29 August 2025 in IEEE Transactions on Instrumentation and Measurement
This paper proposes an enhanced submicron-resolution 3D surface reconstruction framework for in-situ scanning electron microscopy (SEM) characterizing metal deformation evolution, establishing the first practical solution for spatio-temporal 3D surface analysis in dynamic polycrystalline systems. The methodology integrates an optimized gradient computation algorithm with a multi-scale reconstruction (MSR) strategy, specifically addressing high-slope surface recovery challenges under mechanical loading. Key innovations include a cubic compensation term for mitigating steep-region signal occlusion and a hierarchical Gaussian pyramid architecture that preserves more microscopic features through residual propagation while suppressing measurement noise. Experimental validation on IN718 superalloy demonstrates dual-scale characterization capabilities: submicron-resolution reconstructions during early plastic deformation capture slip trace activation dynamics and cross-slip system interactions. In contrast, large-area reconstructions in later stages quantitatively reveal grain boundary collapse patterns and strain gradient distributions preceding crack nucleation. The demonstrated approach bridges slip behavior and dual-scale deformation mechanisms, offering significant implications for material design and failure prediction.