Machine Learning Framework Boosts Search for High-Entropy Heat-Resistant Alloys with Better High-Temperature Yield Strength

Machine Learning Framework Boosts Search for High-Entropy Heat-Resistant Alloys with Better High-Temperature Yield Strength

In an article recently published in the open access journal npj Computer materialsthe researchers discussed the intelligent framework based on machine learning (ML) to find high-entropy refractory alloys with improved high-temperature yield strength.

Study: Intelligent Machine Learning Based Framework for Discovery of High Entropy Heat Resistant Alloys with Improved High Temperature Yield Strength. Image Credit: Quardia/


Promising materials known as High Entropy Alloys (HEAs) have generated a lot of interest. High Temperature Heat Resistant Alloys (RHEA) have been shown through experimental research to have better high temperature strength than superalloys, making them a desirable class of alloys for further study for prospective use. in high efficiency gas turbine engines.

HEAs offer enormous possibilities, but they also pose difficult problems for materials scientists faced with examining a design space with an enormous number of potential compositions. The lack of a broad understanding of the variables that determine the chemical and mechanical characteristics of these complex alloy systems is one of the major challenges hampering the rapid development of HES.

Modern high-resolution imaging techniques can be used to collect atomic and microstructural information, but because they are time-consuming and expensive, they cannot fully investigate and characterize the huge compositional space. Accordingly, the main objective of HEA research has been to formulate guidelines for phase formation as well as atomic and microstructural characteristics.

The literature has many working reports to categorize phases using criteria. However, creating standards to improve mechanical qualities is a topic that has received relatively little research. Recently, several publications have been published that use different ML techniques to predict HEA steps.

About the study

In this study, the authors discussed building a state-of-the-art machine learning system coupled with optimization techniques to intelligently explore the huge compositional space and improve high temperature yield strengths.

The proposed yield stress model significantly outperformed the state-of-the-art method in terms of predicted accuracy, and it also offered quantification of inherent uncertainty through the use of repeated k-fold cross-validation. The linked framework was used to identify RHEAs with higher high-temperature yield strength after building and validating a reliable yield strength prediction model. RHEA compositions with a maximum yield strength at a particular temperature have been designed.

The team intended to replace the experiment cycle only with intelligent ML-based models to filter out HEAs and reduce the search space. As an illustration, this study examined the elastic limits of RHEAs and developed a complete ML model by selecting important descriptors from a set of many descriptors. To find RHEA compositions with improved yield strengths, the direct model was combined with a stochastic genetic algorithm37.

With regard to the determining factors improving the yield strength at both low and high temperature, important information has been obtained.

Researchers analyzed to understand how physical and thermodynamic characteristics contributed to the increase in yield strength. RHEA compositions with suitable yield strengths at particular temperatures were identified using the proposed model based on ML.


Compared to the experimental data, the proposed model had an average absolute error of 147 MPa for NbTaTiV and 224 MPa for CrMoNbV. The improved alloy outperformed the base alloy by 90 MPa at 25°C. The yield strength of the optimal alloy at 1000°C showed a very different temperature dependence, remaining nearly constant between 25 and 800°C. The yield strength of the base alloy has been increased by 13% at 1000°C. However, the ideal alloy at 1000°C had a much lower yield strength at 25°C. Significant improvements have been made over the base alloy compositions at both 25°C and 1000°C.

Concentrations of Ti, Nb and Zr were reduced relative to the base alloy for ideal alloys at 25°C in nearly comparable amounts. However, both Mo and V concentrations increased dramatically. The Ti fraction was increased and the V fraction remained almost unchanged for the ideal alloy at 1000°C.

The composition and descriptors of the alloys optimized for yield strength at 25°C and 1000°C differed significantly, demonstrating that the mechanisms and standards for maximizing low temperature and high temperature strength could differ significantly, and compositions that maximize yield strength at room temperature could not do so for elevated temperatures.


In conclusion, this study used ML and optimization to show how a smart computational framework could predict the yield strength of RHEA and find RHEA compositions that theoretically outperform the initial RHEA. Repeated k-fold cross-validation combined with feature selection has been shown to be an effective method to obtain a more statistically significant prediction of all data points.

The authors combined the robust ML-based yield strength prediction model with a genetic algorithm to find RHEA compositions with improved yield strengths. The algorithm intelligently explored the complex composition space up to the maximum elastic limit given a baseline starting with RHEA.

The idea was illustrated using three different base alloys from the RHEA literature. Optimum alloy compositions have been predicted for both 25°C and 1000°C, with improved yield strengths of up to 80%. Using a broad strategy, the team projected the low and high temperature yield strengths for 252 equiatomic RHEA chemicals. Using the proposed generic approach, the elemental composition was matched to the best candidate for each temperature.

The authors mentioned that the work presented in this study establishes a framework to tackle the enormous task of finding HEAs that experimentally satisfy the requirements of many attributes.


Giles, SA, Sengupta, D., Broderick, SR, et al. Smart framework based on machine learning to discover high-entropy refractory alloys with improved high-temperature yield strength. npj Computer materials 8, 235 (2022).

Disclaimer: The views expressed here are those of the author expressed privately and do not necessarily represent the views of Limited T/A AZoNetwork, the owner and operator of this website. This disclaimer forms part of the terms of use of this website.

#Machine #Learning #Framework #Boosts #Search #HighEntropy #HeatResistant #Alloys #HighTemperature #Yield #Strength

Leave a Comment

Your email address will not be published. Required fields are marked *