Semantic segmentation of synchrotron tomography of multiphase Al-Si alloys using a convolutional neural network with a pixel-wise weighted loss function.

Human-based segmentation of tomographic images can be a tedious time-consuming task. Deep learning algorithms and, particularly, convolutional neural networks have become state of the art techniques for pattern recognition in digital images that can replace human-based image segmentation. However, their use in materials science is beginning to be explored and their application needs to be adapted to the specific needs of this field. In the present work, a convolutional neural network is trained to segment the microstructural components of an Al-Si cast alloy imaged using synchrotron X-ray tomography. A pixel-wise weighted error function is implemented to account for microstructural features which are hard to identify in the tomographs and that play a relevant role for the correct description of the 3D architecture of the alloy investigated. The results show that the total operation time for the segmentation using the trained convolutional neural network was reduced to <1% of the time needed with human-based segmentation.

Revealing the Effect of Local Connectivity of Rigid Phases during Deformation at High Temperature of Cast AlSi12Cu4Ni(2,3)Mg Alloys.

The 3D microstructure and its effect on damage formation and accumulation during tensile deformation at 300 °C for cast, near eutectic AlSi12Cu4Ni2Mg and AlSi12Cu4Ni3Mg alloys has been investigated using in-situ synchrotron micro-tomography, complemented by conventional 2D characterization methods. An increase of Ni from 2 to 3 wt.% leads to a higher local connectivity, quantified by the Euler number χ, at constant global interconnectivity of rigid 3D networks formed by primary and eutectic Si and intermetallics owing to the formation of the plate-like Al-Ni-Cu-rich δ-phase. Damage initiates as micro-cracks through primary Si particles agglomerated in clusters and as voids at matrix/rigid phase interfaces. Coalescence of voids leads to final fracture with the main crack propagating along damaged rigid particles as well as through the matrix. The lower local connectivity of the rigid 3D network in the alloy with 2 wt.% Ni permits localized plastification of the matrix and helps accommodating more damage resulting in an increase of ductility with respect to AlSi12Cu4Ni3Mg. A simple load partition approach that considers the evolution of local connectivity of rigid networks as a function of strain is proposed based on in-situ experimental data.