Quantifying resolution-dependent microstructure and pore-network topology in LDH-modified alkali-activated slag pastes via synchrotron X-ray nano-holotomography and multiscale tomography

This study establishes a multiscale framework that quantitatively links X-ray computed tomography (XCT) resolution with pore-network topology and transport-related parameters in alkali-activated slag (AAS) binders. Laboratory μCT, synchrotron nano-CT, and nano-holotomography (SR-HT) were combined to capture sub-micrometre to micrometre features, supported by a customised multi-step machine-learning segmentation ensuring consistent cross-resolution quantification. Results reveal that XCT spatial resolution strongly governs the perceived topology: porosity increases nearly linearly with voxel refinement, while coarser scans obscure small pores and oversimplify connectivity, leading to underestimated tortuosity and overestimated permeability. Incorporation of layered double hydroxide (LDH) accelerates hydration and refines microstructure, producing denser matrices and reduced permeability, with effects more pronounced in sodium‑carbonate-activated systems than in sodium-sulphate ones. These findings highlight the need to balance resolution and representativity when interpreting 3D pore-networks, and the proposed multiscale XCT–machine-learning workflow offers a transferable route for quantifying resolution-dependent topology–transport relationships in cementitious materials.