Atomic spectrometry update: review of advances in X-ray fluorescence spectrometry

Contents

• 1. Introduction

• 2. Chemical imaging using X-ray spectrometry techniques
  • 2.1 Computed tomography and 3D XRF spectrometry techniques
  • 2.2 Laboratory 2D XRF spectrometry techniques

• 3. Synchrotron and large-scale facilities

• 4. Grazing X-ray spectrometry techniques including TXRF spectrometry

• 5. Hand-held, mobile and online XRF spectrometry techniques
  • 5.1 Hand-held and mobile XRF spectrometry techniques
  • 5.2 Online XRF spectrometry techniques

• 6. Cultural heritage applications

• Abbreviations

• Conflicts of interest

• References

---

Abstract

This review of 89 references covers advances in X-ray fluorescence spectrometry and its special applications, published from April 2024 to March 2025 inclusive. It provides critical insights into developments in instrumentation, methodologies and data handling, representing significant progress in XRF spectrometry. Applications of cultural heritage are also covered. Highlights of this review period include notable research findings. A method was developed to overcome self-absorption effects in confocal XRF spectrometry, enabling quantitative and distortion-free 3D elemental analysis by combining sample density information from μCT with mass attenuation coefficients from absorption measurements. Additionally, compositional data from fundamental parameter quantification of reference-free XRF spectrometry for ‘dark matrix’ elements like C, N and O in the soft X-ray region were incorporated. Using SR-XRF spectrometry, quantitative nano-characterisation of ion beam-implanted samples, particularly focusing on Ga dopants in silicon, was achieved. This method detected a minimum of 3000 Ga atoms per pixel with a 1 s integration (171 nm² spot) and 650 Ga atoms with a 25 s integration, corresponding to LODs of 18 impurities per nm² and 3.8 impurities per nm², respectively. The determination of elemental profiles in size-segregated airborne particulates with high time-dependent resolution (in <1 h) made significant progress with the development of an impactor specifically designed for TXRF spectrometry. A laboratory scanning-free GEXRF spectrometry setup featuring off-the-shelf equipment such as a Cr X-ray tube and a CMOS detector, showed remarkable results, providing measurements close to those obtained from SR facilities, such as the line height of Ti-oxide nanostructures (58 nm) and HfO₂ thicknesses (2.3 nm). We can expect a considerable increase of research in this area in the future. This year, reviews were published highlighting the advancements and applications of hand-held XRF spectrometry techniques over time, specifically for measuring bone lead and for determining the elemental composition of food samples. A comprehensive review provided an overview of synchrotron applications for cultural heritage over the past decade.

---

1. Introduction

This review should be read alongside the previous review¹ and other related ASUs in the series, including advances in environmental analysis,² advances in the analysis of clinical and biological materials, foods and beverages,³ advances in atomic spectrometry and related techniques,⁴ advances in elemental speciation⁵ and advances in the analysis of metals, chemicals and materials.⁶ This review offers a critical perspective on the advancements in instrumentation, methodologies and data handling that mark significant progress in XRF spectrometry. While it does not aim to comprehensively cover all applications of XRF spectrometry techniques, it highlights specific cases where the non-invasive nature and in situ, operando capabilities of XRF spectrometry analysis are particularly valuable and preferred. These applications often involve irreplaceable samples of high scientific and cultural importance, mainly including works of art and archaeological artefacts.

A noteworthy publication is the online textbook, Geostandards and Geoanalytical Research Handbook of Rock and Mineral Analysis. This fully revised and updated edition of the 1987 Handbook of Silicate Rock Analysis by P. J. Potts includes significant updates. Potts, a former respected member of this review team, contributed to the revision. Chapter 6 is divided into two parts. Part 1 (ref. 7) delves into the fundamentals of XRF spectrometry and matrix corrections. Part 2 (ref. 8) discusses WD and ED instrumentation, comparing their types and analytical performance, with a focus on silicate rock analysis. Additionally, the textbook covers TXRF and field portable XRF spectrometry instrumentation, as well as statistical procedures relevant to XRF spectrometry analysis.

2. Chemical imaging using X-ray spectrometry techniques

2.1 Computed tomography and 3D XRF spectrometry techniques

Three-dimensional chemical imaging using scanning XRF spectrometry techniques continues to evolve significantly, with ongoing improvements in both experimental approaches and methods for quantitative data analysis and reconstruction. Combinations with complementary techniques – such as absorption and phase contrast micro/nano-CT, scanning diffraction and scattering methods and ptychography – are increasingly being employed at SR facilities. These advancements are attracting growing interest across a wide array of research disciplines, including life sciences and biomedical imaging, materials science as well as environmental and Earth sciences. The following section reviews two key techniques central to 3D XRF spectrometry: XRF spectrometry-CT and confocal XRF spectrometry.

2.1.1 X-ray fluorescence spectrometry computed tomography. Elemental mapping through virtual cross-sections and fully 3D elemental imaging using XRF spectrometry-CT are predominantly conducted at synchrotron micro- and nanobeam facilities, targeting sample sizes ranging from microns to millimetres. However, recent developments during this review period highlight a growing interest in advanced laboratory-based systems, particularly those optimised for biomedical applications.

An advanced life science application of laboratory based XRF spectrometry-CT was demonstrated by Saladino et al.⁹ to track multifunctional nanocarriers in mice ιn vivo with high spatial and quantitative precision, representing a significant development to advance nanomedicine. The study made a noteworthy contribution by presenting a novel approach to map and quantify the biodistribution of PEGylated ruthenium-loaded liposomes (Ru-Lipo) using dual-mode imaging, combining optical imaging and XRF spectrometry-CT, addressing longstanding challenges in nanoparticle-based drug delivery research. Major focus was organ-specific accumulation of Ru-Lipo, passive tumour targeting and non-invasive quantification of therapeutic agent delivery. The demonstrated XRF spectrometry-CT setup used a 24 keV monochromatic beam from a liquid metal-jet source, which achieved spatial resolution of 100 μm with a voxel size of 200 μm × 200 μm × 200 μm for whole-body in vivo scans. Using an exposure time of 10 ms per scan step, full-body scan times of ca. 15 min could be achieved with a radiation dose of approximately 1 mGy per whole-body image. Individual organs were scanned with 30 projections over 180°, resulting in acquisition times ranging from 1 to 3.5 hours. The Ru-specific XRF spectrometry signatures allowed accurate, quantitative, tissue-level mapping which overcame many limitations of conventional elemental analysis methods like ICP-MS.

Pu et al.¹⁰ reported notable advances in benchtop XRF spectrometry-CT for biomedical imaging by demonstrating a dual-modality system combining XRF with cone-beam CT. Their setup integrated pixelated photon-counting detectors with innovative multipixel event correction algorithms, significantly enhancing spatial resolution and spectral accuracy. The main innovation lies in correcting the artefacts introduced by multipixel events during photon detection, leading to improved quantitative sensitivity and spatial resolution for imaging high-Z tracers like Gd nanoparticles. The employed CdTe photon counting detector (80 × 80 pixels, 200 μm pixel size) achieved a 1 keV energy resolution at 60 keV. By applying an empirical correction to account for the energy lost in charge sharing during multipixel events, the FWHM of the Gd Kα peak was improved from 3.17 keV to 2.12 keV, enhancing spectral accuracy. In their XRF spectrometry-CT imaging studies, the X-ray source was operated at a power of 450 W (150 kV, 3 mA) with a 0.2 mm Cu filter. Using 45 projections with a rotational step of 8° and exposure time of 60 s at each angle, Gd LOD was approximately 0.1 mg mL⁻¹.

In a similar study, Mandot et al.¹¹ introduced a novel benchtop XRF spectrometry-CT system tailored for preclinical applications, particularly for the quantitative imaging of metal-based agents containing Gd and La. The main goal was to address limitations in prior lab-based XRF spectrometry-CT systems, including poor sensitivity, high radiation dose and limitations in detection geometries. The most significant improvement compared to earlier set-ups was achieved by a full-ring configuration of 24 CdTe-based detectors (80 × 80 pixels, 200 μm pixel size), organised hexagonally and paired with a 96-pinhole inverted compound-eye collimator. This setup enables 96 simultaneous views of the XRF spectrometry signal with high spatial and spectral resolution. Spectral resolution FWHM was 0.5 keV at 35 keV and 0.88 keV at 60 keV X-ray energy, respectively. During the scans, the X-ray source was operated using a voltage of 90 kV at a power level of 60 W, to generate a collimated 1 mm pencil beam irradiating the sample. This was sufficient to achieve a LOD of 0.1 mg mL⁻¹ for Gd in acrylic phantoms, representing approximately a 30-fold sensitivity improvement over previous systems. The total dose for a full scan (37 positions, 600 s per position) was estimated to be approximately 3.26 Gy.

In biomedical applications of XRF spectrometry-CT, a considerable effort has been devoted to developing sophisticated image correction approaches based on the use of deep learning. Mahmoodian et al.¹² developed a deep learning framework optimised for denoising low-dose XRF spectrometry-CT images. The authors focused on the development of a so-called Swin–Conv-UNet (SCUNet) architecture that integrated convolutional layers with transformer blocks, targeting the challenge of balancing image quality with radiation exposure. This study showed that optimised deep learning models like SCUNet significantly enhanced XRF spectrometry-CT image quality at reduced radiation doses, particularly in high-noise scenarios. Their model preserved both structural and quantitative accuracy under significantly reduced dose conditions, even in case of a 4-fold dose reduction, advancing the feasibility of safer in vivo XRF spectrometry-CT imaging. In a similar effort, Kusakari et al.¹³ presented a significant advancement in enhancing the image quality of XRF spectrometry-CT by applying a deep image prior (DIP) network, a type of convolutional neural network, as a pre-denoising step on projection images as opposed to the more conventional post-denoising approach. The methodology addressed critical challenges in preclinical XRF spectrometry-CT imaging, particularly under low-signal conditions relevant to in vivo applications. To assess the denoising performance of DIP pre-processing, iodine solutions of known concentrations in PMMA phantoms were used, scanned by an SR beam of 33.2 keV (flux density 10⁸ to 10⁹ photons per mm² per s), collecting 180 projections each with 180 seconds acquisition time. The integration of DIP as a pre-denoising mechanism markedly enhanced XRF spectrometry-CT imaging performance. It boosted the contrast-to-noise-ratio from 3.7 to 4.6 and LOD for I from 0.069 mg mL⁻¹ to 0.035 mg mL⁻¹ without sacrificing spatial resolution, crucial for imaging low-concentration, non-radioactive-labelled compounds in biological samples.

In the field of nanoscale XRF spectrometry-CT studies at an SR facility, Zheng et al.¹⁴ introduced a dual-branch deep learning model to enhance the resolution of XRF spectrometry microscopy images acquired in fly-scan mode. Such scans inherently suffer from spatial resolution loss due to motion blur and probe size limitations. The dual-branch neural network, combining Residual Channel Attention Blocks and Multi-Scale Residual Blocks, enabled a spatial resolution enhancement by a factor 4, reducing effective pixel size from 30 nm to an impressive 7.5 nm in case of 2D imaging, while the reconstruction resolution of XRF spectrometry-CT data improved from 60.8 nm to 14.8 nm compared to raw XRF spectrometry imaging data. The approach also drastically reduced the acquisition time to achieve a given spatial resolution: from 16 h for a high-resolution XRF spectrometry-CT raw image (91 projections at 30 nm) to about 1 h, required to collect a low-resolution acquisition at 120 nm step size. Using the model, high-resolution equivalent reconstructions could be obtained from low-resolution data with 16 times shorter scan times.

An excellent demonstration of high-resolution XRF spectrometry nano-tomography, combined with complementary imaging modalities (confocal microscopy, AFM, TEM), was presented by Duersch et al.¹⁵ for elemental and morphological imaging of nitrogen-fixing cyanobacteria. The work emphasised the importance of obtaining the spatial distribution of elements such as Ca, Fe, Mn, Mo and S in 3D on the nanoscale, which are crucial for understanding the function of nitrogenase and other enzymes. The 2D/3D scanning XRF spectrometry mappings were performed in conjunction with differential phase-contrast imaging at the Bionanoprobe end-station of the Advanced Photon Source (Argonne National Laboratory, USA) using an incident photon energy of 10 keV. Detected elements include Ca, Co, Cr, Cu, Fe, K, Mn, Ni, P, S and Zn. A full 3D XRF spectrometry-CT dataset was obtained from a total of 51 projections with 100 nm step size and 150 ms per pixel dwell time under cryogenic conditions. This revealed that vegetative cells in cyanobacteria contain Ca clusters surrounding smaller P/K clusters near the cell wall, whereas heterocysts showed higher cytosolic Ca, Fe and K levels, reflecting increased micronutrient needs for nitrogen fixation.

The study by Medjoubi et al.¹⁶ illustrated the capabilities of multimodal scanning hard X-ray imaging and tomography at the NANOSCOPIUM beamline at Synchrotron Soleil, for probing biomineralisation processes at multiple length scales. The experimental platform integrated submicron scanning XRF spectrometry with complementary techniques such as XRD, absorption contrast and X-ray excited optical luminescence. These techniques could be performed simultaneously using continuous scanning modes by the SR beam with a focus in the 70 to 500 nm range and energy between 5 and 20 keV. The XRF spectrometry tomography setup allowed for trace elemental mapping down to ppm sensitivity, crucial for detecting biologically relevant metals such as Fe, Mn and Zn. Both sparse (using limited numbers of projections) and high-resolution tomography modes could be used, balancing scan duration with resolution: sparse tomography reached approximately 8.4 μm resolution with 20 projections, while single-slice tomography achieved approximately 1 μm resolution using 360 projections, as demonstrated on mouse renal papillae samples.

An important application of multimodal synchrotron imaging in plant physiology was presented by van der Ent et al.¹⁷ to study Ni-rich laticifers in Pycnandra acuminata, a hyperaccumulator tree. The work demonstrated the combination of synchrotron XRF spectrometry-CT and phase-contrast imaging microtomography (PCI-μCT) for non-invasive, in vivo visualisation of Ni distribution in Pycnandra acuminata samples. The experiments were performed at the P06 beamline of the PETRA III synchrotron facility using a focused beam of 700 nm × 530 nm at 14 keV energy. By reconstructing tomographic slices from multiple projection angles, XRF spectrometry-CT enabled 3D elemental mapping, though the full 3D imaging was limited by self-absorption effects and lengthy acquisition times. Complementary PCI-μCT was employed to overcome these limitations, offering high-resolution structural imaging over extended volumes, albeit without elemental specificity. The dual approach allowed precise correlation of Ni-rich laticifer networks with anatomical features across roots, stems and leaves. Multi-slice and full-field scans revealed intricate articulation and branching of laticifer cells, providing evidence for potential long-distance Ni transport within the investigated plant.

An innovative combination of ptychographic X-ray computed tomography (PXCT) and XRF spectrometry-CT was demonstrated by Weber et al.¹⁸ for the investigation of coking in industrial zeolite-based propane dehydrogenation catalysts. High-resolution (56–61 nm) 3D maps of electron density were provided using PXCT, giving indirect visualisation of coke deposits primarily in the binder phase, while differentiating zeolite from binder components. To assess pore accessibility directly, the catalysts were loaded with copper ions as XRF spectrometry markers, scanned by XRF spectrometry-CT at the NanoMAX beamline (MAX IV, Lund, Sweden). The strong reduction of Cu XRF spectrometry signal (by ca. 94%) for heavily coked samples indicated a massive blocking of the pores, as opposed to high Cu uptake for fresh samples. Notably, the integrated approach facilitated a spatially resolved comparison of coke-induced inaccessibility at the microscale. The combination of PXCT and XRF spectrometry-CT provided complementary structural and chemical information, highlighting regions of inaccessibility to Cu due to coke and providing an exceptionally useful experimental approach to study catalyst deactivation.

2.1.2 Confocal XRF spectrometry. Confocal XRF spectrometry is becoming increasingly well established at both laboratory- and synchrotron-based facilities, with a growing number of depth-resolved 3D elemental analysis applications reported across various domains, including cultural heritage, materials science and industrial research.

Truly quantitative and distortion-free 3D elemental analysis using confocal XRF spectrometry still poses a challenge by, among others, self-absorption effects within the sample. In order to tackle this issue, the combined use of μCT and confocal XRF spectrometry was discussed in detail by Bauer et al.,¹⁹ demonstrating an accurate absorption correction method for the case of dental composite materials investigated by laboratory instrumentation. The method was tested on a 850 μm thick slice of a bovine tooth filled with a commonly used light curing dental composite, containing Ba, Sr and Yb at detectable concentration levels. Measurements were performed using a microfocus X-ray tube at 30 W and 9 W settings (50 kV, 600 μA and 30 kV, 300 μA), achieving a confocal probing acceptance of 47 μm for Ca Kα to 15 μm for Sr Kα. The state-of-the-art voxel-wise absorption correction procedure for the 3D-XRF spectrometry data was based on the Beer–Lambert equation. It combined information about sample density obtained using μCT with mass attenuation coefficients derived from absorption measurements. This was complemented with extra compositional information from fundamental parameter quantification of reference-free XRF spectrometry for dark matrix elements, such as C, N and O, measured in the soft X-ray region.

An improved method based on the combination of confocal XRF and synchrotron μXRF spectrometry for the non-invasive characterisation of paint-layers from prehistoric cave arts was demonstrated by Tapia et al.²⁰ The study investigated red Fe-oxide based colouring matter found on a prehistoric stalactite sample from La Garma cave (Northern Spain), combining confocal XRF spectrometry for depth-resolved elemental profiling and SR-μXRF spectrometry for high-sensitivity elemental imaging. Using a confocal setup with a microfocus X-ray tube at 30 W power (50 kV, 0.6 mA), combined with two polycapillary optics defining an analysed volume of 35–60 μm³, the authors succeeded in resolving 5–10 μm thick layers of the red pigment layer from its calcite support. Via the depth-resolved separation enabled by confocal XRF spectrometry, one could identify the minor and trace elements characteristic of the red pigment, mainly Cr, Mn, Rb, Ti and Zn, while Br, Ga, Sr and Y were more likely related to the support. The results confirmed the feasibility of confocal XRF spectrometry for resolving micrometer-thick prehistoric paint layers, enabling a better differentiation of elemental signatures and demonstrating the future in situ analysis potential of the technique in cave art research.

2.2 Laboratory 2D XRF spectrometry techniques

A new laboratory monochromatic μXRF spectrometry instrument combining a Ga–In liquid jet anode with a multi-layer focusing mirror was presented by Lin et al.²¹ Using the NIST 611 standard for trace elements in glass, LODs ranging from 5.7 to 86.5 ppm were determined for elements in Z range of 20–29. A monochromatic μXRF spectrometry instrument featuring a conventional low power tube (5.3 W, 90 kV) for forensic analysis of tiny glass samples was described by Wang et al.²² The SiC (006) flat crystal monochromator (66.7 mm diameter, 8 mm thickness) was adjusted to the Bragg diffraction angle of the Mo Kα line (17.4 keV). Together with a polycapillary, a focal spot size of ca. 230 μm was achieved. The LOD for Sr, determined from a standard solution, was 51 μg L⁻¹, which is remarkable given the low power of the instrument. This setup successfully differentiated ten different glass samples.

Aligning and combining imaging data sets from different instruments is challenging in micro-characterisation of art works and cultural heritage objects. Gerodimos et al.²³ described in detail the combination of μXRF spectrometry data with multispectral optical imaging data (360 nm to 1200 nm). They studied the 0.25 franc French Gallic cockerel stamp “Gallus gallus domesticus”, painted by Albert Decaris, notable for its bright colours and the pigment ultramarine, which contained the low-Z elements Al and Si. The alignment of the datasets was achieved using the SIFT algorithm. The algorithm first binned the optical reflection spectra to match the μXRF spectrometry pixel size (ca. 50 μm × 50 μm). It then identified anchor points based on features such as contrast for pixel alignments. The optical light data was subjected to k-means clustering, which allowed the identification of areas with different colours. The XRF spectrometry sum spectra of the “blue” cluster confirmed the presence of ultramarine (high Al and Si content). Sub-clustering the reflection data with the μXRF spectrometry data enabled the differentiation between the seal and the stamp.

Tree-rings serve as proxies for studying paleo-environmental conditions, with wood density often being a more reliable temperature proxy than the ring width. The μXRF spectrometry profiles complements density data by providing information on e.g. the nutrient availability. Helama et al.²⁴ collected microdensitometry data and μXRF spectrometry elemental profiles on a set of tree-ring samples obtained from living Scots pine (Pinus sylvestris L.) trees in northern Finnish Lapland. They used a commercial spectrometer operating a rhodium anode, an SDD and a polycapillary optic. The full spectrum intensity closely matched the densitometry data. The intensities of Fe, Mg and Mn correlated well with the densitometry, while elements like Ca, K and S were either anti-correlated or showed no significant correlation. Trojek and Duskova²⁵ also studied dendrodensitometry and dendrochemistry, focusing on the accurate quantification of elemental amounts in varying wood matrices and densities. They introduced a very promising method combining single standard measurement with Monte Carlo simulation. The simulated scattering intensities closely matched the measured scattering across different matrices. Using the μXRF spectrometry data, they successfully reproduced wood densities and calculated actual elemental concentration, including the Mn profile.

In NASA’s Mars 2020 mission (Perseverance rover), the Planetary Instrument for X-ray Lithochemistry (PIXL) provided crucial information on the elemental distribution of Martian geological objects. The spatial distribution of these elements can reveal their origins. Das et al.²⁶ studied the energy-dependent probe spot positions of two systems: the Planetary Flight Model (PFM), a μXRF spectrometer currently performing in situ characterisations of Martian rocks and regolith and the JPL’s lab-based breadboard (LBB). The energy-dependent spot size from the PFM was determined for a limited set of elements: Au, Ba, Cu, Ni, Se, Ta and Ti. To supplement these data, the focal diameters of 18 elements were measured on the LBB instrument including Al, Ca, Cl, Cr, Cu, Fe, Ga, Ge, Mg, Mn, Mo, Na, Ni, Se, Si, Sr, Ti and Zn. Monte Carlo simulations were used to evaluate the energy- dependent coefficients of the spot sizes for both instruments. The breadboard data allowed inference of the trends for the flight hardware, revealing an exponential reduction in spot size with increasing beam energy for both instruments. This advancement allowed for a more precise determination of elemental distribution on Mars, enhancing the interpretation of compositional maps of individual mineral grains produced by PIXL.

3. Synchrotron and large-scale facilities

Over the review period, more than 200 publications reported the use of SR-XRF spectrometry, often in conjunction with complementary X-ray spectroscopic and imaging techniques, to integrate spatially resolved elemental data with information on chemical speciation, structural characteristics and morphological features. The application of μXRF spectrometry continued to grow across a range of disciplines, including biomedical research, environmental and Earth science, materials science and cultural heritage studies. These investigations were predominantly conducted at specialised hard X-ray micro- and nanoprobe facilities, where SR-XRF spectrometry was combined with techniques such as micro- and nanoscale XAS, XRD, ptychography and various tomographic imaging methods to achieve multimodal characterisation at high spatial resolution.

Several comprehensive review articles highlighted advances in SR-XRF spectrometry imaging and associated techniques in various areas of life, material and environmental sciences. Yin et al.²⁷ gave an overview of synchrotron-based X-ray imaging techniques as molecular/elemental probes, including nanoscale XRF spectrometry, with applications across various fields associated with intelligent biomedicine research. Kelkoul et al.²⁸ reviewed studies in neuroscience that correlate metal imaging using SR-XRF spectrometry with protein localisation by other techniques, highlighting the diversity of correlative modalities that have been used, from fluorescence to super-resolution and infrared microscopy. The importance of metal homeostasis in the nervous system and the roles of essential metals like Cu, Fe, Mn and Zn were emphasised while their dysregulation was implicated in neurodegenerative diseases such as Alzheimer’s, Parkinson’s, Huntington’s disease, ALS and multiple sclerosis. Emerging directions included the development of full cryogenic correlative imaging by cryo-fluorescence light microscopy with cryo-SR-XRF spectrometry reaching spatial resolutions of a few tens of nanometres, enabling near-native nanoimaging without sample dehydration.

Recent advances in the use of SR for non-destructive elemental-structural imaging and spectroscopy in the field of seed research at the Canadian Light Source were reviewed by Ashe et al.²⁹ Applying techniques such XRF spectrometry imaging, SR-μCT, mid-IR spectroscopy, XANES and STXM offered critical insights into seed development, composition and health, with applications for crop improvement and seed pathology. To illustrate the range of detectable elements that can be mapped across seeds, XRF spectrometry employing hard (5–20 keV) and soft X-rays (1.7–10 keV) was used. The BioXAS-imaging beamline enabled sensitive XRF spectrometry elemental mapping of biologically relevant elements (e.g. Ca, Cu, Fe, K, Mg, Mn, Zn) with a spatial resolution of 5 μm, whereas the SXRMB was ideally suited to determine the spatial distribution of lower Z elements (e.g. Si, S, P, Cl, K, Ca) at a resolution level of 10 μm.

A detailed overview of the 2D XRF spectrometry imaging capabilities and applications using the NanoMAX nanoprobe beamline at MAX IV Laboratory (Lund, Sweden) was provided by Sala et al.,³⁰ exploring the potential of nanoscale chemical imaging across biology, archaeology and materials science. The XFM nanoprobe was able to detect trace elements down to sub-ng per mm² levels, including As, Ba, Ca, Cl, Fe, Hg, Mn, P, Pb, S, Ti, Zn in the 8–15 keV excitation energy range, resolving features with lateral resolution <100 nm in examples of microalgae (N. polonicum), human dental tissue (Iron Age and modern tooth), geopolymers and Ti-containing electromagnetic compatibility coatings.

A comprehensive review by Schindler et al.³¹ described a large range of bulk and micro-/nano-analytical techniques available to characterise nanomaterial (NM) associations with minerals, organic matter and biological entities including microbes, fungi and plants. Of all elemental analysis techniques reviewed, synchrotron micro-/nano-XRF spectrometry provided the greatest spatial resolution (sub-100 nm) with enhanced sensitivity for trace elements and could map complex, heterogeneous systems such as NM–organic matter or NM–mineral interfaces in soils and biological tissues.

The detection sensitivity of XRF microscopy was assessed by Masteghin et al.³² for the quantitative nano-characterisation of ion beam-implanted samples, particularly focusing on Ga dopants in silicon. Using the Hard X-ray Nanoprobe at Brookhaven National Laboratory’s NSLS-II (Brookhaven, USA), with spot sizes around 13.9 nm × 12.3 nm at an excitation energy of 13.5 keV, the goal was to evaluate whether XRF spectrometry could detect and spatially resolve very low concentrations of impurities, potentially down to the single-atom level. Minimum detectable number of impurities was 3000 Ga atoms per pixel with 1 s integration (171 nm² spot) and 650 Ga atoms detectable with 25 s integration, corresponding to LODs of 18 impurities per nm² and 3.8 impurities per nm², respectively.

A novel operando XRF method was demonstrated by Wu et al.³³ to quantify ion composition and mobility in organic mixed ionic–electronic conductors. Quantifying ionic behaviour under operational conditions in these materials has been challenging due to limitations in existing techniques. The developed operando XRF spectrometry platform provided real-time, non-invasive tracking of ion concentrations during electrochemical cycling over time scales of hours by monitoring Br⁻ and Rb⁺ ions, substituting common cations/anions due to their strong XRF spectrometry signal and reduced absorption effects. The experiments represented the first non-invasive, direct quantification of ion transport phenomena in organic polymers under operando electrochemical conditions.

A novel setup established at the TwinMic beamline of Elettra (Trieste, Italy), combining low-energy XRF spectrometry, STXM and AFM, was reported by Hafner et al.³⁴ This first in-vacuum, in situ multimodal system enabled the simultaneous imaging of topographical, elemental and transmission-based data, addressing limitations of sequential measurements which may hamper interpretation. Using a combination of STXM (for navigation), XRF spectrometry (for elemental composition) and AFM (for topography and morphology), the system could identify and analyse micrometric and submicrometric asbestos fibers deposited on Si₃N₄ membranes with a spatial resolution of ca. 500 nm (XRF/STXM) and 5–15 nm (AFM). The work was performed at an excitation energy of 2.15 keV, detecting Fe (L), Mg, Na, Si and O.

In a detailed ray-tracing simulation study, Kosiorowski et al.³⁵ presented the concept and theoretical evaluation of the future analytical capabilities of PolyX, a new multimodal X-ray imaging beamline currently under development at the SOLARIS National Synchrotron Radiation Centre in Kraków, Poland. This beamline was designed for microimaging and microspectroscopy techniques, with a particular focus on μXRF spectrometry, μXAS and μCT. The source is a 1.31 T bending magnet from a 1.5 GeV electron storage ring. The anticipated X-ray energy range is 4–15 keV, with spot sizes ranging from 10 to 100 μm using polycapillary optics, or down to 2 μm using ellipsoidal monocapillaries. Ray-tracing simulations were conducted using the Monte Carlo codes Polycap for modelling polycapillary optics and XMI-MSIM for predicting the expected XRF spectrometry spectra for specific beam configurations.

4. Grazing X-ray spectrometry techniques including TXRF spectrometry

An excellent overview was provided by Margui et al.³⁶ of the capabilities of TXRF spectrometry as an efficient and sustainable analytical tool in food, cosmetic and pharmaceutical research. They critically discussed sample preparation procedures, including sample amounts, agents and internal standards. The authors concluded that widespread acceptance of TXRF spectrometry by the analytical community should require proper standard operating procedures and fit-for-purpose reference materials. Additionally, Fernández-Ruiz³⁷ highlighted in a review paper the application of TXRF spectrometry in biomedical, biochemical and pharmacological research.

The determination of elemental profiles in airborne particulates is a growing field for TXRF spectrometry applications. Achieving high time-dependent resolution (in <1 h) for monitoring exposure to environmentally harmful elements in ambient aerosols remains challenging. The micro analytical capabilities of TXRF spectrometry allowed the analysis of aerosol samples of <1 μg. This method is particularly powerful when particles are collected on reflectors directly used in the instrument. Challenges included mismatching geometries between the impactor’s impaction plate and nozzle patterns with the requirements of the TXRF spectrometry instrumentation. A novel impactor was developed by Crazzolara and Held³⁸ specifically for quartz reflectors used in TXRF spectrometry analysis. The impactor collected samples across PM10, PM2.5 and PM1 ranges. Attrition and cross-contamination effects were sufficiently low and a spin coating method was developed to avoid bounce off. An impressive low sampling time of 30 min was sufficient to determine e.g. Pb and Ni in ambient aerosols. An optimised TXRF spectrometry procedure allowed Chubarov et al.³⁹ the rapid and efficient analysis of dust particles of masses as low as 50 mg, filtered from thawed snow samples. The procedure demonstrated a good accuracy (87% to 115%) and repeatability (15%).

The determination of minor and trace elements in tobacco products, which can impair health when inhaled, was successfully realised using a suspensions-assisted sample preparation technique prior to TXRF spectrometry analysis.⁴⁰ The powdered samples were suspended in a 2.5 mL of 1% Triton X-100 solution and sonicated for 20 min. After sonication, 10 μL of the internally standardised (Ga) sample was deposited on a quartz glass carrier. Average concentrations of Cd, Cr, Cu, Mn and Pb exceeded the WHO/FAO recommended limits. For instance, the average Cr concentration in chewing tobacco was 4.4 mg kg⁻¹, in Bidi products 3.6 mg kg⁻¹ and in cigarettes 2.5 mg kg⁻¹, all exceeding WHO’s maximum acceptable levels (0.01–1.2 mg kg⁻¹). This pattern was seen for most other studied elements, with the highest concentrations seen in chewing tobacco, followed by Bidi and cigarettes.

An impressive study by Maltsev et al.⁴¹ demonstrated that TXRF spectrometry can precisely determine elemental ratios in individual foraminifera shells originating from Holocene sediments, with a 15% uncertainty attributed to both deposition and instrumental factors. The method included the deposition of the μm-sized shells on the quartz carriers and digestion with 2 N nitric acid. Elemental profiles of 129 shells from 14 sections of a sediment gravity core were obtained. A matrix of elemental concentrations of Ca, Cl, Fe, K, Mn, Sr and Zn, versus core section number and the degree of diagenetic change of the shell was processed using PCA. Score plots based on the elemental matrix showed clear clustering according to the diagenetic alteration of the shells (pristine, minor, moderate and major).

Chemical analysis of gallstones, including their elemental profile, is necessary to ensure optimal treatment of patients suffering from this affliction. Sample quantities are typically small, sometimes only a few milligrams. Shaltout et al.⁴² developed a microwave-assisted digestion procedure for individual gallstone samples, sufficient to determine elements such as Ca, Co, Cr, Cu, Fe, Ge, K, Mn, Ni, P, Pb, Rb, S, Sc, Se, Sr, Ti, V and Zn using TXRF spectrometry. The elements Cd and Mg were only able to be determined by ICP-OES due to their high LODs in TXRF spectrometry. For all the other elements except Cu, Mn, Ti and Zn, the LODs using ICP-OES were too high. Comparison of TXRF spectrometry and ICP-OES results showed good agreement with e.g. Mn values from ICP-OES being 73–98% of the TXRF spectrometry values. The PCA revealed similarities between certain stones.

The microanalytical performance of TXRF spectrometry was exceptionally advantageous in cultural heritage research by Falcone et al.,⁴³ enabling a comprehensive elemental description of historic enamelled ceramics, combined with low-Z element determination by SEM-EDX spectrometry. Both methods are quasi-non-destructive, requiring minimal material (<1 mg) from the cultural heritage artefacts. Calibration of the EDX analysis was facilitated with an Al–Cu standard, while the relative concentrations of elements were determined using TXRF spectrometry. The element oxide percentages from the EDX spectrometry were extended with those calculated from TXRF spectrometry to total 100%. The procedure was validated using soil and pigment reference materials, achieving recoveries in most cases greater than 80% of the certified values. Using PCA and cluster analysis on the TXRF spectrometry results of ancient potteries from archaeological sites of Tamil Nadu, Tamilarasi et al.⁴⁴ identified differences in the origin of the clay used. The information was highly valuable for deeper understanding of the pottery’s provenance and manufacturing methods.

Limited sample amounts often pose a challenge in determining elemental ratios in nanoparticle research. The performance of a novel photocatalytic system comprising CdSe@CdS nanorods as photosensitizers, coated with polydopamine and functionalised with various cobalt-based catalysts was investigated.⁴⁵ A commercial TXRF spectrometer with a Mo X-ray tube and Ti as internal standard was utilised for elemental analysis, requiring only 50 μL of sample. To quantify the functionalisation of the photocatalytic system, the elemental composition of the catalyst (Co) and the photosensitizer (Cd, Se and S) was determined. For the photocatalytic hydrogen production, the Co loading seems to be more critical than the specific compound used. The catalyst-to-photosensitizer ratio was identified as a key parameter for maximising hydrogen production. In another study, the stoichiometry of Cd and Se in quasi-2D nanoplatelets with a thickness of 3.5 monolayers was determined using TXRF spectrometry to learn about their composition-dependent optical properties.⁴⁶ A commercial TXRF spectrometer was used and the suspension was prepared on sapphire substrates. The Cd : Se atomic ratio depended on synthesis temperature, with nanoplatelets synthesized at 200 °C having a ratio of 1.31, closer to the expected 1.33 compared to 2.70 at 180 °C. Photoluminescence excitation spectra from 200 °C products showed no distortion, indicating an ordered electron structure, while 180 °C products exhibited significant distortion.

Medina-González et al.⁴⁷ successfully applied multivariate calibration using PLS analysis combined with machine learning to determine As and Pb in soil using TXRF spectrometry. Efficient analysis was achieved through slurry sampling of the soils. A set of 26 synthetic calibration standards was used to train the algorithm. Compared with the univariate approach, determination errors were reduced from 15.6% to 9.4% for As and from 18.9% to 6.8% for Pb. Using this approach, the assessment of agricultural soils of the Itata Valley (Chile) revealed moderate to considerable contamination with As and Pb.

The challenge of strong line overlap in oceanic polymetallic nodules and crusts exhibiting a complex mineral matrix was addressed by Zhilicheva et al.⁴⁸ They deconvoluted the TXRF spectrometry spectra using least square decomposition of real spectra into the weighted sum of simulated individual elemental sub-spectra. The accuracy of determining 11 elements using this technique was compared with the internal standard and linear calibration approaches. The proposed technique and linear calibration outperformed the internal standard approach for eight of 11 elements e.g. the RMSE of prediction for the Ni determination was 0.030 wt% for the new method, 0.047 wt% for the linear calibration and 0.051 wt% using an internal standard.

The determination of low-Z elements (Z < 13) in mineral-rich samples is challenging and limits the applicability of TXRF spectrometry in this area. Allegretta et al.⁴⁹ applied a specialised in-house built low-Z TXRF spectrometer with excellent results, achieving a complete analysis of major elements from F to Fe with an accuracy of 80% to 120%. The instrument featured a Cr source, a multilayer monochromator, an SDD with an ultrathin Si₃N₄ window and a vacuum sample chamber. Samples were prepared as simple slurries by dispersing 50 mg of powder into 2.5 mL of 1% Triton X-100 water solution and adding Ag as an internal standard.

A systematic evaluation of suitable internal standards for the determination of L-series elements (commonly determined by L-lines in TXRF spectrometry) in water samples was performed by Kubala-Kukus et al.⁵⁰ Single-element solutions of Ag, Ba, Cd, Er, Gd, In, Nd, Pb, Pd, Sb, Sn, Ta, Tl, W and two-element solutions with As, Ba, Cu, Pb, Ti and W were analysed. The performance of the internal standards Co, Ga, Nd and Sr was assessed in terms of LOD. The LODs increased with the element content in the analysed sample, e.g. the LOD for 0.002 mg L⁻¹ of Pb determined using Co as internal standard was 0.2 μg L⁻¹, while for 1.00 mg L⁻¹ of Pb, it was 0.8 μg L⁻¹. Depending on the specific element and internal standard combination, the LODs varied. For instance, the LOD for Pb was as low as 0.2 μg L⁻¹ when using Co as the internal standard and slightly higher at 0.4 μg L⁻¹ when using Ga. The LOD for Pd could be as high as 56 μg L⁻¹ using Ga as internal standard. However, most LODs were in the range of several μg L⁻¹.

The determination of I in saliva using TXRF spectrometry could simplify monitoring this important element for thyroid gland activity. Zambianchi and Zambianchi⁵¹ evaluated two excitation scenarios modelled by Monte Carlo simulation: one with monochromatic excitation and one using only a selective filter. The model resulted in a precise and sensitive I determination in saliva, suggesting that this method might be preferable to the commonly used urine analysis via ICP-MS. Alov et al.⁵² demonstrated the performance of TXRF spectrometry compared to ICP-OES in analysing dietary supplements, with recoveries ranging from 80% to 120% for the elements Ca, Cr, Cu, Fe, K, Mn, Ni, P, Se. V and Zn. Although the precision of TXRF spectrometry was slightly lower (8%) compared to ICP-OES (5%), the sample preparation for TXRF spectrometry was less time- and resource-intensive. Especially, the suspension-assisted preparation using ethylene glycol notably reduced preparation time by a factor of six.

Interestingly, a study on sample preparation of biological tissues available as CRMs for TXRF spectrometry, showed no significant influence from different drying procedures.⁵³ The preparation protocol included acid digestion of apple leaves, tomato leaves, oyster tissue and beef liver reference materials, with 6 μL prepared on a siliconised microscope slide. Five drying approaches were tested: laminar flow chamber, 100 mbar vacuum chamber, 250 W infrared lamp and a hot-plate at 45 °C and 75 °C. A simple acid digestion was also applied in the analysis of the wild carrot plant, Daucus Carota, from various areas around Chicago for biomonitoring environmental pollution by Schmeling et al.⁵⁴ Dried aliquots of the samples (0.1 g) were digested using microwave assisted nitric acid digestion (250 μL 30% hydrogen peroxide, 750 μL of 70% nitric acid). Highest concentrations of Cu, Fe, Mn, Ni, Pb and Zn were taken from an operating railyard with e.g. 14 mg kg⁻¹ of Pb and lowest concentrations in a park area e.g. 1.6 mg kg⁻¹ of Pb.

Biomedical samples sometimes suffer from poor drying behaviour and large residues due to high matrix content. A systematic evaluation of the impact of large specimen size (containing, in this case, between 300 ng and 10 000 ng of Ni and Ga) on the angle-dependent signal intensity below the critical angle was presented by Till et al.⁵⁵ For TXRF spectrometry, the ideal particulate specimen should exhibit a plateau of double excitation below the critical angle. Experiments demonstrated that all scans of specimens above 600 ng significantly deviated from double excitation, showing two mass-dependent changes. Specimens ranging from 600 ng and 1250 ng showed a plateau of reduced intensity. In contrast, specimens above this range (2500 ng −10 000 ng) exhibited significantly reduced intensity at low angles, with an increase up to double excitation just before reaching the critical angle. These phenomena were well reproduced by a fundamental parameter model, which included an approximation of the 3D shape of the specimens (based on confocal laser scanning microscopy findings), excitation geometry, reflectivity of the sample carrier and matrix effects. The model suggested that the width of the 600 ng and 1250 ng specimens, several hundred micrometres, induced dampening of double excitation due to increased absorption of the reflected beam. The 3D model identified areas of the sample suffering most from reduced excitation and suggested that the angle-dependency of signal intensity, with a maximum just below the critical angle for the 2500 ng to 10 000 ng sample was due to sample height.

A He atmospheric-pressure plasma jet treatment was applied to a glass sample carrier, making the surface hydrophilic and causing the sample to spread over a larger area compared to the more common siliconised surface. This treatment, as demonstrated by Tsuji et al.⁵⁶ reduced the height of a wine specimen by about a factor of four. The likely removal of organic surface contamination was confirmed by XPS monitoring of the C 1s signal, which showed low intensity for several hours after the plasma treatment. The treatment resulted in a reduction in RSD from 1.35%–3.78% to 0.28%–3.12% and an improvement in recovery values from ca. 83% to 96%. Additionally, the signal-to-background ratio in the detector field of view was evaluated using 1 mm gold deposits on different carriers. The optimal incident angle for the highest signal-to-background ratio was 0.025° (max. S/N of 60), which was notable since TXRF spectrometry experiments are typically conducted at 0.05 or 0.07°.

An efficient method for identifying adulteration of Chilean wine was developed using the ratio of Compton/Rayleigh scattering observed from the TXRF spectrometry spectra. This ratio allowed for the determination of changes in the effective Z number, enabling the identification of adulterated wine in all cases. To mimic adulteration, Cu, Fe, bentonite, fructose or water were individually added to original wine samples, as demonstrated by Perez et al.⁵⁷

A pre-concentration procedure for the determination of ultra-trace concentrations of Th(IV) species in natural waters using TXRF spectrometry was developed by Saha et al.⁵⁸ using cloud point extraction (CPE) with bis(phosphoramidate). The CPE parameters were optimised and achieved a recovery of 99% and a pre-concentration factor of about 90. The Th complexed in the surfactant-rich phase was prepared in microlitre volumes on sample carriers. Additional oxidative pyrolysis of the prepared phase improved LODs from 0.9 to 0.1 μg L⁻¹. Challenges in determining U in natural waters using TXRF spectrometry were highlighted in a review paper,⁵⁹ identifying the overlap of the U L-lines with Rb and Br lines, elements often present in natural waters. The determination of U in adsorption column resins is also challenging. These resins are used in the in situ leaching method, which facilitates uranium extraction from weakly saline sandstone formations at depths of up to 700 m. Wang et al.⁶⁰ successfully developed a suspension-assisted sample preparation strategy to determine U in the resins. The TXRF spectrometry results obtained with an in-house developed instrument aligned well with those from the laboratory pulse neutron method, exhibiting a linear correlation of y = 1.07x + 1.08 and a correlation coefficient of 0.98. These results make TXRF spectrometry a favourable choice for rapid testing of U in adsorption columns.

Sample preparation procedures for elemental determination in carbon-rich matrices using TXRF spectrometry were extensively tested by Cinosi et al.⁶¹ They evaluated suspension, acid digestion, solid–liquid extraction and ashing based on their complexity and detection capabilities. The suspension-assisted preparation of coconut-based activated carbon showed higher LODs, ranging from 3 mg kg⁻¹ to 0.2 mg kg⁻¹ for elements from K to Sr compared to 0.6 mg kg⁻¹ to 0.03 mg kg⁻¹ obtained by extraction and from 0.16 mg kg⁻¹ to 0.02 mg kg⁻¹ from ashing.

By varying pipetted volumes, drying conditions, surfactant concentrations and temperatures, Wiggershaus et al.⁶² was able to determine elements such as Co, Cr, Cu, Fe, Mn, Ni, or Zn with TXRF spectrometry from 0.1 mg L⁻¹ to 1 mg L⁻¹ in high ionic strength matrices like lithium carbonate and artificial seawater. The samples contained 1000 mg per L Li and 24 000 mg per L NaCl (matrix-analyte ratios of up to 240 000 : 1). The RSDs for most elements were reduced by a factor of 10 by adding Triton X-100 (0.3% or 0.9%) to the sample. The best signal-to-noise ratio for the Ga Kα-line was obtained by drying the sample in a desiccator at ambient pressure over 240 min.

The application of grazing X-ray spectrometry techniques continues to grow. Cara et al.⁶³ demonstrated the effectiveness of a non-destructive reference-free GIXRF spectrometry approach in extracting the density of molecular arrangements in self-assembled monolayers. This technique enabled the determination of surface concentrations of probes on plasmonic nanostructures, addressing a critical factor in assessing substrate performance and enabling meaningful comparisons. The reference-free GIXRF spectrometer aids in establishing molecular surface density standards, suitable for SERS/SEIRA substrates benchmarking. Using selected molecules such as 7-methyl-4-mercaptocoumarin (MMC), 3-mercaptopropionic acid (MPA) and 11-mercaptoundecanoic acid, standard densities could be confidently reproduced on gold. Mainly the same research groups also used molecular dynamic simulation to model MMC and MPA layers.⁶⁴ These developments should boost GIXRF spectrometry applications in laboratories, while SR facilities are also enhancing their GI capabilities. Hemmerle et al.⁶⁵ highlighted GIXRF spectrometry, GISAXS and GIXRD capabilities at the SIRIUS beamline of Synchrotron SOLEIL, dedicated to X-ray scattering and spectroscopy of surfaces and interfaces, covering the tender to mid-hard X-ray range (1.1–13 keV). New compound refractive lenses associated with a transfocator could focus the primary beam down to 10 μm × 10 μm, allowing faster GI measurements at the liquid–air interface and on samples with narrow geometries.

Operando GIXRF spectrometry demonstrated that Cs and Br ion migrate in CsPbBr_3, an emerging semiconductor for optoelectronic applications. Devices based on CsPbBr₃ suffer from poor electrical stability during operation which was believed to be caused by migration of Br⁻, Cs⁺ and Pb²⁺ ions. Using a custom-made GIXRF spectrometer and an operando approach, Rey et al.⁶⁶ applied voltages from 3–8 V and showed that the migration of Cs⁺ and Br⁻ was significant, potentially explaining the observed current instability in the device.

Adding an ED detector to a GIXRD instrument allowed the option for simultaneous quasi-non-destructive characterisation of density, microstructure (texture), composition and thickness of a thin (<1 μm) MoS₂ lubricant film used in extreme environments. To evaluate if the costly RBS analysis could be replaced by the more resource-efficient GIXRD/GIXRF spectrometer combination, Rodriguez et al.⁶⁷ prepared several films using PVD and characterised each by standard methods (e.g. RBS, profilometry, XRD and XRF spectrometry) as references. The GIXRD/GIXRF spectrometer combination proved to be capable of determining density directly by GIXRD from the presence or absence of delayed onset of film scattering intensity, also revealing microstructure and crystalline orientation. The GIXRF spectrometer allowed the determination of Mo and S by modelling the low energy range of the spectra to extract intensity data from the Mo Lα and S Kα peaks. Using composition and density information, the thickness of the deposited films was accurately predicted using established calibration curves.

Staeck et al.⁶⁸ utilised a laboratory scanning-free GEXRF spectrometry setup, featuring off-the-shelf equipment such as a Cr X-ray tube and a CMOS detector to investigate a set of Ti nanostructure samples. Different TiO₂ gratings were measured and the data was compared to SR measurements. The sample model was reconstructed using simulations based on the finite-element method and a Maxwell solver, with Bayesian optimisation used for the optimisation process itself. Notably, the parameters describing the sidewall angles of the grating and the grating line width showed some deviations from the SR data. The setup demonstrated good long-term stability and photon statistics, although the angular resolution limited its analytical potential. The reconstructed sample parameters were mostly close to those obtained from SR facility measurements such as Ti line height (58 nm compared to 55.3 nm) and HfO₂ thickness (2.3 nm compared to 2.30 nm). Another laboratory GEXRF spectrometer was compared with an established hybrid XRR/GIXRF spectrometer combination by Terentev et al.⁶⁹ Analyses were conducted on an Ru/V/B₄C/Ru/Si X-ray waveguide structure. The same fitting procedure was applied to both GEXRF and GIXRF spectrometry simulations. The selected detector proved to be effective for GEXRF spectrometry measurements. Since the footprint of the incident beam remained unchanged with the emission angle, GEXRF spectrometry measurements did not require correction with a geometric factor. This method might also be suitable for topographic scanning.

Sarkar et al.⁷⁰ studied periodic multilayers made of 75 Cr/Sc bilayers, each with a thickness of approximately 4 nm, both with and without B₄C barrier layers of 0.2 nm thickness. The GIXRF spectrometry measurements under standing wave conditions provided precise information on element-specific diffusion at the interfaces of the multilayer structure. Chromium diffusion of ca. 0.5 nm into the layers was observed.

5. Hand-held, mobile and online XRF spectrometry techniques

5.1 Hand-held and mobile XRF spectrometry techniques

Specht et al.⁷¹ reviewed the evolution of XRF spectrometry techniques for measuring bone lead, comparing current and previous methods. The most successful systems utilised K-shell excitation of Pb with a ¹⁰⁹Cd-based radioisotope source for both cortical and trabecular bone. The more user-friendly portable tube-based XRF spectrometry, limited to cortical bone, provided quicker and comparable results. This review assessed their accuracy, identified limitations and discussed potential advances in future techniques.

For determining the elemental composition of food, EDXRF spectrometry was implemented to analyse diverse natural samples, including solids, liquids, powders and bacterial matrices. A comprehensive review⁷² with 86 references covered instrumental aspects, advancements and applications in the food industry over the past 15 years. Recent advancements enhanced the sensitivity of hand-held devices and detectors, improving accuracy for detecting hazardous elements like As, Cd and Pb. The EDXRF spectrometry technique was widely used in authentication, food safety and quality control. The reviewers discussed challenges and future directions, including automation and real-time monitoring to enhance its utility in the global food industry.

Commercial benchtop XRF spectrometry equipment often limits plant science experiments due to sample chamber volume and source-detector geometry. To address this, Santos et al.⁷³ developed an in-house XRF spectrometry setup for in vivo experiments of plants, equipped with a 4 W X-ray tube source with silver anode and SDD. The X-ray tube and the detector were assembled at 45° and 135° from the sample surface and placed at 7.5 and 6.8 mm from the sample, respectively. The sample surface position was determined by a two-laser convergence system. The LODs for Ca, Cu, Fe, K, Mn, Rb, Sr and Zn were comparable or lower compared to commercial instruments and suitable for plant tissue evaluation. A strong linear correlation (ranging from 0.91 to 0.99) was observed between the XRF spectrometry intensities of Ca, Cu, K, Sr and Zn in citrus, coffee and soybean leaves and the corresponding concentrations measured by ICP-OES. A case study using tomato plants and Rb and Sr as tracers for K and Ca, respectively, demonstrated the feasibility of long-term in vivo analysis.

When using EDXRF spectrometry, the choice of the calibration method in food analysis – whether based on matrix-matched standards or CRMs – is crucial for obtaining accurate quantitative results. Benedito et al.⁷⁴ evaluated three calibration methods for Zn determination using portable XRF spectrometry in pelletised common bean samples. Optimised conditions included a titanium–aluminium primary filter and a 10 s irradiation time. Samples with the highest Zn content underwent acid extraction with either HNO₃ or HCl to create synthetic blanks, which were mixed with the original sample to prepare calibration standards. An additional calibration model using plant-based CRMs was also assessed. The calibration model of HCl-extracted and mixed samples showed the best statistical agreement with the reference data (95% confidence level), with a linear correlation factor (r) higher than 0.99 and an acceptable detection limit of 2.23 mg kg⁻¹. Santos et al.⁷⁵ used portable XRF spectrometry to directly determine inorganic macronutrients (Ca, K, Mg and P) in ground-roasted coffee, relevant for nutritional analysis, authentication and origin tracking. Two calibration approaches were tested: plant-based CRMs and matrix-matched calibration using coffee samples and spent coffee grounds. Samples were irradiated for 50 s using two commercial spectrometers under two experimental conditions: X-ray tube voltage of 15 kV and 50 kV. Analyte mass fractions were determined by ICP-OES after digestion with HNO₃ and H₂O₂. Calibration models showed excellent linearity (r ≥ 0.90). The matrix-matched calibration method was the most accurate for elemental determination, particularly for P and K, in coffee and spent coffee grounds. The LODs for P (0.0264 g kg⁻¹, at 15 kV) and K (0.0114 g kg⁻¹, at 50 kV) were considered adequate for the intended analysis. However, the LOD for Mg (1.3574 g kg⁻¹) was insufficient and the root mean square error for Ca (5.12 g kg⁻¹) was too high for this application.

An on-site system for the rapid simultaneous determination of multiple elements was developed, integrating microdroplet technology with a hand-held XRF spectrometer.⁷⁶ This system concentrated heavy metals in microdroplets through homogeneous liquid–liquid extraction and applied hand-held XRF spectrometry. The research focused on heavy metals, specifically Pb. Under optimal conditions, a high extraction percentage of 93.3% and a volume reduction ratio of 120 times (from 23.9 to 0.200 mL) were achieved. Using these conditions, a mixed solution of Cu and Pb was analysed for simultaneous rapid determination. By exposing a 200 μL microdroplet to a focused X-ray beam (9.0 mm diameter), a proportional relationship for Cu and Pb concentrations ranging from 0.0837 to 0.126 mg L⁻¹ was confirmed. The LODs (10σ) were 0.040 mg L⁻¹ for Cu and 0.030 mg L⁻¹ for Pb, enabling stable evaluation of heavy metals at low concentrations (ppb level) with straightforward operation.

Uranium is crucial in the nuclear industry, but its inadvertent release raises health and environmental concerns. Li et al.⁷⁷ proposed a novel method combining magnetic dispersive solid-phase extraction with portable XRF spectrometry for on-site sampling and determination of trace U in real samples. A magnetic covalent organic framework (Fe₃O₄@COF) composite, with excellent chemical stability and a large adsorption capacity of 311 mg g⁻¹, was synthesised and used as an efficient adsorbent for U extraction. The method, using benchtop WDXRF spectrometry, achieved an impressive low LOD of 0.008 μg L⁻¹ with a 50 mL sample volume and a fast adsorption time of 15 minutes, making it suitable for environmental monitoring of UO₂²⁺. Additionally, a hand-held portable XRF spectrometer instrument with an LOD of 0.1 μg L⁻¹ was successfully implemented for quality assessment of real water samples using on-site extraction, yielding accurate results (bias ± 20%) and satisfactory spike recoveries (85–91%).

5.2 Online XRF spectrometry techniques

Emerging near-real-time techniques, such as EDXRF spectrometry, offer advantages for continuous online monitoring and source apportionment of airborne particulate matter. The Horiba PX-375 EDXRF spectrometer monitor was evaluated by Trebs et al.,⁷⁸ who examined the performance evaluation, including LOD, identification and quantification of uncertainty sources and comparison of measurement results from three sites in Luxembourg (urban, semi-urban, rural). Multi-element RMs were used for calibration, with measurements taken during spring and summer 2023. For the toxic elements (Cu, Ni, Pb, Zn) the LODs were below 3 ng m⁻³ at 1 hour time resolution, while higher LODs were observed for the low atomic Z elements (Al, Ca, K, S, Si). Expanded uncertainties ranged from 5% to 25% for elemental concentrations above 20 ng m⁻³ and 60–85% for concentrations below 10 ng m⁻³. Elemental analysis identified S and mineral elements (Al, Ca, Fe, Si) as dominant contributors to PM10 samples. Toxic trace elements (As, Cu, Zn) were enriched at urban and semi-urban sites compared to the rural site. Results explained 51–74% of the gravimetric PM10 mass at the three sites. The use of online monitors significantly enhanced air quality monitoring and source apportionment efforts, leading to a better understanding and management of atmospheric pollution.

6. Cultural heritage applications

An overview of synchrotron applications for cultural heritage over the past decade was presented by Gianoncelli et al.,⁷⁹ citing 105 references. Recent upgrades to SR sources facilitated the construction and design of beamlines optimised for studying cultural heritage materials, requiring specific setups, spatial resolutions and detection limits. In cultural heritage research, integrated approaches combining various techniques were often necessary, even at large facilities, where some beamlines enabled multiple types of measurements at the same analysis point. A list of European SR X-ray beamlines for cultural heritage applications, indicating available techniques, beam size, sample environment and available energy range, was included. The capabilities of SR techniques were demonstrated through a case study of painted ceramics, showcasing how different research questions were addressed using SR-based methods. An impressive overview by Schoeder et al.⁸⁰ described studies performed since 2019 at the PUMA beamline (synchrotron light source SOLEIL, Paris, France), which was designed for heritage research and accessible to all scientific fields. The beamline’s optical layout used a horizontal focusing mirror to prefocus light from the wiggler source for the experimental endstation. It provided a 5 μm × 7 μm microbeam for XAS, XRD, XRF spectrometry and XEOL analysis, or a 20 mm × 5 mm full field when defocused with retracted Kirkpatrick–Baez mirrors. A stable fixed-exit Si(111) monochromator was used to select the wavelength. Numerous experiments, referenced in 24 publications, were conducted at PUMA, notably in archaeology, palaeontology, conservation and art history, also in identifying safer irradiation conditions for heritage samples. Three in-house studies were presented in detail, showing the effects of soil interaction with Palaeolithic ivory, identifying the elemental composition of mineralised textiles and uncovering hidden fossil morphologies.

A review (with 72 references) on Chinese ancient porcelain research using EDXRF spectrometry was presented by Li et al.⁸¹ This technology played a significant role in studying manufacturing processes, spatiotemporal discrimination and the preservation of cultural relics. When combined with XRD and SEM, it offered in-depth chemical analysis, enhancing the understanding of ceramic firing processes. Integrating EDXRF spectrometry with methods like XANES revealed intricate color formation mechanisms. Despite limitations in analysing low-Z elements, the integration of EDXRF spectrometry with advanced techniques and compositional databases improved the accuracy and reliability of artifact research and preservation, revitalising archaeology and cultural heritage studies.

In the study of paintings, macroXRF spectrometry, a well-established tool, generates complex and voluminous datasets that pose significant analytical challenges. Consequently, the use of artificial intelligence networks becomes inevitable to effectively manage and analyse these datasets. Gerodimos et al.⁸² conducted a study where convolutional neural networks (CNNs) were trained using twenty thousand spectra acquired during the macroXRF spectrometry scan of religious icons with the associated Ground-Truth counts per characteristic transition line, as extracted by fundamental parameter analysis. Comparison of CNN-extracted results to Ground-Truth showed remarkable agreement. This successful analysis paved the way for auto-identification of spectral lines, aiding both less-experienced and experienced XRF spectrometry analysts. Preisler et al.⁸³ advanced the field by introducing a deep learning algorithm, trained on a synthetic dataset, to enable fast and accurate analysis of the spectra in macroXRF spectrometry datasets. They not only recovered the correct elemental distributions of paintings but also determined the absolute number of counts for each chemical element in the XRF spectra by introducing scale into the neural network. This approach successfully overcame the limitations of traditional deconvolution methods. When applied to a painting by Raphael, the model achieved superior accuracy in quantifying fluorescence line intensities and effectively eliminated artifacts typically observed in elemental maps generated by conventional methods.

Lach et al.⁸⁴ demonstrated the capabilities and performance of a custom-developed XRF full-field imaging spectrometer for elemental mapping of 3D objects. Utilising pinhole camera optics, which theoretically provides infinite depth of field, enabled measurements of 3D objects without compromising the target spatial resolution. This was clearly shown in two case studies: a wooden fragment of a Rococo altar ornament and a decorated inkwell with cylindrical form. The system operated at limited energy resolution (1.1 keV FWHM at 5.9 keV). In both cases, the depth of the imaged areas varied up to 3 cm without noticeable decrease in spatial resolution. This depth of focus range surpasses considerably that of macroXRF spectrometers using a focused X-ray microbeam.

A low-cost, portable scanning macroXRF spectrometry imaging device was constructed by He et al.,⁸⁵ integrating a tungsten-target X-ray tube (10 W, 50 kV), an SDD, a laser displacement sensor and a 2D scanning platform. It achieved a spatial resolution of 0.55–0.57 mm at a distance of 10 mm with step sizes of 0.02–0.2 mm. Operating at 50 kV and 0.2 mA, the device showed a maximum sensitivity of 481.9 cps (mg g⁻¹)⁻¹ for Zn and a LOD as low as 2.6 ppm. Using a blind deconvolution image reconstruction algorithm, the spatial resolution of the selected test target improved by 35% in FWHM and 24.2% in S/N, achieving an optimal resolution of 0.36 mm. Surface topography correction on a coin reduced the average deviation in Cu elemental maps by 5.7%. Image processing on a piece of the mineral lapis lazuli enhanced the spatial resolution (FWHM) of Ca, K and Sr maps by 34.4%, 36.1% and 33.5%, respectively. Future improvements were suggested to focus on reducing artifacts and enhancing measurement accuracy on rough surfaces, potentially by using a smaller laser spot sensor. Colombo et al.⁸⁶ analysed tissue-specific elemental signatures in fossils preserved in solid resin, liquid glycerin and water using a mobile macroXRF spectrometry scanner, commonly applied for studying historical paintings. The system featured a 30 W rhodium-target X-ray tube (50 kV, 0.6 mA) and a 30 mm² SDD, mounted on an X, Y-motorised stage with an 80 cm × 60 cm travel range and 10 μm minimum step size. The device achieved a beam size of 40 μm and allowed rapid modulation of the beam size without changing the beam-defining optic. Analysis of various fossils revealed that hair and feathers were associated with S and Ti, abdominal tissues with Cu and Zn and stomach contents, e.g., seeds, with Cu, Ni and Zn. A detailed protocol for data acquisition and processing was provided, along with a critical discussion on the application of this approach to palaeontological research. Additionally, the benefits of the mobile macroXRF spectrometry scanner were highlighted, noting that certain fossils cannot be analysed using SR-macroXRF or benchtop μXRF spectrometers due to limitations in their dimensions, weights and conservation requirements.

An overview was provided by Musilek et al.⁸⁷ of the capabilities of various macro- and μ-arrangements of in-house developed XRF spectrometry systems for studying paintings. These systems ranged from fast orientational measurements with a compact hand-held device, to 2D scanning with a collimated beam and 3D analysis with collimated X-ray optics. Measurement with a collimated beam from a laboratory device (up to μXRF spectrometer) offered similar element distinction as a hand-held device but allowed for detailed study of smaller areas. For more detailed analysis, lengthier scanning with μXRF or macroXRF spectrometry was required. Using an automatic mechanical device to fix and move the spectrometer enabled scanning of selected image parts and monitoring of element distribution. While detailed in-depth pigment distribution could not be derived, a spectrometer with a confocal arrangement could differentiate pigment composition heterogeneity at specific locations.

A commercial hand-held XRF spectrometer was combined with a cost-effective automated X–Y remote-controlled motion system for real-time elemental analysis in cultural heritage applications.⁸⁸ Open-source software synchronised the spectrometer’s measurement functions and managed data acquisition and analysis. The system, equipped with a rhodium-target X-ray tube, operated at 6–50 kV and 4.5–195 μA, with a maximum output of 4 W. It used two collimators with 3 mm and 8 mm spot sizes and a custom-made 1 mm bronze collimator for improved spatial resolution. The measuring plane is 30 mm from the X-ray tube, with the primary beam hitting the target at a 45° angle. The detector, with a 20 mm² active area and 8–9 mm distance from the measuring plane, detected photons at a 27° angle. The stage had a resolution of 3 μm, allowing precise positioning with a maximum scanning range of 100 mm and a minimum step size of 100 μm. The spectrometer’s analytical capabilities, including sensitivity, energy resolution, beam spot size and characteristic transition intensity relative to the distance from the target, were thoroughly evaluated. The XRF spectrometry scanner’s potential for real-time imaging was demonstrated by scanning a 19th century religious icon, revealing pigment information and an underlying icon. Its capability to classify metallic objects and perform quantitative analysis in scanning mode was verified using PCA on a coin collection and a pre-installed quantification routine for precious metals. Chiti et al.⁸⁹ designed portable XRF spectrometers, including one with a polycapillary lens focused to 300 μm spot size for analysing small details on glass and pigmented surfaces and another with a 70 kV X-ray tube for improved sensitivity to medium-Z elements, crucial for copper-based artefacts. Both instruments were evaluated for LODs and quantification uncertainties. A case study demonstrated the polycapillary optics’ effectiveness in analysing small details, such as polychromy in vitreous beads, without interference from surrounding materials. However, polycapillary lenses attenuated the high-energy component of the primary beam, worsening LOD and quantification uncertainties for elements around atomic number 50. Another case study on pre- and proto-historic bronzes highlighted the second instrument’s superior analytical performance in quantifying major and minor alloy constituents, aiding in the development of hypotheses on prehistoric metallurgy.

Abbreviations

2D	two dimensional
3D	three dimensional
AFM	atomic force microscopy
ALS	amyotrophic lateral sclerosis
ASU	Atomic spectrometry update
CMOS	complementary metal-oxide-semiconductor
CNN	convolutional neural networks
CPE	cloud point extraction
CRM	certified reference material
CT	computed tomography
DIP	deep image prior
ED	energy dispersive
EDX	energy dispersive X-ray
EDXRF	energy dispersive X-ray fluorescence
FAO	Food and Agriculture Organization of the United Nations
FWHM	full width at half maximum
GEXRF	grazing exit X-ray fluorescence
GI	grazing incidence
GISAXS	grazing incidence small-angle X-ray scattering
GIXRD	grazing incidence X-ray diffraction
GIXRF	grazing incidence X-ray fluorescence
ICP	inductively coupled plasma
IR	infrared
JPL	Jet Propulsion Laboratory
LBB	lab-based breadboard
LOD	limit of detection
MMC	7-methyl-4-mercaptocoumarin
MPA	3-mercaptopropionic acid
MS	mass spectrometry
μCT	microscale computed tomography
μXAS	microscale X-ray absorption spectroscopy
μXRF	microscale X-ray fluorescence
NASA	National Aeronautics and Space Administration
NIST	National Institute of Standards and Technology
NM	nanomaterial
NSLS	National Synchrotron Light Source
OES	optical emission spectrometry
PCA	principal component analysis
PCI	phase-contrast imaging
PEG	polyethylene glycol
PFM	planetary flight model
PIXL	Planetary Instrument for X-ray Lithochemistry
PLS	partial least squares
PM	particulate matter
PM1	particulate matter with a diameter of <1 μm
PM2.5	particulate matter with a diameter of <2.5 μm
PM10	particulate matter with a diameter of <10 μm
PMMA	polymethylmethacrylate
ppb	parts per billion
PUMA	Photons Utilisés pour les Matériaux Anciens
PVD	physical vapour deposition
PXCT	ptychographic X-ray computed tomography
RBS	Rutherford backscattering spectrometry
RM	reference material
RMSE	root mean-square error
RSD	relative standard deviation
SCUNet	Swin–Conv-UNet
SDD	silicon drift detector
SEIRA	surface-enhanced infrared absorption
SEM	scanning electron microscopy
SERS	surface-enhanced Raman spectroscopy
SIFT	scale invariant feature transform
S/N	signal-to-noise ratio
SR	synchrotron radiation
STXM	scanning transmission X-ray microspectroscopy
SXRMB	soft X-ray microcharacterization beamline
TEM	transmission electron microscope
TXRF	total reflection X-ray fluorescence
USA	United States of America
WD	wavelength dispersive
WDXRF	wavelength dispersive X-ray fluorescence
WHO	World Health Organisation
XANES	X-ray absorption near edge structure
XAS	X-ray absorption spectroscopy
XEOL	X-ray excited optical luminescence
XFM	X-ray fluorescence microscopy
XPS	X-ray photoelectric spectroscopy
XRD	X-ray diffraction
XRF	X-ray fluorescence
XRR	X-ray reflectometry
Z	atomic number

Conflicts of interest

There are no conflicts to declare.

References

1. C. Vanhoof, J. R. Bacon, U. E. A. Fittschen and L. Vincze, J. Anal. At. Spectrom., 2024, 39(9), 2152–2164.
2. W. R. L. Cairns, O. T. Butler, O. Cavoura, C. M. Davidson, J. L. Todolí-Torró and M. von der Au, J. Anal. At. Spectrom., 2025, 40(1), 11–69.
3. M. Patriarca, N. Barlow, A. Cross, S. Hill, D. Milde and J. Tyson, J. Anal. At. Spectrom., 2025, 40(3), 541–664.
4. E. H. Evans, J. Pisonero, C. M. M. Smith and R. N. Taylor, J. Anal. At. Spectrom., 2024, 39(5), 1188–1211.
5. R. Clough, C. F. Harrington, S. J. Hill, Y. Madrid and J. F. Tyson, J. Anal. At. Spectrom., 2024, 39(7), 1629–1664.
6. E. Bolea-Fernandez, R. Clough, A. Fisher, B. Gibson and B. Russell, J. Anal. At. Spectrom., 2024, 39(11), 2617–2693.
7. K. E. Turner and E. Webber, Geostand. Geoanal. Res., 2024, 48(3), 505–541.
8. K. E. Turner and E. Webber, Geostand. Geoanal. Res., 2025, 49(1), 51–130.
9. G. M. Saladino, P. H. Chao, B. Brodin, S. D. Li and H. M. Hertz, Nanoscale, 2024, 16(37), 17404–17411.
10. S. Z. Pu, J. D. Song, H. B. Lu, W. L. Zhang and L. Li, Phys. Med. Biol., 2024, 69(21), 215039.
11. S. Mandot, E. M. Zannoni, L. Cai, X. Nie, P. J. La Riviere, M. D. Wilson and L. J. Meng, IEEE Trans. Med. Imaging, 2024, 43(5), 1782–1791.
12. N. Mahmoodian, M. Rezapourian, A. A. Inamdar, K. Kumar, M. Fachet and C. Hoeschen, J. Imaging, 2024, 10(6), 127.
13. S. Kusakari, K. Sato, Y. Tsushima, M. Matsuoka, T. Sasaya, N. Sunaguchi, K. Matsubara, H. Kawashima, K. Hyodo, T. Yuasa and T. Zeniya, Int. J. Comput. Assist. Radiol. Surg., 2024, 20, 665–676.
14. X. Zheng, V. R. Kankanallu, C.-A. Lo, A. Pattammattel, Y. Chu, Y.-C. K. Chen-Wiegart and X. Huang, Optica, 2024, 11(2), 146–154.
15. B. G. Duersch, S. A. Soini, Y. Q. Luo, X. Y. Liu, S. Chen and V. M. Merk, Metallomics, 2024, 16(10), mfae040.
16. K. Medjoubi, K. Benzerara, J. Debrie, E. Tang, D. Bazin, E. Letavernier, K. Desjardins and A. Somogyi, Front. Environ. Chem, 2024, 5, 1339829.
17. A. van der Ent, K. M. Spiers, T. Jaffré, D. Brueckner, S. C. Irvine, J. Garrevoet, P. D. Erskine, V. Gei, E. Montargès-Pelletier, B. Fogliani, G. Echevarria and S. Isnard, Environ. Exp. Bot., 2024, 226, 105877.
18. S. Weber, D. Karpov, M. Kahnt, A. Diaz, Y. Romanenko, S. Kotrel, A. Haas, B. Hinrichsen, N. Bottke, J. D. Grunwaldt, S. Schunk and T. L. Sheppard, ChemCatChem, 2024, 16, e202301298.
19. L. J. Bauer, F. Wieder, V. Truong, F. Foerste, Y. Wagener, A. Jonas, S. Praetz, C. Schlesiger, A. Kupsch, B. R. Mueller, B. Kanngiesser, P. Zaslansky and I. Mantouvalou, Anal. Chem., 2024, 96(21), 8441–8449.
20. J. Tapia, M. Eveno, P. Arias, R. Ontanon, S. Schoder, K. Mueller and I. Reiche, J. Cult. Heritage, 2024, 67, 385–394.
21. X. S. Lin, M. M. Wang, M. Shu, C. Y. Li, H. J. Li, X. Wang, C. Z. Yang and R. Si, J. Anal. At. Spectrom., 2025, 40(3), 747–752.
22. X. Wang, Y. Li, C. Wang, S. Huang, F. Zhou and X. Lin, X-Ray Spectrom., 2025, 54(1), 47–53.
23. T. Gerodimos, I. V. Patakiouta, V. M. Papadakis, D. Exarchos, A. Asvestas, G. Kenanakis, T. E. Matikas and D. F. Anagnostopoulos, J. Imaging, 2024, 10(4), 95.
24. S. Helama, T. Saarinen, T. Suominen, M. Fuentes, T. Frank and B. E. Gunnarson, Dendrochronologia, 2024, 86, 126229.
25. T. Trojek and A. Duskova, Radiat. Phys. Chem., 2024, 218, 111603.
26. A. Das, C. M. Heirwegh, N. Gao, W. T. Elam, L. A. Wade, B. C. Clark III, J. A. Hurowitz, S. J. Vanbommel, M. W. M. Jones and A. C. Allwood, X-Ray Spectrom., 2025, 54(2), 203–213.
27. D. Yin, Y. Liu, Y. Zhang, X. Cai, J. Li, J. Hu, J. Zhang and Y. Zhu, Small Methods, 2025, e2401890.
28. I. Kelkoul, V. P. Muñoz, R. Ortega and A. Carmona, Metallomics, 2025, 17(2), mfaf003.
29. P. Ashe, K. Y. Tu, J. A. Stobbs, J. J. Dynes, M. Vu, H. Shaterian, S. Kagale, K. K. Tanino, J. P. D. Wanasundara, S. Vail, C. Karunakaran and T. D. Quilichini, Front. Plant Sci., 2025, 15, 1395952.
30. S. Sala, K. Rengefors, J. Kiventerä, M. Patanen, L. Gefors, C. Werdinius, S. Winge, K. Broberg, S. Kalbfleisch and K. S. Clauss, Radiat. Phys. Chem., 2025, 229, 112491.
31. M. Schindler, M. A. Alavijeh, M. L. S. Oliveira and L. F. O. Silva, Geostand. Geoanal. Res., 2025, 49(1), 133–159.
32. M. G. Masteghin, T. Gervais, S. K. Clowes, D. C. Cox, V. Zelyk, A. Pattammattel, Y. S. Chu, N. Kolev, T. J. Z. Stock, N. J. Curson, P. G. Evans, M. Stuckelberger and B. N. Murdin, Small Methods, 2024, 8(10), 2301610.
33. R. Wu, X. Ji, Q. Ma, B. D. Paulsen, J. Tropp and J. Rivnay, Sci. Adv., 2024, 10(17), eadn8628.
34. A. Hafner, L. Costa, G. Kourousias, E. Dobovicnik, M. Cautero, L. Gregorat, M. Zizic, V. Bonanni, G. Cautero and A. Gianoncelli, X-Ray Spectrom., 2024, 54, 378.
35. F. Kosiorowski, P. Wróbel, T. Kolodziej, K. M. Sowa, M. Szczerbowska-Boruchowska and P. Korecki, Appl. Sci., 2024, 14(17), 7486.
36. E. Margui, D. Eichert, J. Jablan, F. Bilo, L. E. Depero, A. Pejovic-Milic, A. Gross, H. Stosnach, A. Kubala-Kukus, D. Banas and L. Borgese, J. Anal. At. Spectrom., 2024, 39, 1700–1719.
37. R. Fernández-Ruiz, Int. J. Mol. Sci., 2025, 26(3), 1049.
38. C. Crazzolara and A. Held, Atmos. Meas. Tech., 2024, 17(7), 2183–2194.
39. V. M. Chubarov, A. S. Maltsev, A. A. Amosova, E. V. Chuparina, S. N. Prosekin and J. V. Sokolnikova, X-Ray Spectrom., 2024, 53(5), 374–381.
40. Manjunatha and A. S. Bennal, X-Ray Spectrom., 2024, 54(3), 264–273.
41. A. S. Maltsev, G. V. Pashkova, A. N. Zhilicheva and A. V. Romanova, Spectrochim. Acta, Part B, 2025, 226, 107141.
42. A. A. Shaltout, O. H. Abd-Elkader, P. Lassen and U. A. E. Fittschen, X-Ray Spectrom., 2025, 54, 422.
43. F. Falcone, A. Cinosi, G. Siviero and G. Rosatelli, Spectrochim. Acta, Part B, 2024, 218, 106980.
44. A. Tamilarasi, A. Chandrasekaran, A. S. Bennal and G. R. Annamalai, Spectrosc. Lett., 2025, 58(1), 95–112.
45. M. Boecker, S. Lander, R. Mueller, A.-L. Gaus, C. Neumann, J. Moser, M. Micheel, A. Turchanin, M. v. Delius, C. V. Synatschke, K. Leopold, M. Waechtler and T. Weil, ACS Appl. Nano Mater., 2024, 7, 14146–14153.
46. Y. Koshkinbayev, A. Ospanova, A. Akhmetova, T. Nurakhmetov, A. Kainarbay, K. Zhangylyssov, S. Dorofeev, A. Vinokurov, S. Bubenov and D. Daurenbekov, Nanomaterials, 2024, 14(22), 1794.
47. G. Medina-González, Y. Medina, E. Muñoz, P. Andrade, J. Cruz, Y. Rodriguez-Gallo and A. Matus-Bello, Processes, 2024, 12(8), 1760.
48. A. N. Zhilicheva, G. V. Pashkova, V. M. Chubarov, A. S. Maltsev, D. Kirsanov and V. Panchuk, Spectrochim. Acta, Part B, 2024, 222, 107070.
49. I. Allegretta, D. Krstajic, P. Wobrauschek, P. Kregsamer, D. Ingerle, C. Streli, C. Porfido and R. Terzano, Appl. Clay Sci., 2024, 251, 107326.
50. A. Kubala-Kukus, E. Margui, D. Banas, J. Wudarczyk-Mocko, S. Gozdz, I. Stabrawa, K. Szary, S. Swierczynska, G. Wesolowski, T. Milicevic, D. Relic, F. Bilo and L. Borgese, Spectrochim. Acta, Part B, 2024, 215, 106922.
51. P. Zambianchi and J. K. Zambianchi, Spectrochim. Acta, Part B, 2024, 213, 106882.
52. N. V. Alov, D. V. Danilov, P. Y. Sharanov, V. G. Semenov, V. V. Panchuk, S. S. Savinov and M. M. Khaydukova, X-Ray Spectrom., 2024, 53(3), 181–189.
53. P. Kasprzyk, P. M. Wróbel, W. Tokarczyk and M. Szczerbowska-Boruchowska, Spectrochim. Acta, Part B, 2025, 228, 107187.
54. M. Schmeling, M. Gende and A. Tovar, X-Ray Spectrom., 2025, 54(2), 118–126.
55. H. S. Till, A. Gross and U. E. A. Fittschen, Spectrochim. Acta, Part B, 2025, 225, 107138.
56. K. Tsuji, Y. Tanaka, N. Taniguchi, J.-S. Oh and T. Matsuyama, Spectrochim. Acta, Part B, 2024, 217, 106972.
57. S. Perez, R. Vasquez, G. Pascual, J. Araya, J. Neira and C. L. Cespedes-Acuna, Food Biosci., 2024, 61, 104475.
58. A. Saha, K. Sanyal, K. Kumari, S. B. Deb and M. K. Saxena, J. Anal. At. Spectrom., 2025, 40(2), 447–456.
59. K. Sanyal and S. Dhara, X-Ray Spectrom., 2024, 53(5), 326–339.
60. X. Y. Wang, S. Y. Liao, D. Q. Hei, J. T. Li, L. Chen, P. K. Cai and C. Cheng, Spectrochim. Acta, Part B, 2025, 228, 107169.
61. A. Cinosi, G. Siviero and Z. McCaffrey, Spectrochim. Acta, Part B, 2024, 220, 107017.
62. B. Wiggershaus, E. Franke and C. Vogt, X-Ray Spectrom., 2025, 54(2), 193–202.
63. E. Cara, P. Hoenicke, Y. Kayser, B. Beckhoff, A. M. Giovannozzi, P. Klapetek, A. Zoccante, M. Cossi, L. Tay, L. Boarino and F. F. Lupi, Nanophotonics, 2024, 13(25), 4605–4614.
64. A. Zoccante, E. Cara, F. Ferrarese Lupi, P. Honicke, Y. Kayser, B. Beckhoff, P. Klapetek, D. Marchi and M. Cossi, Phys. Chem. Chem. Phys., 2024, 26, 18799–18807.
65. A. Hemmerle, N. Aubert, T. Moreno, P. Kekicheff, B. Heinrich, S. Spagnoli, M. Goldmann, G. Ciatto and P. Fontaine, J. Synchrotron Radiat., 2024, 31, 162–176.
66. S. Rey, F. Lédée, G. Freychet, N. Vaxelaire, J. Ihrenberger, E. G. Daillon, S. Lombard, C. Anglade, F. Emieux, P. Gergaud and L. Grenet, Appl. Phys. Lett., 2025, 126(11), 113502.
67. M. A. Rodriguez, T. F. Babuska, J. Curry, J. J. M. Griego, M. T. Dugger, S. R. Larson and A. Mings, Powder Diffr., 2024, 39(2), 60–68.
68. S. Staeck, J. Baumann, P. Hönicke, N. Wauschkuhn, F. Spikermann, D. Grötzsch, H. Stiel and B. Kanngiesser, Nanoscale, 2025, 17(6), 3411–3420.
69. T. N. Terentev, M. Gateshki, A. Tiwari, R. de Vries, V. Jovanovic, M. D. Ackermann and I. A. Makhotkin, Thin Solid Films, 2025, 809, 140588.
70. P. Sarkar, A. Biswas, S. Rai, M. H. Modi, G. Ghorai, P. K. Sahoo, S. N. Jha and D. Bhattacharyya, Phys. Scr., 2024, 99(6), 065952.
71. A. J. Specht, C. Hoover and T. Grier, Curr. Environ. Health Rep., 2024, 11, 442–451.
72. S. I. Mamtha, R. Paranthaman, A. Negi and J. A. Moses, J. Food Compos. Anal., 2025, 140, 107216.
73. E. Santos, J. D. Gozetto, E. de Almeida, M. A. S. Brasil, N. G. d. C. da Silva, V. P. Rezende, H. J. F. A. da Silva, J. R. Brandao, G. S. Montanha, J. Lavres and H. W. P. de Carvalho, ACS Agric. Sci. Technol., 2024, 4(4), 471–477.
74. S. K. Benedito, M. M. de Bem, L. R. G. Guilherme, M. A. C. Carneiro and M. B. B. Guerra, J. Food Compos. Anal., 2025, 137, 106851.
75. N. F. Santos, L. R. Guimaraes Guilherme, M. A. Carbone Carneiro and M. B. Bueno Guerra, J. Anal. At. Spectrom., 2024, 39(4), 1152–1159.
76. T. Kato, R. Ando and Y. Takagai, Bull. Chem. Soc. Jpn., 2024, 97(12), uoae122.
77. Y. Li, J. Hu, C. Li and X. Hou, Anal. Chem., 2024, 96(15), 5757–5762.
78. I. Trebs, C. Lett, A. Krein, E. M. Kawaguchi and J. Junk, Atmos. Meas. Tech., 2024, 17(23), 6791–6805.
79. A. Gianoncelli, S. Schoeder, J. R. Plaisier, M. Fugazzotto, G. Barone, A. Russo, P. Mazzoleni and S. Raneri, Heritage, 2024, 7(4), 2118–2137.
80. S. Schoeder, K. Mueller, L. Tranchant, A. Rouquie, P. Gueriau, M. Thoury, E. Berard, T. Okbinoglu, F. Berenguer, C. Iacconi, L. Robbiola, T. Moreno, S. X. Cohen and L. Bertrand, Appl. Phys. A: Mater. Sci. Process., 2024, 130(11), 848.
81. F. Li, S. Qi, S. T. Yang, H. R. Cao, X. H. Liu and F. Cheng, X-Ray Spectrom., 2025 DOI:10.1002/xrs.3475.
82. T. Gerodimos, I. Georvasilis, A. Asvestas, G. P. Mastrotheodoros, A. Likas and D. F. Anagnostopoulos, Chemom. Intell. Lab. Syst., 2024, 250, 105138.
83. Z. Preisler, R. Andolina, A. Busacca, C. Caliri, C. Miliani and F. P. Romano, Sci. Adv., 2024, 10(39), eadp6234.
84. B. Lach, T. Fiutowski, J. M. del Hoyo-Meléndez, S. Koperny, P. Krupska-Wolas, B. Mindur, P. Wiacek, P. M. Wróbel and W. Dabrowski, npj Heritage Sci., 2025, 13, 58.
85. Z. He, Z. An, N. Huang, P. Wang, J. X. Yuan, H. Wang, J. M. Wang and J. He, IEEE Trans. Nucl. Sci., 2024, 71(9), 2151–2161.
86. M. Colombo, T. Lehmann, W. Ensinger and V. Rossi, J. Anal. At. Spectrom., 2025, 40(4), 989–1005.
87. L. Musilek, P. Novotny and T. Trojek, Radiat. Phys. Chem., 2024, 220, 111731.
88. A. Asvestas, D. Chatzipanteliadis, T. Gerodimos, G. P. Mastrotheodoros, A. Tzima and D. F. Anagnostopoulos, Sustainability, 2024, 16(14), 6135.
89. M. Chiti, D. Chiti, F. Chiarelli, R. Donghia, A. Esposito, M. Ferretti and A. Gorghinian, Condens. Matter, 2024, 9(1), 1.

---