Featured
-
-
Article
| Open AccessNeural network kinetics for exploring diffusion multiplicity and chemical ordering in compositionally complex materials
The inherent chemical complexity in compositionally complex materials present a challenge in studying atomic diffusion. Here, the authors introduce a neural network kinetics scheme to effectively address this issue and reveal anomalous diffusion behavior in complex concentrated alloys.
- Bin Xing
- , Timothy J. Rupert
- & Penghui Cao
-
Article
| Open AccessCollective relational inference for learning heterogeneous interactions
Heterogeneous interactions between interactive entities are not well understood due to their complex configurations and many body interactions. Han et al. present a probabilistic-based machine learning method to discover the fundamental laws governing the interactions of heterogeneous systems.
- Zhichao Han
- , Olga Fink
- & David S. Kammer
-
Article
| Open AccessMultilevel design and construction in nanomembrane rolling for three-dimensional angle-sensitive photodetection
Zhang et al. report a quasistatic multilevel finite element model to predict the 3D structures assembled by 2D nanomembranes, validated by large-scale, high-yield, and configurable fabrication. 3D Si/Cr photodetectors assisted by neural network are employed to resolve the incident light angle.
- Ziyu Zhang
- , Binmin Wu
- & Yongfeng Mei
-
Article
| Open AccessModelling atomic and nanoscale structure in the silicon–oxygen system through active machine learning
Understanding the silicon-oxygen system is crucial for various applications. Here, the authors present an interatomic potential covering a wide range of the Si-O configurational space and showcase applications to silica and Si-SiO2 interfaces.
- Linus C. Erhard
- , Jochen Rohrer
- & Volker L. Deringer
-
Article
| Open AccessA comprehensive transformer-based approach for high-accuracy gas adsorption predictions in metal-organic frameworks
Three-dimensional representation learning is efficient in material science. Here, authors proposed a transformer-based framework for multi-purpose gas adsorption prediction. Predicted values correspond with the outcomes of adsorption experiments.
- Jingqi Wang
- , Jiapeng Liu
- & Diannan Lu
-
Article
| Open AccessPore evolution mechanisms during directed energy deposition additive manufacturing
Porosity is a key issue in additive manufacturing (AM). Here, the authors reveal the bubble evolution mechanisms including formation, coalescence, pushing, growth, entrainment, escape, and entrapment during directed energy deposition AM using in situ X-ray imaging and multiphysics modelling.
- Kai Zhang
- , Yunhui Chen
- & Peter D. Lee
-
Article
| Open AccessHigh-throughput computational stacking reveals emergent properties in natural van der Waals bilayers
2D bilayers have recently attracted significant attention due to fundamental properties like interlayer excitons and interfacial ferroelectricity. Here, the authors report a density functional theory approach to identify 2586 stable homobilayer systems and calculate their stacking-dependent electronic, magnetic and vibrational properties.
- Sahar Pakdel
- , Asbjørn Rasmussen
- & Kristian S. Thygesen
-
Article
| Open AccessAn energy-free strategy to elevate anti-icing performance of superhydrophobic materials through interfacial airflow manipulation
Currently, the anti-icing performance limitation of superhydrophobic materials is gradually approached without the assistance of an external field. Here, the authors propose a strategy of microdroplet movement manipulation induced by interfacial airflow for further improving the anti-icing performance.
- Jiawei Jiang
- , Yizhou Shen
- & Haifeng Chen
-
Article
| Open AccessTowards near-term quantum simulation of materials
The use of NISQ devices for useful quantum simulations of materials and chemistry is still mainly limited by the necessary circuit depth. Here, the authors propose to combine classically-generated effective Hamiltonians, hybrid fermion-to-qubit mapping and circuit optimisations to bring this requirement closer to experimental feasibility.
- Laura Clinton
- , Toby Cubitt
- & Evan Sheridan
-
Article
| Open AccessAccelerating process development for 3D printing of new metal alloys
Process development for 3D printing of new metal alloys can be time-consuming and variability in the printing outcome makes it even more challenging. Here, authors demonstrate an in-situ method using high-speed imaging and deep learning to accelerate the process design for a more consistent quality.
- David Guirguis
- , Conrad Tucker
- & Jack Beuth
-
Article
| Open AccessAn invertible, invariant crystal representation for inverse design of solid-state materials using generative deep learning
The lack of invertible and invariant crystal representations hinders the inverse design of crystals. Here the authors develop SLICES, an invertible and invariant representation, empowering property-driven inverse design of crystals using generative AI.
- Hang Xiao
- , Rong Li
- & Lei Wang
-
Article
| Open AccessAutonomous and dynamic precursor selection for solid-state materials synthesis
Solid-state materials synthesis relies on effective precursor design. Here, the authors introduce an algorithm that combines ab-initio computations with insights gained from experimental outcomes to efficiently optimize the selection of precursors.
- Nathan J. Szymanski
- , Pragnay Nevatia
- & Gerbrand Ceder
-
Article
| Open AccessMaterial-agnostic machine learning approach enables high relative density in powder bed fusion products
Exploring laser powder bed fusion in manufacturing, the authors demonstrate a machine learning-based method to optimize processing conditions achieving materials with relative density greater than 98% and experimentally verify its generality for multiple distinct powder materials.
- Jaemin Wang
- , Sang Guk Jeong
- & Byeong-Joo Lee
-
Article
| Open AccessAccelerated discovery of multi-elemental reverse water-gas shift catalysts using extrapolative machine learning approach
Despite the promise that machine learning (ML) can accelerate catalyst development, truly novel catalysts are challenging to find through ML approaches because of their inability to extrapolate. Here, the authors show an extrapolative ML approach to develop new multi-elemental catalysts.
- Gang Wang
- , Shinya Mine
- & Takashi Toyao
-
Article
| Open AccessCapturing dynamical correlations using implicit neural representations
Analysis of experimental data in condensed matter is often challenging due to system complexity and slow character of physical simulations. The authors propose a framework that combines machine learning with theoretical calculations to enable real-time analysis for electron, neutron, and x-ray spectroscopies.
- Sathya R. Chitturi
- , Zhurun Ji
- & Joshua J. Turner
-
Article
| Open AccessMachine learning the microscopic form of nematic order in twisted double-bilayer graphene
Machine learning methods in condensed matter physics are an emerging tool for providing powerful analytical methods. Here, the authors demonstrate that convolutional neural networks can identify nematic electronic order from STM data of twisted double-layer graphene—even in the presence of heterostrain.
- João Augusto Sobral
- , Stefan Obernauer
- & Mathias S. Scheurer
-
Perspective
| Open AccessToward a formal theory for computing machines made out of whatever physics offers
Learning from human brains to build powerful computers is attractive, yet extremely challenging due to the lack of a guiding computing theory. Jaeger et al. give a perspective on a bottom-up approach to engineer unconventional computing systems, which is fundamentally different to the classical theory based on Turing machines.
- Herbert Jaeger
- , Beatriz Noheda
- & Wilfred G. van der Wiel
-
Article
| Open AccessAtomic stiffness for bulk modulus prediction and high-throughput screening of ultraincompressible crystals
Fast and accurate prediction of bulk moduli for diverse materials is challenging. Here, the authors introduce the concept of atomic stiffness to accelerate bulk modulus prediction and high-throughput screening of ultra-incompressible crystals.
- Ruihua Jin
- , Xiaoang Yuan
- & Enlai Gao
-
Article
| Open AccessQuantifying disorder one atom at a time using an interpretable graph neural network paradigm
Level of atomic disorder in materials is critical to understanding the effect of local structure on materials properties. Here the authors present a workflow combining structure-aware graph neural networks and physics-inspired order parameter to characterize structural disorder on a per atom basis.
- James Chapman
- , Tim Hsu
- & Brandon C. Wood
-
Article
| Open AccessLanthanide-doped MoS2 with enhanced oxygen reduction activity and biperiodic chemical trends
Oxygen reduction reaction plays a key role in many applications of MoS2-based materials. Here, using first-principles simulations, the authors find the enhanced oxygen-reduction activity with a biperiodic chemical trend on the lanthanide-doped MoS2.
- Yu Hao
- , Liping Wang
- & Liang-Feng Huang
-
Article
| Open AccessGeneral framework for E(3)-equivariant neural network representation of density functional theory Hamiltonian
Fundamental symmetries are crucial to the deep-learning modeling of physical systems. Here the authors use equivariant neural networks preserving the Euclidean symmetries to accelerate electronic structure calculations by orders of magnitude keeping sub-meV accuracy.
- Xiaoxun Gong
- , He Li
- & Yong Xu
-
Article
| Open AccessThe crucial role of adhesion in the transmigration of active droplets through interstitial orifices
Active fluid droplets are relevant for development of bio-inspired soft materials, however their motion in heterogeneous surrounding environments remains challenging. The authors uncover the role of adhesion forces for a variety of dynamic regimes of active fluid droplet crossing a narrow constriction.
- A. Tiribocchi
- , M. Durve
- & S. Succi
-
Article
| Open AccessNuclear quantum effects on zeolite proton hopping kinetics explored with machine learning potentials and path integral molecular dynamics
The quantum properties of hydrogen atoms in zeolite-catalyzed reactions are generally neglected due to high computational costs. Here, the authors leverage machine learning to derive accurate quantum kinetics for proton transfer reactions in heterogeneous catalysis.
- Massimo Bocus
- , Ruben Goeminne
- & Veronique Van Speybroeck
-
Article
| Open AccessLarge-scale physically accurate modelling of real proton exchange membrane fuel cell with deep learning
Accurate liquid water modelling is challenging. Here the authors use X-ray micro-computed tomography, deep learned super-resolution, multi-label segmentation, and direct multiphase simulation to simulate fuel cell and guide fuel cell design.
- Ying Da Wang
- , Quentin Meyer
- & Ryan T. Armstrong
-
Article
| Open AccessLearning local equivariant representations for large-scale atomistic dynamics
The paper presents a method that allows scaling machine learning interatomic potentials to extremely large systems, while at the same time retaining the remarkable accuracy and learning efficiency of deep equivariant models. This is obtained with an E(3)- equivariant neural network architecture that combines the high accuracy of equivariant neural networks with the scalability of local methods.
- Albert Musaelian
- , Simon Batzner
- & Boris Kozinsky
-
Article
| Open AccessMachine learning models to accelerate the design of polymeric long-acting injectables
Polymer-based long-acting injectable drugs are a promising therapeutic strategy for chronic diseases. Here the authors use machine learning to inform the data-driven development of advanced drug formulations.
- Pauric Bannigan
- , Zeqing Bao
- & Christine Allen
-
Article
| Open AccessQuadrupolar 23Na+ NMR relaxation as a probe of subpicosecond collective dynamics in aqueous electrolyte solutions
Quadrupolar nuclear magnetic relaxometry senses electrical fluctuations around nuclei, but their microscopic interpretation remains elusive. Here, the authors combine experiments and multiscale simulations to interpret relaxation rates in electrolyte solutions and assess commonly used models.
- Iurii Chubak
- , Leeor Alon
- & Benjamin Rotenberg
-
Article
| Open AccessToward the design of ultrahigh-entropy alloys via mining six million texts
The avalanche of publications challenges the norm that researchers extract knowledge from literature to design materials. Here the authors present a text-mining method that is implemented based on the abstracts of 6.4 million papers to enable the design of new high entropy alloys.
- Zongrui Pei
- , Junqi Yin
- & Dierk Raabe
-
Article
| Open AccessDiscovery of two-dimensional binary nanoparticle superlattices using global Monte Carlo optimization
Binary nanoparticle superlattices exhibit different collective optical, magnetic, and electronic properties. Here, the authors develop an efficient global optimization algorithm for the discovery of periodic 2D architectures forming at fluid interfaces.
- Yilong Zhou
- & Gaurav Arya
-
Article
| Open AccessPre-equilibrium biosensors as an approach towards rapid and continuous molecular measurements
Biosensors using ligand-receptor binding tend to operate under equilibrium conditions, but this can make real-time monitoring challenging. Here the authors provide a theoretical foundation for biosensing where ligand concentrations can be continuously measured without needing to reach equilibrium.
- Nicolò Maganzini
- , Ian Thompson
- & Hyongsok Tom Soh
-
Article
| Open AccessAn unconstrained approach to systematic structural and energetic screening of materials interfaces
Predicting structures and stabilities of solid-solid interfaces presents an ongoing and increasingly important challenge for development of new technologies. Here authors report an unconstrained and generally applicable non-periodic screening method for systematic exploration of material´s interfaces.
- Giovanni Di Liberto
- , Ángel Morales-García
- & Stefan T. Bromley
-
Article
| Open AccessAdsorbate chemical environment-based machine learning framework for heterogeneous catalysis
A combination of electronic structure calculations and machine learning strategies is developed to predict structures of complex heterogeneous catalysts in realistic environments, yielding new opportunities for optimization for energy applications.
- Pushkar G. Ghanekar
- , Siddharth Deshpande
- & Jeffrey Greeley
-
Article
| Open AccessDigitally-enhanced lubricant evaluation scheme for hot stamping applications
The digital transformation and Industry 4.0 technologies are rapidly shaping the future of manufacturing. Here, authors use reliable big data to quantitatively evaluate lubricants performance and select desirable candidates for application in target manufacturing processes.
- Xiao Yang
- , Heli Liu
- & Liliang Wang
-
Article
| Open AccessActive learning of reactive Bayesian force fields applied to heterogeneous catalysis dynamics of H/Pt
Uncertainty-aware machine learning models are used to automate the training of reactive force fields. The method is used here to simulate hydrogen turnover on a platinum surface with unprecedented accuracy.
- Jonathan Vandermause
- , Yu Xie
- & Boris Kozinsky
-
Article
| Open AccessAllotropy in ultra high strength materials
Here the authors propose a crystal thermodynamics framework describing the tensor stress induced phase transformations in solids based on nonlinear elasticity and first principles calculations. The proposed approach enables balanced design of high-strength, high-ductility materials.
- A. S. L. Subrahmanyam Pattamatta
- & David J. Srolovitz
-
Article
| Open AccessMachine learning the metastable phase diagram of covalently bonded carbon
Exploration of metastable phases of a given elemental composition is a data-intensive task. Here the authors integrate first-principles atomistic simulations with machine learning and high-performance computing to allow a rapid exploration of the metastable phases of carbon.
- Srilok Srinivasan
- , Rohit Batra
- & Subramanian K.R.S. Sankaranarayanan
-
Article
| Open AccessTowards universal neural network potential for material discovery applicable to arbitrary combination of 45 elements
Existing neural network potentials are generally designed for narrow target materials. Here the authors develop a neural network potential which is able to handle any combination of 45 elements and show its applicability in multiple domains.
- So Takamoto
- , Chikashi Shinagawa
- & Takeshi Ibuka
-
Article
| Open AccessTopological control of liquid-metal-dealloyed structures
Liquid metal dealloying is a method to fabricate bicontinuous composite structures with ultra-high interfacial area for diverse applications. This paper demonstrates how the topology of those structures can be controlled by the choice of melt composition.
- Longhai Lai
- , Bernard Gaskey
- & Alain Karma
-
Article
| Open AccessTheory-guided design of hydrogen-bonded cobaltoporphyrin frameworks for highly selective electrochemical H2O2 production in acid
Guided by high-throughput computational screening, we report the preparation of hydrogen-bonded cobaltoporphyrin frameworks and demonstrate the achievement of high activity and selectivity for electrochemical H2O2 production in acid.
- Xuan Zhao
- , Qi Yin
- & Yanguang Li
-
Article
| Open AccessE(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials
An E(3)-equivariant deep learning interatomic potential is introduced for accelerating molecular dynamics simulations. The method obtains state-of-the-art accuracy, can faithfully describe dynamics of complex systems with remarkable sample efficiency.
- Simon Batzner
- , Albert Musaelian
- & Boris Kozinsky
-
Article
| Open AccessArgyrodite-type advanced lithium conductors and transport mechanisms beyond paddle-wheel effect
Fundamental mechanisms governing the superionic behaviour in solid-state Li-ion batteries are under debate. Here the authors investigate computationally the mechanism of superionic lithium-ion conduction and predict new advanced lithium superionic conductors.
- Hong Fang
- & Puru Jena
-
Article
| Open AccessCrystal structure prediction by combining graph network and optimization algorithm
Predicting crystal structure prior to experimental synthesis is highly desirable. Here the authors propose a machine-learning framework combining graph network and optimization algorithms for crystal structure prediction, which is about three orders of magnitude faster than DFT-based approach.
- Guanjian Cheng
- , Xin-Gao Gong
- & Wan-Jian Yin
-
Article
| Open AccessDensity of states prediction for materials discovery via contrastive learning from probabilistic embeddings
Electrons and phonons give rise to important properties of materials. The machine learning framework Mat2Spec vastly accelerates their computational characterization, enabling discovery of materials for thermoelectrics and solar energy technologies.
- Shufeng Kong
- , Francesco Ricci
- & John M. Gregoire
-
Article
| Open AccessRepresenting individual electronic states for machine learning GW band structures of 2D materials
The study introduces novel methods for representing electronic states as input to machine learning models, which is used to learn high-fidelity band structures of two-dimensional materials from low- fidelity input.
- Nikolaj Rørbæk Knøsgaard
- & Kristian Sommer Thygesen
-
Article
| Open AccessLearning in continuous action space for developing high dimensional potential energy models
Reinforcement learning algorithms are emerging as powerful machine learning approaches. This paper introduces a novel machine-learning approach for learning in continuous action space and applies this strategy to the generation of high dimensional potential models for a wide variety of materials.
- Sukriti Manna
- , Troy D. Loeffler
- & Subramanian K. R. S. Sankaranarayanan
-
Article
| Open AccessVisualization and validation of twin nucleation and early-stage growth in magnesium
The origins of deformation twins in Mg have remained unclear in the past. Here the authors, by combining in situ experimental observations and atomistic simulations, capture the rapid twinning phenomena in Mg crystals and show that twinning occurs through pure atomic shuffle.
- Lin Jiang
- , Mingyu Gong
- & Julie M. Schoenung
-
Article
| Open AccessAugmenting zero-Kelvin quantum mechanics with machine learning for the prediction of chemical reactions at high temperatures
Computational material design often does not account for temperature effects. The present manuscript combines quantum-mechanics based calculations with a machine-learned correction to establish a unified thermodynamics framework for accurate prediction of high temperature reaction free energies in oxides.
- Jose Antonio Garrido Torres
- , Vahe Gharakhanyan
- & Alexander Urban
-
Article
| Open AccessLearning neural network potentials from experimental data via Differentiable Trajectory Reweighting
In machine learning approaches relevant for chemical physics and material science, neural network potentials can be trained on the experimental data. The authors propose a training method applying trajectory reweighting instead of direct backpropagation for improved robustness and reduced computational cost.
- Stephan Thaler
- & Julija Zavadlav
-
Article
| Open AccessCross-property deep transfer learning framework for enhanced predictive analytics on small materials data
Artificial intelligence and machine learning can greatly enhance materials property prediction and discovery. Here the authors propose cross-property transfer learning to build accurate models for dozens of properties with limited data availability.
- Vishu Gupta
- , Kamal Choudhary
- & Ankit Agrawal