Representative three-dimensional structural data for porous media

Obtain representative three-dimensional structural data for porous media despite the cost, limited sampling volume, demanding preparation, configuration complexity, segmentation challenges, and resolution limitations of current high-resolution imaging techniques.

Background

Accurate three-dimensional characterization of porous-media microstructures is important for understanding and predicting macroscopic material properties, including transport behavior in clays, carbonates, and sandstones. The paper explains that pore geometry and connectivity directly influence these properties and applications such as energy systems and geological disposal of radioactive waste.

Existing imaging methods do not readily provide representative three-dimensional datasets. Focused ion beam–scanning electron microscopy offers direct three-dimensional access but is costly, samples only limited volumes, requires demanding preparation and complex instrument configuration, and presents segmentation challenges. X-ray tomography is more accessible but remains limited in spatial resolution. The resulting scarcity of representative three-dimensional ground-truth data motivates computational approaches that synthesize three-dimensional microstructures from two-dimensional images.

References

Obtaining representative 3D structural data, however, remains an open challenge.

— Generating Heterogeneous 3D Geological Microstructures from 2D Images via a Stable Diffusion-Adversarial Model  (2609.20358 - Aouf et al., 17 Sep 2026) in Section 1, Introduction