Toward XRF-like chemical sensitivity in the laboratory: first hyperspectral x-ray imaging demonstration using the...

X-ray spectral imaging is an advanced technique that enables material-sensitive imaging by exploiting energy-dependent interactions with matter. In this context, X-ray fluorescence (XRF) represents the benchmark technique. Despite its excellent elemental sensitivity, however, XRF is intrinsically inefficient because it relies on the detection of isotropically emitted secondary fluorescence induced by a focused beam on the sample. This requirement makes the technique difficult to implement with compact laboratory sources and, especially for tomographic applications, highly brilliant synchrotron sources are required. In a laboratory environment, spectral imaging and tomography are typically performed using photon-counting detectors. While this technology offers significant advantages, the limited number of energy thresholds and coarse energy resolution can hamper the separation of materials with similar attenuation properties. Hyperspectral detectors, featuring sub-keV energy resolution and virtually unlimited spectral binning, provide a technological solution to enable high-sensitivity, chemical-specific imaging in the laboratory. Here we present the first application of the novel CITIUS hyperspectral detector to X-ray micro-CT and radiography at the OptImaTo (Optimal Imaging and Tomography) laboratory (Trieste, Italy), based on a liquid MetalJet source (Excillum, Sweden) with a galinstan anode. Using multiple characteristic emission lines (Ga, In, Sn) and a ⁵⁵Fe source, a sharp energy resolution in the 0.5-0.8 keV range (full width at half maximum) was found, enabling fine energy binning suitable for advanced quantitative material identification. For the first demonstration, planar and tomographic datasets of two multi-material test samples were analyzed using a newly adapted version of the Minimum-Residual Basis Material Decomposition (MR-BMD) algorithm, optimized for hyperspectral detectors providing tens of energy bins with narrow bandwidths. Results show that the laboratory-based hyperspectral approach combined with MR-BMD enables element-sensitive imaging and, remarkably, separates materials with very similar attenuation, such as water and polypropylene. These results demonstrate accurate material identification and quantification, promisingly approaching XRF-like chemical sensitivity in the laboratory.