics simulations, and the virtual crystal approximation

A Full-Dimensional Neural Network Potential-Energy Surface for Water Clusters up to the Hexamer

Contact Info

Product

Resources

About