Image processing and segmentation open source codes applied to FIB-SEM images of ultra-tight gas shales samples: Enhanced pore space representativeness and mineral identification

Geological formations, such as shales and mudstones, are highly compact and challenging to characterize experimentally. Beyond artifacts associated with the chemical composition of the material, certain formations exhibit topological and morphological properties that, during each FIB ablation, generate shading effects (back-pore artifacts). As in the present case, these artifacts may prevent access to a connected pore network during reconstruction and FIB-SEM processing. This issue compounds other challenges that constrain the ability to accurately reconstruct a digital twin of the porous medium of interest. In this study, we present the solutions we have developed to mitigate common artifacts, with a particular focus on addressing back-pore artifacts. This effort aims to advance Digital Rock Physics techniques, extending their analytical capabilities for complex porous media. The workflow outlined here details the processes required to characterize mineralogy and quantify flow properties based on a representative pore-space geometry. At the Focused Ion Beam - Scanning Electron Microscopy (FIB-SEM) scale (nm to μm), characteristic of tight formations, the proposed methodology can be extrapolated to larger volumes. It enables the quantification of transport properties at the pore scale with a high degree of confidence as a function of fluid pressure. The proposed imaging process, in conjunction with the Lattice Boltzmann Method, provides an effective approach for evaluating transport properties within a given pore-scale geometry. Furthermore, it highlights the method's potential for investigating transport behavior as a function of pressure and temperature. • The image processing uses a VSNR filter to correct structured noise including curtaining and charging artifacts. • A machine learning based segmentation returning an accurate and reproducible classification of shale mineral phases. • The back pore artifact is corrected driving to some multiple percolating flow paths in a 10 × 10 × 10 μm 3 rock sample. • The connectivity retrieval permits to access to the 10 nD permeability connected porosity measured with a Lattice Boltzmann Code. • A novel image processing and segmentation methodology for DRP exploitation of tight media using FIB-SEM.