Atomic spectrometry update: review of advances in X-ray fluorescence spectrometry and its special applications

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

Three-dimensional chemical imaging by XRF spectrometry techniques continues to advance in both experimental methods and quantitative data evaluation and reconstruction strategies. These techniques are gaining interest across various research fields, ranging from material science and environmental and Earth sciences to life science and biomedical imaging. Two primary techniques associated with 3D XRF spectrometry are reviewed in this update: XRF spectrometry CT and confocal XRF spectrometry. There has been an increase in the building of in-house specialised 2D XRF spectrometry instruments. Attention to various components, e.g. coating of optics, has improved performance. There was an increase during the review period in the use of SR-XRF spectrometry in conjunction with complementary X-ray spectroscopic and imaging techniques for integrating spatially resolved elemental data with information on speciation and structural and morphological images. Applications of μXRF spectrometry continued to expand in fields such as biomedical, environmental and materials sciences and cultural heritage research. These applications were primarily carried out at specialised hard-X-ray micro- and nano-probe facilities by combining SR-XRF spectrometry with micro- and nano-XAS, XRD analysis, ptychography and various forms of tomographic techniques. The TXRF spectrometry technique continues to be successfully implemented in medical research because of its outstanding performance as a microanalytical method. Changes in the elemental profiles of small organs from, e.g. rats, can be detected. The introduction of a versatile pipetting instrument made possible significant advances in the strategic identification of errors in sample morphology. MacroXRF spectrometry continues to play a significant role in cultural heritage applications. Instrumentation is constantly expanding with new functionalities such as simultaneous measurement with reflectance image spectroscopy and luminescence imaging spectroscopy. The investigation of papyrus fragments was enhanced by upgrading a novel mobile macroXRF spectrometer scanner with new high-performing mechatronics and a high-throughput detection system.

---

1 Introduction

This review describes advances in the XRF spectrometry group of techniques published approximately between April 2023 and March 2024 and follows on from last year’s review.¹ The review is selective with the aim of providing a critical insight into developments in instrumentation, methodologies and data handling that represent a significant advance in XRF spectrometry. It is not the intention of the review to cover comprehensively the applications of XRF spectrometry techniques except in those cases where the non-destructive and remote sensing nature of XRF spectrometry analysis makes it particularly valuable and the method of choice. These applications concern samples which are irreplaceable and of great cultural value such as works of art and archaeological artefacts. For a wider appreciation of the applications of XRF spectrometry, this review should be read in conjunction with other related ASUs in the series, namely: environmental analysis;² clinical and biological materials, foods and beverages;³ advances in atomic spectrometry and related techniques;⁴ elemental speciation;⁵ and metals, chemicals and functional materials.⁶

All the ASU reviews adhere to a number of conventions. An italicised phrase close to the beginning of each paragraph is intended to highlight the subject area of that individual paragraph. A list of abbreviations used in this review appears at the end. It is a convention of ASU that information given in the paper being reported on is presented in the past tense whereas the ASU reviewers offer their critique in the present tense.

2 Chemical imaging using X-ray spectrometry techniques

2.1 Computed tomography and 3D XRF spectrometry techniques

2.1.1 X-ray fluorescence spectrometry computed tomography. Elemental mapping through virtual cross sections of the sample and full 3D elemental imaging by XRF spectrometry-CT are mainly used at synchrotron micro- and nano-beam facilities for the analysis of μm to mm sized samples. The 3rd and, more recently, the 4th generation sources facilitate these applications at sub-μm spatial resolution and are typically used in combination with complementary imaging techniques such as ptychography, absorption and phase-contrast CT, XRD analysis-CT and 2D and 3D XAS.

The potential of an advanced combination of synchrotron XRF spectrometry-CT and XRD analysis-CT was demonstrated by Lanzirotti et al.⁷ by obtaining detailed visualisations of elemental and mineralogical distributions across cross-sections in uncut extraterrestrial samples. Both XRF spectrometry-CT and XRD analysis-CT methods were applied simultaneously to fragments of the CR2 chondrite LaPaz Icefield (LAP) 02342, H5 chondrite MacAlpine Hills (MAC) 88203 and CM2 chondrite Murchison. An 18 keV X-ray beam focused to 1 (vertical) × 2 (horizontal) μm was used to scan the samples continuously at 1 μm every 25 ms and XRF spectrometry and XRD analysis signals were collected for each point over a full angular rotation of 360°. The emitted and diffracted intensities could be reconstructed into 2D slices with a resolution of 1–2 μm thereby highlighting the ability to distinguish, for example, between serpentine minerals and tochilinite–cronstedtite intergrowths without resorting to disruptive sample sectioning or extraction. This method provided new opportunities for investigating samples in a non-destructive manner so preserving microstructure and chemistry.

In a remarkable study, Graefenstein et al.⁸ demonstrated the tomographic XRF spectrometry of frozen-hydrated samples of A. schoenoprasum leaves. The data were collected using a cryogenic XRF spectrometry vacuum setup equipped with a cryogenic sample transfer system. The analysis of vitrified samples was undertaken at a pressure of 10⁻⁶–10⁻⁸ mbar at a typical sample temperature of 120 K. To obtain cross-sectional maps of physiologically relevant elements (Ca, K, Mn, P, S, Zn) at μm-level resolution, the synchrotron beam at PETRA III P06 was focused to 440 × 500 nm using Kirkpatrick–Baez mirrors. The XRF spectrometry sinograms were processed using maximum-likelihood-expectation-maximisation algorithms with advanced self-absorption correction. By applying a tomographic multi-channel-analyser (MCA) hyperspectral-reconstruction method, the researchers generated 2D intensity maps for each channel. They then applied peak fitting by PyMCA to the spectra of individual voxels or groups of voxels for enhanced quantitative analysis. This approach significantly improved the data quality for light elements and trace metals compared to traditional maximum-likelihood algorithms and made possible accurate quantification of elemental concentrations across leaf cross-sections while accounting for depth limitations arising from photon escape.

Another application of XRF spectrometry-CT in plant science was the study of Morina et al.⁹ who investigated the impact of Cd and Zn concentrations on the anti-viral response of the hyperaccumulator plant N. caerulescens when infected with Turnip yellow mosaic virus (TYMV). The investigated plants were cultivated under various metal exposure conditions (deficient, replete, and excess) and examined in terms of their response to viral infection by various biochemical and X-ray imaging techniques. In contrast to non-hyperaccumulator plants for which Zn distribution patterns changed in response to exposure to TYMV, the distribution of Zn in hyperaccumulators did not change. The XRF spectrometry-CT results highlighted that N. caerulescens, in contrast to the non-hyperaccumulator N. ochroleucum, did not significantly change its Zn uptake or sequestration in response to viral infection when exposed to high Zn or low Cd concentrations.

Detailed analyses of Laminaria digitata and Ectocarpus siliculosus were undertaken by Mijovilovich et al.¹⁰ to obtain a better understanding of Fe storage in brown algae. Unlike many other organisms that utilise ferritin for iron storage, brown algae have developed a unique system to accommodate the low iron availability in their marine environments. The accumulation sites of Fe within the algae could be located using advanced imaging techniques such as X-ray microprobe imaging and nanoprobe XRF spectrometry tomography. Iron accumulated predominantly in the cortex of these algae with a subcellular localization, i.e. the Fe was stored within cells (symplastic) rather than in spaces between cells (apoplastic). This finding contrasted with previous findings using conventional chemical fixation methods which suggested apoplastic iron accumulation. In addition, μXANES measurements provided insights into how Fe was bound within the algae; Fe in L. digitata was stored in a mineral non-ferritin core.

The hyperaccumulation of Se was studied by van der Ent et al.¹¹ for three hyperaccumulating taxa (Astragalus bisulcatus, Stanleya pinnata and Neptunia amplexicaulis) by using μXRF spectrometry techniques, including XRF spectrometry-CT. The XRF microscopy experiments were performed at the hard X-ray microprobe beamline P06 (PETRA III) using a 1 × 1 μm focused beam at 17 keV excitation energy. Fundamental similarities between the distributions for the three hyperaccumulators were revealed by the very high quality XRF spectrometry-CT images of Se distribution at the organ, tissue and cellular levels. Maximum Se accumulation was detected in emerging leaves. Whereas preferential localisation of Se occurred in the foliar apoplastic space in Astragalus and Stanleya, there was an absence of Se in xylem vessels of roots, stems and petioles.

The first nanoscale 3D synchrotron XRF spectrometry-CT mapping of a coccolith, presented by Walker et al.,¹² was based on data collected using the hard X-ray nanoprobe I14 at the DLS using a beamsize of 50 nm and an excitation energy of 18 keV. The Sr distributions in S. apsteinii lopadoliths were uneven and had bands of varying Sr to Ca concentration ratios. This could not be explained by standard models for Sr accumulation and fractionation in coccoliths and so demonstrated the need for a deeper understanding of coccolithogenesis.

A beautiful example of complementary XRF spectrometry and ptychographic nano-tomography analyses was presented in a cultural heritage study by Broers et al.¹³ who studied a small paint fragment taken from Rembrandt’s “The Night Watch”. The average 3D spatial resolution was ca. 600 nm for both the XRF spectrometry and the ptychographic datasets when a focused beam of 200 (vertical) × 300 (horizontal) nm was used at 17 keV. When the full 3D datasets were combined, an unknown lead-containing layer was revealed. This presumably acted as a protective impregnation layer applied to the canvas before the quartz-clay ground was applied.

A machine-learning algorithm was developed by Zheng et al.¹⁴ to enhance the spatial resolution of the XRF spectrometry (tomographic) images by mitigating the effects of blurring due to the beam profile. The model was trained by using simulated XRF spectrometry images followed by fine-tuning with experimental data. This second step was necessary to account for differences between simulated and experimental XRF images due to their distinct noise characteristics and the inherent complexities associated with the experimental data. Tomographic reconstructions from enhanced XRF spectrometry projections gave a four-fold improvement in spatial resolution and revealed distinct internal features which were invisible in the original low-resolution XRF spectrometry tomography of a battery sample.

2.1.2 Confocal XRF spectrometry. A comprehensive review on confocal XRF spectrometry by Heimler et al.,¹⁵ emphasised the value of the technique in various scientific fields for which detailed 2D/3D elemental analyses were essential. The technique played a critical role in studies of historical artefacts and artworks and offered significant insights into provenance, composition and methods of production. The critical review presented recent advancements in the technique with detailed descriptions of the commonly used setups together with a discussion on calibration and quantification methods used in confocal XRF spectrometry. The technique was expected to broaden its application area across many fields, making it an increasingly popular choice for researchers requiring detailed and non-destructive 2D/3D elemental analysis.

Morrell et al.¹⁶ discussed the challenges presented in the correlative imaging of exogenous particles within tissue microanatomy and stressed the importance of analysing their concentrations, speciation and distributions to understand better biological responses. To address the limitations of traditional microscopy techniques, the study focused on synchronous X-ray imaging strategies, including confocal-XRF spectrometry, that allowed simultaneous identification of multiple chemical features thereby improving the assessment of spatial relationships between key elements. A new imaging strategy involving lanthanide-tagged antibodies was evaluated through simulations and confocal XRF spectrometry experiments to identify suitable lanthanide tags. This approach was employed to identify simultaneously titanium exposure and CD45 positive cells at sub-cellular resolutions. Significant heterogeneity in the distribution of particles and cells across serial sections highlighted the necessity for synchronous imaging methods. The confocal XRF spectrometry measurements were performed at the DLS beamline I18 using an incident beam energy of 5.7 keV and a beam size of 5 × 5 μm.

Various imaging and spectroscopic techniques (X-ray CT, confocal XRF spectrometry, XRD analysis, IR spectroscopy, EM and Raman spectroscopy) were employed by Young et al.¹⁷ for the analysis of two historic metal artefacts recovered from Kuaua Pueblo (New Mexico, USA). Detailed 2D- and 3D-XRF spectrometry scans were performed to map the distributions of Ca, Cu, Fe, K, Mn, Si and Zn. The main constituent of both artefacts was identified as Cu. A custom-built confocal μXRF spectrometry instrument (30–100 μm steps, 25–30 kV X-ray tube voltage, 0.5–0.8 mA current and 3 s per pixel dwell time) was used in both 2D- and 3D-modes for elemental mapping. The overall scanning times were 16–177 h.

Non-destructive analysis is of utmost importance in forensic science and archaeology. Confocal XRF spectrometry is a valuable technique for these analyses due to its ability to analyse the elemental composition of a sample both on the surface and at depth. A practical application was demonstrated by Mori et al.¹⁸ in the study of two very similar ceramic samples with distinct differences in elemental compositions. Using a confocal volume of ca. 10 × 10 × 10 μm at 17.4 keV, the analysis focused on blue painted areas of the ceramics containing Fe and Mn. Depth profiling of Ca, Co, Fe, K, Mn and Zn revealed variations, particularly in Co and Zn concentrations, which made it possible to differentiate between the sources of these materials.

Matsuyama et al.¹⁹ demonstrated an elegant application of confocal XRF spectrometry for the in situ observation of the electrochemical reaction of Zn primer steel plate in NaCl solution. The motivation was to understand better the corrosion behaviour of Zn-coated steel plates which are widely used because of their resistance to corrosion. The elution of Zn on the anodic side of the steel plate indicated the dissolution of the protective Zn layer due to the corrosive action of the Cl⁻ ions and the electric current. In contrast, Fe was not detected, suggesting that the Zn coating provided effective protection and prevented the underlying steel from corroding. Conversely, neither Fe nor Zn were observed on the cathodic side, indicating that this area was protected. The LOD for Zn in the NaCl solution was 57.9 ppm when using an acquisition time of 30 min.

The use of confocal XRF spectrometry in conjunction with SEM, IR spectroscopy, Raman spectroscopy and XRD analysis was demonstrated by Kubiak et al.²⁰ for the characterisation and optimisation of goethite–spongin composites to be used for the electrochemical sensing of dopamine in human urine samples. The confocal XRF spectrometry measurements were performed using a 30 W Rh-microfocus X-ray tube (50 kV, 600 μA), a polycapillary full lens in the excitation channel for X-ray focusing and a polycapillary half lens in front of a 60 mm² SDD. This setup provided a depth resolution of 69.0 and 31.4 μm at the S Kα and Br Kα line energies, respectively. A full 3D XRF spectrometry analysis of a total sample volume of 500 × 500 × 500 μm with a step size of 5 μm and 50 ms acquisition time per point resulted in an overall measurement time of 63 h.

In an important environmental application, the effectiveness of pine-needle biochar produced by oxygen-limited pyrolysis was studied by Zhang et al.²¹ for the removal of Cr^VIfrom water. This study was the first to provide detailed information on the spatial distribution and speciation of Cr within pine-needle biochar. The confocal XRF spectrometry technique’s large detection depth of 600 μm was ideal for acquiring 3D depth profiles. This comprehensive study provided crucial insights into the mechanisms of Cr^VI removal by pine-needle biochar which will be valuable for environmental remediation.

A new software called voxTrace was based²² on a Monte Carlo ray-tracing approach. It enhanced the accuracy and efficiency of simulating confocal XRF spectrometry spectra by incorporating detailed interactions such as secondary excitation and various scattering processes. One of the key advantages of voxTrace was that it removed the need for peak deconvolution in the analytical process. This was particularly beneficial for materials exhibiting overlapping fluorescence peaks for which traditional methods offered limited accuracy. The code was optimised for high computational efficiency using graphics processing units with Compute Unified Device Architecture (CUDA) technology and allowed for efficient simulation of e.g. elemental mapping and improved depth profiling.

Kranz et al.²³ presented advancements in the development and application of new cone-shaped polycapillary optics for full-field XRF spectrometry systems which can be applied to confocal imaging based on sheet beam excitation. Use of the new optics increased the magnification factor to 41 and thereby achieved a spatial resolution of 2.4 μm. The integration of these innovative optics into full-field XRF spectrometry setups represented a major step forward in the field of XRF spectrometry imaging by significantly improving the spatial resolution achievable by laboratory systems.

2.2 Laboratory 2D XRF spectrometry techniques

Dong et al.²⁴ used a simulation model and experimental studies to optimise the performance of a self-developed microfocus X-ray source. Aperture sizes of 0.5–0.8 mm for the e-beam gave the best results. The absorption images showed that, when probing a line card reference sample, a 0.8 mm aperture gave complete line resolution but a 0.2 mm aperture could not resolve the lines. When a carbon coating was applied to the surface of the 0.2 mm aperture to suppress fluorescence noise from the aperture material, line separation also became possible. Apertures with 1.5 and 2 mm diameters gave poor line resolution and were not considered further.

A monochromatic, low-power (5.3 W) μXRF spectrometry instrument combined²⁵ two polycapillary half lens optics and a crystal monochromator (Si (111) or SiC (006)). When the primary Mo Kα-line was used, the spectral resolution (FWHM) was 270 eV and the spot sizes ca. 40–200 μm. The LODs of Mn and Se in a quasi-matrix-free standard solution were 970 and 52 μg L⁻¹, respectively. A compact polychromatic μXRF spectrometry instrument (Rh target, 50 W X-ray tube) with a polycapillary minilens and a 30 mm² silicon drift detector was built²⁶ for the analysis of nuclear materials held inside a glove box. The focal spot size was 40 μm at 9.7 keV. The performance of this new polychromatic instrument was compared to that of an existing monochromatic instrument. The LODs obtained using the two instruments were comparable; e.g. that for As was 5 and 2 μg g⁻¹ for the polychromatic and monochromatic instruments, respectively. The LODs for Pb and U using the polychromatic instrument were 64 and 61 ng mL⁻¹, respectively.

Yuan et al.²⁷ used an iridium-coated ellipsoidal monocapillary in a low power μXRF spectrometry instrument (Mo target, 30 kV, 400 μA) for measuring the thicknesses of Cu coatings on silicon. Whereas the relative deviation for thicknesses of 0.85–30.00 μm was 3%, that for thicknesses of 30.0–56.5 μm was 5%. The results agreed very well with those obtained by the established SEM method; the relative differences between the two methods being 2–5%. In comparison to the use of an uncoated capillary, the increased excitation flux from the higher captured solid-angle of the coated capillary resulted in a five-fold increase in the Cu signal.

A commercial μXRF spectrometry instrument (Rh-target microbeam X-ray tube; 150 mm² SDD) was applied²⁸ to study calcite precipitates microbially-induced for soil improvement. The optimal segmentation threshold used to determine the volume of Ca and Si particles was optimised using SEM imaging. The Ca contents of 16.4, 18.2 and 17.9% determined with μXRF spectrometry, SEM imaging and bulk chemical analysis using acid extraction, respectively, were in good agreement. The porosity decreased by ca. 10% after the calcite formation. The maximum differences in apparent porosity and apparent calcium carbonate content in different sub-images (N = 16) were 12 and 6%, respectively.

Yang et al.²⁹ evaluated the performance of a commercial μXRF spectrometry instrument used to study elemental distributions in a rod made from an Al–Zn–Mg–Cu alloy. At a dwell time of 30 s, the LODs for Cu, Mg, and Zn were 0.002, 0.068 and 0.007 wt%, respectively. Optimal parameters for measurements at the highest count rates were 30 kV and 600 μA. The results agreed very well with those obtained by established spark-ablation optical emission spectrometry.

3 Synchrotron and large scale facilities

Several comprehensive review articles highlighted advances in SR-XRF spectrometry imaging and associated techniques in various areas of life and environmental sciences. Advances included: X-ray-sensitive probes for in situ bioimaging at the nanoscale;³⁰ multimodal and multiscale correlative elemental imaging over scales from whole tissues to organelles;³¹ synchrotron radiation methods for studying biological interaction of fibres;³² SR X-ray methods for studying mercury neurotoxicology;³³ and synchrotron investigations in environmental radiochemistry research.³⁴ De Pauw et al.³⁵ reviewed laboratory, commercially available and facility-based WDXRF spectrometers.

Buzanich et al.³⁶ gave an overview of the analytical methods and experimental setups available at the BAMline (BESSY II, Berlin, Germany). These were geared towards non-destructive testing in various disciplines including materials science, chemistry, biology, medicine and cultural heritage. In addition to standard μXRF spectrometry and μXAS, the BAMline offered CT applicable to a wide range of sample environments. Several XRF spectrometry modalities were available: μXRF spectrometry mapping (energy range 4–40 keV; spatial resolution as good as 1 μm) with LODs of tens of μg g⁻¹ in fast mapping mode; coded-aperture XRF spectrometry (detection of features down to 100 μm in size; energy range 4–30 keV); high-energy-resolution XRF spectrometry (down to 13 eV); and X-ray sheet-microscopy for obtaining 3D information with a voxel size of 2 × 8 × 8 μm. The paper also outlined proprietary methods for the synthesis of materials and equipment developed specifically for the BAMline.

A new full-field XRF spectrometry imaging (FXI) station at the Synchrotron Light Research Institute (Thailand) was described in detail by Klysubun et al.³⁷ This FXI station used an unfocused synchrotron X-ray beam originating from a bending magnet, measuring 2 (vertical) × 13 (horizontal) mm and with a tunable photon energy (1–10 keV). The FXI experiments utilised an energy-dispersive pn-type CCD array (256 × 256 pixels) equipped with polycapillary optics to achieve a full-frame image size of 12.3 × 12.3 mm with a spatial resolution of 68 μm. The pixel level LODs at the FXI imaging station ranged from 0.3 (K) to 0.03(Zn) wt%.

Conventional μXRF spectrometry typically uses the L lines of REEs which often suffer from interferences, in particular from transition metals. Nagasawa et al.³⁸ developed high-energy μXRF spectrometry and μXAS techniques at the BL37XU beamline in SPring-8, Japan to excite the K-lines of the REEs from La to Dy thereby avoiding interference from other elements. The setup utilised an incident X-ray microbeam (1 × 1 μm, 38–54 keV). The new high-energy approach made it possible to obtain: quantitative REE patterns (La to Dy) in geological materials; distributions of REEs and other elements with a spatial resolution of <1 μm without interferences; and μXAS data at various edges from 7 to 54 keV, including the K-edges of REEs.

A new imaging approach named multielement Z-tag X-ray fluorescence (MEZ-XRF) spectrometry was³⁹ a significant advancement in bioimaging and offered rapid, non-destructive and highly multiplexed imaging capability across molecular to supra-cellular scales. The new approach involved staining of biological samples with Z-tagged affinity reagents containing lanthanides and raster scanning of the stained sample with a focused (ca. 500 nm) 69 keV X-ray beam. The approach was optimised for the detection of lanthanide-based Z-tags in biological samples with LODs as low as 0.3 to 1 ppm and could be successfully applied to the imaging of element-tagged biological molecules in cells and tissues (e.g. breast, tumour, tonsil, appendix).

Li et al.⁴⁰ introduced a novel XRF spectrometry imaging technique that employed illumination of the sample with a structured beam and a generative image-reconstruction-model to achieve nanoscale resolution in mapping compositional heterogeneity. By combining a full-field transmission X-ray microscope and an XRF spectrometry detector, this approach circumvented the complexities of nanoscale X-ray focusing and raster scanning. The potential of the new technique for achieving nanoscale spatially resolved images was highlighted by the imaging of a battery sample containing mixed cathode materials. The compositional variations within the cathode particles were successfully captured with a resolution of ca.100 nm. This new imaging strategy not only enhanced detection sensitivity and spatial resolution but also improved experimental throughput thereby making it highly suitable for a wide range of applications.

Hafner et al.⁴¹ described the design, development, integration and testing of a novel in situ AFM instrument operating under high vacuum in a synchrotron soft X-ray microscopy end-station (TwinMic beamline, Elettra, Trieste, Italy). This was the first demonstration using μXRF spectrometry and AFM together in the soft X-ray regime. The simultaneous acquisition of detailed AFM topography and XRF spectrometry maps provided complementary information with scanning steps of ca. 250 nm.

An interesting combination of nanoscopic XRF spectrometry and X-ray excited optical luminescence (XEOL) using an advanced hard X-ray nanoprobe (ESRF ID16B, Grenoble, France) was employed by Plass et al.⁴² to investigate the dynamics of colour centres in Co-doped ZnO nanowires. The combined scanning approach used an excitation energy of 17.5 keV in “pink-beam” mode (ΔE/E = 10⁻²), a focal spot size of 62 (vertical) × 74 (horizontal) nm and a photon flux of 3 × 10¹⁰ to 6 × 10¹² photons per s. The simultaneous use of XRF spectrometry and XEOL provided insights into compositional and functional variations at the nanoscale and demonstrated potential for a better understanding of emerging quantum technologies.

The characterisation of heterostructured nanowires poses significant challenges, particularly in terms of measuring the strain together with elemental composition at the nanoscale. To address this, Hammarberg et al.⁴³ employed nanofocused scanning X-ray diffraction (nano-XRD) analysis and XRF spectrometry with a 60 nm beam (NanoMAX beamline, MAX IV, Lund, Sweden) to acquire spatially resolved maps of CsPbBr₃ nanowires. The information that could be acquired for the composition, lattice spacing and lattice tilt was critical for understanding the material’s structural and compositional dynamics.

4 Grazing X-ray spectrometry techniques including TXRF spectrometry

Hampel et al.⁴⁴ improved the strategic identification of errors in TXRF spectrometry analysis with respect to the morphology of the dried sample by developing a versatile pL pipette for droplet delivery. The dosing and patterning of samples with 65 elements in variable quantities made it possible to produce many desired element compositions and specimen morphologies. The challenges of cartridge corrosion in acidic solutions and analyte adsorption in basic solutions were overcome through the use of complexing reagents such as EDTA in neutral to alkaline media. This new pipette was used⁴⁵ to study the influence of the formation of ring-like specimens (known as “coffee rings”) on the accuracy of TXRF spectrometry analyses because the signal intensities varied as a function of the position of the ring-like specimens relative to the detector. The extreme condition of complete spatial separation of analyte (prepared as concentric rings) and internal standard (prepared as spot or array in the middle of the sample carrier) was investigated both experimentally and theoretically. Depending on the distance of the ring deposits from the optimal position under the detector, negative biases of 32 (2 mm radius) to 99% (6 mm radius) were determined experimentally. These matched well the theoretically calculated biases of 33 (2 mm) to 95% (6 mm).

The morphology of dried samples (e.g. a coffee ring, concentrated stain or film) depends, among other things, on the polarity of the substrate. When TXRF spectrometry is used to measure low-Z elements in sample solutions with high concentrations of matrix elements, obtaining sufficient signal can be difficult because the XRF is absorbed more by thick residues of the matrix than by thin residues of the analytes. In contrast to the normal approach of overcoming this problem by using hydrophobised reflector surfaces to prevent coffee ring formation and to favour the formation of a small spot-like deposit, Matsuyama et al.⁴⁶ pursued exactly the opposite strategy of preparing hydrophilic surfaces. The glass substrate was treated with an ammonia–hydrogen peroxide mixture (APM) to decompose organic matter and to etch the glass material slightly. The APM-treated region was limited to a diameter of 6 mm by use of a PTFE mask placed on the glass substrate. Their aim was to lower the absorption of low-energy radiation by decreasing the specimen thickness. When compared with results for specimens prepared on classical hydrophobic reflectors, the LODs for analysis of a multielemental standard solution containing Al, Ca and Fe (1000 ppm) and Ba, Cr, Mg, Pb and Sr (100 ppm) were improved from 4 to 1 ppm when using the Al Kα lines and from 1 to 0.8 ppm when using the P Kα lines. No further improvements in the LODs were obtained when the K Kα line or lines with higher energy were used. The same effect was achieved⁴⁷ by using an atmospheric-pressure plasma jet for hydrophilisation of the reflector. When using the Kα lines, the LODs for a multielemental standard solution of Al, Ca, Fe, K and P (100 ppm) and Ba, Cr, Mg, Pb and Sr (10 ppm) were improved from ca. 6 to 2 ppb.

Margui and Torrent⁴⁸ improved a simple and inexpensive method for the determination of Cr^VI in waters using TXRF spectrometry. The LLME procedure was based on the formation of an ion-pair between the cationic part of the surfactant cetyltrimethylammonium bromide and the anionic Cr^VI which was extracted in a few μL of chloroform. No further sample treatment was required. The dynamic range was extended to 5–5000 μg L⁻¹. When applied to the analysis of various matrices (tap water, well water, seawater, galvanic waste and clinker extracts), this procedure gave a very good LOD of 0.9 μg L⁻¹. By modifying graphene oxide with tetraethylenepentamine, Musielak et al.⁴⁹ adsorbed Cr^VI ions from aqueous solutions (pH 3.5) with a high preconcentration factor (100) and high recovery rates (98.5–100%) even though only minimal amounts of adsorbent (10–50 μg mL⁻¹) were used. The graphene oxide/tetraethylenepentamine adsorbed Cr^VI selectively in the presence of Cr^III to give an excellent LOD of 3.5 pg mL⁻¹.

In the determination of trace concentrations of Se^IVand Se^VI in the same sample, Musielak et al.⁵⁰ used dispersive μSPE on thiosemicarbazide-incorporated graphene to prepare samples for TXRF spectrometry analysis. Graphene derivatised in this way adsorbed only Se^IV. Two procedures were used to determine the Se^VI content: either from the difference between total Se and Se^IV concentrations or by reducing the Se^VI in the residue and extracting the Se^IV formed. Various beverage RMs (CRMs or spiked beverages) were analysed. Good recoveries (97–110%) of total Se were obtained. Recoveries of spikes from spiked beverages were also good (97 and 117% for Se^IV and Se^VI, respectively).

Deng et al.⁵¹ used Prussian blue loaded on nanocomposites of magnetic sepiolite for the preconcentration of Tl from waste water. In comparison with the analysis of samples prepared without enrichment, the LODs for TXRF spectrometry analysis were improved ca. 100-fold to 0.84 μg L⁻¹ and the recovery efficiency remained high at >94%.

The determination of lanthanides in waters of high ionic strength by using L-lines is often biased due to line overlaps from the K-lines of other elements. Cloud point extraction with N,N,N′,N′-tetra-octyl-diglycolamide in Triton X-114 was used⁵² to extract the lanthanides present in water CRMs of different ionic strength and spiked with 13–19 μg L⁻¹ of each of the lanthanides. Calibration with PLS regression was used to achieve accurate results. The recoveries were 98–99% except in the presence of some specific ions (e.g. Th⁴⁺ and UO₃²⁺ at 10 mg L⁻¹ or higher) when the recovery was 95%. Other ions (e.g. K⁺ and Na⁺), however, had no effect, even at concentrations up to 2000 mg L⁻¹. Akhmetzhanov et al.⁵³ studied the benefit of PLS regression for the determination of Th and U in the presence of Rb and Sr. In Rb-rich samples, the bias of 30–60% due to line overlap was reduced 2–3 fold by using PLS. However, conventional deconvolution of the spectrum was preferred for Rb-depleted samples because of its simplicity.

The advantages of slurry sampling for TXRF spectrometry analysis include the minimisation of contamination and avoidance of digestions employing hazardous acids like HF. However, the procedure can suffer from matrix effects. A simple acid-digestion procedure was preferred⁵⁴ to suspension-assisted sample preparation using 1% Tritron X-100 as detergent for the determination of Cu, Fe, Mn and Zn in sponges from Lake Baikal. A grinding time of ca. 40 min was required to avoid underestimation of Cu and Fe concentrations. In general, the RSDs for Fe and Mn were higher for analyses employing slurry sampling (up to 40%) than for those involving digestion (four to ten times lower). Samples containing high-density fibres rich in Cr, Cu, Fe and Mn, as identified by BSE imaging, had especially high RSDs. Results for sponges from eight locations around Lake Baikal were comparable with those from earlier studies; e.g. the Zn concentrations in the south basin of 26 to 76 μg g⁻¹ agreed well with the values (28 to 82 μg g⁻¹) obtained previously. In a separate study, the precision of results obtained using slurry sampling were improved⁵⁵ through use of viscosity-modifying additives. Increasing the viscosity of suspensions of copper–nickel sulphide ores through addition of ethylene glycol resulted in RSDs of <10%.

An investigation of influences on line intensities in TXRF spectrometry found⁵⁶ that measurement deadtime resulted in non-linear effects which could not be explained by matrix absorption. Dhara⁵⁷ used a TXRF spectrometry system with optimised low-Z detection using a Cr-tube and high-Z detection using a Rh-tube to detect significant differences between U M and U L lines with respect to their oxidation state. The SDD (resolution 130 eV) was able to resolve the M5N6 and M4N6 U lines. A review paper on the suitability and applications of TXRF spectrometry for characterisation of nuclear materials gave⁵⁸ a comprehensive overview of the subject.

Collection of PM on filters has the disadvantage of requiring disassembly of the sampler prior to analysis of the filters. Takahara et al.⁵⁹ evaluated semi-continuous sampling of PM_2.5 on an hourly basis. Particles were collected on a PTFE roll tape and sections of the tape analysed by XRF spectrometry and TXRF spectrometry under GIXRF spectrometry conditions. The TXRF spectrometry intensities were calibrated using the XRF spectrometry results for Zn. The GIXRF spectrometry analyses gave better LODs than the XRF spectrometry analyses: for example the LOD for K was improved from 18 to 2.5 ng m⁻³ and that for Zn from 6.9 to 0.19 ng m⁻³. A comprehensive review paper by Bilo et al.⁶⁰ compared sample preparation procedures (digestion, ashing) for the EDXRF or TXRF spectrometry analyses of PM collected on filters. The authors concluded that the direct analysis of filters would be preferred but also that further work was required to increase confidence in the analysis of aerosols by XRF spectrometry.

The exceptional microanalytical properties of TXRF spectrometry are also advantageous for the analysis of tissues and body fluids in human health studies. Jablan et al.⁶¹ analysed urine collected from non-professional sportsmen before and after a mountain ultramarathon. Their procedure was validated using the RM Seronorm™ Trace Elements Urine Level 2 for which recoveries ranged from 81 (Zn) to 135% (Pb). The LODs were as low as 7.2 (Rb) and 27 g L⁻¹ (Pb). The concentrations of most elements, but in particular As and Rb, increased during the race and remained higher for up to 12 h after the race. Interestingly though, the concentrations of Pb in the urine decreased after the race.

Only 2.5 mL (ca.1 g) of digested rat tissues was needed⁶² in a study of changes in Ca, Cu, Fe, K, P, S, Se and Zn concentrations in liver, spleen and kidneys from rats on a ketogenic diet. The most significant differences were found in the liver tissue from male rats; the concentrations of Ca, Cu, K, P, S, Se and Zn being significantly lower in rats fed the high fat diet. Interestingly, differences between the sexes were often more pronounced than those between the diets. For example, whereas the Ca concentrations in male rats decreased, those in females increased. In a study on rats implanted with human glioblastoma, significant elemental change was detected,⁶³ especially in the spleen and lungs of the animals, even if the implanted tissue did not develop into a tumour in the rat brain. The concentrations of Ca, Cu, K, P and Zn decreased significantly in spleens of implanted rats. These studies are wonderful examples of how the microanalytical property of TXRF spectrometry can be used to study metabolic difference in rat or mouse organs. This was highlighted⁶⁴ by the first round-robin (four participating laboratories) on the determination of elements in kidney, heart, spleen and lung of rats. The interlaboratory precision was <12% for high-Z elements (Fe, Cu and Zn). Results for the low Z element concentrations determined by TXRF spectrometry deviated significantly from those obtained with the reference methods (ICP-AES, ICP-MS). For example, P could be underestimated by 30%, overestimated by 90% or even not determined at all.

Kubala-Kukus et al.⁶⁵ applied commercial computing programs for the statistical analysis of the large amount of data produced for tissue samples (blood, spleen, kidney) from various sources. By applying survival analysis procedures (the statistics of events expected at different time intervals), they obtained insights on the data distribution. The resulting survival distribution function then yielded information about the part of the distribution below the LODs.

Alam et al.⁶⁶ analysed silicon samples implanted with nm layers of Ni at a depth of ca. 100 nm by using XRR and GIXRF spectrometries simultaneously. The annealing step at 800 °C led to an inward and outward diffusion of Ni, thereby broadening the implantation layer. The Gaussian distribution of the Ni concentration profiles obtained using both techniques was confirmed by RBS. The SIMS analysis could not reproduce the Gaussian distribution in the annealed samples. Non-uniform sputtering up to a few nm depths and the irregular shapes of the surface craters were major sources of errors that caused the SIMS analysis to be generally less accurate at the surface. The Ni K-edge XANES spectra for implanted samples both before and after annealing differed only slightly and strongly indicated an inter-metallic Ni–Si phase. A very good match between the results of XRR-GIXRF spectrometry and RBS analyses was obtained⁶⁷ in a study on Ta/Cr/Pt planar X-ray wave guide material. The materials were composed of layers of Ta (ca. 4 nm thick), Cr (7–20 nm thick) and Pt (ca. 14 nm thick) with Cr concentrations of 5.63 × 10⁻¹⁶ to 16.56 × 10⁻¹⁶ atoms cm⁻².

The GIXRF and GEXRF spectrometry approaches are both powerful tools for extracting information on elemental layers of materials. The main difference between them is that in GIXRF spectrometry the primary beam is influenced by the XSW and suffers significant absorption whereas in GEXRF spectrometry it is the fluorescence that is affected. A comparison of GIXRF spectrometry and GEXRF spectrometry methods used⁶⁸ the reconstruction of a layered Ta/Co/Cu/Co/Ta structure. Results for layer thickness were in agreement, e.g. 4.48 nm for Cu by GIXRF spectrometry and 4.38 nm for GEXRF spectrometry, but agreement was somewhat poorer for concentrations, e.g. 8.800 × 10²² atoms per cm³ for Cu by GIXRF spectrometry and 8.489 × 10²² atoms per cm³ by GEXRF spectrometry. The GEXRF spectrometry reconstructions were in general more precise than those by GIXRF spectrometry. Uncertainties on the model parameters of combined XRR and GIXRF spectrometry data based on the Bootstrap technique were obtained⁶⁹ for the characterisation of a thin layer system. The 63% confidence intervals of the model parameters were calculated by generating 100 Bootstrap samples. The confidence interval values on all the model parameters were very low at ≤5%, so reliable structural parameters for thin-film multilayers could be obtained using the combined XRR-GIXRF spectrometry procedure.

5 Hand-held, mobile and online XRF spectrometry techniques

5.1 Hand-held and mobile XRF spectrometry techniques

In recent years, hand-held XRF spectrometry has found routine use across various applications. However, new innovations continue to provide improved LODs, optimised calibration and enhanced data processing. This year’s ASU highlights these aspects and considers both the advantages and limitations of hand-held XRF spectrometry. The benefits include speed, portability, non-destructive nature and quantitative capabilities. Although it has proven to be cost-effective, understanding its limitations is crucial as covered in several critical reviews. Kobylarz et al.⁷⁰ considered the use of field-portable XRF spectrometry in forensic sciences for which the preservation of evidence is critical. Innovation, affordability, speed and in situ analysis were covered. In a critical assessment, it was anticipated that the XRF spectrometry technique would continue to gain prominence in the ever-evolving field of forensic science. The speed, convenience and precision of analysis by portable XRF spectrometry also hold promise for in situ food product analysis. A review by Frydrych and Jurowski⁷¹ focussed on analytical aspects such as calibration strategies, operating modes, LODs, LOQs and linearity. Available spectrometers and successful analytical calibration strategies were highlighted.

Tavares et al.⁷² reviewed the use of EDXRF spectrometry for quantification of soil nutrients. The advantages and limitations of monitoring nutrients using XRF spectrometry were compared with those of other sensor techniques. The review highlighted knowledge gaps in XRF spectrometry-based assessment of soil-nutrient status. Challenges in establishing effective XRF spectral libraries included mitigation of physical and chemical matrix effects which impacted the relationship between XRF spectrometry signals and total soil element contents. Additionally, the need for agronomic models for converting XRF spectrometry data into available nutrient contents for plants was emphasised. The discussion also considered the in situ application of portable XRF spectrometry equipment.

Roberts et al.⁷³ determined toxic metals in seafood samples using portable XRF spectrometry. The system employed three proprietary DCC optics to provide monochromatic excitation (6.4, 17.4, and 34 keV) and to focus the primary X-ray excitation beam to a spot size of ca. 1 mm. The samples were also analysed using microwave-assisted digestion and ICP-MS/MS analysis. Daily QC performance data for biological CRMs were collected under field conditions for As, Cd, Mn, Pb and Zn. Only six elements in the seafood samples were detectable with only As, Mn, Sr and Zn being consistently quantifiable. Whereas agreement between results obtained by XRF spectrometry and ICP-MS/MS was considered reasonable (±20%) for As, Mn and Sr, agreement for Zn was much poorer. Discrepancies were attributed to sample preparation time, sample heterogeneity and the small spot size of the XRF beam. Therefore, practitioners should recognise limitations, validate specific analytes and receive training in field-based sample preparation for XRF spectrometry analyses.

Bone lead measurements have traditionally been made using K-shell XRF spectrometry with a radioisotope source and a 30 min measurement time. However, low energy hand-held systems utilising an X-ray tube source with silver anode (50 kV, 40 μA) together with silver and iron filters have been employed to give results in just 3 min. Specht et al.⁷⁴ investigated the variability of bone lead across five skeletal sites: mid-tibial shaft, proximal tibia, distal tibia (ankle), ilium and cranium. The average bone lead concentrations (21.6 ± 21.3 μg g⁻¹) did not differ significantly across these sites. The LOD was 2.1 ± 0.5 μg g⁻¹. Although the portable XRF spectrometry device could be used to assess cumulative lead exposure in all the bone types, selective measurement of the tibia would facilitate comparisons across studies and individuals.

A remarkable study was conducted by Tambuzzi et al.⁷⁵ who emphasised the strength of portable XRF spectrometry for forensic investigations. They systematically evaluated traces of metals within marks on skin created during death by electrocution. Railway tracks contain approximately 98% of iron and, therefore, Fe was used as the marker. The ratio of the FeKα peak to the Rayleigh scattering peak of Rh was employed to normalise the different Fe measurements. Encouragingly, XRF spectrometry proved effective not only for analysing skin fragments directly but also for assessing graphite adhesive tapes used to sample skin of living individuals. These findings demonstrated that XRF spectrometry also had potential for diagnosing injuries in survivors of electric shocks.

A novel method developed by Zhang et al.⁷⁶ combined SPE with WDXRF spectrometry for simultaneous determination of 15 REEs in water samples. A phosphonic-acid-functionalised metal–organic framework (MOF) called BPG-MIL-53(Al) adsorbed REEs within 5 min under neutral conditions. After filtration, the MOF with adsorbed REEs formed a thin film on a filter membrane which was analysed directly by XRF spectrometry. Remarkably, the XRF spectrometry intensities for the retained MOF disc remained stable for at least six months. When the MOF sorbent was used for preconcentration, ng mL⁻¹ levels of trace REEs could be quantified with impressive LODs of 0.4 to 4.7 ng mL⁻¹. Interference from the sample matrix was negligible. The method was successfully applied for on-site collection and accurate analysis of REEs in real water samples. Recoveries of spikes (20 ng mL⁻¹) were 80–125% for most REEs.

He et al.⁷⁷ optimised parameters for a portable device that incorporated both XRD and XRF spectrometry techniques. Utilisation of a single X-ray source and detector for both techniques made it possible to acquire both 2D XRD and XRF spectrometry information from a sample simultaneously. This provided high analysis efficiency and consistent measurement points. Details discussed included selection of the X-ray source and the CCD camera detector. Use of two slit designs (bottom-hole and side-hole) was investigated and the slit dimensions were optimised using rectangular holes instead of inclined trapezoidal holes. Theoretical data, derived from a formula to calculate fluorescence X-rays from filters in the X-ray tube spectra, matched well with the experimental data using 0.6 or 2.0 mm Al filters and 10 or 25 μm Ni filters at tube voltages of 10.3 kV and 15.3 kV. Furthermore, a comprehensive analysis platform for the combined XRD/XRF spectrometry system was based on the effects of tube voltage, tube current, exposure time and pixel-binning on single-pixel events of the CCD camera for various sample materials (Al, α-Fe, Zn and ZrO₂(Y₂O₃)).

A hand-held XRF spectrometer was used⁹¹ to analyse 26 copper-based alloy standards using three calibration methods. A customised calibration and PyMCA calibrations yielded more accurate data for both high- and low-Z elements (Ag, As, Bi, Cd, Co, Fe, Mn, Ni, Pb, Sb, Sn and Zn) than the built-in calibration software for “copper alloy”. Notably, the R² values for Fe improved from 0.860 to 0.982 and up to 0.996, respectively, for the two calibrations. The researchers emphasised that, to obtain reliable XRF spectrometry data, RMs should be chosen wisely, measurements should be precise and a QA calibration should be adopted.

5.2 Online XRF spectrometry techniques

Continuous metal monitors are widely used for environmental monitoring and aerosol source appointment. Zhu et al.⁷⁸ developed a reliable multi-point calibration approach using the Primary Standard Aerosol Mass Concentration Calibration System (PAMAS) for the Xact625i Ambient Metals Monitor. Whereas use of the conventional single-point calibration method based on thin-film standards introduced significant bias, especially in the low-concentration range, use of PAMAS to generate aerosols with known concentrations of 20 metals in two concentration ranges (150–1200 and 2.5–30 ng m⁻³) resulted in improved calibration. The slopes of measured versus calibrated concentrations were 1.00 ± 0.03 for all elements (except Al) present at high concentrations as well as for 15 elements present at low-concentrations. The calibrations for the other five elements had significantly steeper slopes. Calibration of the monitor enhanced accuracy thereby giving metal concentrations in agreement with actual ambient levels. This study emphasised the significance of reliable calibration methods for online aerosol monitoring.

An online XRF spectrometry setup was designed⁷⁹ and fine-tuned to analyse the P₂O₅ content for the quality control of the production of phosphate slurry. The setup was tested using two configurations: a low and vertical flow (1.3 L min⁻¹) or a high and horizontal flow (316 L min⁻¹). Reference samples were analysed using both setups to construct calibration curves over P₂O₅ concentration ranges of 13.5 to 18.5% for the horizontal-flow configuration and 14.0 to 15.6% for the vertical-flow setup. The measurement parameters were fine-tuned using the horizontal-flow setup. A good S/N for the PKα spectral line was achieved using an excitation energy of 20 or 25 kV, an excitation current of 600 μA, a sample-detector distance of 18 mm, a measurement time of 60 s per spectrum and an aluminium filter placed between the X-ray tube and the measurement window. The mean absolute errors were 0.38 and 0.87% for the low- and high-flow configurations, respectively. Despite challenges related to the low characteristic energy of P, the complex phosphate slurry matrix and the specific online analysis mode, the study’s outcomes highlighted that XRF spectrometry holds promise for the digitalisation of the chemical analysis of phosphate products.

The Fundamental Parameters Method (FPM) has various limitations, particularly when instrument parameters are unavailable or vary with time. For example, the constant motion of rocks on conveyor belts results in varying angles of incidence and distances. Although neural networks could be used as an alternative to the FPM, they require experimental XRF spectra of samples with known elemental composition, so can have limited use due to the expense involved. Dirks and Poole⁸⁰ proposed a neural network model that learned from limited experimental XRF spectrometry data to be used in combination with an XRF spectra simulator based on fundamental parameters. The model achieved this by learning to invert a forward model which utilised transition energies, element probabilities and parameterised distributions to approximate other fundamental and instrument parameters. Evaluation of a dataset for lithium minerals revealed that the model excelled for both low-Z elements (e.g. Al, K, Li, and Mg) and high-Z elements (e.g. Pb and Sn).

6 Cultural heritage applications

Due to its non-destructive nature, XRF spectrometry has become a valuable tool in forensic investigations. A comprehensive review by Manhas et al.⁸¹ highlighted the significant role of XRF spectrometry in forensic anthropology and archaeology. The discussion included the analysis of various materials (human skeletal and dental remains, soils, ceramics, paintings, coins) as well as species determination.

Nowadays, cultural heritage studies invariably involve the use of macroXRF spectrometry. Nevertheless, macroXRF spectrometry scanning faces limitations in the investigation of 3D objects due to its reliance on consistent measurement geometry for artefact-free data. To address this problem, Alfeld et al.⁸² developed an impressive configuration in which cable-driven robots were used. These cable-driven controls, shown to be suitable by mounting a microscope on the robot platform, offered speed, stability, a spacious working area, flexibility and lightweight design. However, challenges remained. Precise control of the platform necessitated a precise kinematic model which included cable elasticity. In addition, absolute positioning required alternative control methods to reconstruct surface distributions from many point measurements. Such a control system was considered also to be suitable for platform positioning.

Used for the first time in the investigation of artworks, an impressive multimodal scanning system developed by Moreau et al.⁸³ simultaneously combined macroXRF spectrometry, RIS and LIS. This innovative approach generated spatially aligned hyperspectral datasets. The instrument’s specifications and data processing pipeline were described in detail. Its operation for mapping the XRF spectrometry data onto RIS and LIS data in an extended wavelength range (400–2500 and 200–1000 nm, respectively) was demonstrated on an anonymous test painting. Study of a restored area illustrated the complementarity of these techniques for the visualisation and characterisation of pigments, varnish and binders. Occhipinti et al.⁸⁴ combined macroXRF spectrometry and hyperspectral reflectance scanning in the VNIR and SWIR ranges (400–2500 nm) in another innovative integrated portable scanner. This system enabled in situ, rapid and non-invasive sample scanning while maintaining high spectral-resolution and sample throughput. This set-up allowed enhanced or complementary information to be extracted from the same analysis spot with minimal data processing effort and no need for spatial alignment. An evaluation of the qualitative and quantitative performance of the system included in-lab analyses on reference samples and insights from a real case study.

A study by Albrecht et al.⁸⁵ of 19 still-life paintings from the Mauritshuis in The Hague, Netherlands, emphasised the importance of the macro-XRF spectrometry scanner for understanding artists’ compositional techniques. They employed a macroXRF spectrometry scanner equipped with a 30 W Rh-target μ-focus X-ray tube (50 kV, 600 μA), polycapillary lens and a 60 mm² X-Flash SDD. A working distance of ca. 1 cm (spot size of ca. 250 μm) was maintained for all paintings. The chosen step size (250–500 μm) and dwell time (70–80 ms) resulted in a total scan time of around 24 h. Smaller paintings used finer step sizes for higher resolution. Whereas 17^th-century still life painters like Balthasar van der Ast followed detailed underdrawings, Antwerp artists such as Daniel Seghers and Jan Davidsz de Heem began with “abstract” compositions of coloured circles and ovals.

The Laboratory of Molecular and Structural Archaeology at Sorbonne Université/CNRS in Paris developed⁸⁶ a novel mobile instrument called MobiDiff which combined XRD and XRF spectrometry measurements for analysis at the same point in reflection geometry. MobiDiff’s improved degrees of freedom and flexibility enabled measurements to be made on complex-shaped objects. It employed a Pd anode non-monochromatic X-ray source with an SDD for XRF spectrometry measurements and a CuKα monochromatized source with a 1D electronic detector and a 2D imaging plate system for XRD analysis. Technical specifications, design details and application to different types of cultural heritage artefacts including statues were presented. Multilayered decorations could be detected and associated phases and significant variations in stratigraphy between statues identified.

A mobile macroXRF imaging spectroscopy system was updated⁸⁷ with new mechatronics and a 3D multi-detector array consisting of 6 SDDs operating in parallel and arranged in a compact hodoscope geometry to increase significantly the throughput of fluorescence radiation while at the same time minimising the dead time. This resulted in a substantial improvement of the overall sensitivity. Use of this system revealed the layout of ancient Greek papyri from Herculaneum. For the first time, in situ mapping detected very small traces of metallic elements in this type of sample, previously only possible by using advanced techniques available in large facilities. Elemental distribution maps of Pb revealed three different systems of textual layout in ancient papyrus rolls and resolved the dispute around Maas’ Law by demonstrating that slanted text columns were a deliberate aesthetic choice of scribes. By assuming a typical pixel size of 250 μm and taking into account the beamspot diameter of 50 μm, the absolute mass of Pb observed pixel by pixel was determined to be ca. 8 ng per pixel along the ruling-lines (upper limit) and ca. 1 ng per pixel in areas outside the Pb ruling-lines.

Previous studies at the UK synchrotron facility (DLS) on two early 14^th-century paintings by Pietro Lorenzetti revealed a complex orpiment-tinted mordant used to adhere separate layers of silver and gold leaf. In a new study, synchrotron-based μXRD, μXRF spectroscopy and μXANES measurements were undertaken.⁸⁸ The μXRF spectrometry maps, created on cross-sections of the discoloured mordant, revealed the migration of mobile As–Ag–S species. Additionally, μXANES maps of Ag, As and S provided insights into oxidation states. The presence of arsenolite (As₂O₃) throughout the mordant suggested light-induced photooxidation of orpiment. Detection of As³⁺ and As⁵⁺ confirmed orpiment degradation. Altered phases in the mordant layer, identified by μXRD analysis, included acanthite (Ag₂S), xanthoconite (Ag₃AsS₃) and arsenolite (As₂O₃). The darkening of the mordant probably resulted from finely dispersed, grey-colored Ag-bearing acanthite particles (Ag₂S).

Portable XRF spectrometry was demonstrated to be a reliable method for the detection and quantification of elements in biological samples. A hand-held XRF spectrometry scanner was used by Martino et al.⁸⁹ as a rapid in situ approach to identify swiftly the origin of black tiger prawns (Penaeus monodon) across Australia. The portable instrument, equipped with a 50 kV X-ray tube and Rh anode, had previously been calibrated for prawn muscle using accelerator-based ion beam analysis. Operating in Geochem mode with 3 beams, each scanning for 60 s, various factors were examined: provenance (site, jurisdiction, and production method), preparation (cooked vs. raw) and biological traits (carapace length, body weight, sex). The primary variable was the site of harvesting so this approach could be used for establishing the provenance of samples. Raw samples yielded clearer elemental fingerprints than cooked samples. Machine-learning models achieved high accuracy: 87.5% for site, 98% for jurisdiction and 100% for production method. Malo et al.⁹⁰ achieved over 80% accuracy in determining the provenance of seafood using machine-learning models. They emphasised the significance of appropriate calibration factors to take into account variations in wetness and cooking style.

Abbreviations

1D	one dimensional
2D	two dimensional
3D	three dimensional
AES	atomic emission spectrometry
AFM	atomic force microscopy
APM	ammonia hydrogen peroxide mixture
ASU	atomic spectrometry update
BSE	back-scattered electron
CCD	charge coupled device
CNRS	Centre National de la Recherche Scientifique (France)
CRM	certified reference material
CT	computed tomography
CUDA	Compute Unified Device Architecture
DCC	double curved crystal
DLS	diamond light source
EDXRF	energy dispersive X-ray fluorescence
EDTA	ethylenediaminetetraacetic acid
EM	electron microscopy
ESRF	European Synchrotron Radiation Facility
FPM	fundamental parameters method
FWHM	full width at half maximum
FXI	full-field XRF spectrometry imaging
GEXRF	grazing exit X-ray fluorescence
GIXRF	grazing incidence X-ray fluorescence
ICP	inductively coupled plasma
IR	infrared
LIS	luminescence imaging spectroscopy
LLME	liquid–liquid microextraction
LOD	limit of detection
LOQ	limit of quantification
MCA	multi-channel-analyser
MEZ	multielement Z-tag
MOF	metal–organic framework
MS	mass spectrometry
MS/MS	tandem mass spectrometry
μSPE	microscale solid phase extraction
μXANES	microscale X-ray absorption near edge structure
μXAS	microscale X-ray absorption spectroscopy
μXRD	microscale X-ray diffraction
μXRF	microscale X-ray fluorescence
PAMAS	primary standard aerosol mass concentration calibration system
PB	Prussian blue
PLS	partial least squares
PM	particulate matter
PM2.5	particulate matter with a diameter of <2.5 μm
ppb	parts per billion
ppm	parts per million
PTFE	poly(tetrafluoroethylene)
PyMCA	Python microscopy and X-ray computed microtomography analysis
QA	quality assurance
QC	quality control
RBS	Rutherford backscattering spectrometry
REE	rare earth element
RIS	reflectance imaging spectroscopy
RM	reference material
RSD	relative standard deviation
SDD	silicon drift detector
SEM	scanning electron microscopy
SIMS	secondary ion mass spectrometry
S/N	signal-to-noise ratio
SPE	solid phase extraction
SR	synchrotron radiation
SWIR	shortwave infrared
TXRF	total reflection X-ray fluorescence
TYMV	Turnip yellow mosaic virus
VNIR	visible and near infrared
WDXRF	wavelength dispersive X-ray fluorescence
XANES	X-ray absorption near edge structure
XAS	X-ray absorption spectroscopy
XEOL	X-ray excited optical luminescence
XRD	X-ray diffraction
XRF	X-ray fluorescence
XRR	X-ray reflectometry
XSW	X-ray standing waves
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., 2023, 38(9), 1730–1743.
2. J. R. Bacon, O. W. Butler, W. R. L. Cairns, O. Cavoura, J. M. Cook, C. M. Davidson and R. Mertz-Kraus, J. Anal. At. Spectrom., 2024, 39(1), 11–65.
3. M. Patriarca, N. Barlow, A. Cross, S. Hill, A. Robson and J. Tyson, J. Anal. At. Spectrom., 2024, 39(3), 624–698.
4. E. H. Evans, J. Pisonero, C. M. M. Smith and R. N. Taylor, J. Anal. At. Spectrom., 2023, 38(5), 974–999.
5. R. Clough, C. F. Harrington, S. J. Hill, Y. Madrid and J. F. Tyson, J. Anal. At. Spectrom., 2023, 38(7), 1339–1371.
6. R. Clough, A. Fisher, B. Gibson and B. Russell, J. Anal. At. Spectrom., 2023, 38(11), 2215–2279.
7. A. Lanzirotti, S. R. Sutton, M. Newville, A. Brearley and O. Tschauner, Meteorit. Planet. Sci., 2024, 59(2), 395–418.
8. A. Graefenstein, D. Brueckner, C. Rumancev, J. Garrevoet, V. Galbierz, W. H. Schroeder, C. G. Schroer, G. Falkenberg and A. Rosenhahn, Anal. Chem., 2023, 95(27), 10186–10195.
9. F. Morina, A. Mijovilovich, A. Mishra, D. Bruckner, B. Vujic, S. N. H. Bokhari, J. Spak, G. Falkenberg and H. Kupper, Plant Sci., 2023, 336, 111864.
10. A. Mijovilovich, P. Cloetens, A. Lanzirotti, M. Newville, G. Wellenreuther, P. Kumari, C. Katsaros, C. J. Carrano, H. Kupper and F. C. Kupper, Metallomics, 2023, 15(10), mfad058,  DOI:10.1093/mtomcs/mfad058.
11. A. van der Ent, M. Salinitro, D. Brueckner, K. M. Spiers, S. Montanari, A. Tassoni and M. Schiavon, Ann. Bot., 2023, 132(2), 349–361.
12. J. M. Walker, H. J. M. Greene, Y. Moazzam, P. D. Quinn, J. E. Parker and G. Langer, Environ. Sci.: Processes Impacts, 2024 10.1039/D3EM00509G.
13. F. T. H. Broers, I. Verslype, K. W. Bossers, F. Vanmeert, V. Gonzalez, J. Garrevoet, A. van Loon, E. van Duijn, A. Krekeler, N. De Keyser, I. Steeman, P. Noble, K. Janssens, F. Meirer and K. Keune, Sci. Adv., 2023, 9(50) DOI:10.1126/sciadv.adj9394.
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. K. Heimler, C. Gottschalk and C. Vogt, Anal. Bioanal. Chem., 2023, 415(21), 5083–5100.
16. A. P. Morrell, R. A. Martin, H. M. Roberts, H. Castillo-Michel, J. F. W. Mosselmans, K. Geraki, A. T. Warfield, P. Lingor, W. Qayyum, D. Graf, M. Febbraio and O. Addison, Metallomics, 2023, 15(6), mfad030,  DOI:10.1093/mtomcs/mfad030.
17. S. G. Young, J. Valdez, M. Espy, A. Edgar, J. Brett, M. T. Pettes, C. Mathers, M. Barbour and B. M. Patterson, X-Ray Spectrom., 2024, 53(1), 2–15.
18. K. Mori, T. Hourai, T. Matsuyama, S. Zhuo and K. Tsuji, Anal. Sci., 2024, 40(3), 367–373.
19. T. Matsuyama, S. Sonoda, T. Fuchita, S. Sakashita and K. Tsuji, X-Ray Spectrom., 2023, 52(6), 457–462.
20. A. Kubiak, A. Voronkina, M. Pajewska-Szmyt, M. Kotula, B. Lesniewski, A. Ereskovsky, K. Heimler, A. Rogoll, C. Vogt, P. Rahimi, S. Falahi, R. Galli, E. Langer, M. Foerste, A. Charitos, Y. Joseph, H. Ehrlich and T. Jesionowski, Biomimetics, 2023, 8(7), 533.
21. Y. Zhang, N. Liu, P. Liu, Y. Liu and Y. Lei, Arabian J. Chem., 2023, 16(8), 104966.
22. M. Iro, D. Ingerle, S. Hampel, U. Fittschen, V. Dhamgaye, O. Fox and C. Streli, SoftwareX, 2023, 23, 101481.
23. J. Kranz, N. Liebing and A. Bjeoumikhov, X-Ray Spectrom., 2024, 53(1), 79–86.
24. Z. Y. Dong, L. A. Shi, G. Niu, W. B. Wang, J. B. Liu and L. Han, X-Ray Spectrom., 2024, 53, 242.
25. X. Y. Wang, Y. M. Wu and X. Y. Lin, Optik, 2023, 282, 170838.
26. P. Wobrauschek, D. Ingerle, J. Prost, S. Dhara, N. L. Mishra, M. Iro and C. Streli, X-Ray Spectrom., 2024, 53(3), 196–203.
27. T. Yuan, L. Zhang, L. Hua, H. Li, Y. Zhong, B. Liao, C. Li, T. Sun and X. Sun, Nucl. Instrum. Methods Phys. Res., Sect. A, 2024, 1058, 168923.
28. C. Lv, W.-Q. Li, C.-S. Tang, C. Zhu, X.-H. Pan, X. Zhang and B. Shi, Acta Geotech., 2024, 19(1), 115–129.
29. Z. Yang, L. Yang, L. Zhao, D. Huang, J. Li, X. Chen, X. Shen and H. Wang, Talanta, 2024, 269, 125407.
30. S. Wang, Z. Yang, X. Ma, Z. Zhou, Z. Ma, J. Chen and X. Ren, TrAC, Trends Anal. Chem., 2024, 170, 117453.
31. S. Roudeau, A. Carmona and R. Ortega, Curr. Opin. Chem. Biol., 2023, 76, 102372.
32. S. Raneri, L. Fornasini, D. Bersani, S. Mirata and A. Gianoncelli, Period. Mineral., 2023, 92(2), 179–190.
33. A. K. James, B. F. Popescu, M. Weng, G. J. Myers, J. L. O’Donoghue, G. E. Watson, I. J. Pickering and G. N. George, Neurotoxicol, 2023, 99, 129–138.
34. M. A. Denecke, J. Radioanal. Nucl. Chem. Lett., 2024 DOI:10.1007/s10967-10023-09311-10967.
35. E. De Pauw, P. Tack and L. Vincze, J. Anal. At. Spectrom., 2024, 39(2), 310–329.
36. A. G. Buzanich, M. Radtke, K. V. Yusenko, T. M. Stawski, A. Kulow, C. T. Cakir, B. Roder, C. Naese, R. Britzke, M. Sintschuk and F. Emmerling, J. Chem. Phys., 2023, 158(24) DOI:10.1063/1065.0157194.
37. W. Klysubun, P. Tarawarakarn, N. Thamsanong, S. Mahakhod, P. Amonpattaratkit and C. Cholsuk, J. Synchrotron Radiat., 2023, 30, 655–661.
38. M. Nagasawa, O. Sekizawa, K. Nitta, T. Kashiwabara and Y. Takahashi, Minerals, 2023, 13(6), 746.
39. M. Strotton, T. Hosogane, M. di Michiel, H. Moch, Z. Varga and B. Bodenmiller, Nat. Methods, 2023, 20(9), 1310–1322.
40. J. Li, S. Chen, D. Ratner, T. Blu, P. Pianetta and Y. Liu, Proc. Natl. Acad. Sci. U.S.A., 2023, 120(49) DOI:10.1073/pnas.2314542120.
41. A. Hafner, L. Costa, G. Kourousias, V. Bonanni, M. Zizic, A. Stolfa, B. Bazi, L. Vincze and A. Gianoncelli, Analyst, 2024, 149(3), 700–706.
42. C. T. Plass, V. Bonino, M. Ritzer, L. R. Jager, V. Rey-Bakaikoa, M. Hafermann, J. Segura-Ruiz, G. Martinez-Criado and C. Ronning, Adv. Sci., 2023, 10(1) DOI:10.1002/advs.202205304.
43. S. Hammarberg, L. A. B. Marcal, N. Lamers, Z. Zhang, H. Chen, A. Bjorling and J. Wallentin, ACS Appl. Nano Mater., 2023, 6(19), 17698–17705.
44. S. Hampel, F. Sand, D. A. M. Gonzalez, G. Pepponi, G. Helsch, J. Deubener, T. Schirmer, A. Fittschen and U. E. A. Fittschen, Spectrochim. Acta, Part B, 2023, 206, 106716.
45. S. Hampel, F. Sand, H. S. Till and U. E. A. Fittschen, J. Anal. At. Spectrom., 2024, 39(1), 131–140.
46. T. Matsuyama, Y. Tanaka, Y. Mori and K. Tsuji, Talanta, 2023, 265, 124808.
47. T. Matsuyama, Y. Tanaka, N. Taniguchi, J.-S. Oh and K. Tsuji, J. Anal. At. Spectrom., 2024, 39(1), 76–85.
48. E. Margui and L. Torrent, Spectrochim. Acta, Part B, 2023, 204, 106682,  DOI:10.1016/j.sab.2023.106682.
49. M. Musielak, M. Serda, A. Gagor, E. Talik and R. Sitko, Spectrochim. Acta, Part B, 2024, 213, 106863.
50. M. Musielak, E. Musielak and R. Sitko, Food Chem., 2024, 439, 138156.
51. L. Deng, R. Wu, L. Zhang, R. Zhu, J. Zou, C. Gu, Y. Wang, C. Tong and R. Zhu, ACS Appl. Nano Mater., 2024, 7(6), 6409–6417.
52. K. Sanyal, A. Saha, A. Sarkar, S. B. Deb, R. V. Pai and M. K. Saxena, ACS Omega, 2023, 8(44), 41402–41410.
53. T. F. Akhmetzhanov, T. Y. Cherkashina, A. N. Zhilicheva, V. M. Chubarov and G. V. Pashkova, J. Anal. At. Spectrom., 2023, 38(12), 2664–2673.
54. G. V. Pashkova, A. A. Nikonova, S. D. Dylgerova, E. V. Chuparina, A. S. Maltsev, A. N. Zhilicheva, O. Y. Belozerova, L. P. Paradina, O. Y. Glyzina and I. V. Khanaev, X-Ray Spectrom., 2023 DOI:10.1002/xrs.3396.
55. Y. Zhang, Y. He, H. Chen, S. Wei, G. Mo, T. Xu and J. Yuan, J. Anal. At. Spectrom., 2023, 38(12), 2648–2655.
56. L. Bennun, Spectrochim. Acta, Part B, 2023, 203, 106652.
57. S. Dhara, Spectrochim. Acta, Part B, 2023, 201, 106625.
58. K. Sanyal and S. Dhara, Crit. Rev. Anal. Chem., 2024, 1–16.
59. H. Takahara, A. Morikawa, S. Kitayama, T. Matsuyama and K. Tsuji, Anal. Sci., 2024, 40(3), 519–529.
60. F. Bilo, P. Cirelli and L. Borgese, TrAC, Trends Anal. Chem., 2024, 170, 117427.
61. J. Jablan, M. Ortner Hadziabdic, A.-M. Domijan, S. Inic and H. Stosnach, X-Ray Spectrom., 2023 DOI:10.1002/xrs.3415.
62. K. Kawon, M. Rugiel, Z. Setkowicz, K. Matusiak, A. Kubala-Kukus, I. Stabrawa, K. Szary, Z. Rauk and J. Chwiej, Sci. Rep., 2023, 13(1) DOI:10.1038/s41598-41023-45611-41594.
63. A. Wilk, Z. Setkowicz, D. Banas, R. Fernandez-Ruiz, E. Margui, K. Matusiak, P. Wrobel, J. Wudarczyk-Mocko, N. Janik-Olchawa and J. Chwiej, Sci. Rep., 2024, 14(1) DOI:10.1038/s41598-41024-51731-41592.
64. K. Olbrich, A. Kubala-Kukus, E. Margui, R. Fernandez-Ruiz, k. Matusiak, J. Wudarczyk-Mocko, P. Wrobel, Z. Setkowicz and J. Chwiej, Spectrochim. Acta, Part B, 2023, 205, 106695.
65. A. Kubala-Kukus, D. Banas, M. Pajek, J. Braziewicz, S. Gozdz, J. Chwiej, S. Gluszek, P. Jagodzinski, E. Margui, M. Pierzak-Stepien, A. Pniak, Z. Setkowicz, I. Stabrawa, R. Stachura, K. Szary, G. Wesolowski, A. Wojsa and J. Wudarczyk-Mocko, Spectrochim. Acta, Part B, 2024, 213, 106865.
66. M. A. Alam, M. K. Tiwari, D. Devi, S. Tripathi, A. Trivedi, S. Ojha, R. Singh and M. Gupta, Spectrochim. Acta, Part B, 2023, 206, 106707.
67. P. Jonnard, K. Le Guen, R. Delaunay, Y. Menesguen, M.-C. Lepy, E. Briand, D. Schmaus and I. Vickridge, X-Ray Spectrom., 2023, 52(6), 437–446.
68. K. V. Nikolaev, A. I. Safonov, O. A. Kondratev, G. V. Prutskov, I. A. Likhachev, I. A. Subbotin, M. M. Borisov, S. A. Tikhomirov, E. M. Pashaev and S. N. Yakunin, J. Appl. Crystallogr., 2023, 56, 1435–1445.
69. S. Melhem, Y. Menesguen, E. Nolot and M.-C. Lepy, X-Ray Spectrom., 2023, 52(6), 412–422.
70. D. Kobylarz, A. Michalska and K. Jurowski, TrAC, Trends Anal. Chem., 2023, 169, 117355.
71. A. Frydrych and K. Jurowski, TrAC, Trends Anal. Chem., 2023, 166, 117165.
72. T. R. Tavares, B. Minasny, A. McBratney, M. R. Cherubin, G. T. Marques, M. M. Ragagnin, E. E. N. Alves, J. Padarian, J. Lavres and H. W. P. de Carvalho, Geoderma, 2023, 439, 116701.
73. A. A. Roberts, D. Guimaraes, M. W. Tehrani, S. Lin and P. J. Parsons, X-Ray Spectrom., 2023 DOI:10.1002/xrs.3407.
74. A. J. Specht, D. W. Steadman, M. Davis, S. M. Bartell and M. G. Weisskopf, Sci. Total Environ., 2023, 880, 163197.
75. S. Tambuzzi, L. Bonizzoni, F. Di Paola, D. Mazzarelli, G. Caccia and C. Cattaneo, X-Ray Spectrom., 2024, 53(2), 112–120.
76. M. Zhang, J. Hu, Q. Tang, J. Zhang, X. Jiang and X. Hou, Anal. Chem., 2023, 95(38), 14169–14174.
77. Z. He, Y. J. Xu, N. Huang, P. Wang, B. Peng, J. X. Yuan and Z. An, J. Instrum., 2023, 18(9), P09027.
78. Y. Zhu, C. Liu, J. Huo, H. Li, J. Chen, Y. Duan and K. Huang, Sci. Total Environ., 2024, 908, 168274.
79. I. Ben Amar, A. Thomas, C. Bachmann, A. Hafnoaui, H. Griguer, A. Miled and Y. Messaddeq, Sci. Rep., 2023, 13(1), 17925.
80. M. Dirks and D. Poole, X-Ray Spectrom., 2023, 52(3), 142–150.
81. M. Manhas, A. Tomar, M. Tiwari and S. Sharma, X-Ray Spectrom., 2024 DOI:10.1002/xrs.3421.
82. M. Alfeld, P. Tempel and V. van der Wijk, Quantum Beam Sci., 2023, 7(4), 37.
83. R. Moreau, T. Calligaro, L. Pichon, B. Moignard, S. Hermon and I. Reiche, X-Ray Spectrom., 2024, 53, 271.
84. M. Occhipinti, R. Alberti, T. Parsani, C. Dicorato, P. Tirelli, M. Gironda, A. Tocchio and T. Frizzi, X-Ray Spectrom., 2023 DOI:10.1002/xrs.3405.
85. M. Albrecht, S. Meloni, C. Pottasch, A. Vandivere and A. van Loon, X-Ray Spectrom., 2024, 53, 249.
86. V. Poline, P. Bordet, O. Leynaud, A. Prat, R. Bruyere, N. Blanc, F. Lelong and P. Martinetto, Eur. Phys. J. Plus, 2023, 138(3) DOI:10.1140/epjp/s13360-13023-03821-13369.
87. F. P. Romano, E. Puglia, C. Caliri, D. P. Pavone, M. Alessandrelli, A. Busacca, C. G. Fatuzzo, K. J. Fleischer, C. Pernigotti, Z. Preisler, C. Vassallo, G. Verhasselt, C. Miliani and G. Ranocchia, Sci. Rep., 2023, 13(1) DOI:10.1038/s41598-41023-33242-41598.
88. H. Howard, J. Najorka, P. F. Schofield and K. Geraki, Stud. Conserv., 2024, 69(3), 193–208.
89. J. C. Martino, J. Crawford, P. Gadd, J. Sammut, N. Saintilan and D. Mazumder, Food Control, 2023, 151, 109813.
90. N. Malo, D. Mazumder, J. Crawford, P. Gadd, K. Gopi, N. Saintilan and J. Sammut, Foods, 2023, 12(15), 2874.
91. E. Konstantakopoulou, A. C. Municchia, L. Luvidi and M. Ferretti, Condens. Matter, 2024, 9(1), 5,  DOI:10.3390/condmat9010005.

---

Footnote

† Joint review coordinators.

---