Accelerated development of high-strength and high-conductivity Cu-Cr-Ti alloys based on data-driven design and experimental validation
Feng Li, Jiangnan Li, Qiong Lu, Yuanqi You, Zunyan Xu, Liyuan Liu, Fusheng Li, Gao Peng, Jianhong Yi, Caiju Li
Abstract
• Machine learning predicts the effect of a single element on the alloy's conductivity. • Adding Zn to Cu-0.4Cr-0.06Ti simultaneously enhances strength and conductivity. • The addition of Zn promotes the precipitation of the Cr-rich phase. • Zn addition accelerates the Cu/Cr interface transition from coherent to incoherent. Copper alloys, valued for their excellent electrical conductivity and mechanical properties, are widely applied in electronics, power systems, and related fields. However, the extensive diversity and compositional range of alloying elements pose substantial challenges in alloy design. To address this challenge, this study applied a machine learning approach: a Support Vector Regression (SVR) based “composition-conductivity” model was constructed to predict the impact of individual elements on the alloy’s electrical conductivity. According to the prediction results, Zn element was added to Cu-0.4Cr-0.06Ti alloy. Through experimental validation, it was shown that adding 0.05 wt% Zn achieves an ultimate tensile strength of 507 MPa, an electrical conductivity of 79 % IACS, and an elongation of 23 %. Morphology characterization revealed the role of Zn in the alloy: Zn was present in the matrix as a substitutional solid solution, while Cr was present as an interstitial solid solution. The addition of Zn promoted Cr precipitation and accelerated the transformation of Cr-rich phases, altering the interface between the matrix and precipitates from coherent to incoherent, thus reducing lattice distortion. This adjustment in solute elements and interfacial relationships enhanced both electrical conductivity and strength, breaking through the inverted relationship between strength and conductivity of copper alloy Furthermore, this study demonstrated that machine learning-based composition optimization effectively guides experimental design, providing new insights for the development of high-performance copper alloys.