Machine learning-assisted mechanical property prediction and descriptor-property correlation analysis of high-entropy ceramics

Zhou, Qian; Xu, Feng; Gao, Chengzuan; Zhang, Dan; Shi, Xianqing; Yuen, Muk-Fung; Zuo, Dunwen

doi:10.1016/j.ceramint.2022.10.105

Cited by 23 publications

(5 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…used the bond strength in addition to the bond length to predict properties of carbides, nitrides, and carbonitrides, while Liu et al 7 .formulated a modified parameter which also takes the shear modulus mismatch into account. Zhou et al 8 . found that the most important features in predicting hardness and Young’s modulus in carbides with high configurational entropy are the valence electron concentration, the deviation of melting temperatures, and the fraction-weighted mean total energies, while atomic size differences plays a minor, but still significant role.…”

Section: Introductionmentioning

confidence: 99%

Explaining the entropy forming ability for carbides with the effective atomic size mismatch

Kretschmer,

Mayrhofer

2024

Sci Rep

View full text Add to dashboard Cite

To quickly screen for single-phased multi-principal-element materials, a so-called entropy forming ability (EFA) parameter is sometimes used as a descriptor. The larger the EFA, the larger is the propensity to form a single-phase structure. We have investigated this EFA descriptor with atomic relaxations in special-quasi-random structures and discovered that the EFA correlates inversely with the lattice distortion. Large effective atomic size differences lead to multi-phase compounds, and little size differences to single-phase compounds. Instead of configurational entropy, we therefore demonstrate the applicability of the Hume-Rothery rules to phase stability of solid solutions even in compositionally complex ceramics.

show abstract

Section: Introductionmentioning

confidence: 99%

Explaining the entropy forming ability for carbides with the effective atomic size mismatch

Kretschmer,

Mayrhofer

2024

Sci Rep

View full text Add to dashboard Cite

show abstract

“…Tang et al 17 proposed a ML strategy based on bond parameters (bond order, bond ionicity, and bond length) to explore new HECs with excellent mechanical properties, the mean absolute error (MAE) and R 2 of their model were 32.2 GPa and 0.84. Zhou et al 18 developed three ML models (RF, SVR and ANN) to predict the Young's modulus and hardness of various HECs, with MAE of only 15.3 GPa and 1.1 GPa, showing high prediction accuracy.…”

Section: Introductionmentioning

confidence: 99%

Predicting Mechanical Properties of Non-Equimolar High-Entropy Carbides using Machine Learning

Zhao,

Cheng

et al. 2024

Preprint

View full text Add to dashboard Cite

High-entropy carbides (HECs) have garnered significant attention due to their unique mechanic properties. However, the design of novel HECs has been limited by extensive trial-and-error strategies, along with insufficient knowledge and computational capabilities. In this work, the intrinsic correlations between elements in the high-dimensional compositional space of HECs are investigated using high-throughput density functional theory calculations and two machine learning models, which enable us to predict the Young's modulus, hardness and wear resistance only using a chemical formula. Our models demonstrate low root mean square error (11.5GPa) and mean absolute errors (9.0GPa) in predicting the mechanic properties of HECs with arbitrary non-equimolar compositions. We further establish a database of 566,370 HECs and identified 15 novel HECs with best mechanical properties. Our models can rapidly explore the mechanical properties of HECs with descriptor-property correlation analysis, and hence provide an efficient method for accelerating the design of non-equimolar high-entropy materials with desired performance.

show abstract

“…[7][8][9][10] While numerous studies have focused on developing ML models using experimental data, these datasets are relatively small, typically ranging from hundreds to thousands of data points, and cover only a limited portion of the potential design space. [11][12][13] High-throughput density functional theory (DFT) calculations have become a key method for generating extensive materials data. Recent efforts have led to the curation of several large DFT datasets encompassing millions of materials.…”

Section: Introductionmentioning

confidence: 99%

Efficient first principles based modeling via machine learning: from simple representations to high entropy materials

Li,

Choudhary,

DeCost

et al. 2024

J. Mater. Chem. A

View full text Add to dashboard Cite

show abstract

Machine learning-assisted mechanical property prediction and descriptor-property correlation analysis of high-entropy ceramics

Cited by 23 publications

References 41 publications

Explaining the entropy forming ability for carbides with the effective atomic size mismatch

Explaining the entropy forming ability for carbides with the effective atomic size mismatch

Predicting Mechanical Properties of Non-Equimolar High-Entropy Carbides using Machine Learning

Efficient first principles based modeling via machine learning: from simple representations to high entropy materials

Contact Info

Product

Resources

About