Nucleation of recrystallisation is critical for the microstructure of thermomechanically processed metals. Even though some specific features in the microstructure are known to be preferential nucleation sites – such as grain boundaries, shear bands and large second phase particles, it is today not possible to predict where nucleation will take place – a missing piece in the jigsaw puzzle may be the spatial distribution of the local plastic strain. The aim of this work is therefore to study possible relationships between local plastic strain and nucleation probability and furthermore relate that to the specific nucleation site, using a multimodal multiscale approach.

Nucleation is investigated in an AA1050 sample, containing Al-Fe-Si and Si particles. The sample was mapped in 3D during tensile deformation as well as during subsequent heat treatments at the European Synchrotron Radiation Facility (ESRF). The tensile deformation was followed by phase contrast tomography (PCT), while the microstructure evolution was mapped by diffraction contrast tomography (DCT) and scanning 3D x-ray diffraction (3DXRD) as the sample was heat treated. The spatial distribution of local plastic strain is quantified by digital volume corelation (DVC). A new forward reconstruction implementation allowed reconstruction of the deformed microstructure, while the standard grain tracking method appears to properly reconstruct nuclei. This combined knowledge provides unique data which for the first time allow pinpointing the active nucleation and relate that to the local plastic strains as well as to the deformed microstructure in 3D, and to follow the growth of the nuclei during difference time steps. It is discussed how these data may be used to advance recrystallization modelling, so that future models will become able of also correctly predicting the nucleation of recrystallization.