Based on ab initio calculations, a Finnis-Sinclair-type interatomic potential for W-W interactions has been developed. The modified potential is able to reproduce the correct formation energies of self-interstitial atom defects in tungsten, offering a significant improvement over the Ackland-Thetford tungsten potential. Using the modified potential, the thermal expansion is calculated in a temperature range from 0 K to 3500 K, and the results are in reasonable agreement with the experimental data, thus overcoming the shortcoming of the negative thermal expansion using the Derlet-Nguyen-Manh-Dudarev tungsten potential. The present W-W potential is also applied to study in detail the diffusion of self-interstitial atoms in tungsten. It is revealed that the initial self-interstitial atom initiates a sequence of W atom displacements and replacements in the <111> direction. An Arrhenius fit to the diffusion data at temperatures below 550 K indicates a migration energy of 0.022 eV, in reasonable agreement with the experimental data.
Recently, we developed a method to construct polynomial interatomic potentials from ab-initio calculations in order to accurately describe laser excited solids [PRL 124, 085501 (2020)]. However, ab-initio methods, and therefore analytical potentials derived from them, commonly do not provide an accurate prediction of the melting temperature. In order to reproduce the experimental melting properties, but keeping the accuracy in the laser excited case, we present here an approach to modify few key coefficients of polynomial interatomic potentials constructed from ab-initio data. We show that, wi…
We develop an Fe-C-H interatomic potential based on the modified embedded-atom method (MEAM) formalism based on density functional theory to enable large-scale modular dynamics simulations of carbon steel and hydrogen.
A new simulation approach of field evaporation is presented. The model combines classical electrostatics with molecular dynamics (MD) simulations. Unlike previous atomic-level simulation approaches, our method does not rely on an evaporation criterion based on thermal activation theory, instead, electric-field-induced forces on atoms are explicitly calculated and added to the interatomic forces. Atoms then simply move according to the laws of classical molecular dynamics and are "evaporated" when the external force overcomes interatomic bonding. This approach thus makes no ad-hoc assumptions c…
We develop an electronic-temperature dependent interatomic potential $Φ(T_\text{e})$ for unexcited and laser-excited silicon. The potential is designed to reproduce ab initio molecular dynamics simulations by requiring force- and energy matching for each time step. $Φ(T_\text{e})$ has a simple and flexible analytical form, can describe all relevant interactions and is applicable for any kind of boundary conditions (bulk, thin films, clusters). Its overall shape is automatically adjusted by a self-learning procedure, which finally finds the global minimum in the parameter space. We show that $Φ…
The development of modern ab initio methods has rapidly increased our understanding of physics, chemistry and materials science. Unfortunately, intensive ab initio calculations are intractable for large and complex systems. On the other hand, empirical force fields are less accurate with poor transferability even though they are efficient to handle large and complex systems. The recent development of machine-learning based neural-network (NN) for local atomic environment representation of density functional theory (DFT) has offered a promising solution to this long-standing challenge. Si is on…
In this paper we present new parameters of the TB-SMA interatomic potentials for the Pt/Cu(111) surface alloy. The parameters are fitted using both the experimental and {\it ab initio} data. The potentials reproduce not only the bulk properties of copper and platinum, but also the energy characteristics of the Pt/Cu(111) surface alloy. The potentials can be used for the simulations of the growth of the Pt/Cu(111) surface alloy on the atomic scale.
Machine learning interatomic potentials (MLIPs) are routinely used atomic simulations, but generating databases of atomic configurations used in fitting these models is a laborious process, requiring significant computational and human effort. A computationally efficient method is presented to generate databases of atomic configurations that contain optimal information on the small-displacement regime of the potential energy surface of bulk crystalline matter. Utilising non-diagonal supercell (NDSC), an automatic process is suggested for ab initio data generation. MLIPs were fitted for Al, W, …
Molecular dynamics simulations provide a versatile framework to study interfacial heat transport, but their accuracy remains limited by the accuracy of available interatomic potentials. In the past, researchers have adopted the use of analytic potentials and simple mixing rules to model interfacial systems, with minimal justification for their use. On the other hand, contemporary machine learned interatomic potentials have greater complexity, but have not seen rigorous validation of interfacial heat transport properties. Moreover, when fitting to ab initio data, it is not known whether interfa…
A well-known drawback of state-of-the-art machine-learning interatomic potentials is their poor ability to extrapolate beyond the training domain. For small-scale problems with tens to hundreds of atoms this can be solved by using active learning which is able to select atomic configurations on which a potential attempts extrapolation and add them to the ab initio-computed training set. In this sense an active learning algorithm can be viewed as an on-the-fly interpolation of an ab initio model. For large-scale problems, possibly involving tens of thousands of atoms, this is not feasible becau…
Universal Interatomic Potentials as Configuration-Space Generators for One-Shot and Iterative Fine-Tuning of Ab Initio-Accurate Material-Specific Models
While several studies confirmed that machine-learned potentials (MLPs) can provide accurate free energies for determining phase stabilities, the abilities of MLPs for efficiently constructing a full phase diagram of multi-component systems are yet to be established. In this work, by employing neural network interatomic potentials (NNPs), we demonstrate construction of the MgO-CaO eutectic phase diagram with temperatures up to 3400 K, which includes liquid phases. The NNP is trained over trajectories of various solid and liquid phases at several compositions that are calculated within the densi…
Ligand-Controlled Phonon Dynamics in CsPbBr3 Nanocrystals Revealed by Machine-Learned Interatomic Potentials
Georgia Institute of Technology（佐治亚理工学院）；Queens College, City University of New York（纽约市立大学皇后学院）；The Graduate Center, City University of New York（纽约市立大学研究生中心）
Examination of thermal expansion of two-dimensional (2D) nanomaterials is a challenging theoretical task with either ab-initio or classical molecular dynamics simulations. In this regard, while ab-initio molecular dynamics (AIMD) simulations offer extremely accurate predictions, but they are excessively demanding from computational point of view. On the other side, classical molecular dynamics simulations can be conducted with affordable computational costs, but without predictive accuracy needed to study novel materials and compositions. Herein, we explore the thermal expansion of several car…
Machine learning interatomic potentials (MLIPs) based on a large dataset obtained by density functional theory (DFT) calculation have been developed recently. This study gives both conceptual and practical bases for the high accuracy of MLIPs, although MLIPs have been considered to be simply an accurate black-box description of atomic energy. We also construct the most accurate MLIP of the elemental Ti ever reported using a linearized MLIP framework and many angular-dependent descriptors, which also corresponds to a generalization of the modified embedded atom method (MEAM) potential.
Finite temperature calculations, based on ab initio molecular dynamics (AIMD) simulations, are a powerful tool able to predict material properties that cannot be deduced from ground state calculations. However, the high computational cost of AIMD limits its applicability for large or complex systems. To circumvent this limitation we introduce a new method named Machine Learning Assisted Canonical Sampling (MLACS), which accelerates the sampling of the Born--Oppenheimer potential surface in the canonical ensemble. Based on a self-consistent variational procedure, the method iteratively trains a…
Application-specific machine-learned interatomic potentials: exploring the trade-off between DFT convergence, MLIP expressivity, and computational cost
The neural-network interatomic potential for crystalline and liquid Si has been developed using the forward stepwise regression technique to reduce the number of bases with keeping the accuracy of the potential. This approach of making the neural-network potential enables us to construct the accurate interatomic potentials with less and important bases selected systematically and less heuristically. The evaluation of bulk crystalline properties, and dynamic properties of liquid Si show good agreements between the neural-network potential and ab-initio results.
Ab-initio molecular dynamics simulation of hydrogen diffusion in $α$-iron
采用第一性原理原子分子动力学模拟，在微正则系综和正则系综下研究了间隙氢在$α$ -铁中的扩散。$θ=1/16 and 1/2 have been considered by locating interstitial hydrogen atoms at random positions in a $ 2 \times 2 \times 2$超晶胞之间的氢与铁比率。我们发现，平均最佳吸收位置和扩散势垒依赖于interestial的浓度。铁的德拜温度随间隙氢浓度的增加而单调下降，证明了大量扩散氢原子的存在显著削弱了铁-铁原子间相互作用势。
Accurate evaluation of the thermal conductivity of a material can be a challenging task from both experimental and theoretical points of view. In particular for the nanostructured materials, the experimental measurement of thermal conductivity is associated with diverse sources of uncertainty. As a viable alternative to experiment, the combination of density functional theory (DFT) simulations and the solution of Boltzmann transport equation is currently considered as the most trusted approach to examine thermal conductivity. The main bottleneck of the aforementioned method is to acquire the a…
Thermal and mechanical properties of two-dimensional nanomaterials are commonly studied by calculating force constants using the density functional theory (DFT) and classical molecular dynamics (MD) simulations. Although DFT simulations offer accurate estimations, the computational cost is high. On the other hand, MD simulations strongly depend on the accuracy of interatomic potentials. Here, we investigate thermal conductivity and elastic modulus of nitrogenated holey graphene (C2N) using passively fitted machine-learning interatomic potentials (MLIPs), which depend on computationally inexpen…
Comparison of intermediate-range order in GeO$_2$ glass: molecular dynamics using machine-learning interatomic potential vs.\ reverse Monte Carlo fitting to experimental data
Simulation of crack propagation in alumina with ab-initio based polarizable force field
我们提出了一个有效的原子相互作用势的结晶α-Al 2 O 3产生的程序potfit。静电相互作用的沃尔夫直接，成对求和方法与球形截断。利用Tangney-Scandolo原子间力场方法计算了氧原子的极化率。势被优化以再现在松弛和应变配置以及Al 2 O3的{0001}、{10-10}和{11-20}表面中的力、能量和应力。力场生成的细节，并证明其有效性。我们应用开发的潜力，调查裂纹扩展α-Al 2 O 3单晶。
Module for ab initio structure evolution (MAISE) is an open-source package for materials modeling and prediction. The code's main feature is an automated generation of neural network (NN) interatomic potentials for use in global structure searches. The systematic construction of Behler-Parrinello-type NN models approximating ab initio energy and forces relies on two approaches introduced in our recent studies. An evolutionary sampling scheme for generating reference structures improves the NNs' mapping of regions visited in unconstrained searches, while a stratified training approach enables t…
Accelerating the Training and Improving the Reliability of Machine-Learned Interatomic Potentials for Strongly Anharmonic Materials through Active Learning
We report that single interatomic potential, developed using Gaussian regression of density functional theory calculation data, has high accuracy and flexibility to describe phonon transport with ab initio accuracy in two different atomistic configurations: perfect crystalline Si and crystalline Si with vacancies. The high accuracy of second- and third-order force constants from the Gaussian approximation potential (GAP) are demonstrated with phonon dispersion, Grüneisen parameter, three-phonon scattering rate, phonon-vacancy scattering rate, and thermal conductivity, all of which are very clo…
