Hierarchical nanoporous gold (HNPG) is a uniform network of ligaments with a typical ligament size of 15 – 110 nm and pore size between 5 to 20 nm. This material exhibits unique mechanical properties, e.g. enhanced stiffness and strength in comparison to geometrically similar structures with only one characteristic length scale [1]. However, the materialographic investigation of the HNPG is challenging and requires its accurate reconstruction.   

Focused ion beam (FIB) tomography is a destructive technique used to collect three-dimensional (3D) structural information at a resolution of a few nanometers. For FIB tomography, a material sample is degraded by layer-wise milling. After removing each layer, the new surface is imaged by a scanning electron microscope (SEM), providing a consecutive series of cross-sectional images of the HNPG structure.  

For nanoporous materials, the reconstruction of the 3D microstructure of the material, from the information collected during FIB tomography, is impaired by the so-called shine-through effect. This effect prevents a unique mapping between voxel intensity values and material phase (e.g., solid or void) and often substantially reduces the accuracy of conventional methods for image segmentation.  

In this poster, we demonstrate the use of machine learning-based algorithms for the 3D reconstruction of HNPG. A significant bottleneck in using machine-learning models is the availability of sufficient training data. To overcome this, we present an approach to generate synthetic training data in the form of FIB-SEM images generated by Monte Carlo simulations [2]. We show that machine learning-based methods enhance the segmentation performance significantly. 

References 

[1] S. Shi, et al., Science, 2021, 371, 1026-1033. 

[2] Sardhara, et al., Frontiers in Materials, 2022