Correction to:npj Computational Materials https://doi.org/10.1038/s41524-023-00987-9,published online 18 March 2023 The original version of this Article contained typos in both the PDF and the HTML versions.In the fou...Correction to:npj Computational Materials https://doi.org/10.1038/s41524-023-00987-9,published online 18 March 2023 The original version of this Article contained typos in both the PDF and the HTML versions.In the fourth and fifth sentences of the first paragraph of the‘Case study of three example stable materials’section of the Results,the incorrect expression“β=β”has been replaced by“α=β”.展开更多
The adsorption energies of molecular adsorbates on catalyst surfaces are key descriptors in computational catalysis research.For the relatively large reaction intermediates frequently encountered,e.g.,in syngas conver...The adsorption energies of molecular adsorbates on catalyst surfaces are key descriptors in computational catalysis research.For the relatively large reaction intermediates frequently encountered,e.g.,in syngas conversion,a multitude of possible binding motifs leads to complex potential energy surfaces(PES),however.This implies that finding the optimal structure is a difficult global optimization problem,which leads to significant uncertainty about the stability of many intermediates.To tackle this issue,we present a global optimization protocol for surface adsorbate geometries which trains a surrogate machine learning potential on-the-fly.The approach is applicable to arbitrary surface models and adsorbates and minimizes both human intervention and the number of required DFT calculations by iteratively updating the training set with configurations explored by the algorithm.We demonstrate the efficiency of this approach for a diverse set of adsorbates on the Rh(111)and(211)surfaces.展开更多
Phase-field method(PFM)has become a mainstream computational method for predicting the evolution of nano and mesoscopic microstructures and properties during materials processes.The paper briefly reviews latest progre...Phase-field method(PFM)has become a mainstream computational method for predicting the evolution of nano and mesoscopic microstructures and properties during materials processes.The paper briefly reviews latest progresses in applying PFM to understanding the thermodynamic driving forces and mechanisms underlying microstructure evolution in metallic materials and related processes,including casting,aging,deformation,additive manufacturing,and defects,etc.Focus on designing alloys by integrating PFM with constitutive relations and machine learning.Several examples are presented to demonstrate the potential of integrated PFM in discovering new multi-scale phenomena and high-performance alloys.The article ends with prospects for promising research directions.展开更多
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.Rece...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.展开更多
Predicting the thermal conductivity of glasses from first principles has hitherto been a very complex problem.The established Allen-Feldman and Green-Kubo approaches employ approximations with limited validity—the fo...Predicting the thermal conductivity of glasses from first principles has hitherto been a very complex problem.The established Allen-Feldman and Green-Kubo approaches employ approximations with limited validity—the former neglects anharmonicity,the latter misses the quantum Bose-Einstein statistics of vibrations—and require atomistic models that are very challenging for first-principles methods.Here,we present a protocol to determine from first principles the thermal conductivityκ(T)of glasses above the plateau(i.e.,above the temperature-independent region appearing almost without exceptions in theκ(T)of all glasses at cryogenic temperatures).The protocol combines the Wigner formulation of thermal transport with convergence-acceleration techniques,and accounts comprehensively for the effects of structural disorder,anharmonicity,and Bose-Einstein statistics.We validate this approach in vitreous silica,showing that models containing less than 200 atoms can already reproduceκ(T)in the macroscopic limit.We discuss the effects of anharmonicity and the mechanisms determining the trend ofκ(T)at high temperature,reproducing experiments at temperatures where radiative effects remain negligible.展开更多
In data-driven materials design where the target materials have limited data,the transfer machine learning from large known source materials,becomes a demanding strategy especially across different crystal structures....In data-driven materials design where the target materials have limited data,the transfer machine learning from large known source materials,becomes a demanding strategy especially across different crystal structures.In this work,we proposed a deep transfer learning approach to predict thermodynamically stable perovskite oxides based on a large computational dataset of spinel oxides.The deep neural network(DNN)source domain model with“Center-Environment”(CE)features was first developed using the formation energy of 5329 spinel oxide structures and then was fine-tuned by learning a small dataset of 855 perovskite oxide structures,leading to a transfer learning model with good transferability in the target domain of perovskite oxides.Based on the transferred model,we further predicted the formation energy of potential 5329 perovskite structures with combination of 73 elements.Combining the criteria of formation energy and structure factors including tolerance factor(0.7<t≤1.1)and octahedron factor(0.45<μ<0.7),we predicted 1314 thermodynamically stable perovskite oxides,among which 144 oxides were reported to be synthesized experimentally,10 oxides were predicted computationally by other literatures,301 oxides were recorded in the Materials Project database,and 859 oxides have been first reported.Combing with the structure-informed features the transfer machine learning approach in this work takes the advantage of existing data to predict new structures at a lower cost,providing an effective acceleration strategy for the expensive high-throughput computational screening in materials design.The predicted stable novel perovskite oxides serve as a rich platform for exploring potential renewable energy and electronic materials applications.展开更多
We address the problem of predicting the zero-temperature dynamical stability (DS) of a periodic crystal without computing its fullphonon band structure. Here we report the evidence that DS can be inferred with good r...We address the problem of predicting the zero-temperature dynamical stability (DS) of a periodic crystal without computing its fullphonon band structure. Here we report the evidence that DS can be inferred with good reliability from the phonon frequencies atthe center and boundary of the Brillouin zone (BZ). This analysis represents a validation of the DS test employed by theComputational 2D Materials Database (C2DB). For 137 dynamically unstable 2D crystals, we displace the atoms along an unstablemode and relax the structure. This procedure yields a dynamically stable crystal in 49 cases. The elementary properties of these newstructures are characterized using the C2DB workflow, and it is found that their properties can differ significantly from those of theoriginal unstable crystals, e.g., band gaps are opened by 0.3 eV on average. All the crystal structures and properties are available inthe C2DB. Finally, we train a classification model on the DS data for 3295 2D materials in the C2DB using a representation encodingthe electronic structure of the crystal. We obtain an excellent receiver operating characteristic (ROC) curve with an area under thecurve (AUC) of 0.90, showing that the classification model can drastically reduce computational efforts in high-throughput studies.展开更多
To reduce global warming,many countries are shifting to sustainable energy production systems.Solid oxide electrolysis cells(SOECs)are being considered due to their high hydrogen generation efficiency.However,low fara...To reduce global warming,many countries are shifting to sustainable energy production systems.Solid oxide electrolysis cells(SOECs)are being considered due to their high hydrogen generation efficiency.However,low faradaic efficiency in scaling SOEC technology affects costs and limits large-scale adoption of hydrogen as fuel.This review covers SOECs’critical aspects:current state-of-the-art anode,cathode,and electrolyte materials,operational and materials parameters affecting faradaic efficiency,and computational modeling techniques to resolve bottlenecks affecting SOEC faradaic efficiency.展开更多
Metal additive manufacturing(AM)offers flexibility and cost-effectiveness for printing complex parts but is limited to few alloys.Qualifying new alloys requires process parameter optimisation to produce consistent,hig...Metal additive manufacturing(AM)offers flexibility and cost-effectiveness for printing complex parts but is limited to few alloys.Qualifying new alloys requires process parameter optimisation to produce consistent,high-quality components.High-resolution X-ray computed tomography(XCT)has not been effective for this task due to artifacts,slow scan speed,and costs.We propose a deep learning-based approach for rapid XCT acquisition and reconstruction of metal AM parts,leveraging computer-aided design models and physics-based simulations of nonlinear interactions between X-ray radiation and metals.This significantly reduces beam hardening and common XCT artifacts.We demonstrate high-throughput characterisation of over a hundred AlCe alloy components,quantifying improvements in characterisation time and quality compared to high-resolution microscopy and pycnometry.Our approach facilitates investigating the impact of process parameters and their geometry dependence in metal AM.展开更多
Maximally-localized Wannier functions(MLWFs)are broadly used to characterize the electronic structure of materials.Generally,one can construct MLWFs describing isolated bands(e.g.valence bands of insulators)or entangl...Maximally-localized Wannier functions(MLWFs)are broadly used to characterize the electronic structure of materials.Generally,one can construct MLWFs describing isolated bands(e.g.valence bands of insulators)or entangled bands(e.g.valence and conduction bands of insulators,or metals).Obtaining accurate and compact MLWFs often requires chemical intuition and trial and error,a challenging step even for experienced researchers and a roadblock for high-throughput calculations.Here,we present an automated approach,projectability-disentangled Wannier functions(PDWFs),that constructs MLWFs spanning the occupied bands and their complement for the empty states,providing a tight-binding picture of optimized atomic orbitals in crystals.Key to the algorithm is a projectability measure for each Bloch state onto atomic orbitals,determining if that state should be kept identically,discarded,or mixed into the disentanglement.We showcase the accuracy on a test set of 200 materials,and the reliability by constructing 21,737 Wannier Hamiltonians.展开更多
Many-body perturbation theory methods,such as the G0W0 approximation,are able to accurately predict quasiparticle(QP)properties of several classes of materials.However,the calculation of the QP band structure of two-d...Many-body perturbation theory methods,such as the G0W0 approximation,are able to accurately predict quasiparticle(QP)properties of several classes of materials.However,the calculation of the QP band structure of two-dimensional(2D)semiconductors is known to require a very dense BZ sampling,due to the sharp q-dependence of the dielectric matrix in the long-wavelength limit(q→0).In this work,we show how the convergence of the QP corrections of 2D semiconductors with respect to the BZ sampling can be drastically improved,by combining a Monte Carlo integration with an interpolation scheme able to represent the screened potential between the calculated grid points.The method has been validated by computing the band gap of three different prototype monolayer materials:a transition metal dichalcogenide(MoS2),a wide band gap insulator(hBN)and an anisotropic semiconductor(phosphorene).The proposed scheme shows that the convergence of the gap for these three materials up to 50meV is achieved by using k-point grids comparable to those needed by DFT calculations,while keeping the grid uniform.展开更多
Computational modeling of physical processes in metal-organic frameworks(MOFs)is highly challenging due to the presence of spatial heterogeneities and complex operating conditions which affect their behavior.Density f...Computational modeling of physical processes in metal-organic frameworks(MOFs)is highly challenging due to the presence of spatial heterogeneities and complex operating conditions which affect their behavior.Density functional theory(DFT)may describe interatomic interactions at the quantum mechanical level,but is computationally too expensive for systems beyond the nanometer and picosecond range.Herein,we propose an incremental learning scheme to construct accurate and data-efficient machine learning potentials for MOFs.The scheme builds on the power of equivariant neural network potentials in combination with parallelized enhanced sampling and on-the-fly training to simultaneously explore and learn the phase space in an iterative manner.With only a few hundred single-point DFT evaluations per material,accurate and transferable potentials are obtained,even for flexible frameworks with multiple structurally different phases.The incremental learning scheme is universally applicable and may pave the way to model framework materials in larger spatiotemporal windows with higher accuracy.展开更多
A desired prerequisite when performing a quantum mechanical calculation is to have an initial idea of the atomic positions within an approximate crystal structure.The atomic positions combined should result in a syste...A desired prerequisite when performing a quantum mechanical calculation is to have an initial idea of the atomic positions within an approximate crystal structure.The atomic positions combined should result in a system located in,or close to,an energy minimum.However,designing low-energy structures may be a challenging task when prior knowledge is scarce,specifically for large multi-component systems where the degrees of freedom are close to infinite.In this paper,we propose a method for identification of low-energy crystal structures within multi-component systems by combining cluster expansion and crystal structure predictions with density-functional theory calculations.Crystal structure prediction searches are applied to the Mo_(2)AlB_(2) and Sc2AlB_(2) ternary systems to identify candidate structures,which are subsequently used to explore the quaternary(pseudo-binary)(Mo_(x)Sc_(1–x))2AlB_(2) system through the cluster expansion formalism utilizing the ground-state search approach.Furthermore,we show that utilizing low-energy structures found within the cluster expansion ground-state search as seed structures within crystal structure predictions of(Mo_(x)Sc_(1–x))2AlB_(2) can significantly reduce the computational demands.With this combined approach,we not only correctly identified the recently discovered Mo_(4/3)Sc_(2/3)AlB_(2) i-MAB phase,comprised of in-plane chemical ordering of Mo and Sc and with Al in a Kagomélattice,but also predict additional low-energy structures at various concentrations.This result demonstrates that combining crystal structure prediction with cluster expansion provides a path for identifying low-energy crystal structures in multi-component systems by employing the strengths from both frameworks.展开更多
Electron-electron correlations play central role in condensed matter physics,governing phenomena from superconductivity to magnetism and numerous technological applications.Two-dimensional(2D)materials with flat elect...Electron-electron correlations play central role in condensed matter physics,governing phenomena from superconductivity to magnetism and numerous technological applications.Two-dimensional(2D)materials with flat electronic bands provide natural playground to explore interaction-driven physics,thanks to their highly localized electrons.The search for 2D flat band materials has attracted intensive efforts,especially now with open science databases encompassing thousands of materials with computed electronic bands.Here we automate the otherwise daunting task of materials search and classification by combining supervised and unsupervised machine learning algorithms.To this end,convolutional neural network was employed to identify 2D flat band materials,which were then subjected to symmetry-based analysis using a bilayer unsupervised learning algorithm.Such hybrid approach of exploring materials databases allowed us to construct a genome of 2D materials hosting flat bands and to reveal material classes outside the known flat band paradigms.展开更多
We report a high throughput computational search for two-dimensional ferroelectric materials.The starting point is 252 pyroelectric materials from the computational 2D materials database(C2DB)and from these we identif...We report a high throughput computational search for two-dimensional ferroelectric materials.The starting point is 252 pyroelectric materials from the computational 2D materials database(C2DB)and from these we identify 63 ferroelectrics.In particular we find 49 materials with in-plane polarization,8 materials with out-of-plane polarization and 6 materials with coupled in-plane and out-of-plane polarization.Most of the known 2D ferroelectrics are recovered by the screening and the far majority of the predicted ferroelectrics are known as bulk van der Waals bonded compounds,which makes them accessible by direct exfoliation.For roughly 25%of the materials we find a metastable state in the non-polar structure,which may imply a first order transition to the polar phase.Finally,we list the magnetic pyroelectrics extracted from the C2DB and focus on the case of VAgP2Se6,which exhibits a three-state switchable polarization vector that is strongly coupled to the magnetic excitation spectrum.展开更多
Understanding the physical picture of Li ion transport in the current ionic conductors is quite essential to further develop lithium superionic conductors for solid-state batteries.The traditional practice of directly...Understanding the physical picture of Li ion transport in the current ionic conductors is quite essential to further develop lithium superionic conductors for solid-state batteries.The traditional practice of directly extrapolating room temperature ion diffusion properties from the high-temperature(>600 K)ab initio molecular dynamics simulations(AIMD)simulations by the Arrhenius assumption unavoidably cause some deviations.Fortunately,the ultralong-time molecular dynamics simulation based on the machine-learning interatomic potentials(MLMD)is a more suitable tool to probe into ion diffusion events at low temperatures and simultaneously keeps the accuracy at the density functional theory level.Herein,by the low-temperature MLMD simulations,the non-linear Arrhenius behavior of Li ion was found for Li3ErCl6,which is the main reason for the traditional AIMD simulation overestimating its ionic conductivity.The 1μs MLMD simulations capture polyanion rotation events in Li_(7)P_(3)S_(11) at room temperature,in which four[PS_(4)]^(3−)tetrahedra belonging to a part of the longer-chain[P_(2)S_(7)]4−group are noticed with remarkable rotational motions,while the isolated group[PS_(4)]^(3−)does not rotate.However,no polyanion rotation is observed in Li10GeP_(2)S12,β-Li3PS_(4),Li3ErCl6,and Li3YBr6 at 300 K during 1μs simulation time.Additionally,the ultralong-time MLMD simulations demonstrate that not only there is no paddle-wheel effect in the crystalline Li_(7)P_(3)S_(11) at room temperature,but also the rotational[PS_(4)]^(3−)polyanion groups have weakly negative impacts on the overall Li ion diffusion.The ultralong-time MLMD simulations deepen our understanding of the relationship between the polyanion rotation and cation diffusion in ionic conductors at room environments.展开更多
This review discussed the dilemma of small data faced by materials machine learning.First,we analyzed the limitations brought by small data.Then,the workflow of materials machine learning has been introduced.Next,the ...This review discussed the dilemma of small data faced by materials machine learning.First,we analyzed the limitations brought by small data.Then,the workflow of materials machine learning has been introduced.Next,the methods of dealing with small data were introduced,including data extraction from publications,materials database construction,high-throughput computations and experiments from the data source level;modeling algorithms for small data and imbalanced learning from the algorithm level;active learning and transfer learning from the machine learning strategy level.Finally,the future directions for small data machine learning in materials science were proposed.展开更多
The design of alloys for use in gas turbine engine blades is a complex task that involves balancing multiple objectives and constraints.Candidate alloys must be ductile at room temperature and retain their yield stren...The design of alloys for use in gas turbine engine blades is a complex task that involves balancing multiple objectives and constraints.Candidate alloys must be ductile at room temperature and retain their yield strength at high temperatures,as well as possess low density,high thermal conductivity,narrow solidification range,high solidus temperature,and a small linear thermal expansion coefficient.Traditional Integrated Computational Materials Engineering(ICME)methods are not sufficient for exploring combinatorially-vast alloy design spaces,optimizing for multiple objectives,nor ensuring that multiple constraints are met.In this work,we propose an approach for solving a constrained multi-objective materials design problem over a large composition space,specifically focusing on the Mo-Nb-Ti-V-W system as a representative Multi-Principal Element Alloy(MPEA)for potential use in next-generation gas turbine blades.Our approach is able to learn and adapt to unknown constraints in the design space,making decisions about the best course of action at each stage of the process.As a result,we identify 21 Pareto-optimal alloys that satisfy all constraints.Our proposed framework is significantly more efficient and faster than a brute force approach.展开更多
We introduce Crystal Edge Graph Attention Neural Network (CEGANN) workflow that uses graph attention-based architecture tolearn unique feature representations and perform classification of materials across multiple sc...We introduce Crystal Edge Graph Attention Neural Network (CEGANN) workflow that uses graph attention-based architecture tolearn unique feature representations and perform classification of materials across multiple scales (from atomic to mesoscale) anddiverse classes ranging from metals, oxides, non-metals to hierarchical materials such as zeolites and semi-ordered mesophases.CEGANN can classify based on a global, structure-level representation such as space group and dimensionality (e.g., bulk, 2D,clusters, etc.). Using representative materials such as polycrystals and zeolites, we demonstrate its transferability in performing localatom-level classification tasks, such as grain boundary identification and other heterointerfaces. CEGANN classifies in (thermal)noisy dynamical environments as demonstrated for representative zeolite nucleation and growth from an amorphous mixture.Finally, we use CEGANN to classify multicomponent systems with thermal noise and compositional diversity. Overall, our approachis material agnostic and allows for multiscale feature classification ranging from atomic-scale crystals to heterointerfaces tomicroscale grain boundaries.展开更多
Restricted use of hazardous environmental chemicals is one important challenge that the semiconductor industry needs to face to improve its sustainability.Ovonic threshold switching(OTS)ternary compound materials used...Restricted use of hazardous environmental chemicals is one important challenge that the semiconductor industry needs to face to improve its sustainability.Ovonic threshold switching(OTS)ternary compound materials used in memory selector devices contain As and Se.Engineering these elements out of these materials requires significant research effort.To facilitate this process,we performed systematic material screening for As/Se-free ternary materials,based on ab-initio simulations.To limit the large amount of possible chemical compositions to fewer promising candidates,we used physics-based material parameter filters like material stability,electronic properties,or change in polarizability.The OTS gauge concept is introduced as a computed parameter to estimate the probability of a material to show an OTS behavior.As a result,we identified 35 As/Se-free ternary alloy compositions for stand-alone OTS memory applications,as well as 12 compositions for RRAM selector applications.This work aims seeding the development of As/Se-free OTS materials.展开更多
文摘Correction to:npj Computational Materials https://doi.org/10.1038/s41524-023-00987-9,published online 18 March 2023 The original version of this Article contained typos in both the PDF and the HTML versions.In the fourth and fifth sentences of the first paragraph of the‘Case study of three example stable materials’section of the Results,the incorrect expression“β=β”has been replaced by“α=β”.
文摘The adsorption energies of molecular adsorbates on catalyst surfaces are key descriptors in computational catalysis research.For the relatively large reaction intermediates frequently encountered,e.g.,in syngas conversion,a multitude of possible binding motifs leads to complex potential energy surfaces(PES),however.This implies that finding the optimal structure is a difficult global optimization problem,which leads to significant uncertainty about the stability of many intermediates.To tackle this issue,we present a global optimization protocol for surface adsorbate geometries which trains a surrogate machine learning potential on-the-fly.The approach is applicable to arbitrary surface models and adsorbates and minimizes both human intervention and the number of required DFT calculations by iteratively updating the training set with configurations explored by the algorithm.We demonstrate the efficiency of this approach for a diverse set of adsorbates on the Rh(111)and(211)surfaces.
基金Also supported by National Natural Science Foundation of China(Nos.52074246,52201146,52205429,52275390)National Defense Basic Scientific Research Program of China(No.JCKY2020408B002)Key Research and Development Program of Shanxi Province(202102050201011).Many thanks to Dr.XL Tian of North University of China for her kind effort and time in checking,processing,and editing,and Professor L.Q.Chen of Pennsylvania State University for his invitation and critical feedback.
文摘Phase-field method(PFM)has become a mainstream computational method for predicting the evolution of nano and mesoscopic microstructures and properties during materials processes.The paper briefly reviews latest progresses in applying PFM to understanding the thermodynamic driving forces and mechanisms underlying microstructure evolution in metallic materials and related processes,including casting,aging,deformation,additive manufacturing,and defects,etc.Focus on designing alloys by integrating PFM with constitutive relations and machine learning.Several examples are presented to demonstrate the potential of integrated PFM in discovering new multi-scale phenomena and high-performance alloys.The article ends with prospects for promising research directions.
文摘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.
基金N.M.acknowledges funding from the Swiss National Science Foundation under the Sinergia grant no.189924M.S.acknowledges support from Gonville and Caius College,and from the SNSF project P500PT_203178Part of the calculations presented in this work have been performed using computational resources provided by the Cambridge Tier-2 system operated by the University of Cambridge Research Computing Service(www.hpc.cam.ac.uk)funded by EPSRC Tier-2 capital grant EP/T022159/1.
文摘Predicting the thermal conductivity of glasses from first principles has hitherto been a very complex problem.The established Allen-Feldman and Green-Kubo approaches employ approximations with limited validity—the former neglects anharmonicity,the latter misses the quantum Bose-Einstein statistics of vibrations—and require atomistic models that are very challenging for first-principles methods.Here,we present a protocol to determine from first principles the thermal conductivityκ(T)of glasses above the plateau(i.e.,above the temperature-independent region appearing almost without exceptions in theκ(T)of all glasses at cryogenic temperatures).The protocol combines the Wigner formulation of thermal transport with convergence-acceleration techniques,and accounts comprehensively for the effects of structural disorder,anharmonicity,and Bose-Einstein statistics.We validate this approach in vitreous silica,showing that models containing less than 200 atoms can already reproduceκ(T)in the macroscopic limit.We discuss the effects of anharmonicity and the mechanisms determining the trend ofκ(T)at high temperature,reproducing experiments at temperatures where radiative effects remain negligible.
基金This work was supported by the National Natural Science Foundation of China[Nos.22177067]Sino-German Mobility Program[No.M-0209]+3 种基金the Shanghai Rising-Star Program[No.20QA1403400]the Key Basic Research Program of Science and Technology Commission of Shanghai Municipality(20JC1415300)This work was also supported the Key Research Project of Zhejiang Laboratory(No.2021PE0AC02)Shanghai Technical Service Center for Advanced Ceramics Structure Design and Precision Manufacturing(No.20DZ2294000).The authors acknowledge the Beijing Super Cloud Computing Center,Hefei Advanced Computing Center,and Shanghai University for providing HPC resources.
文摘In data-driven materials design where the target materials have limited data,the transfer machine learning from large known source materials,becomes a demanding strategy especially across different crystal structures.In this work,we proposed a deep transfer learning approach to predict thermodynamically stable perovskite oxides based on a large computational dataset of spinel oxides.The deep neural network(DNN)source domain model with“Center-Environment”(CE)features was first developed using the formation energy of 5329 spinel oxide structures and then was fine-tuned by learning a small dataset of 855 perovskite oxide structures,leading to a transfer learning model with good transferability in the target domain of perovskite oxides.Based on the transferred model,we further predicted the formation energy of potential 5329 perovskite structures with combination of 73 elements.Combining the criteria of formation energy and structure factors including tolerance factor(0.7<t≤1.1)and octahedron factor(0.45<μ<0.7),we predicted 1314 thermodynamically stable perovskite oxides,among which 144 oxides were reported to be synthesized experimentally,10 oxides were predicted computationally by other literatures,301 oxides were recorded in the Materials Project database,and 859 oxides have been first reported.Combing with the structure-informed features the transfer machine learning approach in this work takes the advantage of existing data to predict new structures at a lower cost,providing an effective acceleration strategy for the expensive high-throughput computational screening in materials design.The predicted stable novel perovskite oxides serve as a rich platform for exploring potential renewable energy and electronic materials applications.
基金The Center for Nanostructured Graphene(CNG)is sponsored by the Danish National Research Foundation,Project DNRF103This project has received funding from the European Research Council(ERC)under the European Union’s Horizon 2020 research and innovation program grant agreement no.773122(LIMA)K.S.T.is a Villum Investigator supported by VILLUM FONDEN(grant no.37789).
文摘We address the problem of predicting the zero-temperature dynamical stability (DS) of a periodic crystal without computing its fullphonon band structure. Here we report the evidence that DS can be inferred with good reliability from the phonon frequencies atthe center and boundary of the Brillouin zone (BZ). This analysis represents a validation of the DS test employed by theComputational 2D Materials Database (C2DB). For 137 dynamically unstable 2D crystals, we displace the atoms along an unstablemode and relax the structure. This procedure yields a dynamically stable crystal in 49 cases. The elementary properties of these newstructures are characterized using the C2DB workflow, and it is found that their properties can differ significantly from those of theoriginal unstable crystals, e.g., band gaps are opened by 0.3 eV on average. All the crystal structures and properties are available inthe C2DB. Finally, we train a classification model on the DS data for 3295 2D materials in the C2DB using a representation encodingthe electronic structure of the crystal. We obtain an excellent receiver operating characteristic (ROC) curve with an area under thecurve (AUC) of 0.90, showing that the classification model can drastically reduce computational efforts in high-throughput studies.
基金Work supported through the INL Laboratory Directed Research&Development(LDRD)Program under DOE Idaho Operations Office Contract DE-AC07-05ID14517.
文摘To reduce global warming,many countries are shifting to sustainable energy production systems.Solid oxide electrolysis cells(SOECs)are being considered due to their high hydrogen generation efficiency.However,low faradaic efficiency in scaling SOEC technology affects costs and limits large-scale adoption of hydrogen as fuel.This review covers SOECs’critical aspects:current state-of-the-art anode,cathode,and electrolyte materials,operational and materials parameters affecting faradaic efficiency,and computational modeling techniques to resolve bottlenecks affecting SOEC faradaic efficiency.
基金This manuscript has been authored by UT-Battelle,LLC,under contract DE-AC05-00OR22725 with the US Department of Energy(DOE)Research sponsored by the US Department of Energy,Office of Energy Efficiency and Renewable Energy,Advanced Manufacturing Office and Technology Commercialisation Fund(TCF-21-24881)+1 种基金under contract DE-AC05-00OR22725 with UT-Battelle,LLCThe US government retains and the publisher,by accepting the article for publication,acknowledges that the US government retains a nonexclusive,paid-up,irrevocable,worldwide license to publish or reproduce the published form of this manuscript,or allow others to do so,for US government purposes.DOE will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan(http://energy.gov/downloads/doe-public-access-plan).
文摘Metal additive manufacturing(AM)offers flexibility and cost-effectiveness for printing complex parts but is limited to few alloys.Qualifying new alloys requires process parameter optimisation to produce consistent,high-quality components.High-resolution X-ray computed tomography(XCT)has not been effective for this task due to artifacts,slow scan speed,and costs.We propose a deep learning-based approach for rapid XCT acquisition and reconstruction of metal AM parts,leveraging computer-aided design models and physics-based simulations of nonlinear interactions between X-ray radiation and metals.This significantly reduces beam hardening and common XCT artifacts.We demonstrate high-throughput characterisation of over a hundred AlCe alloy components,quantifying improvements in characterisation time and quality compared to high-resolution microscopy and pycnometry.Our approach facilitates investigating the impact of process parameters and their geometry dependence in metal AM.
基金We acknowledge financial support from the NCCR MARVEL(a National Centre of Competence in Research,funded by the Swiss National Science Foundation,grant No.205602)the Swiss National Science Foundation(SNSF)Project Funding(grant 200021E_206190“FISH4DIET”)The work is also supported by a pilot access grant from the Swiss National Supercomputing Centre(CSCS)on the Swiss share of the LUMI system under project ID“PILOT MC EPFL-NM 01”,a CHRONOS grant from the CSCS on the Swiss share of the LUMI system under project ID“REGULAR MC EPFL-NM 02”,and a grant from the CSCS under project ID s0178.
文摘Maximally-localized Wannier functions(MLWFs)are broadly used to characterize the electronic structure of materials.Generally,one can construct MLWFs describing isolated bands(e.g.valence bands of insulators)or entangled bands(e.g.valence and conduction bands of insulators,or metals).Obtaining accurate and compact MLWFs often requires chemical intuition and trial and error,a challenging step even for experienced researchers and a roadblock for high-throughput calculations.Here,we present an automated approach,projectability-disentangled Wannier functions(PDWFs),that constructs MLWFs spanning the occupied bands and their complement for the empty states,providing a tight-binding picture of optimized atomic orbitals in crystals.Key to the algorithm is a projectability measure for each Bloch state onto atomic orbitals,determining if that state should be kept identically,discarded,or mixed into the disentanglement.We showcase the accuracy on a test set of 200 materials,and the reliability by constructing 21,737 Wannier Hamiltonians.
基金This work was partially supported by SUPER(Supercomputing Unified Platform-Emilia-Romagna)from Emilia-Romagna PORFESR 2014-2020 regional fundsWe also thank MaX-MAterials design at the eXascale-a European Centre of Excellence,funded by the European Union programme H2020-INFRAEDI-2018-1(Grant No.824143),HORIZON-EUROHPC-JU-2021-COE-1,Grant No.101093324.We also acknowledge financial support from ICSC-Centro Nazionale di Ricerca in High Performance Computing,Big Data and Quantum Computing,funded by European Union-NextGenerationEU-PNRR,Missione 4 Componente 2 Investimento 1.4.D.V.also thanks the Italian national programme PRIN20172017BZPKSZ“Excitonic insulator in two-dimensional long-range interacting systems”.Computational time on the Galileo machine at CINECA was provided by the Italian ISCRA programme.
文摘Many-body perturbation theory methods,such as the G0W0 approximation,are able to accurately predict quasiparticle(QP)properties of several classes of materials.However,the calculation of the QP band structure of two-dimensional(2D)semiconductors is known to require a very dense BZ sampling,due to the sharp q-dependence of the dielectric matrix in the long-wavelength limit(q→0).In this work,we show how the convergence of the QP corrections of 2D semiconductors with respect to the BZ sampling can be drastically improved,by combining a Monte Carlo integration with an interpolation scheme able to represent the screened potential between the calculated grid points.The method has been validated by computing the band gap of three different prototype monolayer materials:a transition metal dichalcogenide(MoS2),a wide band gap insulator(hBN)and an anisotropic semiconductor(phosphorene).The proposed scheme shows that the convergence of the gap for these three materials up to 50meV is achieved by using k-point grids comparable to those needed by DFT calculations,while keeping the grid uniform.
基金S.V.and M.C.C.wish to thank the Research Foundation-Flanders(FWO)for doctoral fellowships(grant nos.11H6821N and 11D0420N respectively)The resources and services used in this work were provided by VSC(Flemish Supercomputer Center),funded by the Research Foundation-Flanders(FWO)and the Flemish Government。
文摘Computational modeling of physical processes in metal-organic frameworks(MOFs)is highly challenging due to the presence of spatial heterogeneities and complex operating conditions which affect their behavior.Density functional theory(DFT)may describe interatomic interactions at the quantum mechanical level,but is computationally too expensive for systems beyond the nanometer and picosecond range.Herein,we propose an incremental learning scheme to construct accurate and data-efficient machine learning potentials for MOFs.The scheme builds on the power of equivariant neural network potentials in combination with parallelized enhanced sampling and on-the-fly training to simultaneously explore and learn the phase space in an iterative manner.With only a few hundred single-point DFT evaluations per material,accurate and transferable potentials are obtained,even for flexible frameworks with multiple structurally different phases.The incremental learning scheme is universally applicable and may pave the way to model framework materials in larger spatiotemporal windows with higher accuracy.
基金M.D.acknowledges support from the Swedish Research council through project 2019-05047J.R.acknowledges funding from the Knut and Alice Wallenberg(KAW)Foundation for a Fellowship/Scholar Grant and Project funding(KAW 2020.0033)+1 种基金The calculations were conducted using supercomputer resources provided by the Swedish National Infrastructure for Computing(SNIC)the National Supercomputer Center(NSC)and the High Performance Computing Center North(HPC2N),partially founded by the Swedish Research Council through grant agreement no.2018-05973.
文摘A desired prerequisite when performing a quantum mechanical calculation is to have an initial idea of the atomic positions within an approximate crystal structure.The atomic positions combined should result in a system located in,or close to,an energy minimum.However,designing low-energy structures may be a challenging task when prior knowledge is scarce,specifically for large multi-component systems where the degrees of freedom are close to infinite.In this paper,we propose a method for identification of low-energy crystal structures within multi-component systems by combining cluster expansion and crystal structure predictions with density-functional theory calculations.Crystal structure prediction searches are applied to the Mo_(2)AlB_(2) and Sc2AlB_(2) ternary systems to identify candidate structures,which are subsequently used to explore the quaternary(pseudo-binary)(Mo_(x)Sc_(1–x))2AlB_(2) system through the cluster expansion formalism utilizing the ground-state search approach.Furthermore,we show that utilizing low-energy structures found within the cluster expansion ground-state search as seed structures within crystal structure predictions of(Mo_(x)Sc_(1–x))2AlB_(2) can significantly reduce the computational demands.With this combined approach,we not only correctly identified the recently discovered Mo_(4/3)Sc_(2/3)AlB_(2) i-MAB phase,comprised of in-plane chemical ordering of Mo and Sc and with Al in a Kagomélattice,but also predict additional low-energy structures at various concentrations.This result demonstrates that combining crystal structure prediction with cluster expansion provides a path for identifying low-energy crystal structures in multi-component systems by employing the strengths from both frameworks.
基金This research was supported by the European Research Council(ERC)under the European Union’s Horizon 2020 research and innovation program(Grant Agreement No.865590)the Royal Society International Exchanges 2019 Cost Share Program(IEC\R2\192001)+1 种基金A.B.acknowledges the Commonwealth Scholarship Commission in the UK for financial assistance.Q.Y.acknowledges the funding from Leverhulme Early Career Fellowship ECF-2019-612Dame Kathleen Ollerenshaw Fellowship from the University of Manchester,and Royal Society University Research Fellowship URF\R1\221096.
文摘Electron-electron correlations play central role in condensed matter physics,governing phenomena from superconductivity to magnetism and numerous technological applications.Two-dimensional(2D)materials with flat electronic bands provide natural playground to explore interaction-driven physics,thanks to their highly localized electrons.The search for 2D flat band materials has attracted intensive efforts,especially now with open science databases encompassing thousands of materials with computed electronic bands.Here we automate the otherwise daunting task of materials search and classification by combining supervised and unsupervised machine learning algorithms.To this end,convolutional neural network was employed to identify 2D flat band materials,which were then subjected to symmetry-based analysis using a bilayer unsupervised learning algorithm.Such hybrid approach of exploring materials databases allowed us to construct a genome of 2D materials hosting flat bands and to reveal material classes outside the known flat band paradigms.
基金M.K.and T.O.were supported by the Danish Independent Research Foundation,Grant number 9040-00269B.U.PT.O.were supported by the Villum foundation,Grant No.00028145.
文摘We report a high throughput computational search for two-dimensional ferroelectric materials.The starting point is 252 pyroelectric materials from the computational 2D materials database(C2DB)and from these we identify 63 ferroelectrics.In particular we find 49 materials with in-plane polarization,8 materials with out-of-plane polarization and 6 materials with coupled in-plane and out-of-plane polarization.Most of the known 2D ferroelectrics are recovered by the screening and the far majority of the predicted ferroelectrics are known as bulk van der Waals bonded compounds,which makes them accessible by direct exfoliation.For roughly 25%of the materials we find a metastable state in the non-polar structure,which may imply a first order transition to the polar phase.Finally,we list the magnetic pyroelectrics extracted from the C2DB and focus on the case of VAgP2Se6,which exhibits a three-state switchable polarization vector that is strongly coupled to the magnetic excitation spectrum.
基金This work is supported by the Young Scientists Fund of the National Natural Science Foundation of China(22209074)the Fundamental Research Funds for the Central Universities(NO.NS2022059,NO.NS2021039)+1 种基金the Talent Research Startup Funds of Nanjing University of Aeronautics and Astronautics,the Jiangsu Funding Program for Excellent Postdoctoral Talent,and the Selected Funding for Scientific and Technological Innovation Projects for Overseas Students in NanjingThis work is partially supported by High Performance Computing Platform of Nanjing University of Aeronautics and Astronautics.
文摘Understanding the physical picture of Li ion transport in the current ionic conductors is quite essential to further develop lithium superionic conductors for solid-state batteries.The traditional practice of directly extrapolating room temperature ion diffusion properties from the high-temperature(>600 K)ab initio molecular dynamics simulations(AIMD)simulations by the Arrhenius assumption unavoidably cause some deviations.Fortunately,the ultralong-time molecular dynamics simulation based on the machine-learning interatomic potentials(MLMD)is a more suitable tool to probe into ion diffusion events at low temperatures and simultaneously keeps the accuracy at the density functional theory level.Herein,by the low-temperature MLMD simulations,the non-linear Arrhenius behavior of Li ion was found for Li3ErCl6,which is the main reason for the traditional AIMD simulation overestimating its ionic conductivity.The 1μs MLMD simulations capture polyanion rotation events in Li_(7)P_(3)S_(11) at room temperature,in which four[PS_(4)]^(3−)tetrahedra belonging to a part of the longer-chain[P_(2)S_(7)]4−group are noticed with remarkable rotational motions,while the isolated group[PS_(4)]^(3−)does not rotate.However,no polyanion rotation is observed in Li10GeP_(2)S12,β-Li3PS_(4),Li3ErCl6,and Li3YBr6 at 300 K during 1μs simulation time.Additionally,the ultralong-time MLMD simulations demonstrate that not only there is no paddle-wheel effect in the crystalline Li_(7)P_(3)S_(11) at room temperature,but also the rotational[PS_(4)]^(3−)polyanion groups have weakly negative impacts on the overall Li ion diffusion.The ultralong-time MLMD simulations deepen our understanding of the relationship between the polyanion rotation and cation diffusion in ionic conductors at room environments.
基金This work was supported by the National Natural Science Foundation of China(No.52102140)Shanghai Pujiang Program(No.21PJD024)the Key Research Project of Zhejiang Laboratory(No.2021PE0AC02).
文摘This review discussed the dilemma of small data faced by materials machine learning.First,we analyzed the limitations brought by small data.Then,the workflow of materials machine learning has been introduced.Next,the methods of dealing with small data were introduced,including data extraction from publications,materials database construction,high-throughput computations and experiments from the data source level;modeling algorithms for small data and imbalanced learning from the algorithm level;active learning and transfer learning from the machine learning strategy level.Finally,the future directions for small data machine learning in materials science were proposed.
基金The authors acknowledge the support from the U.S.Department of Energy(DOE)ARPA-E ULTIMATE Program through Project DE-AR0001427 and DEVCOM-ARL under Contract No.W911NF2220106(HTMDEC)B.V.acknowledges the support of NSF through Grant No.DGE-1545403+1 种基金D.K.acknowledges the support of NSF through Grant No.CDSE-2001333R.A.acknowledges the support from Grants No.NSF-CISE-1835690 and NSF-DMREF-2119103.High-throughput CALPHAD and DFT calculations were carried out partly at the Texas A&M High-Performance Research Computing(HPRC)Facility.ARPA-E supported the applications of theory in this work.In contrast,the theory development(KKR-CPA and SCRAPs by DDJ/PS)at Ames National Laboratory were supported by the U.S.DOE,Office of Science,Basic Energy Sciences,Materials Science and Engineering Department.Ames Laboratory is operated by Iowa State University for the U.S.DOE under contract DE-AC02-07CH11358.
文摘The design of alloys for use in gas turbine engine blades is a complex task that involves balancing multiple objectives and constraints.Candidate alloys must be ductile at room temperature and retain their yield strength at high temperatures,as well as possess low density,high thermal conductivity,narrow solidification range,high solidus temperature,and a small linear thermal expansion coefficient.Traditional Integrated Computational Materials Engineering(ICME)methods are not sufficient for exploring combinatorially-vast alloy design spaces,optimizing for multiple objectives,nor ensuring that multiple constraints are met.In this work,we propose an approach for solving a constrained multi-objective materials design problem over a large composition space,specifically focusing on the Mo-Nb-Ti-V-W system as a representative Multi-Principal Element Alloy(MPEA)for potential use in next-generation gas turbine blades.Our approach is able to learn and adapt to unknown constraints in the design space,making decisions about the best course of action at each stage of the process.As a result,we identify 21 Pareto-optimal alloys that satisfy all constraints.Our proposed framework is significantly more efficient and faster than a brute force approach.
基金The authors acknowledge support from the US Department of Energy through BES award DE-SC0021201This material is based on work supported by the DOE,Office of Science,BES Data,Artificial Intelligence and Machine Learning at DOE Scientific User Facilities programme(MLExchange).Use of the Center for Nanoscale Materials,an Office of Science user facility,was supported by the US Department of Energy,Office of Science,Office of Basic Energy Sciences,under Contract No.DE-AC02-06CH11357+2 种基金This research also used resources from the Argonne Leadership Computing Facility at Argonne National Laboratory,which is supported by the Office of Science of the U.S.Department of Energy under contract DE-AC02-06CH11357This research used resources of the National Energy Research Scientific Computing Centera DOE Office of Science User Facility supported by the Office of Science of the U.S.Department of Energy under Contract No.DE-AC02-05CH11231.We gratefully acknowledge the computing resources provided via high-performance computing clusters operated by the Laboratory Computing Resource Center(LCRC)at Argonne National Laboratory.
文摘We introduce Crystal Edge Graph Attention Neural Network (CEGANN) workflow that uses graph attention-based architecture tolearn unique feature representations and perform classification of materials across multiple scales (from atomic to mesoscale) anddiverse classes ranging from metals, oxides, non-metals to hierarchical materials such as zeolites and semi-ordered mesophases.CEGANN can classify based on a global, structure-level representation such as space group and dimensionality (e.g., bulk, 2D,clusters, etc.). Using representative materials such as polycrystals and zeolites, we demonstrate its transferability in performing localatom-level classification tasks, such as grain boundary identification and other heterointerfaces. CEGANN classifies in (thermal)noisy dynamical environments as demonstrated for representative zeolite nucleation and growth from an amorphous mixture.Finally, we use CEGANN to classify multicomponent systems with thermal noise and compositional diversity. Overall, our approachis material agnostic and allows for multiscale feature classification ranging from atomic-scale crystals to heterointerfaces tomicroscale grain boundaries.
基金This work was carried out in the framework of the imec Core CMOS-Active Memory Program.T.R.acknowledges the support by Research Foundation-Flanders(FWO)for providing the funding via strategic basic research PhD fellowship(grant no.1SD4721).
文摘Restricted use of hazardous environmental chemicals is one important challenge that the semiconductor industry needs to face to improve its sustainability.Ovonic threshold switching(OTS)ternary compound materials used in memory selector devices contain As and Se.Engineering these elements out of these materials requires significant research effort.To facilitate this process,we performed systematic material screening for As/Se-free ternary materials,based on ab-initio simulations.To limit the large amount of possible chemical compositions to fewer promising candidates,we used physics-based material parameter filters like material stability,electronic properties,or change in polarizability.The OTS gauge concept is introduced as a computed parameter to estimate the probability of a material to show an OTS behavior.As a result,we identified 35 As/Se-free ternary alloy compositions for stand-alone OTS memory applications,as well as 12 compositions for RRAM selector applications.This work aims seeding the development of As/Se-free OTS materials.