Large igneous provinces（LIPs）generally refer to the different types of the igneous rocks,which intrude in a short time,ranging in area from 50000 to 100000 km;（Sheth,2007;Bryan et al.,2008）.While the mafic large
Machine learning interatomic potentials(MLIPs)have achieved remarkable accuracy on standard benchmarks,yet their ability to reproduce molecular kinetics,critical for reaction rate calculations,remains largely unexplor...Machine learning interatomic potentials(MLIPs)have achieved remarkable accuracy on standard benchmarks,yet their ability to reproduce molecular kinetics,critical for reaction rate calculations,remains largely unexplored.We introduce Landscape17,a dataset of complete kinetic transition networks(KTNs)for the six molecules of the rMD17 dataset,computed using hybrid-level density functional theory.Each KTN contains minima,transition states,and approximate steepest-descent paths,along with energies,forces,and Hessian eigenspectra at stationary points.We develop a comprehensive test suite to evaluate theMLIPs’ability to reproduce these reference landscapes and apply it to state-of-the-art architectures.Our results reveal limitations in current MLIPs:all models considered miss over half of the DFT transition state paths and generate stable unphysical structures throughout the potential energy surface.Data augmentation with pathway configurations improves reproduction of DFT potential energy surfaces,resulting in significant improvement in global kinetics.However,these models still produce many spurious stable structures,indicating that current MLIP architectures face underlying challenges in capturing the topology of molecular potential energy surfaces.The Landscape17 benchmark provides a straightforward,lightweight,but demanding test of MLIPs for kinetic applications,and we propose this test for validation of next-generation MLIPs targeting reaction discovery and rate prediction.展开更多
Emerging machine learning interatomic potentials(MLIPs)offer a promising solution for large-scale accurate material simulations,but stringent tests related to the description of vibrational dynamics in molecular cryst...Emerging machine learning interatomic potentials(MLIPs)offer a promising solution for large-scale accurate material simulations,but stringent tests related to the description of vibrational dynamics in molecular crystals remain scarce.Here,we develop a general MLIP by leveraging the graph neural network-based MACE architecture and active-learning strategies to accurately capture vibrational dynamics across a range of polyacene-based molecular crystals,namely naphthalene,anthracene,tetracene and pentacene.Through careful error propagation,we show that these potentials are accurate and enable the study of anharmonic vibrational features,vibrational lifetimes,and vibrational coupling.In particular,we investigate large-scale host-vip systems based on these molecular crystals,showing the capacity of molecular-dynamics-based techniques to explain and quantify vibrational coupling between host and vip nuclear motion.Our results establish a framework for understanding vibrational signatures in large-scale complex molecular systems and thus represent an important step for engineering vibrational interactions in molecular environments.展开更多
Machine learning interatomic potentials(MLIPs)have introduced a new paradigm for atomic simulations.Recent advancements have led to universal MLIPs(uMLIPs)that are pre-trained on diverse datasets,providing opportuniti...Machine learning interatomic potentials(MLIPs)have introduced a new paradigm for atomic simulations.Recent advancements have led to universal MLIPs(uMLIPs)that are pre-trained on diverse datasets,providing opportunities for universal force fields and foundational machine learning models.However,their performance in extrapolating to out-of-distribution complex atomic environments remains unclear.In this study,we highlight a consistent potential energy surface(PES)softening effect in three uMLIPs:M3GNet,CHGNet,and MACE-MP-0,which is characterized by energy and force underprediction in atomic-modeling benchmarks including surfaces,defects,solid-solution energetics,ion migration barriers,phonon vibration modes,and general high-energy states.The PES softening behavior originates primarily from the systematically underpredicted PES curvature,which derives from the biased sampling of near-equilibrium atomic arrangements in uMLIP pre-training datasets.Our findings suggest that a considerable fraction of uMLIP errors are highly systematic,and can therefore be efficiently corrected.We argue for the importance of a comprehensive materials dataset with improved PES sampling for next-generation foundational MLIPs.展开更多
Weintroduce acomputational framework leveraging universal machine learning interatomic potentials(MLIPs)to dramatically accelerate the calculation of photoluminescence(PL)spectra of atomic or molecular emitters with a...Weintroduce acomputational framework leveraging universal machine learning interatomic potentials(MLIPs)to dramatically accelerate the calculation of photoluminescence(PL)spectra of atomic or molecular emitters with ab initio accuracy.By replacing the costly density functional theory(DFT)computation of phonon modes with much faster MLIP phonon mode calculations,our approach achieves speed improvements exceeding an order of magnitude with minimal precision loss.We benchmark the approach using a dataset comprising ab initio emission spectra of 791 color centers spanning various types of crystal point defects in different charge and magnetic states.The method is also applied to a molecular emitter adsorbed on a hexagonal boron nitride surface.Across all the systems,we find excellent agreement for both the Huang-Rhys factor and the PL lineshapes.This application of universal MLIPs bridges the gap between computational efficiency and spectroscopic fidelity,opening pathways to high-throughput screening of defect-engineered materials.Ourwork not only demonstrates accelerated calculation of PL spectra with DFT accuracy,but also makes such calculations tractable for more complex materials.展开更多
Machine learning interatomic potentials(MLIPs)often neglect long-range interactions,such as electrostatic and dispersion forces.In this work,we introduce a straightforward and efficient method to account for long-rang...Machine learning interatomic potentials(MLIPs)often neglect long-range interactions,such as electrostatic and dispersion forces.In this work,we introduce a straightforward and efficient method to account for long-range interactions by learning a hidden variable from local atomic descriptors and applying an Ewald summation to this variable.We demonstrate that in systems including charged and polar molecular dimers,bulk water,and water-vapor interface,standard short-ranged MLIPs can lead to unphysical predictions even when employing message passing.The long-range models effectively eliminate these artifacts,with only about twice the computational cost of short-range MLIPs.展开更多
Adversarial approaches,which intentionally challenge machine learning models by generating difficult examples,are increasingly being adopted to improve machine learning interatomic potentials(MLIPs).While already prov...Adversarial approaches,which intentionally challenge machine learning models by generating difficult examples,are increasingly being adopted to improve machine learning interatomic potentials(MLIPs).While already providing great practical value,little is known about the actual prediction errors of MLIPs on adversarial structures and whether these errors can be controlled.We propose the Calibrated Adversarial Geometry Optimization(CAGO)algorithm to discover adversarial structures with userassigned errors.Through uncertainty calibration,the estimated uncertainty of MLIPs is unified with real errors.By performing geometry optimization for calibrated uncertainty,we reach adversarial structures with the user-assigned target MLIP prediction error.Integrating with active learning pipelines,we benchmark CAGO,demonstrating stable MLIPs that systematically converge structural,dynamical,and thermodynamical properties for liquid water and water adsorption in a metal-organic framework within only hundreds of training structures,where previously many thousands were typically required.展开更多
Extended defects such as dislocation networks and general grain boundaries are ubiquitous in metals,and accurate modeling these extensive defects is crucial to elucidate their deformation mechanisms.However,existing m...Extended defects such as dislocation networks and general grain boundaries are ubiquitous in metals,and accurate modeling these extensive defects is crucial to elucidate their deformation mechanisms.However,existing machine learning interatomic potentials(MLIPs)often fall short in adequately describing these defects,as their large characteristic scales exceed the computational limits of firstprinciples calculations.To address this challenge,wepresent acomputational frameworkcombining a defect genome constructed via empirical interatomic potential-guided sampling,with an automated reconstruction technique that enables accurate first-principles modeling of general defects by converting atomic clusters into periodic configurations.The effectiveness of this approach was validated through simulations of nanoindentation,tensile deformation,and fracture in BCC tungsten.This framework enhances the modeling accuracy of extended defects in crystalline materials and provides a robust foundation for advancing MLIP development by leveraging defect genomes strategically.展开更多
Machine learning interatomic potentials(MLIPs)have substantially advanced atomistic simulations in materials science and chemistry by balancing accuracy and computational efficiency.While leading MLIPs rely on represe...Machine learning interatomic potentials(MLIPs)have substantially advanced atomistic simulations in materials science and chemistry by balancing accuracy and computational efficiency.While leading MLIPs rely on representing atomic environments using spherical tensors,Cartesian representations offer potential advantages in simplicity and efficiency.Here,we introduce the Cartesian Atomic Moment Potential(CAMP),an approach to building MLIPs entirely in Cartesian space.CAMP constructs atomic moment tensors from neighboring atoms and employs tensor products to incorporate higher body-order interactions,providing a complete description of local atomic environments.Integrated into a graph neural network(GNN)framework,CAMP enables physically motivated,systematically improvable potentials.The model demonstrates excellent performance across diverse systems,including periodic structures,small organic molecules,and two-dimensional materials,achieving accuracy,efficiency,and stability in molecular dynamics simulations that rival or surpass current leadingmodels.CAMPprovides apowerful tool for atomistic simulations to accelerate materials understanding and discovery.展开更多
Modeling the response of material and chemical systems to electric fields remains a longstanding challenge.Machine learning interatomic potentials(MLIPs)offer an efficient and scalable alternative to quantum mechanica...Modeling the response of material and chemical systems to electric fields remains a longstanding challenge.Machine learning interatomic potentials(MLIPs)offer an efficient and scalable alternative to quantum mechanical methods,but do not by themselves incorporate electrical response.Here,we show that polarization and Born effective charge(BEC)tensors can be directly extracted from longrange MLIPs within the Latent Ewald Summation(LES)framework,solely by learning from energy and force data.Using this approach,we predict the infrared spectra of bulk water under zero or finite external electric fields,ionic conductivities of high-pressure superionic ice,and the phase transition and hysteresis in ferroelectric PbTiO_(3)perovskite.This work thus extends the capability of MLIPs to predict electrical response–without training on charges or polarization or BECs–and enables accurate modeling of electric-field-driven processes in diverse systems at scale.展开更多
We propose a strategy for generating unbiased and systematically extendable training data for machine learning interatomic potentials(MLIP)for multicomponent alloys,called Automated Small SYmmetric Structure Training ...We propose a strategy for generating unbiased and systematically extendable training data for machine learning interatomic potentials(MLIP)for multicomponent alloys,called Automated Small SYmmetric Structure Training or ASSYST.Based on exploring the full space of random crystal structures with space groups,it facilitates the construction of training sets for MLIPs in an automatic way without prior knowledge of the material in question.The advantages of this approach are that only cells consisting of few atoms(≈10)are needed for the DFT training set,and the size and completeness of the data set can be systematically controlled with very few parameters.We validate that potentials fitted this way can accurately describe a wide range of binary and ternary phases,random alloys,as well as point and extended defects,that have not been part of the training set.Finally,we estimate the binary phase diagrams with good experimental agreement.We demonstrate that the overall excellent performance is not a coincidence,but a consequence of the extensive sampling in phase space of ASSYST.Overall,this means that ASSYST will enable the largely autonomous generation of highquality DFT reference data and MLIPs.展开更多
Understanding the mechanisms of oxygen anion electrochemical reactions within crystals has long perplexed electrochemical scientists and hindered the structural design and composition optimization of Li-ion cathode ma...Understanding the mechanisms of oxygen anion electrochemical reactions within crystals has long perplexed electrochemical scientists and hindered the structural design and composition optimization of Li-ion cathode materials.Machine learning interatomic potentials(MLIP)are transforming the landscape by enabling high-accuracy atomistic modeling on a large scale in materials science and chemistry.The diversity and comprehensiveness of the dataset are fundamental to building a high-accuracy MLIP.Here,we constructed a Li_(1.2–x)Mn_(0.6)Ni_(0.2O_(2))(x=0–1.04)dataset that includes over 15,000 chemical non-equilibrium and chemical equilibrium structures.Using this dataset,we trained an MLIP model(multistate equilibrium potential,named MSEP)with test accuracies of 0.008 eV/atom and 0.153 eV/Åfor energy and force,respectively.Through MSEP-MD simulations,we identify a kinetically viable O-redox mechanism in which the formation of transient interlayer O_(2)^(2−),O_(2)^(−)or O_(3)^(−)intermediates drives out-of-plane Mn and Ni migration,resulting in O_(2)molecules forming within the bulk structure.O3−intermediates have a certain ability to capture O_(2),which may help alleviate the formation of lattice O_(2).展开更多
基金supported by the Chinese National Basic Research 973 Program (2011CB403105)the National Geological Survey Program (121201010000150014)
文摘Large igneous provinces（LIPs）generally refer to the different types of the igneous rocks,which intrude in a short time,ranging in area from 50000 to 100000 km;（Sheth,2007;Bryan et al.,2008）.While the mafic large
基金performed using resources provided by the Cambridge Service for Data Driven Discovery (www.csd3.cam.ac.uk), provided by Dell EMC and Intel using Tier-2 funding from the Engineering and Physical Sciences Research Council(capital grant EP/T022159/1), and DiRAC funding from the Science and Technology Facilities Council (www.dirac.ac.uk). V.C. acknowledges the computational resources obtained through the University of Cambridge EPSRC Core Equipment Award (EP/X034712/1) and EPSRC IAA award number G116766. The authors thank Gábor Csányi for useful discussions.
文摘Machine learning interatomic potentials(MLIPs)have achieved remarkable accuracy on standard benchmarks,yet their ability to reproduce molecular kinetics,critical for reaction rate calculations,remains largely unexplored.We introduce Landscape17,a dataset of complete kinetic transition networks(KTNs)for the six molecules of the rMD17 dataset,computed using hybrid-level density functional theory.Each KTN contains minima,transition states,and approximate steepest-descent paths,along with energies,forces,and Hessian eigenspectra at stationary points.We develop a comprehensive test suite to evaluate theMLIPs’ability to reproduce these reference landscapes and apply it to state-of-the-art architectures.Our results reveal limitations in current MLIPs:all models considered miss over half of the DFT transition state paths and generate stable unphysical structures throughout the potential energy surface.Data augmentation with pathway configurations improves reproduction of DFT potential energy surfaces,resulting in significant improvement in global kinetics.However,these models still produce many spurious stable structures,indicating that current MLIP architectures face underlying challenges in capturing the topology of molecular potential energy surfaces.The Landscape17 benchmark provides a straightforward,lightweight,but demanding test of MLIPs for kinetic applications,and we propose this test for validation of next-generation MLIPs targeting reaction discovery and rate prediction.
基金support from the Cluster of Excellence“CUI:Advanced Imaging of Matter”—EXC 2056—project ID 390715994BiGmax,the Max Planck Society Research Network on Big-Data-Driven Materials-Science and the Max Planck-New York City Center for Non-Equilibrium Quantum Phenomena.The Flatiron Institute is a division of the Simons Foundation+1 种基金We also acknowledge support from the European Research Council MSCA-ITN TIMES under grant agreement 101118915S.S.and P.L.acknowledge support from the UFAST International Max Planck Research School.
文摘Emerging machine learning interatomic potentials(MLIPs)offer a promising solution for large-scale accurate material simulations,but stringent tests related to the description of vibrational dynamics in molecular crystals remain scarce.Here,we develop a general MLIP by leveraging the graph neural network-based MACE architecture and active-learning strategies to accurately capture vibrational dynamics across a range of polyacene-based molecular crystals,namely naphthalene,anthracene,tetracene and pentacene.Through careful error propagation,we show that these potentials are accurate and enable the study of anharmonic vibrational features,vibrational lifetimes,and vibrational coupling.In particular,we investigate large-scale host-vip systems based on these molecular crystals,showing the capacity of molecular-dynamics-based techniques to explain and quantify vibrational coupling between host and vip nuclear motion.Our results establish a framework for understanding vibrational signatures in large-scale complex molecular systems and thus represent an important step for engineering vibrational interactions in molecular environments.
基金funded by the U.S.Department of Energy,Office of Science,Office of Basic Energy Sciences,Materials Sciences and Engineering Division under Contract No.DE-AC0205CH11231(Materials Project program KC23MP)supported by the computational resources provided by the Extreme Science and Engineering Discovery Environment(XSEDE),supported by National Science Foundation grant number ACI1053575+1 种基金the National Energy Research Scientific Computing Center(NERSC),a U.S.Department of Energy Office of Science User Facility located at Lawrence Berkeley National Laboratoryand the Swift Cluster resource provided by the National Renewable Energy Laboratory(NREL).
文摘Machine learning interatomic potentials(MLIPs)have introduced a new paradigm for atomic simulations.Recent advancements have led to universal MLIPs(uMLIPs)that are pre-trained on diverse datasets,providing opportunities for universal force fields and foundational machine learning models.However,their performance in extrapolating to out-of-distribution complex atomic environments remains unclear.In this study,we highlight a consistent potential energy surface(PES)softening effect in three uMLIPs:M3GNet,CHGNet,and MACE-MP-0,which is characterized by energy and force underprediction in atomic-modeling benchmarks including surfaces,defects,solid-solution energetics,ion migration barriers,phonon vibration modes,and general high-energy states.The PES softening behavior originates primarily from the systematically underpredicted PES curvature,which derives from the biased sampling of near-equilibrium atomic arrangements in uMLIP pre-training datasets.Our findings suggest that a considerable fraction of uMLIP errors are highly systematic,and can therefore be efficiently corrected.We argue for the importance of a comprehensive materials dataset with improved PES sampling for next-generation foundational MLIPs.
基金funding from the Horizon Europe MSCA Doctoral network grant n.101073486, EUSpecLabfunded by the European Union, and from the Novo Nordisk Foundation Data Science Research Infrastructure 2022 Grant: A high-performance computing infrastructure for data-driven research on sustainable energy materials, Grant no. NNF22OC0078009+1 种基金F.N. has received funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie Grant Agreement No. 899987K.S.T. is a Villum Investigator supported by VILLUM FONDEN (grant no. 37789).
文摘Weintroduce acomputational framework leveraging universal machine learning interatomic potentials(MLIPs)to dramatically accelerate the calculation of photoluminescence(PL)spectra of atomic or molecular emitters with ab initio accuracy.By replacing the costly density functional theory(DFT)computation of phonon modes with much faster MLIP phonon mode calculations,our approach achieves speed improvements exceeding an order of magnitude with minimal precision loss.We benchmark the approach using a dataset comprising ab initio emission spectra of 791 color centers spanning various types of crystal point defects in different charge and magnetic states.The method is also applied to a molecular emitter adsorbed on a hexagonal boron nitride surface.Across all the systems,we find excellent agreement for both the Huang-Rhys factor and the PL lineshapes.This application of universal MLIPs bridges the gap between computational efficiency and spectroscopic fidelity,opening pathways to high-throughput screening of defect-engineered materials.Ourwork not only demonstrates accelerated calculation of PL spectra with DFT accuracy,but also makes such calculations tractable for more complex materials.
文摘Machine learning interatomic potentials(MLIPs)often neglect long-range interactions,such as electrostatic and dispersion forces.In this work,we introduce a straightforward and efficient method to account for long-range interactions by learning a hidden variable from local atomic descriptors and applying an Ewald summation to this variable.We demonstrate that in systems including charged and polar molecular dimers,bulk water,and water-vapor interface,standard short-ranged MLIPs can lead to unphysical predictions even when employing message passing.The long-range models effectively eliminate these artifacts,with only about twice the computational cost of short-range MLIPs.
基金supported by the Research Council of Norway through the Centre of Excellence Hylleraas Centre for Quantum Molecular Sciences(Grant 262695)the Young Researcher Talent grants 344993 and 354100+2 种基金We acknowledge the EuroHPC Joint Undertaking for awarding this project access to the EuroHPC supercomputer LUMI,hosted by CSC(Finland)and the LUMI consortium through a EuroHPC Regular Access call(Grants EHPC-REG-2023R02-088,EHPC-REG-2023R03-146)Support was also received from the Centre for Advanced Study in Oslo,Norway,which funded and hosted the SLB Young CAS Fellow research project during the academic year of 23/24 and 24/25Part of the simulations were performed on resources provided by Sigma2—the Norwegian National Infrastructure for High-Performance Computing and Data Storage(grant numbers NN4654K and NS4654K).
文摘Adversarial approaches,which intentionally challenge machine learning models by generating difficult examples,are increasingly being adopted to improve machine learning interatomic potentials(MLIPs).While already providing great practical value,little is known about the actual prediction errors of MLIPs on adversarial structures and whether these errors can be controlled.We propose the Calibrated Adversarial Geometry Optimization(CAGO)algorithm to discover adversarial structures with userassigned errors.Through uncertainty calibration,the estimated uncertainty of MLIPs is unified with real errors.By performing geometry optimization for calibrated uncertainty,we reach adversarial structures with the user-assigned target MLIP prediction error.Integrating with active learning pipelines,we benchmark CAGO,demonstrating stable MLIPs that systematically converge structural,dynamical,and thermodynamical properties for liquid water and water adsorption in a metal-organic framework within only hundreds of training structures,where previously many thousands were typically required.
基金sponsored by Nederlandse Organisatie voor WetenschappelijkOnderzoek(The Netherlands Organization for Scientific Research,NWO)domain Science for the use of supercomputer facilities.
文摘Extended defects such as dislocation networks and general grain boundaries are ubiquitous in metals,and accurate modeling these extensive defects is crucial to elucidate their deformation mechanisms.However,existing machine learning interatomic potentials(MLIPs)often fall short in adequately describing these defects,as their large characteristic scales exceed the computational limits of firstprinciples calculations.To address this challenge,wepresent acomputational frameworkcombining a defect genome constructed via empirical interatomic potential-guided sampling,with an automated reconstruction technique that enables accurate first-principles modeling of general defects by converting atomic clusters into periodic configurations.The effectiveness of this approach was validated through simulations of nanoindentation,tensile deformation,and fracture in BCC tungsten.This framework enhances the modeling accuracy of extended defects in crystalline materials and provides a robust foundation for advancing MLIP development by leveraging defect genomes strategically.
基金supported by the National Science Foundation under Grant No.2316667supported by the Center for HPC at the University of Electronic Science and Technology of China.This work also uses computational resources provided by the Research Computing Data Core at the University of Houston.
文摘Machine learning interatomic potentials(MLIPs)have substantially advanced atomistic simulations in materials science and chemistry by balancing accuracy and computational efficiency.While leading MLIPs rely on representing atomic environments using spherical tensors,Cartesian representations offer potential advantages in simplicity and efficiency.Here,we introduce the Cartesian Atomic Moment Potential(CAMP),an approach to building MLIPs entirely in Cartesian space.CAMP constructs atomic moment tensors from neighboring atoms and employs tensor products to incorporate higher body-order interactions,providing a complete description of local atomic environments.Integrated into a graph neural network(GNN)framework,CAMP enables physically motivated,systematically improvable potentials.The model demonstrates excellent performance across diverse systems,including periodic structures,small organic molecules,and two-dimensional materials,achieving accuracy,efficiency,and stability in molecular dynamics simulations that rival or surpass current leadingmodels.CAMPprovides apowerful tool for atomistic simulations to accelerate materials understanding and discovery.
基金funding from the BIDMaP Postdoctoral Fellowship.
文摘Modeling the response of material and chemical systems to electric fields remains a longstanding challenge.Machine learning interatomic potentials(MLIPs)offer an efficient and scalable alternative to quantum mechanical methods,but do not by themselves incorporate electrical response.Here,we show that polarization and Born effective charge(BEC)tensors can be directly extracted from longrange MLIPs within the Latent Ewald Summation(LES)framework,solely by learning from energy and force data.Using this approach,we predict the infrared spectra of bulk water under zero or finite external electric fields,ionic conductivities of high-pressure superionic ice,and the phase transition and hysteresis in ferroelectric PbTiO_(3)perovskite.This work thus extends the capability of MLIPs to predict electrical response–without training on charges or polarization or BECs–and enables accurate modeling of electric-field-driven processes in diverse systems at scale.
基金funding from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) through the Collaborative Research Center 1394 (SFB 1394, No. 409476157) and Project No. 405621160MP would like to thank Prince Matthews for setting up the hcp grain boundaries, Sarath Menon for providing support for CALPHY34Bengt Hallstedt for providing plots of the Mg/Ca and Al/Ca phase diagrams from his assessments, Chad Sinclair together with SFB 1394 for funding a research stay at UBC Vancouver where part of this work was conducted, as well Mira Todorova and Ali Tehranchi for fruitful discussions and Ralf Drautz for critical reading of the manuscript.
文摘We propose a strategy for generating unbiased and systematically extendable training data for machine learning interatomic potentials(MLIP)for multicomponent alloys,called Automated Small SYmmetric Structure Training or ASSYST.Based on exploring the full space of random crystal structures with space groups,it facilitates the construction of training sets for MLIPs in an automatic way without prior knowledge of the material in question.The advantages of this approach are that only cells consisting of few atoms(≈10)are needed for the DFT training set,and the size and completeness of the data set can be systematically controlled with very few parameters.We validate that potentials fitted this way can accurately describe a wide range of binary and ternary phases,random alloys,as well as point and extended defects,that have not been part of the training set.Finally,we estimate the binary phase diagrams with good experimental agreement.We demonstrate that the overall excellent performance is not a coincidence,but a consequence of the extensive sampling in phase space of ASSYST.Overall,this means that ASSYST will enable the largely autonomous generation of highquality DFT reference data and MLIPs.
基金supported by the National Key R&D Program of China(2022YFB3807200)the National Natural Science Foundation of China,NSFC(22133005,22403103)+2 种基金the Project funded by China Postdoctoral Science Foundation(2022M723276 and GZB20230793)Sponsored by the Shanghai Sailing Program(23YF1454900)and the Shanghai Post-doctoral Excellence Program(2022660).
文摘Understanding the mechanisms of oxygen anion electrochemical reactions within crystals has long perplexed electrochemical scientists and hindered the structural design and composition optimization of Li-ion cathode materials.Machine learning interatomic potentials(MLIP)are transforming the landscape by enabling high-accuracy atomistic modeling on a large scale in materials science and chemistry.The diversity and comprehensiveness of the dataset are fundamental to building a high-accuracy MLIP.Here,we constructed a Li_(1.2–x)Mn_(0.6)Ni_(0.2O_(2))(x=0–1.04)dataset that includes over 15,000 chemical non-equilibrium and chemical equilibrium structures.Using this dataset,we trained an MLIP model(multistate equilibrium potential,named MSEP)with test accuracies of 0.008 eV/atom and 0.153 eV/Åfor energy and force,respectively.Through MSEP-MD simulations,we identify a kinetically viable O-redox mechanism in which the formation of transient interlayer O_(2)^(2−),O_(2)^(−)or O_(3)^(−)intermediates drives out-of-plane Mn and Ni migration,resulting in O_(2)molecules forming within the bulk structure.O3−intermediates have a certain ability to capture O_(2),which may help alleviate the formation of lattice O_(2).
