MengChu Yang 1* , Yong Du 1, Yuling Liu 1, Yong Liu 2

1 State Key Laboratory of Powder Metallurgy, Central South University, Changsha 410083, China
2 Jiangxi Province Key Laboratory of Light Alloy, Nanchang University, Nanchang 330031, China

*22025116@csu.edu.cn

Al-Mg-Zn(-Cu) alloys serve as key materials in the aerospace and automotive lightweighting industries due to their outstanding balance of strength, plasticity, and corrosion resistance[1,2]. However, due to the limited understanding of precipitation mechanisms, composition design largely relies on the traditional "trial-and-error" approach, leading to low efficiency in alloy development. This study proposes and implements a novel alloy design strategy that integrates data-driven and physics-based modeling.

Firstly, machine learning algorithms and CALPHAD phase diagram calculations[3,4] are integrated to construct a quantitative predictive model linking composition, phase formation, and properties. Machine learning rapidly extracts critical features and underlying patterns from historical data, enabling efficient composition screening, while CALPHAD provides theoretical guidance on phase stability and precipitation sequences from a thermodynamic equilibrium perspective. The synergy of these two methods significantly accelerates the composition optimization process and substantially reduces experimental trial-and-error costs. Building on this, techniques including synchrotron small-angle X-ray scattering[5,6], high-resolution transmission electron microscopy and differential scanning calorimetry are employed to systematically reveal the microstructural evolution and precipitation sequences of the alloys at different aging stages. The precipitation kinetics are further simulated using the KWN mode[7,8], clarifying the growth and coarsening behaviors of precipitates.

Moreover, first-principles calculations are applied to uncover the mechanisms by which Cu influences the nucleation, growth and stability of precipitates, thereby effectively elucidating the physical principles underlying the "black-box" machine learning model and achieving a deep integration of data-driven prediction and physics-based interpretation. This work not only provides a robust methodological framework for the rapid development of high-performance Al-Mg-Zn-Cu alloys but also deepens the understanding of precipitation strengthening mechanisms in complex alloy systems through the coupled multiscale simulation and experiment, establishing a theoretical and technical foundation for the intelligent design of materials.

References

[1] L. Stemper, M.A. Tunes, P. Oberhauser, P.J. Uggowitzer, S. Pogatscher, Agehardening response of AlMgZn alloys with Cu and Ag additions, Acta Mater. 2020, 195, 541–554.
[2] L. Stemper, M.A. Tunes, R. Tosone, P.J. Uggowitzer, S. Pogatscher, On the potential of aluminum crossover alloys, Prog. Mater. Sci. 2022, 124,100873.
[3] F. Y. Gao, Y. L. Liu, L. G. Zhang, Y. Du, H. J. Seifert, Thermodynamic investigation of the Al2O3-CaO-Fe2O3-FeO system: Heat capacity measurement, modeling, and application, J. Mater. Sci. Technol. 2026, 260, 53-66.
[4] Y. C. Shi, Y. F. Wang, M. F. Tang, Y. P. Zeng, Y. L. Liu, B. Hu, Y. Du, Thermodynamic calculation of the Mn-Sb and Mn-Bi-Sb systems, Mater. Today Commun. 2025, 48, 113423.
[5] T. Marlaud, A. Deschamps, F. Bley, W. Lefebvre, B. Baroux, Acta Mater. 2010, 58, 248–260.
[6] A. Deschamps, C. Sigli, T. Mourey, F. de Geuser, W. Lefebvre, B. Davo, Acta Mater. 2012, 60, 1917-1928.
[7] Y. Yang, J. Wang, M. R. G. Ferdowsi, S. R. Kada, T. Dorin, M. R. Barnett, M. Perez, Acta Mater. 2024, 281, 120392.
[8] J. Tang, M. C. Liu, G. W. Bo, F. L. Jiang, C. H. Luo, J. Teng, D. F. Fu, H. Zhang, J. Mater. Sci. Technol. 2022, 116, 130-150.

Dr. Mengchu Yang