Quantitative prediction of interfacial void evolution linking process parameters to joint performance in Ti-6Al-4V diffusion bonding

Interfacial void evolution is critical to the service reliability of diffusion-bonded Ti-6Al-4 V components with complex internal cavities or microchannel structures, which are widely used in aerospace, marine, and energy industries. This study establishes a quantitative framework linking process parameters, void characteristics, and mechanical performance of Ti-6Al-4 V diffusion-bonded joints through multi-scale experiments, mechanical modeling, and machine learning. The diffusion-bonded joints with designed void characteristics were prepared by tailoring process parameters and surface topography, from which the interfacial bonding ratio and key void descriptors were extracted. The bonded zone achieved metallurgical bonding through the synergistic action of dynamic recrystallization, grain boundary migration, and elemental interdiffusion. Quantitative “S-curve” relationships were established between the bonding ratio and both the tensile strength and fracture elongation of the joints. Furthermore, the void-tip stress concentration coefficient and the ratio of void spacing to width are proposed as supplementary metrics for a more comprehensive assessment of interfacial quality. Among four machine-learning models evaluated, the meta-learning approach exhibited the highest accuracy and convergence efficiency in predicting void morphology and bonding ratio under multi-parameter coupling. This work provides a theoretical and data-driven framework for precisely controlling interfacial integrity and optimizing the performance of diffusion-bonded components.