Abstract
We apply machine learning algorithms to optimize thermodynamic and elastic properties of multicomponent Fe-Cr alloys with additions of Ni, Mo, Al, W, V, and Nb. The target properties are mixing enthalpy, Young's elastic modulus, and the ratio between shear and bulk moduli, which is often used as a phenomenological criterion for a material's ductility. We thoroughly analyze the descriptors that provide the robust performance of the machine learning models. Next, the iterative active learning method is used for the optimization of the chemical composition to simultaneously improve both thermodynamic stability and the elastic properties of Fe-Cr-based alloys. As a result, we predict compositions of thermodynamically stable alloys with improved mechanical properties, demonstrating the high potential of data-driven computational design in the field of materials for nuclear energy applications.
Highlights
Fe-Cr-based bcc alloys are widely used as important structural and industrial steels
Theoretical predictions, which take advantage of state-of-the-art first-principles simulations combined with alternative data-driven methods, such as machine learning, are believed to significantly accelerate discovery of novel materials and shorten the time it takes to bring them to practical applications [2]
Considering a task of optimization of stability and elastic properties of multicomponent Fe-Cr-based alloys, an important material system; e.g., for nuclear energy applications, we have investigated the applicability of the machine learning algorithms for efficient search of unique promising alloys
Summary
Fe-Cr-based bcc alloys are widely used as important structural and industrial steels. They are known for their resistance to corrosion and irradiation-induced swelling at high temperatures [1]. This makes the Fe-Cr-based steels a suitable candidate material for harsh environments in, for example, nuclear and fusion reactors. At the same time, increasing demands for materials performance, e.g., for the generation of nuclear reactors, call for a development of novel steels. At the same time, increasing the number of alloy components could greatly increase the complexity of purely experimental materials design, leading to longer development time and higher costs. Theoretical predictions, which take advantage of state-of-the-art first-principles simulations combined with alternative data-driven methods, such as machine learning, are believed to significantly accelerate discovery of novel materials and shorten the time it takes to bring them to practical applications [2]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
