arXiv subjects

Alexander V. Shapeev

Publications and source records attributed to Alexander V. Shapeev.

At least 37 records · Page 2Linked to original sources

arXiv2020

Exceptional piezoelectricity, high thermal conductivity and stiffness and promising photocatalysis in two-dimensional MoSi2N4 family confirmed by first-principles

Chemical vapor deposition has been most recently employed to fabricate centimeter-scale high-quality single-layer MoSi2N4 (Science; 2020;369; 670). Motivated by this exciting experimental advance, herein we conduct extensive first-principles based simulations to explore the stability, mechanical properties, lattice thermal conductivity, piezoelectric and flexoelectric response, and photocatalytic and electronic features of MA2Z4 (M = Cr, Mo, W; A = Si, Ge; Z = N, P) monolayers. The considered nanosheets are found to exhibit dynamical stability and remarkably high mechanical properties. Moreover, they show diverse electronic properties from antiferromagnetic metal to half metal and to semiconductors with band gaps ranging from 0.31 to 2.57 eV. Among the studied nanosheets, the MoSi2N4 and WSi2N4 monolayers yield appropriate band edge positions, high electron and hole mobilities, and strong visible light absorption, highly promising for applications in optoelectronics and photocatalytic water splitting. The MoSi2N4 and WSi2N4 monolayers are also predicted to show outstandingly high lattice thermal conductivity of 440 and 500 W/mK, respectively. For the first time we show that machine learning interatomic potentials trained over small supercells can be employed to examine the flexoelectric and piezoelectric properties of complex structures. As the most exciting finding, WSi2N4, CrSi2N4 and MoSi2N4 are found to exhibit the highest piezoelectric coefficients, outperforming all other-known 2D materials. Our results highlight that MA2Z4 nanosheets not only undoubtedly outperform the transition metal dichalcogenides family but also can compete with graphene for applications in nanoelectronics, optoelectronic, energy storage/conversion and thermal management systems.

cond-mat.mtrl-sci↗

arXiv2020

Accelerating first-principles estimation of thermal conductivity by machine-learning interatomic potentials: A MTP/ShengBTE solution

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 anharmonic interatomic force constants using the computationally demanding DFT calculations. In this work we propose a substantially accelerated approach for the evaluation of anharmonic interatomic force constants via employing machine-learning interatomic potentials (MLIPs) trained over short ab-initio molecular dynamics trajectories. The remarkable accuracy of the proposed accelerated method is confirmed by comparing the estimated thermal conductivities of several bulk and two-dimensional materials with those computed by the full-DFT approach. The MLIP-based method proposed in this study can be employed as a standard tool, which would substantially accelerate and facilitate the estimation of lattice thermal conductivity in comparison with the commonly used full-DFT solution.

cond-mat.mtrl-sci↗

arXiv2020

The MLIP package: Moment Tensor Potentials with MPI and Active Learning

The subject of this paper is the technology (the "how") of constructing machine-learning interatomic potentials, rather than science (the "what" and "why") of atomistic simulations using machine-learning potentials. Namely, we illustrate how to construct moment tensor potentials using active learning as implemented in the MLIP package, focusing on the efficient ways to sample configurations for the training set, how expanding the training set changes the error of predictions, how to set up ab initio calculations in a cost-effective manner, etc. The MLIP package (short for Machine-Learning Interatomic Potentials) is available at https://mlip.skoltech.ru/download/.

physics.comp-ph↗

arXiv2020

Machine-Learning Interatomic Potentials Enable First-Principles Multiscale Modeling of Lattice Thermal Conductivity in Graphene/Borophene Heterostructures

One of the ultimate goals of computational modeling in condensed matter is to be able to accurately compute materials properties with minimal empirical information. First-principles approaches such as the density functional theory (DFT) provide the best possible accuracy on electronic properties but they are limited to systems up to a few hundreds, or at most thousands of atoms. On the other hand, classical molecular dynamics (CMD) simulations and finite element method (FEM) are extensively employed to study larger and more realistic systems, but conversely depend on empirical information. Here, we show that machine-learning interatomic potentials (MLIPs) trained over short ab-initio molecular dynamics trajectories enable first-principles multiscale modeling, in which DFT simulations can be hierarchically bridged to efficiently simulate macroscopic structures. As a case study, we analyze the lattice thermal conductivity of coplanar graphene/borophene heterostructures, recently synthesized experimentally (Sci. Adv. 2019; 5: eaax6444), for which no viable classical modeling alternative is presently available. Our MLIP-based approach can efficiently predict the lattice thermal conductivity of graphene and borophene pristine phases, the thermal conductance of complex graphene/borophene interfaces and subsequently enable the study of effective thermal transport along the heterostructures at continuum level. This work highlights that MLIPs can be effectively and conveniently employed to enable first-principles multiscale modeling via hierarchical employment of DFT/CMD/FEM simulations, thus expanding the capability for computational design of novel nanostructures.

cond-mat.mtrl-sci↗

arXiv2020

High thermal conductivity in semiconducting Janus and non-Janus diamanes

Most recently, F-diamane monolayer was experimentally realized by the fluorination of bilayer graphene. In this work we elaborately explore the electronic and thermal conductivity responses of diamane lattices with homo or hetero functional groups, including: non-Janus C2H, C2F and C2Cl diamane and Janus counterparts of C4HF, C4HCl and C4FCl. Noticeably, C2H, C2F, C2Cl, C4HF, C4HCl and C4FCl diamanes are found to show electronic diverse band gaps of, 3.86, 5.68, 2.42, 4.17, 0.86, and 2.05 eV, on the basis of HSE06 method estimations. The thermal conductivity of diamane nanosheets was acquired using the full iterative solutions of the Boltzmann transport equation, with substantially accelerated calculations by employing machine-learning interatomic potentials in obtaining the anharmonic force constants. According to our results, the room temperature lattice thermal conductivity of graphene and C2H, C2F, C2Cl, C4HF, C4HCl and C4FCl diamane monolayers are estimated to be 3636, 1145, 377, 146, 454, 244 and 196 W/mK, respectively. The underlying mechanisms resulting in significant effects of functional groups on the thermal conductivity of diamane nanosheets were thoroughly explored. Our results highlight the substantial role of functional groups on the electronic and thermal conduction responses of diamane nanosheets.

cond-mat.mtrl-sci↗

arXiv2020

Nanoporous C3N4, C3N5 and C3N6 nanosheets; Novel strong semiconductors with low thermal conductivities and appealing optical/electronic properties

Carbon nitride two-dimensional (2D) materials are among the most attractive class of nanomaterials, with wide range of application prospects. As a continuous progress, most recently, two novel carbon nitride 2D lattices of C3N5 and C3N4 have been successfully experimentally realized. Motivated by these latest accomplishments and also by taking into account the well-known C3N4 triazine-based graphitic carbon nitride structures, we predicted two novel C3N6 and C3N4 counterparts. We then conducted extensive density functional theory simulations to explore the thermal stability, mechanical, electronic and optical properties of these novel nanoporous carbon-nitride nanosheets. According to our results all studied nanosheets were found to exhibit desirable thermal stability and mechanical properties. Non-equilibrium molecular dynamics simulations on the basis of machine learning interatomic potentials predict ultralow thermal conductivities for these novel nanosheets. Electronic structure analyses confirm direct band gap semiconducting electronic character and optical calculations reveal the ability of these novel 2D systems to adsorb visible range of light. Extensive first-principles based results by this study provide a comprehensive vision on the stability, mechanical, electronic and optical responses of C3N4, C3N5 and C3N6 as novel 2D semiconductors and suggest them as promising candidates for the design of advanced nanoelectronics and energy storage/conversion systems.

physics.comp-ph↗

arXiv2020

Exploring Phononic Properties of Two-Dimensional Materials using Machine Learning Interatomic Potentials

Phononic properties are commonly studied by calculating force constants using the density functional theory (DFT) simulations. Although DFT simulations offer accurate estimations of phonon dispersion relations or thermal properties, but for low-symmetry and nanoporous structures the computational cost quickly becomes very demanding. Moreover, the computational setups may yield nonphysical imaginary frequencies in the phonon dispersion curves, impeding the assessment of phononic properties and the dynamical stability of the considered system. Here, we compute phonon dispersion relations and examine the dynamical stability of a large ensemble of novel materials and compositions. We propose a fast and convenient alternative to DFT simulations which derived from machine-learning interatomic potentials passively trained over computationally efficient ab-initio molecular dynamics trajectories. Our results for diverse two-dimensional (2D) nanomaterials confirm that the proposed computational strategy can reproduce fundamental thermal properties in close agreement with those obtained via the DFT approach. The presented method offers a stable, efficient, and convenient solution for the examination of dynamical stability and exploring the phononic properties of low-symmetry and porous 2D materials.

cond-mat.mtrl-sci↗

arXiv2020

Elinvar effect in $β-$Ti simulated by on-the-fly trained moment tensor potential

A combination of quantum mechanics calculations with machine learning (ML) techniques can lead to a paradigm shift in our ability to predict materials properties from first principles. Here we show that on-the-fly training of an interatomic potential described through moment tensors provides the same accuracy as state-of-the-art {\it ab inito} molecular dynamics in predicting high-temperature elastic properties of materials with two orders of magnitude less computational effort. Using the technique, we investigate high-temperature bcc phase of titanium and predict very weak, Elinvar, temperature dependence of its elastic moduli, similar to the behavior of the so-called GUM Ti-based alloys [T. Sato {\ it et al.}, Science {\bf 300}, 464 (2003)]. Given the fact that GUM alloys have complex chemical compositions and operate at room temperature, Elinvar properties of elemental bcc-Ti observed in the wide temperature interval 1100--1700 K is unique.

cond-mat.mtrl-sci↗

arXiv2020

Ab initio analysis of some Ge-based 2D nanomaterials

The structural, electronic and dynamical properties of a group of 2D germanium-based compounds, including GeC, GeN, GeO, GeSi, GeS, GeSe, and germanene, are investigated by employing first-principles calculations. The most stable structure of each of these systems is identified after considering the most probable configurations and performing accurate phonon calculations. We introduce a new phase of germanene, which we name the tile germanene, which is significantly more stable than the known hexagonal germanene. We apply the modern modified Becke-Johnson (mBJ) and DFT1/2 schemes to obtain an accurate band structure for our selected 2D materials. It is seen that GeO and GeC exhibit the highest band gaps of more than 3 eV in this group of materials. Moreover, we argue that, in contrast to the semi-metallic nature of hexagonal germanene, the tile germanene is a very good conductor. The band edges of our semiconducting 2D materials are accurately aligned to the vacuum level to address the potential photocatalytic application of this system for water splitting and carbon dioxide reduction. The optical properties, including dielectric functions, refractive index, reflectivity, and Loss function of the samples are investigated in the framework of the Bethe-Salpeter approach.

cond-mat.mtrl-sci↗

arXiv2019

Ring Polymer Molecular Dynamics and Active Learning of Moment Tensor Potential for Gas-Phase Barrierless Reactions: Application to S + H2

Ring polymer molecular dynamics (RPMD) has proven to be an accurate approach for calculating thermal rate coefficients of various chemical reactions. For wider application of this methodology, efficient ways to generate the underlying full-dimensional potential energy surfaces (PESs) and the corresponding energy gradients are required. Recently, we have proposed a fully automated procedure based on combining the original RPMDrate code with active learning for PES on-the-fly using moment tensor potential and successfully applied it to two representative thermally activated chemical reactions [I. S. Novikov, Y. V. Suleimanov, A. V. Shapeev, Phys. Chem. Chem. Phys., 29503-29512 (2018)]. In this work, using a prototype insertion chemical reaction S + H$_2$, we show that this procedure works equally well for another class of chemical reactions. We find that the corresponding PES can be generated by fitting to less than 1500 automatically generated structures while the RPMD rate coefficients show deviation from the reference values within the typical convergence error of RPMDrate. We note that more structures are accumulated during the real-time propagation of the dynamic factor (the recrossing factor) as opposed to the previous study. We also observe that relatively flat free energy profile of the along the reaction coordinate before entering the complex-formation well can cause issues with locating the maximum of the free energy surface for less converged PESs. However, the final RPMD rate coefficient is independent of the position of the dividing surface that makes it invulnerable to this problem, keeping the total number of necessary structures within a few thousand. Our work concludes that, in future, the proposed methodology can be applied to realistic complex chemical reactions with various energy profiles.

physics.chem-ph↗

arXiv2019

A Performance and Cost Assessment of Machine Learning Interatomic Potentials

Machine learning of the quantitative relationship between local environment descriptors and the potential energy surface of a system of atoms has emerged as a new frontier in the development of interatomic potentials (IAPs). Here, we present a comprehensive evaluation of ML-IAPs based on four local environment descriptors --- Behler-Parrinello symmetry functions, smooth overlap of atomic positions (SOAP), the Spectral Neighbor Analysis Potential (SNAP) bispectrum components, and moment tensors --- using a diverse data set generated using high-throughput density functional theory (DFT) calculations. The data set comprising bcc (Li, Mo) and fcc (Cu, Ni) metals and diamond group IV semiconductors (Si, Ge) is chosen to span a range of crystal structures and bonding. All descriptors studied show excellent performance in predicting energies and forces far surpassing that of classical IAPs, as well as predicting properties such as elastic constants and phonon dispersion curves. We observe a general trade-off between accuracy and the degrees of freedom of each model, and consequently computational cost. We will discuss these trade-offs in the context of model selection for molecular dynamics and other applications.

physics.comp-ph↗

arXiv2019

Machine-learned Interatomic Potentials for Alloys and Alloy Phase Diagrams

We introduce machine-learned potentials for Ag-Pd to describe the energy of alloy configurations over a wide range of compositions. We compare two different approaches. Moment tensor potentials (MTP) are polynomial-like functions of interatomic distances and angles. The Gaussian Approximation Potential (GAP) framework uses kernel regression, and we use the Smooth Overlap of Atomic Positions (SOAP) representation of atomic neighbourhoods that consists of a complete set of rotational and permutational invariants provided by the power spectrum of the spherical Fourier transform of the neighbour density. Both types of potentials give excellent accuracy for a wide range of compositions and rival the accuracy of cluster expansion, a benchmark for this system. While both models are able to describe small deformations away from the lattice positions, SOAP-GAP excels at transferability as shown by sensible transformation paths between configurations, and MTP allows, due to its lower computational cost, the calculation of compositional phase diagrams. Given the fact that both methods perform as well as cluster expansion would but yield off-lattice models, we expect them to open new avenues in computational materials modeling for alloys.

cond-mat.mtrl-sci↗

arXiv2019

Machine-learned multi-system surrogate models for materials prediction

Surrogate machine-learning models are transforming computational materials science by predicting properties of materials with the accuracy of ab initio methods at a fraction of the computational cost. We demonstrate surrogate models that simultaneously interpolate energies of different materials on a dataset of 10 binary alloys (AgCu, AlFe, AlMg, AlNi, AlTi, CoNi, CuFe, CuNi, FeV, NbNi) with 10 different species and all possible fcc, bcc and hcp structures up to 8 atoms in the unit cell, 15\,950 structures in total. We find that the deviation of prediction errors when increasing the number of simultaneously modeled alloys is less than 1\,meV/atom. Several state-of-the-art materials representations and learning algorithms were found to qualitatively agree on the prediction errors of formation enthalpy with relative errors of $<$2.5\% for all systems.

cond-mat.mtrl-sci↗

arXiv2019

Accelerating crystal structure prediction by machine-learning interatomic potentials with active learning

In this letter we propose a new methodology for crystal structure prediction, which is based on the evolutionary algorithm USPEX and the machine-learning interatomic potentials actively learning on-the-fly. Our methodology allows for an automated construction of an interatomic interaction model from scratch replacing the expensive DFT with a speedup of several orders of magnitude. Predicted low-energy structures are then tested on DFT, ensuring that our machine-learning model does not introduce any prediction error. We tested our methodology on a problem of prediction of carbon allotropes, dense sodium structures and boron allotropes including those which have more than 100 atoms in the primitive cell. All the the main allotropes have been reproduced and a new 54-atom structure of boron have been found at very modest computational efforts.

cond-mat.mtrl-sci↗

arXiv2018

Improving accuracy of interatomic potentials: more physics or more data? A case study of silica

In this paper we test two strategies to improving the accuracy of machine-learning potentials, namely adding more fitting parameters thus making use of large volumes of available quantum-mechanical data, and adding a charge-equilibration model to account for ionic nature of the SiO2 bonding. To that end, we compare Moment Tensor Potentials (MTPs) and MTPs combined with the charge-equilibration (QEq) model (MTP+QEq) fitted to a density functional theory dataset of alpha-quartz SiO2-based structures. In order to make a meaningful comparison, in addition to the accuracy, we assess the uncertainty of predictions of each potential. It is shown that adding the QEq model to MTP does not make any improvement over the MTP potential alone, while adding more parameters does improve the accuracy and uncertainty of its predictions.

physics.comp-ph↗

arXiv2018

Automated Calculation of Thermal Rate Coefficients using Ring Polymer Molecular Dynamics and Machine-Learning Interatomic Potentials with Active Learning

We propose a methodology for fully automated calculation of thermal rate coefficients of gas phase chemical reactions, which is based on combining the ring polymer molecular dynamics (RPMD) with the machine-learning interatomic potentials actively learning on-the-fly. Based on the original computational procedure implemented in the RPMDrate code, our methodology gradually and automatically constructs the potential energy surfaces (PESs) from scratch with the data set points being selected and accumulated during the RPMDrate simulation. Such an approach ensures that our final machine-learning model provides reliable description of the PES which avoids artifacts during exploration of the phase space by RPMD trajectories. We tested our methodology on two representative thermally activated chemical reactions studied recently by RPMDrate at temperatures within the interval of 300--1000~K. The corresponding PESs were generated by fitting to only a few thousands automatically generated structures (less than 5000) while the RPMD rate coefficients retained the deviation from the reference values within the typical convergence error of RPMDrate. In future, we plan to apply our methodology to chemical reactions which proceed via complex-formation thus providing a completely general tool for calculating RPMD thermal rate coefficients for any polyatomic gas phase chemical reaction.

physics.chem-ph↗

arXiv2018

Accelerating high-throughput searches for new alloys with active learning of interatomic potentials

We propose an approach to materials prediction that uses a machine-learning interatomic potential to approximate quantum-mechanical energies and an active learning algorithm for the automatic selection of an optimal training dataset. Our approach significantly reduces the amount of DFT calculations needed, resorting to DFT only to produce the training data, while structural optimization is performed using the interatomic potentials. Our approach is not limited to one (or a small number of) lattice types (as is the case for cluster expansion, for example) and can predict structures with lattice types not present in the training dataset. We demonstrate the effectiveness of our algorithm by predicting the convex hull for the following three systems: Cu-Pd, Co-Nb-V, and Al-Ni-Ti. Our method is three to four orders of magnitude faster than conventional high-throughput DFT calculations and explores a wider range of materials space. In all three systems, we found unreported stable structures compared to the AFLOW database. Because our method is much cheaper and explores much more of materials space than high-throughput methods or cluster expansion, and because our interatomic potentials have a systematically improvable accuracy compared to empirical potentials such as EAM, etc., it will have a significant impact in the discovery of new alloy phases, particularly those with three or more components.

cond-mat.mtrl-sci↗

arXiv2018

Machine learning of molecular properties: locality and active learning

In recent years the machine learning techniques have shown a great potential in various problems from a multitude of disciplines, including materials design and drug discovery. The high computational speed on the one hand and the accuracy comparable to that of DFT on another hand make machine learning algorithms efficient for high-throughput screening through chemical and configurational space. However, the machine learning algorithms available in the literature require large training datasets to reach the chemical accuracy and also show large errors for the so-called outliers - the out-of-sample molecules, not well-represented in the training set. In the present paper we propose a new machine learning algorithm for predicting molecular properties that addresses these two issues: it is based on a local model of interatomic interactions providing high accuracy when trained on relatively small training sets and an active learning algorithm of optimally choosing the training set that significantly reduces the errors for the outliers. We compare our model to the other state-of-the-art algorithms from the literature on the widely used benchmark tests.

physics.chem-ph↗