Designing refractory high-entropy alloys(RHEAs)for high-temperature(HT)applications is an outstanding challenge given the vast possible composition space,which contains billions of candidates,and the need to optimize ...Designing refractory high-entropy alloys(RHEAs)for high-temperature(HT)applications is an outstanding challenge given the vast possible composition space,which contains billions of candidates,and the need to optimize across multiple objectives.Here,we present an approach that accelerates the discovery of RHEA compositions with superior strength and ductility by integrating machine learning(ML),genetic search,cluster analysis,and experimental design.We iteratively synthesize and characterize 24 predicted compositions after six feedback loops.Four compositions show outstanding combinations of HT yield strength and room-temperature(RT)ductility spanning the ranges of 714–1061 MPa and 17.2%–50.0%fracture strain,respectively.We identify an attractive alloy system,ZrNbMoHfTa,particularly the composition Zr_(0.13)Nb_(0.27)Mo_(0.26)Hf_(0.13)Ta_(0.21),which demonstrates a yield approaching 940 MPa at 1200℃ and favorable RT ductility with 17.2%fracture strain.The high yield strength at 1200℃ exceeds that reported for RHEAs,with 1200℃ exceeding the service temperature limit for nickel(Ni)-based superalloys.Our ML-based approach makes it possible to rapidly optimize multiple properties for materials design,thus overcoming the common problems of limited data and a vast composition space in complex materials systems while satisfying multiple objectives.展开更多
The state of the art for data-driven creep rupture life predictions incorporates microstructural and deformation characteristics into machine learning.However,the microstructures and deformation mechanisms for unknown...The state of the art for data-driven creep rupture life predictions incorporates microstructural and deformation characteristics into machine learning.However,the microstructures and deformation mechanisms for unknown alloys are inaccessible and uncertain before experiments are carried out,and therefore prevents extrapolations of the learned models.展开更多
Machine learning assisted design of materials is so far based on features selected by considering the accuracy of model predictions,and those features do not necessarily ensure a high efficiency in searching for new m...Machine learning assisted design of materials is so far based on features selected by considering the accuracy of model predictions,and those features do not necessarily ensure a high efficiency in searching for new materials.Here we estimate the efficiency of active learning loop by resampling method using available data as an alternative criterion for selection.The selected features allow an optimization of targeted property with as few new experiments as possible.Input those features into machine learning,we synthesized new high entropy alloys(HEAs)with strengths 2.8–3.0 GPa within five experimental iterations.The alloy AlVCrCoNiMo is found to possess compressive specific yield strengths of 397,144 and 105(MPa cm^(3))/g at 25,800 and 900℃,respectively.The specific yield strength of AlVCrCoNiMo alloy at 800℃is about twice that of the commercial Inconel 718 and the typical refractory HEA of VNbMoTaW.A unique microstructure consisting of multi-scale hierarchical B2 precipitates with coherent interfaces to the BCC matrix strengthens the alloy.Our strategy of maximizing active learning efficiency provides a recipe for selecting features that accelerate the optimization of targeted property.展开更多
Accelerating materials discovery crucially relies on strategies that efficiently sample the search space to label a pool of unlabeled data.This is important if the available labeled data sets are relatively small comp...Accelerating materials discovery crucially relies on strategies that efficiently sample the search space to label a pool of unlabeled data.This is important if the available labeled data sets are relatively small compared to the unlabeled data pool.Active learning with efficient sampling methods provides the means to guide the decision making to minimize the number of experiments or iterations required to find targeted properties.We review here different sampling strategies and show how they are utilized within an active learning loop in materials science.展开更多
Functional stability of superelasticity is crucial for practical applications of shape memory alloys.It is degraded by a Lüders-like deformation with elevated local stress concentration under tensile load.By incr...Functional stability of superelasticity is crucial for practical applications of shape memory alloys.It is degraded by a Lüders-like deformation with elevated local stress concentration under tensile load.By increasing the degree of solute supersaturation and applying appropriate thermomechanical treatments,a Ti-Ni alloy with nanocrystallinity and dispersed nanoprecipitates is obtained.In contrast to conventional Ti-Ni alloys,the superelasticity in the target alloy is accompanied by homogeneous deformation due to the sluggish stress-induced martensitic transformation.The alloy thus shows a fully recoverable strain of 6%under tensile stress over 1 GPa and a large adiabatic temperature decrease of 13.1 K under tensile strain of 4.5%at room temperature.Moreover,both superelasticity and elastocaloric effect exhibit negligible degradation in response to applied strain of 4%during cycling.We attribute the improved functional stability to low dislocation activity resulting from the suppression of localized deformation and the combined strengthening effect of nanocrystalline structure and nanoprecipitates.Thus,the design of such a microstructure enabling homogeneous deformation provides a recipe for stable superelasticity and elastocaloric effect.展开更多
There is growing interest in applying machine learning techniques in the field of materials science.However,the interpretation and knowledge extracted from machine learning models is a major concern,particularly as fo...There is growing interest in applying machine learning techniques in the field of materials science.However,the interpretation and knowledge extracted from machine learning models is a major concern,particularly as formulating an explicit model that provides insight into physics is the goal of learning.In the present study,we propose a framework that utilizes the filtering ability of feature engineering,in conjunction with symbolic regression to extract explicit,quantitative expressions for the band gap energy from materials data.We propose enhancements to genetic programming with dimensional consistency and artificial constraints to improve the search efficiency of symbolic regression.We show how two descriptors attributed to volumetric and electronic factors,from 32 possible candidates,explicitly express the band gap energy of Na Cl-type compounds.Our approach provides a basis to capture underlying physical relationships between materials descriptors and target properties.展开更多
The singular change of the order parameter at the first order martensitic transformation(MT)temperature restricts the caloric response to a narrow temperature range.Here the MT is tuned into a sluggish strain glass tr...The singular change of the order parameter at the first order martensitic transformation(MT)temperature restricts the caloric response to a narrow temperature range.Here the MT is tuned into a sluggish strain glass transition by defect doping and a large elastocaloric effect appears in a wide temperature range.Moreover,an inverse elastocaloric effect is observed in the strain glass alloy with history of zerofield cooling and is attributed to the slow dynamics of the nanodomains in response to the external stress.This study offers a design recipe to expand the temperature range for good elastocaloric effect.展开更多
An issue of current interest in the use of machine learning models to predict compositions of materials is their reliability in predicting outcomes with elements not included in the training data.We show that the phas...An issue of current interest in the use of machine learning models to predict compositions of materials is their reliability in predicting outcomes with elements not included in the training data.We show that the phase diagram of the ceramic(Ba_(1-x-y)Ca_(x)Sr_(y))(Ti_(1-u-v-w)Zr_(u)Sn_(v)Hf_(w))O_(3)can be accurately predicted if the feature values of unknown elements do not exceed the range of values for existing elements in the training data.In particular,we employ physical features as descriptors and compositions as weights to show that by excluding an element,such as Zr,Sn or Hf from the training set and treating it as an unknown element,the machine learning model accurately predicts the property only if the feature values of the unknown element does not exceed the range of values of existing elements in the training set.By adding a small amount of data for the unknown element restores the prediction accuracy.We demonstrate this for BaTiO_(3)ceramics doped with rare earth elements where the prediction accuracy is restored if the physical feature space is suitably enlarged with training data.The prediction error increases with the Euclidean distance of the testing sample relative to the nearest training sample in the physical feature space.Our work provides an effective strategy for extending machine learning models for material compositions beyond the scope of available data.展开更多
Finding high temperature superconductors(HTS)has been a continuing challenge due to the difficulty in predicting the transition temperature(Tc)of superconduc-tors.Recently,the efficiency of predicting Tc has been grea...Finding high temperature superconductors(HTS)has been a continuing challenge due to the difficulty in predicting the transition temperature(Tc)of superconduc-tors.Recently,the efficiency of predicting Tc has been greatly improved via ma-chine learning(ML).Unfortunately,prevailing ML models have not shown adequate generalization ability to find new HTS,yet.In this work,a graph neural network model is trained to predict the maximal Tc(Tc max)of various materials.Our model reveals a close connection between Tc max and chemical bonds.It sug-gests that shorter bond lengths are favored by high Tc,which is in coherence with previous domain knowledge.More importantly,it also indicates that chemical bonds consisting of some specific chemical elements are responsible for high Tc,which is new even to the human experts.It can provide a convenient guidance to the materials scientists in search of HTS.展开更多
Recent advancements in machine learning(ML)have revolutionized the field of high-performance materials design.However,developing robust ML models to decipher intricate structure-property relationships in materials rem...Recent advancements in machine learning(ML)have revolutionized the field of high-performance materials design.However,developing robust ML models to decipher intricate structure-property relationships in materials remains challenging,primarily due to the limited availability of labeled datasets with well-characterized crystal structures.This is particularly pronounced in materials where functional properties are closely intertwined with their crystallographic symmetry.Weintroduce a selfsupervised probabilistic model(SSPM)that autonomously learns unbiased atomic representations and the likelihood of compounds with given crystal structures,utilizing solely the existing crystal structure data from materials databases.SSPM significantly enhances the performance of downstream ML models by efficient atomic representations and accurately captures the probabilistic relationships between composition and crystal structure.We showcase SSPM’s capability by discovering shapememory alloys(SMAs).Amongst the top 50 predictions,23 have been confirmed as SMAs either experimentally or theoretically,and a previously unknown SMA candidate,MgAu,has been identified.展开更多
Machine learning has been widely used to guide the search for new materials by learning the patterns underlying available data.However,the pure prediction-oriented search is often biased to interpolation due to the li...Machine learning has been widely used to guide the search for new materials by learning the patterns underlying available data.However,the pure prediction-oriented search is often biased to interpolation due to the limited data in a large unexplored space.Here we present a sampling framework towards extrapolation,that integrates unsupervised clustering,interpretable analysis,and similarity evaluation to sample target candidates with improved properties from a vast search space.展开更多
This special issue of MGE advances focuses on the revolutionary impact of machine learning(ML)on materials science.As we navigate the threshold of a new era in scientific innovation,this issue collates a series of res...This special issue of MGE advances focuses on the revolutionary impact of machine learning(ML)on materials science.As we navigate the threshold of a new era in scientific innovation,this issue collates a series of research articles that epitomize machine learning as a foundational pillar in materials science and engineering.The synergy between ML and conventional materials science methodologies not only accelerates the discovery of novel materials but also refines the prediction of material properties and streamlines manufacturing processes.These advances offer unparalleled opportunities for technological progress and sustainability.We,as the vip editors,are excited to present these contributions that introduce new methodologies and enhance our understanding of material behavior through the prism of advanced analytics and computational power.展开更多
Martensitic transformation plays a pivotal role in strengthening and hardening of steels,yet an accu-rate interatomic potential for a comprehensive description of the martensitic phase formation in Fe-C alloys is lack...Martensitic transformation plays a pivotal role in strengthening and hardening of steels,yet an accu-rate interatomic potential for a comprehensive description of the martensitic phase formation in Fe-C alloys is lacking.Herein,we developed a deep learning-based interatomic potential to perform molecu-lar dynamics(MD)simulations to study the martensitic phase transformation across a range of carbon(C)concentrations.The results revealed that an increased C concentration leads to a suppressed phase boundary movement and a decelerated phase transformation rate.To overcome the timescale limitations inherent in MD simulations,metadynamics sampling was employed to accelerate the simulations of C dif-fusion.We found that C atoms tend to cluster at distances equivalent to the lattice parameter of Fe with the same sublattice occupation,leading to local lattice tetragonality.Such C-ordered structures effectively inhibit dislocation movement and enhance strength.The stress field induced by dislocations facilitates a higher degree of ordering,and the formation of C-ordered structures was identified as a potentially cru-cial strengthening mechanism for martensitic steels.The consistency between our simulation results and reported experimental observations underscores the effectiveness of the developed DP model in simu-lating martensitic phase transformation in Fe-C alloys,providing detailed insights into the mechanisms underlying this process.This work not only advances the understanding of martensitic phase transforma-tions in Fe-C alloys but also establishes a powerful computational framework for designing steels with optimized mechanical properties through the precise control of carbon ordering and dislocation behavior.展开更多
One of the main challenges in materials discovery is efficiently exploring the vast search space for targeted properties as approaches that rely on trial-and-error are impractical.We review how methods from the inform...One of the main challenges in materials discovery is efficiently exploring the vast search space for targeted properties as approaches that rely on trial-and-error are impractical.We review how methods from the information sciences enable us to accelerate the search and discovery of new materials.In particular,active learning allows us to effectively navigate the search space iteratively to identify promising candidates for guiding experiments and computations.The approach relies on the use of uncertainties and making predictions from a surrogate model together with a utility function that prioritizes the decision making process on unexplored data.We discuss several utility functions and demonstrate their use in materials science applications,impacting both experimental and computational research.We summarize by indicating generalizations to multiple properties and multifidelity data,and identify challenges,future directions and opportunities in the emerging field of materials informatics.展开更多
Data provides a foundation for machine learning,which has accelerated data-driven materials design.The scientific literature contains a large amount of high-quality,reliable data,and automatically extracting data from...Data provides a foundation for machine learning,which has accelerated data-driven materials design.The scientific literature contains a large amount of high-quality,reliable data,and automatically extracting data from the literature continues to be a challenge.We propose a natural language processing pipeline to capture both chemical composition and property data that allows analysis and prediction of superalloys.Within 3 h,2531 records with both composition and property are extracted from 14,425 articles,coveringγ′solvus temperature,density,solidus,and liquidus temperatures.A data-driven model forγ′solvus temperature is built to predict unexplored Co-based superalloys with highγ′solvus temperatures within a relative error of 0.81%.We test the predictions via synthesis and characterization of three alloys.A web-based toolkit as an online open-source platform is provided and expected to serve as the basis for a general method to search for targeted materials using data extracted from the literature.展开更多
Atomic simulations provide an effective means to understand the underlying physics of structural phase transformations.However,this remains a challenge for certain allotropic metals due to the failure of classical int...Atomic simulations provide an effective means to understand the underlying physics of structural phase transformations.However,this remains a challenge for certain allotropic metals due to the failure of classical interatomic potentials to represent the multitude of bonding.Based on machine-learning(ML)techniques,we develop a hybrid method in which interatomic potentials describing martensitic transformations can be learned with a high degree of fidelity from ab initio molecular dynamics simulations(AIMD).Using zirconium as a model system,for which an adequate semiempirical potential describing the phase transformation process is lacking,we demonstrate the feasibility and effectiveness of our approach.Specifically,the ML-AIMD interatomic potential correctly captures the energetics and structural transformation properties of zirconium as compared to experimental and density-functional data for phonons,elastic constants,as well as stacking fault energies.Molecular dynamics simulations successfully reproduce the transformation mechanisms and reasonably map out the pressure–temperature phase diagram of zirconium.展开更多
Alloy synthesis and processing determine the design of alloys with desired microstructure and properties.However,using data science to identify optimal synthesis-design routes from a specified set of starting material...Alloy synthesis and processing determine the design of alloys with desired microstructure and properties.However,using data science to identify optimal synthesis-design routes from a specified set of starting materials has been limited by large-scale data acquisition.Text mining has made it possible to convert scientific text into structured data collections.Still,the complexity,diversity,and flexibility of synthesis and processing expressions,and the lack of annotated corpora with a gold standard severely hinder accurate and efficient extraction.Here we introduce a semi-supervised text mining method to extract the parameters corresponding to the sequence of actions of synthesis and processing.We automatically extract a total of 9853 superalloy synthesis and processing actions with chemical compositions from a corpus of 16,604 superalloy articles published up to 2022.These have then been used to capture an explicitly expressed synthesis factor for predictingγ′phase coarsening.The synthesis factor derived from text mining significantly improves the performance of the data-drivenγ′size prediction model.The method thus complements the use of data-driven approaches in the search for relationships between synthesis and structures.展开更多
Optimizing several properties simultaneously based on small data-driven machine learning in complex black-box scenarios can present difficulties and challenges.Here we employ a triple-objective optimization algorithm ...Optimizing several properties simultaneously based on small data-driven machine learning in complex black-box scenarios can present difficulties and challenges.Here we employ a triple-objective optimization algorithm deduced from probability density functions of multivariate Gaussian distributions to optimize theγ′volume fraction,size,and morphology in CoNiAlCr-based superalloys.The effectiveness of the algorithm is demonstrated by synthesizing alloys with desiredγ/γ′microstructure and optimizingγ′microstructural parameters.In addition,the method leads to incorporating refractory elements to improveγ/γ′microstructure in superalloys.After four iterations of experiments guided by the algorithm,we synthesize sixteen alloys of relatively high creep strength from~120,000 candidates of which three possess highγ′volume fraction(>54%),smallγ′size(<480 nm),and high cuboidalγ′fraction(>77%).展开更多
Optimizing more than one property is inevitable in designing new materials;however,some properties are usually improved at the expense of others.Multiobjective optimization methods in engineering and computer science ...Optimizing more than one property is inevitable in designing new materials;however,some properties are usually improved at the expense of others.Multiobjective optimization methods in engineering and computer science have proven to be an effective means to optimize several different properties simultaneously.Here,we reviewed these approaches including scalarization,evolutionary algorithms,and especially Bayesian optimization.Their promising applications to a number of materials problems are also discussed in the paper.展开更多
基金financial support of the National Key Research and Development Program of China(2021YFB3802100)the National Natural Science Foundation of China(52203293)the Innovation Centre of Nuclear Materials Fund(ICNM-2022-ZH-02).
文摘Designing refractory high-entropy alloys(RHEAs)for high-temperature(HT)applications is an outstanding challenge given the vast possible composition space,which contains billions of candidates,and the need to optimize across multiple objectives.Here,we present an approach that accelerates the discovery of RHEA compositions with superior strength and ductility by integrating machine learning(ML),genetic search,cluster analysis,and experimental design.We iteratively synthesize and characterize 24 predicted compositions after six feedback loops.Four compositions show outstanding combinations of HT yield strength and room-temperature(RT)ductility spanning the ranges of 714–1061 MPa and 17.2%–50.0%fracture strain,respectively.We identify an attractive alloy system,ZrNbMoHfTa,particularly the composition Zr_(0.13)Nb_(0.27)Mo_(0.26)Hf_(0.13)Ta_(0.21),which demonstrates a yield approaching 940 MPa at 1200℃ and favorable RT ductility with 17.2%fracture strain.The high yield strength at 1200℃ exceeds that reported for RHEAs,with 1200℃ exceeding the service temperature limit for nickel(Ni)-based superalloys.Our ML-based approach makes it possible to rapidly optimize multiple properties for materials design,thus overcoming the common problems of limited data and a vast composition space in complex materials systems while satisfying multiple objectives.
基金financially supported by the National Key Research and Development Program of China(No.2021YFB3702601)the National Science and Technology Major Project of China(No.J2019-VI-0023-0140)+1 种基金the National Natural Science Foundation of China(No.52002326)the Natural Science Foundation of Chongqing(No.cstc2021jcyj-msxmX0602)。
文摘The state of the art for data-driven creep rupture life predictions incorporates microstructural and deformation characteristics into machine learning.However,the microstructures and deformation mechanisms for unknown alloys are inaccessible and uncertain before experiments are carried out,and therefore prevents extrapolations of the learned models.
基金supported by the National Key Re-search and Development Program of China(Nos.2021YFB3802102 and 2022YFB3707500)the National Natural Science Founda-tion of China(No.52203293).
文摘Machine learning assisted design of materials is so far based on features selected by considering the accuracy of model predictions,and those features do not necessarily ensure a high efficiency in searching for new materials.Here we estimate the efficiency of active learning loop by resampling method using available data as an alternative criterion for selection.The selected features allow an optimization of targeted property with as few new experiments as possible.Input those features into machine learning,we synthesized new high entropy alloys(HEAs)with strengths 2.8–3.0 GPa within five experimental iterations.The alloy AlVCrCoNiMo is found to possess compressive specific yield strengths of 397,144 and 105(MPa cm^(3))/g at 25,800 and 900℃,respectively.The specific yield strength of AlVCrCoNiMo alloy at 800℃is about twice that of the commercial Inconel 718 and the typical refractory HEA of VNbMoTaW.A unique microstructure consisting of multi-scale hierarchical B2 precipitates with coherent interfaces to the BCC matrix strengthens the alloy.Our strategy of maximizing active learning efficiency provides a recipe for selecting features that accelerate the optimization of targeted property.
基金the National Key Research and Development Program of China(Grant No.2017YFB0702401)the National Natural Science Foundation of China(Grant Nos.51571156,51671157,51621063,and 51931004).
文摘Accelerating materials discovery crucially relies on strategies that efficiently sample the search space to label a pool of unlabeled data.This is important if the available labeled data sets are relatively small compared to the unlabeled data pool.Active learning with efficient sampling methods provides the means to guide the decision making to minimize the number of experiments or iterations required to find targeted properties.We review here different sampling strategies and show how they are utilized within an active learning loop in materials science.
基金the support of National Key Research and Development Program of China(2021YFB3802104)National Natural Science Foundation of China(Grant Nos.51931004,52173228,52271190 and 51571156)the 111 project 2.0(BP2018008)。
文摘Functional stability of superelasticity is crucial for practical applications of shape memory alloys.It is degraded by a Lüders-like deformation with elevated local stress concentration under tensile load.By increasing the degree of solute supersaturation and applying appropriate thermomechanical treatments,a Ti-Ni alloy with nanocrystallinity and dispersed nanoprecipitates is obtained.In contrast to conventional Ti-Ni alloys,the superelasticity in the target alloy is accompanied by homogeneous deformation due to the sluggish stress-induced martensitic transformation.The alloy thus shows a fully recoverable strain of 6%under tensile stress over 1 GPa and a large adiabatic temperature decrease of 13.1 K under tensile strain of 4.5%at room temperature.Moreover,both superelasticity and elastocaloric effect exhibit negligible degradation in response to applied strain of 4%during cycling.We attribute the improved functional stability to low dislocation activity resulting from the suppression of localized deformation and the combined strengthening effect of nanocrystalline structure and nanoprecipitates.Thus,the design of such a microstructure enabling homogeneous deformation provides a recipe for stable superelasticity and elastocaloric effect.
基金financially supported by the National Key Research and Development Program of China(No.2016YFB0700500)the Guangdong Province Key Area R&D Program(No.2019B010940001)。
文摘There is growing interest in applying machine learning techniques in the field of materials science.However,the interpretation and knowledge extracted from machine learning models is a major concern,particularly as formulating an explicit model that provides insight into physics is the goal of learning.In the present study,we propose a framework that utilizes the filtering ability of feature engineering,in conjunction with symbolic regression to extract explicit,quantitative expressions for the band gap energy from materials data.We propose enhancements to genetic programming with dimensional consistency and artificial constraints to improve the search efficiency of symbolic regression.We show how two descriptors attributed to volumetric and electronic factors,from 32 possible candidates,explicitly express the band gap energy of Na Cl-type compounds.Our approach provides a basis to capture underlying physical relationships between materials descriptors and target properties.
基金financially supported by the National Key Research and Development Program of China(No.2017YFB0702401)the National Natural Science Foundation of China(Nos.51671157,51571156,and 51931004)the 111 project 2.0(No.BP2018008)。
文摘The singular change of the order parameter at the first order martensitic transformation(MT)temperature restricts the caloric response to a narrow temperature range.Here the MT is tuned into a sluggish strain glass transition by defect doping and a large elastocaloric effect appears in a wide temperature range.Moreover,an inverse elastocaloric effect is observed in the strain glass alloy with history of zerofield cooling and is attributed to the slow dynamics of the nanodomains in response to the external stress.This study offers a design recipe to expand the temperature range for good elastocaloric effect.
基金supported by grants from National Natural Science Foundation of China(52303295,52173217)National Key Research and Development Program of China(2022YFB3807401)111 project(B170003)and Yunnan Fundamental Research Projects(grant NO.202301BE070001-031).
文摘An issue of current interest in the use of machine learning models to predict compositions of materials is their reliability in predicting outcomes with elements not included in the training data.We show that the phase diagram of the ceramic(Ba_(1-x-y)Ca_(x)Sr_(y))(Ti_(1-u-v-w)Zr_(u)Sn_(v)Hf_(w))O_(3)can be accurately predicted if the feature values of unknown elements do not exceed the range of values for existing elements in the training data.In particular,we employ physical features as descriptors and compositions as weights to show that by excluding an element,such as Zr,Sn or Hf from the training set and treating it as an unknown element,the machine learning model accurately predicts the property only if the feature values of the unknown element does not exceed the range of values of existing elements in the training set.By adding a small amount of data for the unknown element restores the prediction accuracy.We demonstrate this for BaTiO_(3)ceramics doped with rare earth elements where the prediction accuracy is restored if the physical feature space is suitably enlarged with training data.The prediction error increases with the Euclidean distance of the testing sample relative to the nearest training sample in the physical feature space.Our work provides an effective strategy for extending machine learning models for material compositions beyond the scope of available data.
基金the financial support of National Key Research and Development Program of China(2022ZD0117805)Guangdong Province Key Area Research and Development Program(2019B010940001).
文摘Finding high temperature superconductors(HTS)has been a continuing challenge due to the difficulty in predicting the transition temperature(Tc)of superconduc-tors.Recently,the efficiency of predicting Tc has been greatly improved via ma-chine learning(ML).Unfortunately,prevailing ML models have not shown adequate generalization ability to find new HTS,yet.In this work,a graph neural network model is trained to predict the maximal Tc(Tc max)of various materials.Our model reveals a close connection between Tc max and chemical bonds.It sug-gests that shorter bond lengths are favored by high Tc,which is in coherence with previous domain knowledge.More importantly,it also indicates that chemical bonds consisting of some specific chemical elements are responsible for high Tc,which is new even to the human experts.It can provide a convenient guidance to the materials scientists in search of HTS.
基金supported by the National Natural Science Foundation of China(51931004,52322103,52350710205,and 52171011)Key Technologies R&D Program(2022YFB3707600)+1 种基金the 111 project 2.0(BP2018008)T.Q.L.acknowledges the support from the China Scholarship Council(202306280011).
文摘Recent advancements in machine learning(ML)have revolutionized the field of high-performance materials design.However,developing robust ML models to decipher intricate structure-property relationships in materials remains challenging,primarily due to the limited availability of labeled datasets with well-characterized crystal structures.This is particularly pronounced in materials where functional properties are closely intertwined with their crystallographic symmetry.Weintroduce a selfsupervised probabilistic model(SSPM)that autonomously learns unbiased atomic representations and the likelihood of compounds with given crystal structures,utilizing solely the existing crystal structure data from materials databases.SSPM significantly enhances the performance of downstream ML models by efficient atomic representations and accurately captures the probabilistic relationships between composition and crystal structure.We showcase SSPM’s capability by discovering shapememory alloys(SMAs).Amongst the top 50 predictions,23 have been confirmed as SMAs either experimentally or theoretically,and a previously unknown SMA candidate,MgAu,has been identified.
基金supported by the National Key Research and Development Program of China(2021YFB3702604)the National Natural Science Foundation of China(52002326)the Research Fund of the State Key Laboratory of Solidification Processing(NPU),China(grant no.2023-TS-12).
文摘Machine learning has been widely used to guide the search for new materials by learning the patterns underlying available data.However,the pure prediction-oriented search is often biased to interpolation due to the limited data in a large unexplored space.Here we present a sampling framework towards extrapolation,that integrates unsupervised clustering,interpretable analysis,and similarity evaluation to sample target candidates with improved properties from a vast search space.
文摘This special issue of MGE advances focuses on the revolutionary impact of machine learning(ML)on materials science.As we navigate the threshold of a new era in scientific innovation,this issue collates a series of research articles that epitomize machine learning as a foundational pillar in materials science and engineering.The synergy between ML and conventional materials science methodologies not only accelerates the discovery of novel materials but also refines the prediction of material properties and streamlines manufacturing processes.These advances offer unparalleled opportunities for technological progress and sustainability.We,as the vip editors,are excited to present these contributions that introduce new methodologies and enhance our understanding of material behavior through the prism of advanced analytics and computational power.
基金the National Key Research and Devel-opment Program of China(No.2022YFB3709000)the National Natural Science Foundation of China(Nos.52101019,52122408,52071023,52474397)+1 种基金support from the Fundamental Research Funds for the Central Universities(University of Science and Technology Beijing,No.FRF-TP-2021-04C1,and 06500135)supported by USTB MatCom of Beijing Advanced Innovation Center for Materials Genome Engineering.
文摘Martensitic transformation plays a pivotal role in strengthening and hardening of steels,yet an accu-rate interatomic potential for a comprehensive description of the martensitic phase formation in Fe-C alloys is lacking.Herein,we developed a deep learning-based interatomic potential to perform molecu-lar dynamics(MD)simulations to study the martensitic phase transformation across a range of carbon(C)concentrations.The results revealed that an increased C concentration leads to a suppressed phase boundary movement and a decelerated phase transformation rate.To overcome the timescale limitations inherent in MD simulations,metadynamics sampling was employed to accelerate the simulations of C dif-fusion.We found that C atoms tend to cluster at distances equivalent to the lattice parameter of Fe with the same sublattice occupation,leading to local lattice tetragonality.Such C-ordered structures effectively inhibit dislocation movement and enhance strength.The stress field induced by dislocations facilitates a higher degree of ordering,and the formation of C-ordered structures was identified as a potentially cru-cial strengthening mechanism for martensitic steels.The consistency between our simulation results and reported experimental observations underscores the effectiveness of the developed DP model in simu-lating martensitic phase transformation in Fe-C alloys,providing detailed insights into the mechanisms underlying this process.This work not only advances the understanding of martensitic phase transforma-tions in Fe-C alloys but also establishes a powerful computational framework for designing steels with optimized mechanical properties through the precise control of carbon ordering and dislocation behavior.
基金We are grateful to the Laboratory Directed Research and Development(LDRD)program(project#20180660ER)the Center for Nonlinear Studies at Los Alamos National Laboratory for support.
文摘One of the main challenges in materials discovery is efficiently exploring the vast search space for targeted properties as approaches that rely on trial-and-error are impractical.We review how methods from the information sciences enable us to accelerate the search and discovery of new materials.In particular,active learning allows us to effectively navigate the search space iteratively to identify promising candidates for guiding experiments and computations.The approach relies on the use of uncertainties and making predictions from a surrogate model together with a utility function that prioritizes the decision making process on unexplored data.We discuss several utility functions and demonstrate their use in materials science applications,impacting both experimental and computational research.We summarize by indicating generalizations to multiple properties and multifidelity data,and identify challenges,future directions and opportunities in the emerging field of materials informatics.
基金This work is financially supported by the National Key Research and Development Program of China(2020YFB0704503,2016YFB0700500)Guangdong Province Key Area R&D Program(2019B010940001)+1 种基金111 Project(B170003)USTB MatCom of Beijing Advanced Innovation Center for Materials Genome Engineering.
文摘Data provides a foundation for machine learning,which has accelerated data-driven materials design.The scientific literature contains a large amount of high-quality,reliable data,and automatically extracting data from the literature continues to be a challenge.We propose a natural language processing pipeline to capture both chemical composition and property data that allows analysis and prediction of superalloys.Within 3 h,2531 records with both composition and property are extracted from 14,425 articles,coveringγ′solvus temperature,density,solidus,and liquidus temperatures.A data-driven model forγ′solvus temperature is built to predict unexplored Co-based superalloys with highγ′solvus temperatures within a relative error of 0.81%.We test the predictions via synthesis and characterization of three alloys.A web-based toolkit as an online open-source platform is provided and expected to serve as the basis for a general method to search for targeted materials using data extracted from the literature.
基金This work was supported by Key Technologies R&D Program(2017YFB0702401)the National Natural Science Foundation of China(51320105014,51621063,and 51501141)+1 种基金Los Alamos National Laboratory(ASC/PEM and LDRD)the ERC grant“Hecate”,and the China Postdoctoral Science Foundation(2015M580843).
文摘Atomic simulations provide an effective means to understand the underlying physics of structural phase transformations.However,this remains a challenge for certain allotropic metals due to the failure of classical interatomic potentials to represent the multitude of bonding.Based on machine-learning(ML)techniques,we develop a hybrid method in which interatomic potentials describing martensitic transformations can be learned with a high degree of fidelity from ab initio molecular dynamics simulations(AIMD).Using zirconium as a model system,for which an adequate semiempirical potential describing the phase transformation process is lacking,we demonstrate the feasibility and effectiveness of our approach.Specifically,the ML-AIMD interatomic potential correctly captures the energetics and structural transformation properties of zirconium as compared to experimental and density-functional data for phonons,elastic constants,as well as stacking fault energies.Molecular dynamics simulations successfully reproduce the transformation mechanisms and reasonably map out the pressure–temperature phase diagram of zirconium.
基金This work is financially supported by the National Key Research and Development Program of China(2021YFB3702403,2022YFB3707502)National Natural Science Foundation of China(52201061,U22A20106)+1 种基金Fundamental Research Funds for the Central Universities(FRF-TP-22-008A1)USTB MatCom of Beijing Advanced Innova-tion Center for Materials Genome Engineering,and the CNNC Science Fund for Talented Young Scholars(FY222506000902).
文摘Alloy synthesis and processing determine the design of alloys with desired microstructure and properties.However,using data science to identify optimal synthesis-design routes from a specified set of starting materials has been limited by large-scale data acquisition.Text mining has made it possible to convert scientific text into structured data collections.Still,the complexity,diversity,and flexibility of synthesis and processing expressions,and the lack of annotated corpora with a gold standard severely hinder accurate and efficient extraction.Here we introduce a semi-supervised text mining method to extract the parameters corresponding to the sequence of actions of synthesis and processing.We automatically extract a total of 9853 superalloy synthesis and processing actions with chemical compositions from a corpus of 16,604 superalloy articles published up to 2022.These have then been used to capture an explicitly expressed synthesis factor for predictingγ′phase coarsening.The synthesis factor derived from text mining significantly improves the performance of the data-drivenγ′size prediction model.The method thus complements the use of data-driven approaches in the search for relationships between synthesis and structures.
基金The authors acknowledge the financial support of National Key Research and Development Program of China(2022YFB3707502 and 2021YFB3802100)Guangdong Major Project of Basic and Applied Basic Research(2019B030302011).
文摘Optimizing several properties simultaneously based on small data-driven machine learning in complex black-box scenarios can present difficulties and challenges.Here we employ a triple-objective optimization algorithm deduced from probability density functions of multivariate Gaussian distributions to optimize theγ′volume fraction,size,and morphology in CoNiAlCr-based superalloys.The effectiveness of the algorithm is demonstrated by synthesizing alloys with desiredγ/γ′microstructure and optimizingγ′microstructural parameters.In addition,the method leads to incorporating refractory elements to improveγ/γ′microstructure in superalloys.After four iterations of experiments guided by the algorithm,we synthesize sixteen alloys of relatively high creep strength from~120,000 candidates of which three possess highγ′volume fraction(>54%),smallγ′size(<480 nm),and high cuboidalγ′fraction(>77%).
基金the support of National Key Research and Development Program of China(2021YFB3802103).
文摘Optimizing more than one property is inevitable in designing new materials;however,some properties are usually improved at the expense of others.Multiobjective optimization methods in engineering and computer science have proven to be an effective means to optimize several different properties simultaneously.Here,we reviewed these approaches including scalarization,evolutionary algorithms,and especially Bayesian optimization.Their promising applications to a number of materials problems are also discussed in the paper.
