Author Correction: Characterization of domain distributions by second harmonic generation in ferroelectrics

Correction to:npj Computational Materials https://doi.org/10.1038/s41524-018-0095-6,published online 24 July 2018 In this article the affiliation details for author Yinghao Chu were incorrectly given as‘Department of Materials Science and Engineering,National Chiao Tung University,30010 Hsinchu,Taiwan’but should have been‘Department of Materials Science and Engineering,National Chiao Tung University,30010 Hsinchu,Taiwan,China’.The original article has been corrected.

Deep material network via a quilting strategy: visualization for explainability and recursive training for improved accuracy

Recent developments integrating micromechanics and neural networks offer promising paths for rapid predictions of the response of heterogeneous materials with similar accuracy as direct numerical simulations.The deep material network is one such approaches,featuring a multi-layer network and micromechanics building blocks trained on anisotropic linear elastic properties.Once trained,the network acts as a reduced-order model,which can extrapolate the material’s behavior to more general constitutive laws,including nonlinear behaviors,without the need to be retrained.However,current training methods initialize network parameters randomly,incurring inevitable training and calibration errors.Here,we introduce a way to visualize the network parameters as an analogous unit cell and use this visualization to“quilt”patches of shallower networks to initialize deeper networks for a recursive training strategy.The result is an improvement in the accuracy and calibration performance of the network and an intuitive visual representation of the network for better explainability.

Polycrystalline morphology and mechanical strength of nanotube fibers

Correlating mechanical performance with mesoscale structure is fundamental for the design and optimization of light and strong fibers(or any composites),most promising being those from carbon nanotubes.In all forms of nanotube fiber production strategies,due to tubes’mutual affinity,some degree of bundling into liquid crystal-like domains can be expected,causing heterogeneous load transfer within and outside these domains,and having a direct impact on the fiber strength.By employing large-scale coarse-grained simulations,we demonstrate that the strength s of nanotube fibers with characteristic domain size D scales as s~1/D,while the degree of longitudinal/axial disorder within the domains(akin to a smectic↔nematic phase transition)can substantially mitigate this dependence.

Phase-field framework with constraints and its applications to ductile fracture in polycrystals and fatigue

Modeling of ductile fracture in polycrystalline structures is challenging,since it requires integrated modeling of cracks,crystal plasticity,and grains.Here we extend the typical phase-field framework to the situations with constraints on the order parameters,and formulate two types of phase-field models on ductile fracture.The Type-Ⅰ model incorporates three sets of order parameters,which describe the distributions of cracks,plastic strain,and grains,respectively.Crystal plasticity is employed within grain interiors accommodated by J_(2)plasticity at grain boundaries.The applications of the Type-Ⅰ model to single crystals and bicrystals demonstrate the influences of grain orientations and grain boundaries on crack growth.In the Type-Ⅱ model,J_(2)plasticity is assumed for the whole system and grain structures are neglected.Taking advantage of the efficiency of the fast Fourier transform,our Type-Ⅱ model is employed to study low cycle fatigue.Crack closure and striation-like patterning of plastic strain are observed in the simulations.Crack growth rate is analyzed as a function of the J-integral,and the simulated fatigue life as a function of plastic strain agrees with the Coffin–Manson relation without a priori assumption.

A public database of thermoelectric materials and system-identified material representation for data-driven discovery

Thermoelectric materials have received much attention as energy harvesting devices and power generators.However,discovering novel high-performance thermoelectric materials is challenging due to the structural diversity and complexity of the thermoelectric materials containing alloys and dopants.For the efficient data-driven discovery of novel thermoelectric materials,we constructed a public dataset that contains experimentally synthesized thermoelectric materials and their experimental thermoelectric properties.For the collected dataset,we were able to construct prediction models that achieved R^(2)-scores greater than 0.9 in the regression problems to predict the experimentally measured thermoelectric properties from the chemical compositions of the materials.Furthermore,we devised a material descriptor for the chemical compositions of the materials to improve the extrapolation capabilities of machine learning methods.Based on transfer learning with the proposed material descriptor,we significantly improved the R^(2)-score from 0.13 to 0.71 in predicting experimental ZTs of the materials from completely unexplored material groups.

Accelerated identification of equilibrium structures of multicomponent inorganic crystals using machine learning potentials

The discovery of multicomponent inorganic compounds can provide direct solutions to scientific and engineering challenges,yet the vast uncharted material space dwarfs synthesis throughput.While the crystal structure prediction(CSP)may mitigate this frustration,the exponential complexity of CSP and expensive density functional theory(DFT)calculations prohibit material exploration at scale.Herein,we introduce SPINNER,a structure-prediction framework based on random and evolutionary searches.Harnessing speed and accuracy of neural network potentials(NNPs),the program navigates configurational spaces 10^(2)–10^(3) times faster than DFT-based methods.Furthermore,SPINNER incorporates algorithms tuned for NNPs,achieving performances exceeding conventional algorithms.In blind tests on 60 ternary compositions,SPINNER identifies experimental(or theoretically more stable)phases for~80%of materials.When benchmarked against data-mining or DFT-based evolutionary predictions,SPINNER identifies more stable phases in many cases.By developing a reliable and fast structure-prediction framework,this work paves the way to large-scale,open exploration of undiscovered inorganic crystals.

Study of disorder in pulsed laser deposited double perovskite oxides by first-principle structure prediction

Double perovskite oxides,with generalized formula A_(2)BB'O_(6),attract wide interest due to their multiferroic and charge transfer properties.They offer a wide range of potential applications such as spintronics and electrically tunable devices.However,great practical limitations are encountered,since a spontaneous order of the B-site cations is notoriously hard to achieve.In this joint experimental-theoretical work,we focused on the characterization of double perovskites La2TiFeO6 and La_(2)VCuO_(6) films grown by pulsed laser deposition and interpretation of the observed B-site disorder and partial charge transfer between the B-site ions.A random structure sampling method was used to show that several phases compete due to their corresponding configurational entropy.In order to capture a representative picture of the most relevant competing microstates in realistic experimental conditions,this search included the potential formation of non-stoichiometric phases as well,which could also be directly related to the observed partial charge transfer.We optimized the information encapsulated in the potential energy landscape,captured via structure sampling,by evaluating both enthalpic and entropic terms.These terms were employed as a metric for the competition of different phases.This approach,applied herein specifically to La_(2)TiFeO_(6),highlights the presence of highly entropic phases above the ground state which can explain the disorder observed frequently in the broader class of double perovskite oxides.

Searching for a route to synthesize in situ epitaxial Pr_(2)Ir_(2)O_(7) thin films with thermodynamic methods

In situ growth of pyrochlore iridate thin films has been a long-standing challenge due to the low reactivity of Ir at low temperatures and the vaporization of volatile gas species such as IrO_(3)(g)and IrO_(2)(g)at high temperatures and high PO_(2).To address this challenge,we combine thermodynamic analysis of the Pr-Ir-O_(2)system with experimental results from the conventional physical vapor deposition(PVD)technique of co-sputtering.Our results indicate that only high growth temperatures yield films with crystallinity sufficient for utilizing and tailoring the desired topological electronic properties and the in situ synthesis of Pr_(2)Ir_(2)O_(7)thin films is fettered by the inability to grow with PO_(2)on the order of 10 Torr at high temperatures,a limitation inherent to the PVD process.Thus,we suggest techniques capable of supplying high partial pressure of key species during deposition,in particular chemical vapor deposition(CVD),as a route to synthesis of Pr_(2)Ir_(2)O_(7).

Efficient Gaussian process regression for prediction of molecular crystals harmonic free energies

We present a method to accurately predict the Helmholtz harmonic free energies of molecular crystals in high-throughput settings.This is achieved by devising a computationally efficient framework that employs a Gaussian Process Regression model based on local atomic environments.The cost to train the model with ab initio potentials is reduced by starting the optimization of the framework parameters,as well as the training and validation sets,with an empirical potential.This is then transferred to train the model based on density-functional theory potentials,including dispersion-corrections.We benchmarked our framework on a set of 444 hydrocarbon crystal structures,comprising 38 polymorphs and 406 crystal structures either measured in different conditions or derived from these polymorphs.Superior performance and high prediction accuracy,with mean absolute deviation below 0.04 kJ mol^(−1) per atom at 300 K is achieved by training on as little as 60 crystal structures.Furthermore,we demonstrate the predictive efficiency and accuracy of the developed framework by successfully calculating the thermal lattice expansion of aromatic hydrocarbon crystals within the quasi-harmonic approximation,and predict how lattice expansion affects the polymorph stability ranking.

Three-dimensional acetylenic modified graphene for highperformance optoelectronics and topological materials

Seeking carbon phases with versatile properties is one of the fundamental goals in physics,chemistry,and materials science.Here,based on the first-principles calculations,a family of three-dimensional(3D)graphene networks with abundant and fabulous electronic properties,including rarely reported dipole-allowed truly direct band gap semiconductors with suitable band gaps(1.07–1.87 eV)as optoelectronic/photovoltaic materials and topological nodal-ring semimetals,are proposed through stitching different graphene layers with acetylenic linkages.Remarkably,the optical absorption coefficients in some of those semiconducting carbon allotropes express possibly the highest performance among all of the semiconducting carbon phases known to date.On the other hand,the topological states in those topological nodal-ring semimetals are protected by the time-reversal and spatial symmetry and present nodal rings and nodal helical loops topological patterns.Those newly revealed carbon phases possess low formation energies and excellent thermodynamic stabilities;thus,they not only host a great potential in the application of optoelectronics,photovoltaics,and quantum topological materials etc.,but also can be utilized as catalysis,molecule sieves or Liion anode materials and so on.Moreover,the approach used here to design novel carbon allotropes may also give more enlightenments to create various carbon phases with different applications.

Tension-free Dirac strings and steered magnetic charges in 3D artificial spin ice

3D nano-architectures presents a new paradigm in modern condensed matter physics with numerous applications in photonics,biomedicine,and spintronics.They are promising for the realization of 3D magnetic nano-networks for ultra-fast and low-energy data storage.Frustration in these systems can lead to magnetic charges or magnetic monopoles,which can function as mobile,binary information carriers.However,Dirac strings in 2D artificial spin ices bind magnetic charges,while 3D dipolar counterparts require cryogenic temperatures for their stability.Here,we present a micromagnetic study of a highly frustrated 3D artificial spin ice harboring tension-free Dirac strings with unbound magnetic charges at room temperature.We use micromagnetic simulations to demonstrate that the mobility threshold for magnetic charges is by 2 eV lower than their unbinding energy.By applying global magnetic fields,we steer magnetic charges in a given direction omitting unintended switchings.The introduced system paves the way toward 3D magnetic networks for data transport and storage.

Causal analysis of competing atomistic mechanisms in ferroelectric materials from high-resolution scanning transmission electron microscopy data

Machine learning has emerged as a powerful tool for the analysis of mesoscopic and atomically resolved images and spectroscopy in electron and scanning probe microscopy,with the applications ranging from feature extraction to information compression and elucidation of relevant order parameters to inversion of imaging data to reconstruct structural models.However,the fundamental limitation of machine learning methods is their correlative nature,leading to extreme susceptibility to confounding factors.Here,we implement the workflow for causal analysis of structural scanning transmission electron microscopy(STEM)data and explore the interplay between physical and chemical effects in a ferroelectric perovskite across the ferroelectric–antiferroelectric phase transitions.

Author Correction:Machine-learned impurity level prediction for semiconductors:the example of Cd-based chalcogenides

The authors became aware of a mistake in the original version of this Article.Specifically,some of the band gap values plotted and reported in Fig.1c and Table SI-1 were incorrect.This error originated because two different types of k-point meshes were used in DFT computations performed on CdTe,CdSe and CdS:one which is gamma-centered and one which is not gamma-centered.

Recent developments in computational modelling of nucleation in phase transformations

Nucleation is one of the most common physical phenomena in physical,chemical,biological and materials sciences.Owing to the complex multiscale nature of various nucleation events and the difficulties in their direct experimental observation,development of effective computational methods and modeling approaches has become very important and is bringing new light to the study of this challenging subject.Our discussions in this manuscript provide a sampler of some newly developed numerical algorithms that are widely applicable to many nucleation and phase transformation problems.We first describe some recent progress on the design of efficient numerical methods for computing saddle points and minimum energy paths,and then illustrate their applications to the study of nucleation events associated with several different physical systems.

Design and discovery of materials guided by theory and computation

Computational materials science and engineering has emerged as an interdisciplinary subfield spanning materials science and engineering,condensed matter physics,chemistry,mechanics and engineering in general.Modern materials research often requires a close integration of computation and experiments in order to fundamentally understand the materials structures and properties and their relation to synthesis and processing.A number of computational methods and tools at different spatiotemporal scales are now well established,ranging from electronic structure calculations based on density functional theory,1,2 atomic molecular dynamics3,4 and Monte Carlo techniques,5 phase-field method6–9 to continuum macroscopic approaches.Over the last few years,computational materials activities have been steadily moving from technique development and purely computational studies of materials towards discovering and designing new materials guided by computation,machine learning and data mining or by a closely tied combination of computational predictions and experimental validation.This movement is being further accelerated by the recent initiatives by various government agencies in the United States,Europe,China and other countries to pursue the materials genome initiative,10 integrated computational materials engineering11–13 as well as the‘Big Data’initiative.

Modelling charge transport and electro-optical characteristics of quantum dot light-emitting diodes

Quantum dot light-emitting diodes(QD-LEDs)are considered as competitive candidate for next-generation displays or lightings.Recent advances in the synthesis of core/shell quantum dots(QDs)and tailoring procedures for achieving their high quantum yield have facilitated the emergence of high-performance QD-LEDs.Meanwhile,the charge-carrier dynamics in QD-LED devices,which constitutes the remaining core research area for further improvement of QD-LEDs,is,however,poorly understood yet.Here,we propose a charge transport model in which the charge-carrier dynamics in QD-LEDs are comprehensively described by computer simulations.The charge-carrier injection is modelled by the carrier-capturing process,while the effect of electric fields at their interfaces is considered.The simulated electro-optical characteristics of QD-LEDs,such as the luminance,current density and external quantum efficiency(EQE)curves with varying voltages,show excellent agreement with experiments.Therefore,our computational method proposed here provides a useful means for designing and optimising high-performance QD-LED devices.

Atomistic Line Graph Neural Network for improved materials property predictions

Graph neural networks(GNN)have been shown to provide substantial performance improvements for atomistic material representation and modeling compared with descriptor-based machine learning models.While most existing GNN models for atomistic predictions are based on atomic distance information,they do not explicitly incorporate bond angles,which are critical for distinguishing many atomic structures.Furthermore,many material properties are known to be sensitive to slight changes in bond angles.We present an Atomistic Line Graph Neural Network(ALIGNN),a GNN architecture that performs message passing on both the interatomic bond graph and its line graph corresponding to bond angles.We demonstrate that angle information can be explicitly and efficiently included,leading to improved performance on multiple atomistic prediction tasks.We ALIGNN models for predicting 52 solid-state and molecular properties available in the JARVIS-DFT,Materials project,and QM9 databases.ALIGNN can outperform some previously reported GNN models on atomistic prediction tasks by up to 85%in accuracy with better or comparable model training speed.

A deep neural network regressor for phase constitution estimation in the high entropy alloy system Al-Co-Cr-Fe-Mn-Nb-Ni

High Entropy Alloys(HEAs)are composed of more than one principal element and constitute a major paradigm in metals research.The HEA space is vast and an exhaustive exploration is improbable.Therefore,a thorough estimation of the phases present in the HEA is of paramount importance for alloy design.Machine Learning presents a feasible and non-expensive method for predicting possible new HEAs on-the-fly.A deep neural network(DNN)model for the elemental system of:Mn,Ni,Fe,Al,Cr,Nb,and Co is developed using a dataset generated by high-throughput computational thermodynamic calculations using Thermo-Calc.The features list used for the neural network is developed based on literature and freely available databases.A feature significance analysis matches the reported HEAs phase constitution trends on elemental properties and further expands it by providing so far-overlooked features.The final regressor has a coefficient of determination(r^(2))greater than 0.96 for identifying the most recurrent phases and the functionality is tested by running optimization tasks that simulate those required in alloy design.The DNN developed constitutes an example of an emulator that can be used in fast,real-time materials discovery/design tasks.

TransPolymer: a Transformer-based language model for polymer property predictions

Accurate and efficient prediction of polymer properties is of great significance in polymer design.Conventionally,expensive and time-consuming experiments or simulations are required to evaluate polymer functions.Recently,Transformer models,equipped with self-attention mechanisms,have exhibited superior performance in natural language processing.However,such methods have not been investigated in polymer sciences.Herein,we report TransPolymer,a Transformer-based language model for polymer property prediction.Our proposed polymer tokenizer with chemical awareness enables learning representations from polymer sequences.Rigorous experiments on ten polymer property prediction benchmarks demonstrate the superior performance of TransPolymer.Moreover,we show that TransPolymer benefits from pretraining on large unlabeled dataset via Masked Language Modeling.Experimental results further manifest the important role of self-attention in modeling polymer sequences.We highlight this model as a promising computational tool for promoting rational polymer design and understanding structure-property relationships from a data science view.

Ultrafast laser-driven topological spin textures on a 2D magnet

Ultrafast laser excitations provide an efficient and low-power consumption alternative since different magnetic properties and topological spin states can be triggered and manipulated at the femtosecond(fs)regime.However,it is largely unknown whether laser excitations already used in data information platforms can manipulate the magnetic properties of recently discovered two-dimensional(2D)van der Waals(vdW)materials.Here we show that ultrashort laser pulses(30−85 fs)can not only manipulate magnetic domains of 2D-XY CrCl_(3)ferromagnets,but also induce the formation and control of topological nontrivial meron and antimeron spin textures.We observed that these spin quasiparticles are created within~100 ps after the excitation displaying rich dynamics through motion,collision and annihilation with emission of spin waves throughout the surface.Our findings highlight substantial opportunities of using photonic driving forces for the exploration of spin textures on 2D magnetic materials towards magneto-optical topological applications.