Accurate interatomic potential for the nucleation in liquid Ti-Al binary alloy developed by deep neural network learning method

Zhai, B.; Wang, H.P.

doi:10.1016/j.commatsci.2022.111843

Cited by 17 publications

(5 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the previous section, the MLIP predictive performance was demonstrated through practices commonly employed by researchers while validating MLIPs and empirical potentials. ,,, In this section, the MLIP’s performance on phenomena proven difficult to model with empirical potentials, such as thermal decomposition and surface reconstruction, are summarized. The aim of these studies is to showcase the MLIP’s ability to model complex phenomena and spur future investigations into SiC behaviors (e.g., silicon carbide epitaxy).…”

Section: Resultsmentioning

confidence: 99%

A Genetic Algorithm Trained Machine-Learned Interatomic Potential for the Silicon–Carbon System

MacIsaac,

Bavdekar,

Spearot

et al. 2024

J. Phys. Chem. C

View full text Add to dashboard Cite

A linear regression-based machine learned interatomic potential (MLIP) was developed for the silicon−carbon system. The MLIP was predominantly trained on structures discovered through a genetic algorithm, encompassing the entire silicon−carbon composition space, and uses as its foundation the Ultra-Fast Force Fields (UF 3 ) formulation. To improve MLIP performance, the learning algorithm was modified to include higher spline interpolation resolution in regions with large potential energy surface curvature. The developed MLIP demonstrates exceptional predictive performance, accurately estimating energies and forces for structures across the silicon−carbon composition and configuration space. The MLIP predicts structural, energetic, and elastic properties of silicon carbide (SiC) with high precision and captures fundamental volume-pressure and volumetemperature relationships. Uniquely, this silicon−carbon MLIP is adept at modeling complex high-temperature phenomena, including the peritectic decomposition of SiC and carbon dimer formation during SiC surface reconstruction, which cannot be captured with prior classical interatomic potentials for this material.

show abstract

Section: Resultsmentioning

confidence: 99%

A Genetic Algorithm Trained Machine-Learned Interatomic Potential for the Silicon–Carbon System

MacIsaac,

Bavdekar,

Spearot

et al. 2024

J. Phys. Chem. C

View full text Add to dashboard Cite

show abstract

“…However, the high computational costs of density functional theory (DFT) calculations limit their applications in molecular systems, whose unit cells usually contain dozens or even hundreds of atoms. Recently developed machine learning potentials (MLPs), such as the Gaussian approximation potential (GAP) [1][2][3][4][5][6] , moment tensor potential (MTP) [7][8][9] , and DeepMD-kit package [10][11][12][13] , can perform DFT-level calculations at the computational cost of classical force fields. CSP methods, such as Universal Structure Predictor: Evolutionary Xtallography (USPEX) [14,15] and Crystal structure AnaLYsis by Particle Swarm Optimization (CALPSO) [16] , combining MLPs [17][18][19][20][21][22][23] will undoubtedly become the main approaches in the search for new materials.…”

Section: Introductionmentioning

confidence: 99%

Searching new cocrystal structures of CL-20 and HMX via evolutionary algorithm and machine learning potential

Ye,

Guo,

Chai

et al. 2024

Journal of Materials Informatics

View full text Add to dashboard Cite

In this work, we report the discovery of energy cocrystals using an efficient iterative workflow combining an evolutionary algorithm and a machine learning potential (MLP). The compound 2,4,6,8,10,12-Hexanitro-2,4,6,8,10,12-hexaazaisowurtzitane (CL-20) has attracted significant attention owing to its higher energy density than traditional energetic materials. However, the higher sensitivity has limited its applications. An important way to reduce its sensitivity involves cocrystal engineering with traditional explosives. Many cocrystal structures are expected to be composed of these two components. We developed an efficient iterative workflow to explore the phase space of CL-20 and 1,3,5,7-tetranitro-1,3,5,7-tetraazacyclooctane (HMX) cocrystals using an evolutionary algorithm and an MLP. The algorithm was based on the Universal Structure Predictor: Evolutionary Xtallography (USPEX) software, and the MLP was the reactive force field with neural networks (ReaxFF-nn ) model. A set of high-density cocrystal structures was produced through this workflow; these structures were further checked via first-principles geometry optimizations. After careful screening, we identified several high-density cocrystal structures with densities of up to 1.937 g/cm3 and HMX: CL-20 ratios of 1:1 and 1:2. The influence of hydrogen bonds on the formation of high-density cocrystals was also discussed, and a roughly linear relationship was found between energy and density.

show abstract

“…Among the various machine‐learning methods, the combination of deep‐learning method and molecular dynamics has accelerated the exploration of new structures 29 . For complex chemical structures, such as high‐entropy alloys and large protein molecules, the deep‐learning potential molecular dynamic (DPMD) has revealed their SPC in larger spatial and time scales 30–33 . Recent studies have shown that DPMD, when trained with a well‐curated dataset, can yield accurate predictions for dynamic properties like viscosity of alloys 34 .…”

Section: Introductionmentioning

confidence: 99%

“…29 For complex chemical structures, such as high-entropy alloys and large protein molecules, the deep-learning potential molecular dynamic (DPMD) has revealed their SPC in larger spatial and time scales. [30][31][32][33] Recent studies have shown that DPMD, when trained with a well-curated dataset, can yield accurate predictions for dynamic properties like viscosity of alloys. 34 DPMD combines elements of machine learning and molecular dynamics to capture the complex dynamics of materials.…”

Section: Introductionmentioning

confidence: 99%

Unraveling medium‐range order and melting mechanism of ZIF‐4 under high temperature

Shi,

Liu,

Yue

et al. 2024

J Am Ceram Soc.

View full text Add to dashboard Cite

Glass formation in zeolitic imidazolate frameworks (ZIFs) has garnered significant attention in the field of metal–organic frameworks (MOFs) in recent years. Numerous works have been conducted to investigate the microscopic mechanisms involved in the melting–quenching process of ZIFs. Understanding the density variations that occur during the melting process of ZIFs is crucial for comprehending the origins of glass formation. However, conducting large‐scale simulations has been challenging due to limitations in computational resources. In this work, we used deep‐learning methods to accurately construct a potential function that describes the atomic‐scale melting behavior of ZIF‐4. The results revealed the spatial heterogeneity associated with the formation of low‐density phases during the melting process of ZIF‐4. This work discusses the advantages and limitations of applying deep‐learning simulation methods to complex structures like ZIFs, providing valuable insights for the development of machine‐learning approaches in designing MOF glasses.

show abstract

Accurate interatomic potential for the nucleation in liquid Ti-Al binary alloy developed by deep neural network learning method

Cited by 17 publications

References 45 publications

A Genetic Algorithm Trained Machine-Learned Interatomic Potential for the Silicon–Carbon System

A Genetic Algorithm Trained Machine-Learned Interatomic Potential for the Silicon–Carbon System

Searching new cocrystal structures of CL-20 and HMX via evolutionary algorithm and machine learning potential

Unraveling medium‐range order and melting mechanism of ZIF‐4 under high temperature

Contact Info

Product

Resources

About