LÜ Pingping, SUN Ruixia, SUN Xuan, ZHANG Cong, ZHANG Ruijie, WANG Yongwei, YIN Haiqing, QU Xuanhui
【Objective】The development of advanced metallic materials requires simultaneous consideration of mechanical properties, service performance, manufacturing feasibility, and cost. Strongly coupled and nonlinear relationships among composition, processing, microstructure, and properties make conventional trial-and-error approaches inefficient. This review aimed to clarify the methodological framework of inverse design for metallic materials, summarize advances across different design levels, and identify challenges and future opportunities, with particular attention to intelligent design in powder metallurgy.
【Method】Relevant studies were retrieved from Web of Science and Engineering Village using keywords including “inverse design” and “alloy”. The literature was organized around target specification, design-space definition, data collection and feature engineering, forward-model development, physically constrained candidate generation, and computational and experimental validation. Two principal routes were examined: optimization-based search coupled with forward models and conditional generation of candidate materials. Applications were reviewed in alloy composition design, processing-parameter determination, microstructure generation, and integrated composition-processing-microstructure design. Forward prediction and inverse design in powder metallurgy were discussed separately. The roles of large language models, thermodynamic calculations, first-principles calculations, and multiscale simulations were also examined.
【Result】The reviewed studies show that inverse design is a constrained search for potentially multiple feasible solutions rather than a simple mathematical inversion of composition-property relationships. Forward models provide candidate evaluations, whereas optimization or conditional generation enables the identification of compositions, processing conditions, or structures from specified targets. Composition design increasingly addresses competing property requirements while incorporating phase stability, density, and cost constraints. Processing design connects target properties with thermal and deformation histories through phase transformation, recrystallization, precipitation, and defect evolution. Microstructure generation expands the design variables to grain morphology, crystallographic orientation, phase distribution, and pore architecture, although experimental realization remains less established than computational validation. Integrated design links composition and processing with microstructural descriptors, providing a route to more physically interpretable candidate selection. In powder metallurgy, existing applications include prediction of sintered density, dimensional changes, and mechanical properties, together with inverse selection of powder systems, alloying additions, compaction conditions, and sintering schedules. Recent bidirectional models extend this approach to spark plasma sintering by incorporating microstructural and property requirements. However, inverse-design studies remain relatively limited compared with forward-prediction applications. Across these areas, thermodynamic and kinetic constraints help exclude physically implausible candidates, while experimental characterization and property testing establish whether predicted improvements are attainable. Large language models support literature-data extraction and knowledge organization, but their outputs require verification.
【Conclusion】Inverse design of metallic materials is evolving from isolated parameter optimization toward coordinated composition-processing-microstructure design. Its engineering value depends on physically consistent representations, reliable model evaluation, manufacturing constraints, and experimental feedback. Priorities include standardized datasets, uncertainty-aware candidate selection, integration of thermodynamic and kinetic knowledge, and active-learning loops connecting computation with experiments. For powder metallurgy, closer integration of powder characteristics, densification behavior, pore evolution, and final properties is particularly important for developing reliable inverse-design frameworks. These advances provide a basis for accelerating materials development while balancing performance, manufacturability, and resource efficiency.