A new approach to design multicomponent metallic glasses using the mendeleev number. Issue 24 (17th December 2022)

• Record Type: Journal Article
• Authors: Bajpai, Anurag; Bhatt, Jatin; Gurao, Nilesh P.; Biswas, Krishanu
• Is Part Of: Philosophical magazine. Volume 102:Issue 24(2022)
• Page Start: 2554
• Page End: 2571
• Publication Date: 2022-12-17
• Subjects: Multicomponent metallic glasses -- atomic clusters -- mendeleev number -- machine learning -- random forest; Condensed matter -- Periodicals; Physics -- Periodicals; Matière condensée -- Périodiques; Physique -- Périodiques; 530.41
• DOI: 10.1080/14786435.2022.2121868
• Languages: English
• ISSNs: 1478-6435

Abstract

ABSTRACT: Designing novel Multicomponent Metallic Glasses (MMGs) based on empirical parameters such as enthalpy of mixing ( Δ H m i x ) and configurational entropy ( Δ S m i x ) is a time-consuming exercise that requires various assumptions, limiting the capability to predict new MMG compositions. The current study involves constructing a modified Mendeleev Number ( M N P ) element scale based on many important elemental properties that impact the glass forming phenomena. Machine learning (ML) was used to assess the competence of the proposed M N P to predict MMGs. The ML findings demonstrate that proposed M N P can be utilised as a salient attribute to predict MMGs with 87.8% cross-validation accuracy. Further, the mean square variation in the M N P of the alloy constituents ( Δ M N P ) provides a delineated zone of glass forming multicomponent alloys. In summary, the research work presents a novel phenomenological coordinate system that can effectively predict new MMGs while avoiding the limitations of empirical parameters based design strategies.