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

Christine Vanhoof a, Jeffrey R. Bacon b, Ursula E. A. Fittschen c and Laszlo Vincze d
aFlemish Institute for Technological Research (VITO), Boeretang 200, B-2400 Mol, Belgium. E-mail:
b59 Arnhall Drive, Westhill, Aberdeenshire, AB32 6TZ, UK. E-mail:
cClausthal University of Technology, Institute of Inorganic and Analytical Chemistry, Arnold-Sommerfeld-Strasse 4, D-38678 Clausthal-Zellerfeld, Germany
dDepartment of Chemistry, Ghent University, Krijgslaan 281 S12, B-9000 Ghent, Belgium

Received 29th June 2021

First published on 15th July 2021


That the boundaries of analysis using XRF spectrometry performed at SR sources continues to be pushed back is evidenced by the nanoscale spatial resolution with near single-atom sensitivity for the most efficiently detected elements in nanobeam applications and high speed 2D/3D imaging capabilities. The 4th generation SR facilities currently under development, such as the ESRF Extremely Brilliant Source (EBS), will be at the forefront of these advances. A clear trend is that SR-XRF spectrometry is increasingly applied with complementary X-ray spectroscopic/imaging techniques to combine spatially resolved elemental information with speciation and structural/morphological imaging. The wide range of applications of scanning (sub)microXRF spectrometry for elemental imaging published in the period covered by this review included biomedical, environmental, materials science and cultural heritage studies. Most applications involved XAS and XRD methods at hard X-ray micro- and nano-probe facilities. Full field microXRF spectrometry has made very promising advances with respect to the optics used. Coded apertures have potential for overcoming the low count rates that often restrict the full potential of laboratory-based full-field setups. The availability of commercial full-field detectors will increase the user community and thereby foster advancement in full field microXRF spectrometry. The TXRF spectrometry of ambient air is becoming more and more sophisticated and the advantages of this micro analytical tool over ICP-MS in terms of short sampling times with high particle size resolution are becoming ever more apparent. For example, the optimal sampling time for aerosols for TXRF analysis was well below 12 hours, whereas that for ICP-MS analysis was about 24 hours. High-quality grazing incidence XRF analysis in the laboratory has become more feasible with the development of prototype TXRF instrumentation and the availability of commercial XRD setups with energy dispersive detectors. Portable XRF spectrometry has undergone significant technological improvements in recent years and is now applied in a wide range of applications. This is reflected in a significant number of valuable review papers dealing with different aspects of the portable XRF technique. The growth in the use of macroXRF scanning systems in cultural heritage investigations has required development of new software and methodologies for efficient handling of the huge data files generated. In several contributions the possibilities of a new scanning station equipped with real time macroXRF spectrometry was demonstrated.

1. Introduction

This review describes advances in the XRF spectrometry group of techniques published approximately between April 2020 and March 2021. 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 ASU reviews in the series, namely: environmental analysis;1 clinical and biological materials, foods and beverages;2 advances in atomic spectrometry and related techniques;3 elemental speciation;4 and metals, chemicals and functional materials.5

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; the same abbreviation is used both for singular and plural forms. It is a convention of ASU that information given in the paper being reported on is presented in the past tense whereas the views of the ASU reviewers are presented in the present tense.

2. Chemical imaging using X-ray techniques

2.1 Computed tomography and 3D XRF techniques

2.1.1 X-ray fluorescence computed tomography (XRF-CT). Although the application of XRF-CT using laboratory microfocus sources is still a novelty and relatively rare, the technique is now routinely applied at SR micro- and nano-probe facilities not only for 2D cross-sectional mapping of (micro)samples but also for full 3D (quantitative) elemental imaging. The 3rd and upcoming 4th generation sources facilitate these applications at sub-μm resolution as used regularly in combination with complementary imaging modalities, such as absorption and phase-contrast CT, XRD-CT, 2D and 3D XAS and ptychography.

A review on such multimodal X-ray nanotomography presented by Gürsoy and Jacobsen6 described material science and bioimaging applications and discussed future opportunities and challenges of 3D elemental and structural nanoimaging. The diffraction-limited SR storage rings becoming available offered coherent flux gains of a hundredfold or more and so were identified as the major driving force of nanoscale 3D analysis. Challenges needing to be addressed were those associated with the relatively long measuring times, the sample positioning and rotational stage accuracy and the accumulated dose and radiation effects on the increasingly smaller specimens.

An impressive demonstration of combined cryogenic X-ray holography and X-ray fluorescence nanotomography was given by Gramaccioni et al.7 using the ESRF ID16A beamline to reconstruct the 3D distributions of mass density and molar concentration of major and minor elements in frozen-hydrated macrophage cells. In-line holograms were collected with a pixel size of 40 nm and a field of view of 80 μm. The XRF nanotomograms corresponded to a relatively course resolution (120–130 nm) and limited number of observation angles (30) collected using the fly-scan approach. Apart from the nanoscale spatial resolution achieved, an especially interesting aspect of this study was the quantitative nature of the 3D elemental distributions of Ca (0–0.4 M), Cl (0–0.1 M), Fe (0–2 mM), K (0–0.2 M), Mn (0–0.8 mM), P (0–0.3 M), S (0–0.12 M) and Zn (0–1.0 mM).

In the field of cellular bioimaging, a proof-of-concept for 2D and 3D elemental analyses was presented by De Samber et al.8 using a novel cryoloop SR-XRF setup at the PETRA III P06 beamline for the analysis of cryofrozen islets of Langerhans cells which play a key role in glucose homeostasis. Step sizes of 500 nm were used with a scanning angle range of 0–180° and a dwell time of 50 ms per pixel for XRF tomography. When a self-supporting pellet of NIST SRM 1577C (bovine liver) was scanned with a dwell time of 100 ms per sampling point, typical LOD were in the low to sub-ppm range.

The elegant use of XRF-CT and XRD-CT in combination achieved a spatial resolution better than 120 nm as demonstrated by Palle et al.9 in an experiment optimised for studies on nanostructured hierarchical materials such as biomineralised bone. The XRF/XRD-CT measurements were conducted at the ESRF ID13 beamline using an X-ray beam focused by Multilayer Laue Lenses to 32 × 31 nm2 (FWHM) at an X-ray energy of 12.7 keV and intensity of 1.25 × 1010 photons per s. The tomographic data sets were collected with an exposure time of 100 ms per pixel and over an angular range of 180°. The XRD and XRF projection images were recorded simultaneously at each angular position.

Vaughan et al.10 described the newly refurbished ESRF ID15A beamline which was optimised for operando and time-resolved experiments using diffraction and total scattering in imaging modes including CT. Although the beamline was developed mainly for XRD-based imaging experiments, rapid alternation between the different techniques is possible and the combination with XRF-CT is easily achievable. The XRF detector could be used to measure fluorescence lines with energies up to 80 keV, thereby including the K-lines of the lanthanides, and to detect simultaneously relatively low-energy lines excited with high-energy beams, thereby giving access to the K-lines of rows 4 to 7 of the periodic table. A unique feature of this facility was that it took full advantage of the ESRF EBS upgrade by combining high energy (up to 120 keV) with sub-μm-focusing capability (300 nm) and a high photon flux of 1011–1012 photons per s.

2.1.2 Confocal XRF spectrometry. An interesting new approach presented by Nakano et al.11 was based on a confocal line XRF excitation and detection system that provided elemental information from a larger area and at a higher intensity than possible using conventional confocal (point-based) XRF analysis. The primary X-rays were focused as a line instead of a point by passing them through Soller slits. The fluorescent X-rays emitted from the irradiated region were recorded by an ED detector equipped with Soller slits. The spatial resolutions were 107 (lateral) and 62 (depth) μm (FWHM). Although these values were larger than those typically achieved by confocal-XRF spectrometry, the XRF intensities were ca. 33 times higher than those achievable by 3D-XRF laboratory systems using polycapillary-based point focusing for both the excitation and detection channels. This new setup with line focusing was considered to be particularly effective for the analysis of wide-area samples with layered structures.

The polychromatic excitation typically used for confocal μXRF measurements in the laboratory results in poor peak-to-background ratios due to scattered Bremsstrahlung and, in addition, makes quantification and fundamental parameter-based corrections difficult. A novel monochromatic-excitation-based confocal-XRF system described by Ingerle et al.12 addressed these problems without sacrificing spatial resolution. The source assembly consisted of a 2 kW water-cooled Mo fine-focus X-ray tube and a parallel beam-mirror which produced a quasi-parallel monochromatic beam with an energy of 17.4 keV. The confocal setup itself consisted of two polycapillary half-lenses, one for the source side and the other, used in combination with a 50 mm2 SDD, for the detector side. Both polycapillaries had a focus size of ca. 15 μm (FWHM) at the Mo-Kα line. The LOD were established for selected elements from Ca to Pb in both confocal and non-confocal detection modes using NIST SRM 621 (soda-lime container) and 1412 (multicomponent glass). Those for As were 1 and 20 μg g−1 in non-confocal and confocal modes, respectively.

The key step of sensitivity calibration of a confocal-XRF setup was addressed by Zhou et al.13 by calibrating the confocal volume for both mono- and poly-chromatic 3D μXRF setups. The theoretical model employed was verified using a microfocus X-ray source with a molybdenum target and polycapillary lenses with focal sizes of ca. 32–33 μm at 17.4 keV. A similar procedure was described by Prokeš and Trojek14 for the calibration of the confocal μXRF setup constructed at the Czech Technical University in Prague for non-destructive analyses of cultural heritage objects, in particular paintings. The procedure involved the determination of the energy and relative position-dependent sensitivity functions in the XRF energy range of 4–11 keV by using a set of thin metal foils (Co, Cu, Fe, Ni, Pb, Ti, Zn) with thicknesses of 5–7 μm. The derived Gaussian-like functions that quantified the sensitivity of the spectrometer for the elements of interest represented the response functions influencing the shape of the measured depth profiles.

The ray-tracing code for mono- and poly-capillary optics presented by Tack et al.15 could be used to predict the performance of future confocal XRF setups in terms of 3D spatial resolution and sensitivity. In addition to supporting the X-ray photon ray-tracing of ideal (straight, conical, ellipsoidal) channel profiles and external shapes for polycapillary optics, the code also allowed for the simulation of photon propagation through arbitrarily shaped devices to account for the small deviations from the ideal shape encountered in real-world examples. The simulation was able to take into account beam polarisation effects and so-called ‘leak events’ of high-energy X-rays penetrating and escaping through the capillary walls.

The 3D-printing approach used by Zaitsev et al.16 for production of confocal collimators for radionuclide diagnostics and XRF analysis was of interest because of the simplicity, reliability and wide availability of commercial 3D-printers. Use of the confocal collimator produced increased the sensitivity by 2–7 times and improved resolution ca. 9 times when compared with use of single channel 3D-printed collimators of equivalent dimensions.

2.2 Laboratory-based 2D XRF techniques

Full-field microXRF spectrometry can be a versatile tool for studying elemental changes in situ but the low transmission of common optics (pinholes and polycapillaries) can often make its detection power insufficient. Kulow et al.17 used coded apertures of 7.5% openness (ratio of the open area to the entire mask area) to achieve a remarkable improvement in the performance of a setup equipped with a colour X-ray camera. Count rates were up to a factor of 9.5 higher than those achievable with polycapillary optics. All the algorithms applied were suitable for reconstructing the elemental distributions of the samples. Xiong et al.18 developed a setup operable from 12 to 30 keV in which a key component was a commercially available optical CCD camera operated at −80 °C and with a 1 MHz read-out. The spectral resolution obtained (FWHM: 275 eV at 14.9 keV (Y Kα)) was considered acceptable. The spatial resolution (effective pixel diameter) was 135 and 350 μm using pinholes of 100 and 300 μm, respectively. An impressive factor-of-4 reduction of background signal and a 10% increase in analytical signal was achieved by An et al.19 by using a polycapillary optic instead of a pinhole. The detector consisted of an electronic component capable of reading individual pixels and bonded to a 256 × 256 array silicon chip. The major drawback of this setup was the poor energy resolution with a FWHM of 1.12 keV at 8.04 keV (Cu Kα).

A time-efficient procedure for identifying the polycapillary position of the focal spot was achieved20 by using a CMOS camera to study directly the shape of the output beam of a polycapillary in a low-power (2.4 W) setup. A knife-edge scan was still necessary to determine the focal spot size (ca. 53 μm). Hampai et al.21 used a CCD camera and a scintillator counter in their time-efficient procedure for selecting the optimal optics for their setup. They characterised the beam shape and transmission of several polycapillary optics. Sun et al.22 improved the transmission 5-fold and the focal spot size (ca. 9.3 μm Fe Kα) significantly by combining two optics. The first, a polycapillary optic, focused the beam from a rotating-anode tube onto the second optic, a tapered monocapillary optic, which focused the beam more sharply. The monocapillary optic was custom-made from a nested-glass tube.

3. Synchrotron and large-scale facilities

3.1 Bioimaging

Comprehensive reviews emphasised advances in the elemental imaging of biological matter and focused either on (sub-)cellular elemental/structural imaging or the visualisation of specific elements in biological systems. In their critical review, Matsuyama et al.23 gave an historical overview of the development of sub-μm SR-XRF spectrometry for bioimaging and illustrated the imaging technologies with examples from SPring-8, based on the use of sub-100 nm focusing systems. The use of nanoXRF spectrometry in conjunction with coherent X-ray diffraction and ptychography for nanoscale elemental/structural imaging was discussed. Leary and Ralle24 focused on the latest advances for the visualisation of Cu in mammalian systems, in particular for studying the homeostasis of Cu in the brain and the role of Cu in neurodegenerative diseases. Studies discussed included the detection of unusual Cu-storage vesicles originally identified in mouse brain tissues and the involvement of Cu in Alzheimer’s disease. The review highlighted the relevance of correlative sub-cellular imaging using XRF microscopy together with other imaging techniques (e.g. with SR-based ptychography) and also considered possibilities offered by the 4th generation of synchrotron sources being developed.

Falchini et al.25 developed quantitative elemental imaging approaches for application to the important topic of cancer research. Their quantitative strategy was based on the fundamental parameter method which considered excitation by a polychromatic incident microbeam and took into account sample matrix effects. The excitation spectrum was established through spectral reconstruction based on the scattering pattern from a thin plastic calibration standard in conjunction with the use of two ionisation chambers. The quantitative procedure was applied to 15 sections of mammary adenocarcinoma tissues (30 μm thickness) fixed in paraffin. Scans over 5 × 5 mm sample areas in air with a spatial resolution of 20 μm and acquisition time of 5 s per pixel gave a LOD of 240 ppm w/w for P.

Also in the context of cancer research, Summers et al.26 investigated the intracellular distribution of selected elements in rat brain glioma cells following 8HQ-treatment with and without CuII co-treatments. When CuII was added, all three 8HQ compounds investigated demonstrated greater cytotoxicity towards cancer cells. The XRF imaging was performed on cells deposited on silicon nitride substrates and analysed on beamline 2 ID-D at the Advanced Photon Source, Argonne National Laboratory (Lemont, IL, USA) with a storage ring electron current of 100 mA. The incident X-ray energy was set to 14.5 keV for simultaneous excitation of the K-lines of key elements of interest (Br, Cu, Fe, Zn). The XRF and scattering intensities, monitored using a single-element SDD, were collected at a spatial resolution of ca. 250 nm and with a dwell time of 400 ms.

Archanjo et al.27 studied 78 samples of tumour tissues from patients (50–70 years old) with oral cancer squamous cell carcinoma (OCSCC) in order to establish a link between trace element concentrations and the occurrence of head and neck cancers, in particular in individuals with a record of tobacco or alcohol use. The presence of As, Br, Cl, Cr, Mg, Mn and Ni in OCSCC samples was associated with smoking, there being a significant correlation between relapse and the detection of Cl and Cr. The tumour tissue samples (mean thickness 450 μm) were analysed at the D09-XRF beamline at the Brazilian Synchrotron Light Laboratory (Campinas, São Paulo, Brazil) with a white beam (2 mm2) in the energy range 4–24 keV and a dwell time of 20 s.

A significant number of nanoscale SR-XRF studies are of neurodegenerative diseases. A nanoscale imaging approach employed by Genoud et al.28 combined XRF microscopy (XFM) and X-ray ptychography to correlate elemental contents with the structures of complex neuropathological microfeatures (Lewy bodies; aggregations of superoxide dismutase 1; neuromelanin in Parkinson’s disease tissues taken post-mortem). The lower Cu[thin space (1/6-em)]:[thin space (1/6-em)]Zn concentration ratios found in tissues from subjects with Parkinson’s disease suggested impaired Cu binding. The data were obtained using the Bionanoprobe of the Advanced Photon Source (Argonne National Laboratory, IL, USA) with a 10 keV incident beam focused by a stacked Fresnel-zone plate to a ca. 100 nm spot on the sample. The scans were performed in fly-scan mode with an 80 nm pixel size and 50 ms dwell time per pixel.

Joppe et al.29 used sub-μm scanning XRF spectrometry and XRD at the ESRF ID13 beamline to study the structure and elemental composition of neuromelanin-positive neurons in substantia nigra tissue from subjects with Parkinson’s disease. A total of 53 neurons were analysed for the intracellular presence of As, Br, Cu, Fe, Mn, Se and Zn. Surprisingly large inter- and intra-individual variances in trace element concentrations were detected, bringing into question the existence of a clearly defined and general dyshomeostasis. A 240 × 250 nm beam focused by two sets of sub-μm-focusing Si-compound refractive lenses was used with a dwell time of 1 s per pixel.

An impressive quasi-correlative approach for analytical nanoimaging was presented by Lemelle et al.30 for research on Parkinson’s disease. Trace-element distributions were established for subcellular structures inside neuronal cell bodies located in the substantia nigra tissue. The method was based on quasi-correlative XRF nano-imaging of 500 nm-thick sections taken from a rat ventral midbrain in combination with TEM-images from adjacent 80 nm thick sections. The ultrastructure provided quantitative trace-element contents down to organelle levels in dopaminergic neurons. The usefulness of this approach for exploring dysfunctions at organelle levels was demonstrated by the observation of elemental (Fe and S) dyshomeostasis in cytoplasmic granules that overexpressed alpha-synuclein protein associated with Parkinson’s disease. The XRF measurements were performed using the scanning XRF microscopy setup of the ID16A nano-imaging beamline of the ESRF with a nanobeam (23 (horizontal) × 37 (vertical) nm) at an excitation energy of 17 keV and a photon flux of ca. 3 × 1011 photons per s. The highest-resolution elemental maps from the neuronal structures were obtained with a pixel size of 25 nm and a dwell time of 50 ms per pixel.

Rumancev et al.31 demonstrated the use of a recently developed cryogenic vacuum chamber optimised for trace element bioimaging by applying it to the analysis of frozen hydrated HeLa cells. The analyses were performed with a large-acceptance SDD and continuous scanning at the hard X-ray micro/nano-probe beamline P06 at PETRA III. Details were given for the cryogenic XRF-chamber and the sample transfer system. The data acquisition concept was also presented. The conditions used for the analysis of cryogenically fixed cells were a sample temperature of 140 K, a chamber base pressure of 10−8 mbar and a beam focus of 600 × 500 nm at a photon energy of 13 keV. When a fly-scan approach (5 μm s−1) was used, sufficiently large areas could be scanned to correlate results with phase contrast microscopy images.

Kourousias et al.32 demonstrated an interesting dynamic scanning XRF approach for reducing the acquisition time for wide scale mapping. The method, applicable to, for example, large-area brain tissue elemental-imaging, involved skipping individual sampling points if only a background signal was detected and by dynamically adjusting the acquisition time (or other scan settings) based on pre-defined analytical conditions. Such “compressive sensing XRF” scans could reduce data collection times dramatically by skipping or accelerating data collection from less relevant sample regions, thereby making possible those challenging experiments which are subject to time constraints when employing traditional scanning strategies. A 57% reduction in acquisition time was achieved using this approach when a soybean root specimen was imaged with sub-μm spatial resolution at the TwinMic beamline of Elettra (Trieste, Italy).

3.2 Environmental science

Environmental science studies using SR-based XRF, XAS and XRD micro- and nano-beam techniques remain in the forefront of elemental and chemical-state imaging applications.

A considerable number of studies devoted to soil science applications addressed important questions such as the fate of bioavailable nutrients and their mobility in the environment. In particular, the long-term supply and cycling of P in soils are of global environmental and agricultural concern. Adediran et al.33 combined synchrotron XRF microscopy with multi-elemental co-localisation and P K-edge XANES spectroscopy for the 2D imaging of P retention and speciation at the microscale in two types of forest soil. The μXRF and μXANES experiments were performed at the LUCIA beamline of SOLEIL (Saint-Aubin, France) and used a combination of a Si(111) double crystal monochromator and Kirkpatrick–Baez (K–B) mirrors system to provide a monochromatic energy-tunable microbeam (2.5 × 2.5 μm). The photon flux was 6 × 1011 photons per s at 2.6 keV.

The same procedure was employed34 with sub-μm resolution at the ESRF ID21 beamline to study the leaching of colloids and NP from agricultural soils, particular emphasis being placed on the behaviour of P and S. High-resolution μXRF images of elements were acquired with a dwell time of 100 ms using step sizes of 1.0 × 1.0 and 0.5 × 0.5 μm for the coarse and fine particles, respectively, and were used in conjunction with P and S K-edge XANES data.

In a study of P in agricultural soils of the mid-Atlantic region of the USA, Gamble et al.35 used μXRF mapping to distinguish P species adsorbed on Al oxides from those adsorbed on Fe oxides and μXANES to determine soil P speciation at a spatial resolution of <20 μm. The point μXANES (P K-edge) analyses and μXRF mapping of P, Al and Si were performed on beamline 14-3 at the Stanford Synchrotron Radiation Lightsource (California, USA) using an excitation energy of 2.24 keV (10 μm step size and 50 ms dwell time per pixel). The μXRF mapping of Ca, Fe and Si (5 μm step size and 20 ms dwell time per pixel) was performed on beamline 2-3 using 12 keV excitation.

A potentially simple way to modify and control the P availability in biochar using a low-cost potassium additive was investigated by Buss et al.36 as a means to address the global issue of P loss from agricultural systems. The optimised P and K biochar fertiliser was produced by doping sewage sludge with a low-cost mineral (2 and 5% potassium acetate) and pyrolysing at 700 °C. Highly soluble potassium hydrogen phosphate was identified up to 200–300 μm below the biochar surface using SR-XRF mapping in a helium-purged sample environment at a spatial resolution of 2–3 μm and P K-edge XANES at beamline I18 of the Diamond Light Source (Oxfordshire, UK).

Mitsunobu et al.37 developed a novel soil chamber for non-destructive ‘live’ soil imaging by using both μXRF spectrometry and μXAS to observe directly the local behaviour of metal(loid)s in waterlogged soils such as flooded paddy soils. The new soil chamber was made of titanium plate and glass with low gas permeability and high corrosion resistance. The soil chamber was successfully used in the investigation of the behaviour of As, a common redox-sensitive toxic element in flooded paddy soils, Fe and Mn. The μXRF analyses were performed using an excitation energy of 12 keV and at a spatial resolution of 5 μm at beamline 4 A at the KEK Photon Factory (Japan).

3.3 Earth and planetary sciences

A nice example of employing spatially resolved SR-XRF and XANES analyses was presented by Maruoka et al.38 who investigated the enrichment of chalcophile elements in Cretaceous–Paleogene boundary clays (from Stevns Klint, Denmark) associated with the end-of-the-Cretaceous impact event 66.0 million years ago. The concentrations of several chalcophile elements including Ag, Cu and Pb correlated with that of Ir, suggesting that these elements were supplied to the oceans by processes related to the end-of-the-Cretaceous asteroid impact. The SR-XRF images revealed that Ag and Cu existed as trace elements either in pyrite grains or as 1–10 μm-sized discrete phases enriched in Ag or Cu. The SR-XRF data were obtained by using beam line 37XU at SPring-8 (Hyogo, Japan) with a focused beam of 0.5 × 0.3 μm at 14 keV for sub-μm measurements or 1.8 × 1.3 μm at 30 keV for larger mappings.

Ash fallouts from volcanic super-eruptions have a potentially toxic impact on biocalcifier planktic microorganisms. Lemelle et al.39 demonstrated a method to detect past bioactive metal releases in ocean surface water. The use of nanoXRF imaging revealed successfully a specific Zn- and Mn-rich banding pattern in the test walls of Globorotalia menardii planktic foraminifers extracted from the Young Toba Tuff layer which corresponded to Toba’s super-eruption 74[thin space (1/6-em)]000 years ago. The nanoXRF experiments were performed at the leading synchrotron hard X-ray nano-analysis ID16B beamline (ESRF) using a highly focused beam of 55 × 60 nm with a flux of 5 × 1011 photons per s at 17.4 keV.

A very interesting in situ SR-XRF/XAS study was conducted by Louvel et al.40 with the aim of understanding better the behaviour of halogens (Br, Cl, I) and associated trace-elements in subduction zones. By identifying the boundaries of the geochemical cycle of these volatiles, it will be possible to quantify the halogen fluxes to the atmosphere via volcanic degassing. Diamond anvil cells were used to generate conditions with temperatures of up to 840 °C and pressures of up to 2.2 GPa. The partitioning of Br between coexisting aqueous fluids and hydrous granitic melts was observed in situ by μSR-XRF spectrometry. The Br speciation in slab-derived fluids was determined by XAS. The measurements were performed at the microXAS beamline (X05LA) of the Swiss Light Source (Villigen, Switzerland) using a microbeam of 5 × 8 μm with an incident energy tuned around 13.6 keV by a Si(111) double-crystal monochromator to provide a photon flux of 2 × 1011 photons per s.

3.4 Material science

NanoXRF spectrometry combined with XAS and XRD is now routinely used for advanced material characterisation at the 3rd- and the newest 4th-generation facilities. An excellent example of XRF-imaging at nanoscale resolution combined with nanoXANES was reported by Segura-Ruiz et al.41 who studied Ge1−xSnx alloys optimised for the fabrication of electronic and optoelectronic monolithically integrated devices on CMOS platforms. These high-performance devices require high quality GeSn layers free of Sn precipitates. The study demonstrated that Sn precipitation observed in thick GeSn layers grown directly on Ge buffers could be fully suppressed with Ge1−xSnx step-graded buffers with Sn concentrations as high as 16%. The nanoXRF measurements were performed using a 33.8 keV pink beam (4 × 1010 photons per s, ΔE/E = ca. 10−2) down to a spot size of 51 × 60 nm. The nanoXANES measurements were undertaken with a monochromatic beam (6 × 108 photons per s, ΔE/E = ca. 10−4).

Particularly interesting operando XRF measurements were presented by Dawkins et al.42 who demonstrated a methodology for measuring electrolyte concentrations during battery operation by tracking solution-phase concentration profiles in a model LiFePO4/Li cell. The Kα emissions of As, Cr and Fe were monitored to identify the steel current collector in the cell, the LiAsF6 electrolyte and the LiFePO4 based on linescans with a step size 10 μm s−1 over a length of 1 mm. The temporal resolution was ca. 100 s. The experiment was performed at the Canadian Light Source (Saskatoon, Canada).

Based on a similar strategy, operando and spatially and temporally resolved synchrotron X-ray fluorescence mapping measurements using an aqueous Zn/α-MnO2 cell provided43 direct evidence of a Mn dissolution–deposition faradaic mechanism governing the electrochemistry. The simultaneous visualisation and quantification of the Mn distribution in the electrolyte revealed the formation of an aqueous Mn species during discharge and a depletion during charging. The operando μXRF experiments were performed using a custom-designed electrochemical cell at the XFM beamline (4-BM) at the National Synchrotron Light Source II (Brookhaven, NY, USA). Regions of interest within the cell were rastered using a 10 μm step and a 50 ms dwell time per pixel for coarse maps, and a 2 μm step and a 100 ms dwell time per pixel for fine-resolution maps.

The nanoscale characterisation of photovoltaic materials represents an emerging class of applications employing nanoXRF spectrometry combined with complementary spectroscopic/imaging techniques. Ritzer et al.44 used X-ray and electron microscopies to study the composition and microstructure of Cu2ZnSn(S,Se)4 (CZTSSe) solar cell absorbers with different nominal compositions and performance at the nanoscale. The experiments showed the coexistence of the CZTSSe absorber material and ZnS(Se) secondary phase segregations, the size and number of which increased when the synthesis conditions become depleted in Cu and richer in Zn. The nanoXRF measurements allowed subtle compositional changes at the nm level to be observed. The potential of this approach was demonstrated by identification of CZTSSe domains less poor in Cu and less rich in Zn than the nominal composition and of grain boundaries with increased Cu, decreased Zn and unchanged Sn concentrations. The nanoXRF imaging was conducted at the nano-analysis ID16B beamline of the ESRF using a nanobeam with a spot size of 54 × 52 nm (scanning step size of 50 nm) at an excitation energy of 29.6 keV.

A helium continuous-flow mini-cryostat was demonstrated by Steinmann et al.45 for low-temperature X-ray nanoanalysis at the same beamline as the previous study. The cryostat was designed to perform simultaneous collection of XRF and X-ray excited optical luminescence data in the temperature range of 3–50 K. The performance of the mini-cryostat was demonstrated through 2D scans of core/shell InGaN/GaN multi-quantum well wires acquired with a pixel size of 70 × 70 nm and a dwell time of 500 ms per pixel. The cryostat could operate for at least 3 days at 20 K but only for 2 days at 35 K working temperature. The high spatial and temperature stability of the cryo-system was compatible with the 50 nm spatial resolution provided by the beamline at temperatures as low as 6 K. This new instrument was expected to reduce substantially the risk of radiation damage of sensitive samples when using the new ESRF EBS.

4. Grazing X-ray techniques including TXRF spectrometry

A comprehensive review by de la Calle et al.46 on the binding of volatile species to NP included a detailed description of trapping strategies such as complexation with ligands directly on the carrier, addition of a polymer dispersion and immobilisation by redox reactions. An overview was given of the problems encountered in TXRF analysis arising from losses due to volatilisation of species such as As, Hg, Se and halides. Of the various NP considered, carbon nanotubes were favoured because they gave a lower background than metal and metal oxide NP.

Significant advances have been made in the determination of volatile and particulate species commonly present in ambient air at low concentrations. Threshold concentrations of Hg in the workplace are 20 μg m−3 (ref. 47) and background levels ca. 10 ng m−3 (ref. 48) so the European Air Quality Index accepts PM2.5 and PM10 concentrations of up to 10 and 20 μg m−3, respectively, as an indication of good air quality. As TXRF spectrometry is a micro analytical tool, it would be well suited for the determination of heavy metals at such low concentrations if issues of sample collection and calibration could be overcome. Böttger et al.49 used silver NP to study the efficiency of Hg collection from ambient air. The LOD for 24 h sampling and 500 s measurement time was 18 ng m−3 when 10 nm silver NP stabilised with citrate and BH4 were used. This is considered acceptable when compared with the LOD of ca. 2 ng m−3 obtained by dedicated Hg analysers. Not only particle size but also the nature of the stabilising ions had a significant influence on the collection efficiency. Silver nitrate that had been exposed to light exhibited the best Hg collection capabilities, indicating that a redox reaction can be used to increase Hg binding further.

Borgese et al.50 improved the analysis of air particulate matter samples protected between two layers of polymer by initially scanning the incident angle and then selecting an optimal position. The RSD of 10% obtained was significantly better than that of ca. 16% obtained when a fixed-angle spectrometer was used. The combination of size-segregated air-particulate-matter sampling using a May impactor and a laboratory-based TXRF system using a WOBI-module enabled Osán et al.51 to undertake contamination-free and high throughput (1–4 h aerosol collection time) sampling and analysis of sub-μg amounts of particulate matter for major and trace metals and Br, Cl and S.

Seeger et al.52 presented a comparison of TXRF spectrometry and ICP-MS methods for the determination of elements in air particulate matter. Samples for analysis using TXRF spectrometry were obtained from aerosols impacted directly on acrylic glass and silicon wafer carriers (Dekati and May impactor) whereas the samples for ICP-MS analysis were impacted on filters and digested prior to analysis. Optimal sampling times were considerably less than 12 h for analysis by TXRF spectrometry and about 24 h for analysis using ICP-MS. The elemental concentrations determined by the two methods differed by 18–230% for elements at concentrations of ca. 1 ng m−3 (Ni, Pb, V) and by 3–62% for more abundant (2–110 ng m−3) elements (Cr, Cu, Fe, Mn, Zn). This was considered acceptable for such low concentrations. Overall, TXRF spectrometry offered shorter sampling intervals of <12 h with a LOD of ca. 10 pg m−3.

The designs of particle impactors vary to achieve efficient size-cutoff of the particle size fractions to be collected. Commonly they are constructed for filter sampling and the impaction plates cannot be easily exchanged for TXRF reflectors. Matsuyama et al.53 overcame this problem by collecting the aerosols on filters which were subsequently dissolved in acetone and deposited directly on a silicon wafer. Use of the thin residue obtained reduced the LOD by a factor of ca. 1.7. Fomba et al.54 evaluated two sample preparation procedures (digestion and direct analysis of polycarbonate filters; extraction from quartz filters) to determine elements in ambient air particulate matter and cloud water. Recoveries from NIST SRM 2783 (PM2.5 on polycarbonate membrane) were 99–101% for both procedures. The LOD of 0.3 ng cm−2 for As was determined from the analysis of a blank filter.

Suspension-assisted analysis is one of the outstanding advantages of TXRF spectrometry over most other trace element atomic spectrometry approaches. However, careful consideration needs to be given to the preparation of a representative aliquot, the analyte homogeneity in the specimen, IS distribution and the critical thickness when developing suspension-assisted sample preparation procedures. The particle size and distribution affect all these aspects. By using a 200 W ultrasonic homogeniser to reduce the mean diameter particle size of petroleum coke samples from ca. 10 μm to 1 μm, Fernández-Ruiz et al.55 improved the recovery of S from ca. 50 to 94%. Results from the angle scans indicated that not only absorption effects but also the inhomogeneous distribution of S and the Ti IS contributed to the lower recovery from the specimens which contained a fraction of larger sized particles. The high density (ca. 4–6 g cm−3) of ore samples makes them more difficult to stabilise in an aqueous suspension. Pashkova et al.56 evaluated suspension-assisted sample preparation and IS calibration to study iron, ferromanganese, manganese and copper–nickel sulfide ores as well as ocean ferromanganese nodules. The samples were ground using a planetary ball mill (tungsten carbide) and a suspension (4 mg mL−1) prepared in a 1% Triton X-100 solution by sonication. Adjusting the surface density of the specimen to below the critical value (defined as I/I0 ≥ 90%) of Al (e.g. 40 μg cm−2 for the magnetite matrix) accounted only for the difference in absorption between the analyte and the IS rather than for the total absorption. Whereas reducing the mean particle diameter of the iron and manganese ore samples from about 11–19 μm to ca. 3 μm reduced the analysis RSD from 20–30% to 3–8%, surprisingly, the RSD of the copper sulfide ore analysis were not significantly improved by particle size reduction. This finding was attributed to the poor wettability of the copper sulfide ores which led to a less homogeneous distribution of the Ga IS in the copper sulfide ore than in the other ores. In a related study, similar improvements were obtained57 by wet milling a clay CRM to reduce particle sizes even further (to 1–2 μm). Akhmetzhanov et al.58 compared multi- and univariate calibration for both TXRF and WDXRF analyses for the determination of REE in ores with the aim of assessing whether the cost-efficient suspension-assisted TXRF procedure could be used as an alternative to the WDXRF procedure which employed pressed pellets. A set of calibration samples with variable concentrations of REE was used to improve the accuracy. Two multivariate calibration models based on PLS and PCR were used to predict the content of REE in the samples. Cerium, La and Nd could be determined in most of the samples with single digit percentage bias and not exceeding 30%. However, Pr and Sm could not be determined using these multivariate models at levels below 100 ppm. The multivariate calibration approach using TXRF spectrometry therefore provided results comparable with those typically obtained by WDXRF spectrometry.

Instrumental developments included59 modification of an instrument with a silver anode X-ray tube to excite the L2 shell of U efficiently. The U Lβ1 peak was used to determine U (at ca. 0.5 ppm) in the presence of large amounts of Rb and Sr (10 ppm) for the on-site screening of environmental samples. A portable low-power TXRF instrument equipped with a single-digit watt tube was developed by Nagai et al.60 The LOD of Mg in vacuum was 0.81 ng L−1.

Calibration of the determination of As V pre-concentrated from water samples was optimised by Sanyal et al.61 They used IS of either gold NP synthesised in situ or those complexed with N-methyl-D-glucamine (NMDG) which was anchored to the quartz carrier. The NMDG bound both the AsV and the NP. The intensity of the As Kα line was normalised to the Au Lα line to account for varying surface coverage by the NMDG. Recoveries, from ground and tap waters spiked with as little as 0.5 ng mL−1 of the AsIII and AsV species, were 98–112% and the LOD was 0.05 ng mL−1 (1000 s live time). Synthesising the IS as part of the specimen removed the risk of IS loss and so simplified the TXRF analysis significantly.

Shao et al.62 used the Ar present in the ambient air surrounding the sample and detector as an IS in the suspension-assisted TXRF analysis of plant CRMs. All elemental-line net-intensities were normalised to the Ar Kα net intensity and sensitivities obtained from calibration curves plotting the normalised signals against the concentrations. The recoveries of generally 90–110% were comparable to those obtained using calibration with Ga or V IS. The RSDs were 1–10% for all determinations.

The important issue of specimen morphology has been the subject of a number of studies. It has been modelled,63 absorption effects have been evaluated64 and preparation strategies (surface modification and volume reduction) proposed to control it.65 Sugioka et al.66 suggested a procedure in which diamond-like carbon-coated quartz carriers were coated with a hydrophobic film and then part of the film was removed to accommodate a 200 μL water sample. If the sample was vibrated during drying, the ring-like residue transformed into a spot-like residue, thereby giving a significant three-fold increase in sensitivity. In the direct TXRF analysis of beverages sweetened with sugar or synthetic compounds, the reduction in the intensity of low-energy characteristic fluorescence lines matched67 the calculated absorption well.

Grazing incidence XRF spectrometry is a versatile tool for studying shallow and deep implants in functional materials. Czyzycki et al.68 discussed state-of-the-art implant analysis and assessed the use of GI-XRF spectrometry together with model calculations to study Ar implanted into Si-wafers at ca. 250 nm. Although the model was not suitable in its present form for the GI-XRF analysis of multi-layer structures, it was considered that the inclusion of second-order X-ray phenomena would improve suitability. Soft-X-ray-based GI-XRF analysis and a modelling scheme was used69 to perform an element-sensitive reconstruction of a lamellar grating and three dimensional nanostructures. Many GI-XRF models must take into account the change in solid angle of detection that occurs when the incident angle is scanned. Hönicke et al.70 characterised the effective solid angle of detection and the incident photon-flux of a Bruker S4 T-STAR instrument using a well-characterised 12 nm thick Ni layer on a Si wafer surface. The procedure was validated by the depth-profiling of two ultra-shallow (6–8 nm depth) As-ion-implant samples. The characterisation of NP is an ongoing challenge and usually requires several methods to obtain an accurate picture. Unterumsberger et al.71 combined experimental GI-XRF spectrometry of Pt/Ti core–shell NP (2–3 nm) and modelled XSW field intensities to determine surface coverage of the NP and to study the effect of the NP on the XSW. The core–shell NP modified the intensity distribution of the XSW field with increasing surface coverage as expected from the calculations. In a study of periodic nanostructures, Nikolaev et al.72 used dynamical diffraction theory as the basis for a computational scheme applied to GI-XRF experimental results. The model calculations simulated GI-XRF data from structures with specific element distributions both in-plane and with depth.

The combination of GI-XRF spectrometry and XRR to distinguish layers of equal density but different composition (and vice versa) was highlighted in last year’s ASU.73 Ingerle et al.74 demonstrated that a commercial XRD instrument with XRR capabilities can be upgraded quite easily for GI-XRF analysis by adding a SDD. The performance was validated by analysis of a 50 nm Ni film on silicon and applying the JGIXA software. The intensity modulation of the characteristic XRF lines induced by the XSW was usually limited to the first nm at the surface of the sample so the accuracy of the composition deduced by GI-XRF spectrometry degraded with depth. Nolot et al.75 used a multilayer (Mo/Si)*N Bragg mirror to improve the sensitivity of GI-XRFS-XRR analysis. The multilayered substrates generated a XSW field at angles significantly higher than the critical angle. Composition- and structure-matched RM were used by Torrengo et al.76 to determine the key parameters of a GI-XRFS-XRR setup with the aim of improving analytical accuracy. Application to a study of titanium- and tungsten-nitride thin films gave a recovery of 96% and a RSD of ≤20% for the determination of the Ti/W ratio. The GI-XRFS-XRR approach allowed identification of sublayers of different Ti and W composition, which could not be determined by XRR alone or by sputtering based-techniques such as plasma profiling-TOF-MS or TOF-SIMS.

5. Hand-held, mobile and on-line XRF techniques

5.1 Hand-held and mobile XRF techniques

Portable XRF spectrometry has undergone significant technological improvements in recent years and is now applied in a wide range of studies. The period covered by this ASU saw publication of a significant number of valuable review papers dealing with different aspects of the pXRF technique. That of Adams et al.77 focused on the use of pXRF spectrometry for determination of the concentrations of light elements such as Al, Mg, Na and Si. Limitations in hardware design such as excitation energy and current; counting capability and dead time requirements; detector resolution and size and operating-environment were discussed. Moreover, extensive datasets were presented for empirical investigations under various conditions (air, vacuum, helium flush) and using different detector windows. In a review by Laperche and Lemière,78 the most important issues to be taken into account while using pXRF instruments for the analysis of minerals and geomaterials included the analytical mode and parameters; protective films; sample geometry and density, especially for light elements; analytical interferences between elements; physical effects of the matrix and sample condition. Both reviews highlighted that a strong understanding of pXRF characteristics was required to avoid reporting erroneous results.

For pXRF spectrometry to be applied to the analysis of plant samples, challenges such as low sample density and relatively high moisture content need to be addressed. Zhou et al.79 reviewed the effect of water content, protective film effect, critical penetration depth of X-rays, sample thickness, measurement time and particle size on XRF results and provided guidance for application to plant analysis.

Portable XRF technology is commonly applied to the assessment of the elemental composition of soil but the technique is being increasingly implemented as a tool for predicting other soil parameters of interest (e.g. soil CEC, soil salinity). A nice overview of the applicability of the pXRF technique in that field was presented by Weindorf and Chakraborty80 who reviewed existing procedures, data quality and processing, sample applications and case studies in which data were correlated to the soil salinity or even combined with remote sensing data. Tavares et al.81 developed a method to overcome matrix effects for the prediction of key soil fertility attributes (clay, CEC, exchangeable-Ca and -K concentrations) in soil samples. In order to mitigate the matrix effect, combination of Compton normalisation (CN) with multivariate regressions (e.g. MLR and PLSR) was evaluated and compared to a modelling approach using both univariate and multivariate regression analyses without CN. In addition, the proposed methodology was compared to modelling based on a pre-programmed measurement package approach (Geo Exploration package). The combined use of CN with multivariate regressions (MLR or PLSR) achieved excellent prediction results for cation exchange capacity (R2 = 0.87) and exchangeable-Ca (R2 ≥ 0.96) and exchangeable-K concentrations (R2 ≥ 0.94). The prediction of clay concentrations was similar (0.89 ≤ R2 ≤ 0.92) with or without CN. The authors suggested using multivariate regressions combined with CN to remove the soil matrix effects thereby providing optimal predictions of all the key soil fertility parameters examined.

The combined use of pXRF and DRIFT-MIR spectroscopies with machine learning models was assessed by Towett et al.82 for the determination of total elemental compositions in organic amendments. Randomised cross validation trials (using 33% of all organic amendment samples and Forest or XGBoost models) confirmed the good predictive value of the XRF and MIR calibrations for most elements. Although pXRF spectrometry cannot be used to measure elements such as C or N directly, the application of the models to XRF data gave excellent results (R2 > 0.86) for both elements. This approach has the potential of providing data for nutrient management at minimal analytical cost. To determine elemental concentrations in moist livestock manure samples, Sapkota et al.83 used a whole-spectrum approach coupled with random forest regression not only to predict the moisture content of the sample up to 70% moisture but also to correct moisture effects on pXRF measurements, thereby making in-field analysis feasible. Oven-dried manure samples (n = 40) were ground and adjusted to five moisture contents in the range <10% to 70% (w/w). Samples were scanned for 180 s using a commercially available pXRF instrument. The spectrum in the energy range from 14 to 15 keV had strong predictive power (R2 = 0.98) for moisture content. The random forest approach increased R2 between XRF data and wet chemical methods data from <0.66 to >0.90 for K, Mg and P and from 0.78 to 0.98 for Fe when compared with linear, non-linear and Lucas-Tooth and Price equations (n = 200). This procedure had potential for the analysis of other matrices such as moist soil and plant samples.

5.2 On-line XRF techniques

Source apportionment was performed for size-resolved PM2.5and PM10 using an on-line XRF spectrometer during two consecutive winters (2018 and 2019) in Delhi (India).84 Aerosols were collected on a Teflon filter tape at a flow rate of 16.7 L min−1. The instrument switched between PM2.5 and PM10 sampling thereby resulting in a gap of 30 minutes between each analysis. The XRF calibration of 26 elements from Al to Pb was achieved using thin-film standards. The response reproducibility during calibration throughout the campaign was within ±5%. Measurements obtained with a high time-resolution made it possible to characterise plumes from emission sources and to establish diurnal patterns. These XRF data provided insights into key PM sources in Delhi such as dust, inorganic constituents (e.g. Cl) and public health risks due to the extremely high concentrations of carcinogenic (As, Cr, Ni, Pb) and other elements (e.g. Cu, Mn, V, Zn). Nine different aerosol sources were identified during both winters using positive matrix factorisation.

Mineralogical information about flotation slurries is crucial for process control. Wang et al.85 determined the concentrations of minerals from a tungsten processing plant by fusing XRF and NIR data obtained for element concentrations using an on-stream slurry analyser. The XRF system (rhodium target X-ray tube, 50 W) was equipped with a fast SDD having a high resolution (123 eV at 5.9 keV), large probe window (70 mm2) and a short measurement time (4 μs). The XRF parameters were optimised by Monte Carlo simulation and a multivariate statistical method was employed to perform the element-to-mineral conversion. In the fusion of pre-processed Fe and W channel XRF data with NIR data, the prediction ability was best for scheelite (R2 = 0.907), wolframite (R2 = 0.924) and calcite (R2 = 0.845). Such data integrations allowed the on-stream and quantitative identification of slurry mineral contents, in particular for scheelite, wolframite, fluorite and calcite which are essential minerals in tungsten ore beneficiation.

5.3 Planetary exploration

A miniaturised, low-budget XRF spectrometer (the Active X-Ray Spectrometer (AXRS)) was developed86 at the John Hopkins University Applied Physics Laboratory for the in situ study of the bulk rock geochemistry of lithophile materials undertaken on planetary missions. The AXRS sensor head had a mass of 360 g, fitted within an envelope of 11 × 6 × 5 cm and consumed less than 5 W during operation. A robust calibration and geochemical analysis campaign showed that AXRS could achieve an average 10% relative uncertainty and precision for the concentrations of major and minor elements from Na to Fe. Measurements were made with powdered samples analogous to the highly processed regoliths to be expected on the surfaces of objects on airless planets. Potential applications for AXRS included the identification of igneous rocks on planets (Mars, Venus) with volcanic activity, discerning subtle petrologic variations in lunar materials and examination of carbonaceous chondrites to resolve mineralogy on C-type asteroids and similar bodies. Nittler et al.87 presented global maps of Mg/Si, Al/Si, S/Si, Ca/Si, and Fe/Si ratios derived from XRF spectrometry data collected throughout MESSENGER’s orbital mission of Mercury. The data selection and map generation procedures, for which detailed descriptions were given, allowed smoothed versions of the maps to be produced and subsequently archived in NASA’s Planetary Data System. The spatial resolution was 10–15% better than that obtained for previous mapping because of the inclusion of higher-resolution data from late in the mission. Several examples were provided of how the XRF maps could be used to investigate elemental variations in the geological features on Mercury at spatial scales ranging from single 100 km diameter craters to large impact basins.

6. Cultural heritage applications

The continued increase in the use of macroXRF systems in cultural heritage investigations has seen easy-to-use open-source software becoming available to handle the metadata files generated. Lins et al.88 developed an XRF imaging software (XIMuS) for multiple samples using an intuitive and simple graphical user interface. The software structure consisted of the program core, a datacube file containing all relevant information of the dataset and the Mosaic application used for stitching blocks of data together. Elemental mapping was made simple and the fast, balanced or precise calculations available allowed supporting parallel-computing. Ratios between elements in specific regions of the image could be measured while a region-derived spectrum was shown and updated live. Image correlation supported the use of threshold filters and/or region selection. The authors demonstrated the introduction of novel features such as data stitching and a constantly updated database. Dai et al.89 presented an XRF image inpainting approach to address the long scanning times of macroXRF systems. This speeded up the scanning process but did not compromise construction of high quality XRF images. The proposed adaptive image-sampling-algorithm was applied to the RGB image of the scanning target to generate the sampling mask. The XRF scanner was then driven according to the sampling mask to scan a subset of the total image pixels. This scanned XRF image was inpainted by fusing the RGB image to reconstruct successfully the full scan XRF image.

A recently found portrait of Leonardo da Vinci (oil on paint) was studied by Caliri et al.90 using in situ macroXRF scanning spectrometry to characterise the chemical nature of pigments used in the palette and to confirm compatibility with the date (17th century) established by scholars. The mobile LANDIS-X scanner, developed at the LANDIS laboratory of ISPC-CNR and INFN-LNS in Catania (Italy) and based on novel real-time technology, was used to analyse the entire surface of the portrait in a unique continuous scan (675 × 449 pixels) at 10 cm s−1 speed and lasting less than 2 h. A spatial resolution of 1 mm for elemental images was obtained with a dwell time of 10 ms per pixel. The X-ray tube operated at 50 kV and 0.3 mA. The visualisation of the chemical elements on the pictorial support was made in real time during the scanning. The original colour of some degraded regions of the painting was revealed, suggesting an appropriate conservation policy. In addition, pictorial details of a hidden figure painted below the Leonardo da Vinci portrait were unveiled and the identification of several pictorial details, such as the vest and its original colour, obtained by using Cu- and Hg-based pigments. Bicchieri et al.91 used the same mobile LANDIS-X scanner to investigate two important drawings that Leonardo da Vinci completed with the metalpoint technique. They used macroXRF, μXRF and confocal-XRF spectrometries to obtain the 2D and 3D elemental imaging of the artworks. MacroXRF measurements performed on the Portrait of a girl (1485) established that the metalpoint used had Cu as its main alloying element but typically used metals such as Ag or Au were not present. In the same drawing, the scatter plots of Ca and P confirmed the use of white bone for the preparatory layer. In the case of the drawing Study of the front legs of a horse (1490), macroXRF spectrometry revealed that the paper was prepared with apatite and then coloured with a superficial layer of indigo mixed with white lead. In contrast, use of Raman spectroscopy would easily detect the white lead but would fail to detect the apatite. The metalpoint used by Leonardo da Vinci in this particular drawing was a ternary Cu–Ag–Au alloy. In order to inform the restoration of a huge oil canvas (266 × 361 × 3 cm) by Agostino Bosia, Nervo et al.92 applied multispectral techniques associated with point (XRF spectrometry, FORS) and microstratigraphic analyses to reconstruct accurately the artist’s creative process and to identify the materials used. By combining the standard analytical protocol with macroXRF scanning the number of samples could significantly be reduced. This analysis was carried out on a selected area (130 × 150 cm) of the “Costruzione del viadotto” with the X-ray tube operated at 50 kV and 0.3 mA, a scanning speed of 10 cm s−1 and a spatial resolution of 1 mm obtained with a dwell time of 10 ms per pixel. The whole area was analysed in 5.4 h and the elemental images were elaborated in real time during the scan. The element distributions of Ca, Co, Cr, Pb and Zn, used in combination with data obtained by other techniques, showed the localisation and the distribution of the applied pigments in the painting.

The use of SR sources using beamlines with micrometric focusing optics makes it possible to view specific details in the composition of a pigment. A painting from the 17th century suspected of being counterfeit was analysed by Pereira et al.93 using the SR scanning macroXRF technique in combination with complementary analytical techniques such as handheld XRF spectrometry. Elements such as Ba, Ca, Fe, Mn, Pb, Ti and Zn were detected using a commercially available handheld XRF system. Elemental maps of the signature region (mapping area of 120 × 120 mm) were acquired using the LNLS D09B μXRF beamline with an energy of 16 keV, a pixel size of 100 μm and a dwell time of 100 ms per pixel. Measurements were made in the flyscan mode. The elemental maps constructed using the Ca and Cl K-lines provided information that did not fit the period of the artwork. Elemental maps of Ba, Fe, Ti and Zn (using K-lines) and of Pb (using the L-lines) visualised the way pigments were used in the creation of the painting thereby indicating that the work was probably a forgery.

Characterisation of pigments present in an important painting, Portrait of a Young Man with a Golden Chain by Rembrandt and/or his atelier from the 17th century, was made in situ by Molari et al.94 using a portable EDXRF spectrometry system which consisted of a silver target X-ray tube, a 3 mm silver collimator and an SDD. The pigments were identified based on the colours associated with the presence of key elements in the XRF spectra. The various materials identified included the filler (chalk), pigments (lead white, bone black, vermilion, azurite) and ochre pigments (red and yellow ochres and brown earths). All the results obtained in this study were consistent both with materials available to artists in the 17th century and with pigments previously identified in paintings by Rembrandt and his atelier. The variable blue and yellow pigments found in four paintings by Caspar David Friedrich by using in situ XRF imaging revealed95 different creative periods of the painter. The commercially available mobile XRF scanner used to acquire the XRF maps consisted of a rhodium target X-ray tube, an SDD (25 mm2) and a motorised stage of 10 × 10 cm. The X-ray tube was operated at 40 kV and 100 μA. The scanning was performed with a dwell time of 1 s per pixel and step sizes of ca. 0.5 mm horizontally and 1.0 mm vertically. Smalt (As, Co, K, Ni, Si) was the oldest blue pigment which could be identified in the four paintings. Prussian blue (Fe, K) slowly superseded smalt in the 18th century and in the 19th century cobalt blue (Co, Ni) was introduced. The yellow pigments were lead chromate (Cr, Pb), cadmium yellow (Cd) and yellow and red earth (Fe). The prevalence of lead white in the central motif, The Lamb of God, of the famous Ghent Altarpiece by Hubert and Jan Van Eyck meant that Van der Snickt et al.96 were unable to visualise Van Eycks’ original underdrawing of the Lamb and any design changes or subsequent overpainting by later restorers when using a single spectral imaging modality. However, the chemical contrast needed could be achieved by using both elemental macroXRF and molecular IR reflectance imaging spectroscopies followed by analysis of the resulting data cubes. The techniques applied were described in detail and their benefits, limits and complementarity discussed.

The macroXRF technique is often combined with other techniques to elucidate the layers of ancient wall-painting fragments or panelled vaults. Vadrucci et al.97 combined macroXRF spectrometry with PIXE to study eight fragments of mural paintings of Villa della Piscina (Rome, Italy). The macroXRF instrument consisted of a measuring head fitted with a molybdenum target X-ray tube and an SDD (17 mm2), mounted on a three-axis precision positioning stage (300 × 150 mm). A dynamic positioning system controlled and adjusted the working distance during the motion of the measuring head. The operating conditions were 25 kV voltage, 30 μA current, 2 mm s−1 scanning velocity and 500 μm pixel size. An 800 μm collimator on the X-ray tube was used. The macroXRF maps allowed the pigments used in the paintings to be identified from their elemental compositions. The blue/greenish/greyish hues were likely obtained with different qualities of Egyptian Blue (Cu, Ca and Si). Iron-oxide pigments were identified in all brown/reddish areas. Lead-based compounds and cinnabar were occasionally used in some samples. Laclavetine et al.98 studied multiple layers of paintings from a small part of the large vault (0.4% of the 32 m2) in the church of Le Quillio (France). The examination was performed by extracting, processing and cross-referencing the information from the complementary data obtained from macroXRF scanning as well as multi- and hyperspectral imaging. Elemental mappings were performed with an XRF scanner with a rhodium target X-ray tube (50 kV and 600 μA), polycapillary optics and an SDD (60 mm2). The target was placed horizontally at an average working distance of 2.4 cm. The beam diameter and the distance between each point were both 450 μm and the dwell time 100 ms per pixel. All XRF analyses were performed through hermetically sealed polyethylene bags in which panels were kept. By characterising the mineral composition of the different layers, it was possible to attribute the layers to the original painting (end of the 15th century), to an overpaint (17th century) and to a large retouching or restoration (19th century).

A lightweight macroXRF scanner with a molybdenum target X-ray tube (4 W) and an SDD (50 mm2) was used in a study of manuscripts (a set of choir books) preserved in the Abbey of San Giorgio Maggiore on a Venetian island (Italy).99 Use of a telemeter for the continuous control of the sample-instrument distance during the scan was particularly helpful in scanning the uneven surface of the manuscripts. The measuring head was mounted on three linear motor stages (300 × 150 × 50 mm). The whole scanner including the control computer was lighter (<10 kg) and more compact (60 × 50 × 50 cm) than other systems with similar performance. The operating conditions were 25 kV tube voltage, 50 μA filament current, 1 mm s−1 scanning velocity and 500 μm equivalent-pixel size with a 800 μm diameter collimator. This compact macroXRF system provided high-quality maps of three illuminated initials, thereby proving its potential. Christiansen et al.100 used SR macro- and μXRF, μXRD and μFTIR spectrometries to highlight the compositions of 12 red and 12 black inks preserved on ancient Egyptian papyri from the Roman period. The distribution of lead compounds detected in most red inks and some black inks supported the theory that they were probably employed for their drying properties rather than as colouring inks. These findings indicated the need for a reassessment of the composition of lead-based components in ancient Mediterranean pigments.

The complementary use of the non-invasive techniques macroXRF imaging and Raman spectroscopy was demonstrated by Bicchieri et al.101 through analysis of nine pieces of Chinese paper money (14th–19th century). The LANDIS-X scanner was the same one used in the study of Leonardo Da Vinci paintings.91 For these measurements, the sample-to-detector distance was 16.5 mm, the beam spot-size ca. 120 μm, the scanning speed 10 mm s−1 and the dwell time 50 ms. The composition of the dyes could be used to authenticate the paper money and revealed that seven of the banknotes were counterfeit.

The potential of macroXRF scanning was demonstrated102 through a study of a series of ink inscriptions on a set of artefacts from Stradivari’s workshop. The scanner was equipped with a rhodium target X-ray tube, polycapillary optics and a SDD (60 mm2). The target was placed horizontally at a working distance of 12 mm. The beam presented a diameter of 100 μm, the distance between each point was 100 μm and the time per point 200 ms. The use of XRF imaging techniques not only improved the readability of inscriptions because of the enhanced contrast offered by chemical maps but also allowed the evaluation of Cu, Fe and Zn contents of the inks. Lins et al.103 developed an innovative algorithm for evaluating large macroXRF datasets and determining the thickness of an overlapping layer. The authors developed further the differential-attenuation method used for single-spot XRF measurements for application to macroXRF scanning. MacroXRF elemental distribution maps of a gilded copper-based buckle from the 16th–17th century found in Rome (Italy) were fundamental in identifying and choosing sampling areas to calculate the thickness of the gilding layer. This non-invasive approach provided a mean thickness value of 1.24 ± 0.43 μm, with a maximum of 2.20 μm for the amalgam-gilded layer, whereas previous SEM-EDS analysis had given values of 1.65 and 3.00 μm in two different analysed regions.

Cagno et al.104 compared the performance of four non-invasive techniques (macroXRF, UV-VIS-NIR and Raman spectroscopies and IR thermography) that provided the accuracy, flexibility, time-efficiency and transportability required for in situ characterisation of leaded glass windows (17th century). The macroXRF system consisted of an XRF measurement head mounted on an XY motor stage with a travel range of about 60 × 57 cm. The measurement head was equipped with a SDD and a 50 W rhodium target X-ray tube. The diameter of the primary beam was reduced to a focal spot of 200 μm. Use of a step size of 800 μm and a dwell time of 500 ms resulted in a relatively slow scan speed. The macroXRF technique provided the most detailed chemical information. In particular, the number of plausible glass families could be reduced considerably by using the ratio between the network modifier (K) and network stabiliser (Ca) and the level of colourants and decolourisers (As, Fe, Mn). The power of the new LANDIS-X scanning station equipped with a real-time macroXRF system was demonstrated by Cavaleri et al.105 by analysing the coffin lid of Neskhonsuennekhy in the Museo Egizio collection (Turin, Italy). Pigments, glazes and mixtures were studied as well as the Pb impurities in the copper source used to make the Egyptian blue pigment. The ancient polychromy of the lid was reconstructed and the delicate underlying drawings rediscovered through combination of data from macroXRF spectrometry and multispectral imaging techniques such as UV fluorescence, IR reflectography and, especially, visible induced luminescence.

The high spatial and energy resolution of SR-based techniques is beneficial in the study of archaeological artefacts. A selection of Attic and western-Greek black-glazed vessels from two archaeological excavations in Gela (Sicily, Italy) were studied by Gianoncelli et al.,106 using SR μXRF imaging and μXANES at the newly opened PUMA beamline of the French national synchrotron facility (SOLEIL). The X-rays were focused to a spot of 3 × 3 μm on the sample through use of a Kirkpatrick–Baez mirror. Samples were scanned at two different excitation energies, 18 keV and 7.3 keV. A scan at 18 keV with an acquisition time of 0.5 s was used to identify elements such as As, Br, Ni, Pb and Zn whereas a scan at 7.3 keV with an acquisition time of 1 s excited only the lower Z elements such as Ca, K and Si. In both cases, a step size of 5 μm was chosen. The XANES spectra were collected in fluorescence mode with an acquisition time of 5 s. The region around the edge from 7.08 to 7.2 keV was scanned with high resolution in steps of 0.5 eV. Chemical and mineralogical markers in the glaze used for provenance identification were Fe, Mn, Zn and iron speciation. The μXRF mappings highlighted a variety of black-glaze compositions, including some characterised by high levels of Zn.

The estimation of the coating thickness of gilded objects using a commercial XRF spectrometer was studied by Sabbarese et al.107 using four metals (copper, iron, lead, silver) covered with increasing thicknesses of gold. For double layers, the thickness of the covering layer was assessed by (a) the ratio between the most intense fluorescence lines of the covered element, (b) a cross-ratio between two lines of the elements in the two layers and (c) the PLS regression method. All three approaches provided comparable data for coating thicknesses. The same researchers used108 a series of 23 reference silver layers with known thicknesses to obtain calibration curves for the estimation of the thickness of silver objects. A systematic determination of the intensity of Ag Kα, Kβ and Lα and their ratios was performed for different sheet thicknesses and for the three elements (copper, iron, lead) placed under the silver. Although all the methods gave consistent results within the expected thickness range, the PLS regression methodology was the most accurate. The novelty of these studies was a clear comparison between the various methodologies.

Portable XRF spectrometry remains an important analytical tool for investigating artefacts.

Costa et al.109 evaluated two commercially available handheld XRF instruments using tesserae from the Mosaico de los Amores (Linares, Spain). The first XRF system (40 kV, 11 μA and a real time of 60 s) was equipped with an 10 mm2 SDD and a rhodium target X-ray tube delivering a polychromatic X-ray beam of 3 × 3 mm. The second XRF instrument was equipped with a 20 mm2 SDD and a rhodium target X-ray tube delivering a 3 × 3 mm collimated X-ray beam. A two-beam mode (Geochem Mode) was used to analyse the selected tesserae with a real time of 300 s. In this mode, spectra were recorded sequentially using two different conditions: (a) 10 kV, 200 μA and no filter; and (b) 40 kV and 100 μA and an aluminium filter. Although both systems could be recommended for in situ campaigns in cultural heritage studies, the two-beam mode of the second system allowed a better detection of low-Z elements, as was to be expected. Melquiades et al.110 combined the use of pXRF measurements and Monte Carlo spectral simulation for quantitative evaluation of 16 golden artefacts from the Museo Nacional de Etnografia y Folklore (La Paz, Bolivia). The spectra simulation was conducted with the open source XMI-MSIM software. The method was efficient in the quantification of Au, Ag and Cu artefacts alloy composition. The Cu content in the samples analysed was <5% whereas that of Au was 13–100%.

7. Abbreviations

2Dtwo dimensional
3Dthree dimensional
ASUAtomic Spectrometry Update
AXRSactive X-ray spectrometer
CCDcharge coupled device
CECcation exchange capacity
CMOScomplementary metal oxide semiconductor
CNCompton normalisation
CRMcertified reference material
CTcomputed tomography
CZTSSecopper zinc tin sulfur selenium
DRIFTdiffuse reflectance infrared Fourier transform
EBSExtremely Brilliant Source
EDenergy dispersive
EDSenergy dispersive X-ray spectrometry
EDXRFenergy dispersive X-ray fluorescence
ESRFEuropean Synchrotron Radiation Facility
FORSfibre optics reflectance spectrometry
FTIRFourier transform infrared
FWHMfull width at half maximum
GI-XRFgrazing incidence X-ray fluorescence
ICPinductively coupled plasma
INFN-LNSIstituto Nazionale di Fisica Nucleare, Laboratori Nazionali del Sud
ISinternal standard
ISPC-CNRIstituto di Scienze del Patrimonio Culturale, Consiglio Nazionale delle Ricerche
JGIXAjava grazing incidence X-ray analysis
LANDISlaboratory of non destructive analysis in situ
LNLSLaboratório Nacional de Luz Síncrotron
LODlimit of detection
LUCIAline for ultimate characterisation by imaging and absorption
MIRmid infrared
MLRmultiple linear regression
MSmass spectrometry
μmicro (followed by technique name)
NASANational Aeronautics and Space Administration
NIRnear infrared
NISTNational Institute of Standards and Technology
NMDG N-methyl-D-glucamine
OCSCCoral cancer squamous cell carcinoma
PCRprincipal component regression
PETRApositron-electron tandem ring accelerator
PIXEparticle-induced X-ray emission
PLSRpartial least squares regression
PLSpartial least squares
PMxparticulate matter (with an aerodynamic diameter of up to x μm)
PUMAPhotons Utilisés pour les Matériaux Anciens
pXRFportable X-ray fluorescence
REErare earth element
RGBred green blue
RMreference material
RSDrelative standard deviation
SDDsilicon drift detector
SEMscanning electron microscopy
SIMSsecondary ion mass spectrometry
SOLEILSource Optimisée de Lumière d’Énergie Intermédiaire du LURE
SRsynchrotron radiation
SRMstandard reference material
TEMtransmission electron microscopy
TXRFtotal reflection X-ray fluorescence
WDXRFwavelength dispersive X-ray fluorescence
XANESX-ray absorption near edge structure
XASX-ray absorption spectroscopy
XFMX-ray fluorescence microscopy
XRDX-ray diffraction
XRFX-ray fluorescence
XRFSX-ray fluorescence spectrometry
XRRX-ray reflectometry
XSWX-ray standing wave
Z atomic number

8. Conflicts of interest

There are no conflicts to declare.

9. References

  1. J. R. Bacon, O. T. Butler, W. R. L. Cairns, O. Cavoura, J. M. Cook, C. M. Davidson and R. Mertz-Kraus, J. Anal. At. Spectrom., 2021, 36(1), 10–55 Search PubMed .
  2. M. Patriarca, N. Barlow, A. Cross, S. Hill, A. Robson, A. Taylor and J. Tyson, J. Anal. At. Spectrom., 2021, 36(3), 452–511 Search PubMed .
  3. E. H. Evans, J. Pisonero, C. M. Smith and R. N. Taylor, J. Anal. At. Spectrom., 2020, 35(5), 830–851 Search PubMed .
  4. R. Clough, C. F. Harrington, S. J. Hill, Y. Madrid and J. F. Tyson, J. Anal. At. Spectrom., 2020, 35(7), 1236–1278 Search PubMed .
  5. S. Carter, R. Clough, A. Fisher, B. Gibson, B. Russell and I. Whiteside, J. Anal. At. Spectrom., 2020, 35(11), 2410–2474 Search PubMed .
  6. D. Gürsoy and C. Jacobsen, MRS Bull., 2020, 45, 272–276 Search PubMed .
  7. C. Gramaccioni, Y. Yang, A. Pacureanu, N. Vigano, A. Procopio, P. Valenti, L. Rosa, F. Berlutti, S. Bohic and P. Cloetens, Anal. Chem., 2020, 92(7), 4814–4819 Search PubMed .
  8. B. De Samber, M. Bensellam, S. J. M. Van Malderen, F. Seiboth, D. Brückner, J. Garrevoet, G. Falkenberg, J. C. Jonas and L. Vincze, J. Anal. At. Spectrom., 2020, 35(7), 1368–1379 Search PubMed .
  9. J. Palle, N. K. Wittig, A. Kubec, S. Niese, M. Rosenthal, M. Burghammer, T. A. Grünewald and H. Birkedal, J. Struct. Biol., 2020, 212(3) DOI:10.1016/j.jsb.2020.107631 .
  10. G. B. M. Vaughan, R. Baker, R. Barret, J. Bonnefoy, T. Buslaps, S. Checchia, D. Duran, F. Fihman, P. Got, J. Kieffer, S. A. J. Kimber, K. Martel, C. Morawe, D. Mottin, E. Papillon, S. Petitdemange, A. Vamvakeros, J.-P. Vieux and M. Di Michiel, J. Synchrotron Radiat., 2020, 27, 515–528 Search PubMed .
  11. H. Nakano, S. Sonoda, T. Matsuyama, S. Komatani and K. Tsuji, X-Ray Spectrom., 2021, 50(3), 224–230 Search PubMed .
  12. D. Ingerle, J. Swies, M. Iro, P. Wobrauschek, C. Streli and K. Hradil, Rev. Sci. Instrum., 2020, 91(12) DOI:10.1063/5.0028830 .
  13. P. Zhou, X. R. Ma, S. Zhang, T. X. Sun and Z. G. Liu, Chin. Phys. B, 2020, 29(2) DOI:10.1088/1674-1056/ab1671c .
  14. R. Prokeš and T. Trojek, Radiat. Prot. Dosim., 2019, 186(2–3), 268–273 Search PubMed .
  15. P. Tack, T. Schoonjans, S. Bauters and L. Vincze, Spectrochim. Acta, Part B, 2020, 173 DOI:10.1016/j.sab.2020.105974 .
  16. S. I. Zaitsev, O. V. Trofimov, Y. L. Shabelnikova and M. V. Chukalina, Instrum. Exp. Tech., 2020, 63(3), 425–429 Search PubMed .
  17. A. Kulow, A. G. Buzanich, U. Reinholz, F. Emmerling, S. Hampel, U. E. A. Fittschen, C. Streli and M. Radtke, J. Anal. At. Spectrom., 2020, 35(7), 1423–1434 Search PubMed .
  18. G. C. Xiong, W. B. Jia, Q. Shan, X. L. Zhang, X. R. Tang and J. Li, Rev. Sci. Instrum., 2020, 91(12) DOI:10.1063/5.0024461 .
  19. S. An, D. Krapohl, B. Norlin and G. Thungström, J. Instrum., 2020, 15(12) DOI:10.1088/1748-0221/15/12/P12033 .
  20. Q. Yang, Z. P. Deng, J. L. Shi, H. Q. Ao and W. C. Lai, Nucl. Instrum. Methods Phys. Res., Sect. A, 2020, 963 DOI:10.1016/j.nima.2020.163751 .
  21. D. Hampai, V. Guglielmotti, A. Marcelli, G. Cappuccio and S. B. Dabagov, Radiat. Phys. Chem., 2020, 174 DOI:10.1016/j.radphyschem.2020.108965 .
  22. X. P. Sun, X. Y. Zhang, Y. B. Wang, S. K. Shao, Y. F. Li, S. Q. Peng, Z. G. Liu and T. X. Sun, Spectrochim. Acta, Part B, 2020, 165 DOI:10.1016/j.sab.2020.105770 .
  23. S. Matsuyama, K. Maeshima and M. Shimura, J. Anal. At. Spectrom., 2020, 35(7), 1279–1294 Search PubMed .
  24. S. C. Leary and M. Ralle, Curr. Opin. Chem. Biol., 2020, 55, 19–25 Search PubMed .
  25. G. E. Falchini, A. Malezan, M. E. Poletti, E. Soria, M. Pasqualini and R. D. Perez, Radiat. Phys. Chem., 2021, 179 DOI:10.1016/j.radphyschem.2020.109157 .
  26. K. L. Summers, N. V. Dolgova, K. B. Gagnon, G. J. Sopasis, A. K. James, B. Lai, N. J. Sylvain, H. H. Harris, H. K. Nichol, G. N. George and I. J. Pickering, Metallomics, 2020, 12(12), 1979–1994 Search PubMed .
  27. A. B. Archanjo, A. de Assis, M. M. de Oliveira, S. O. Mendes, A. R. Borçoi, L. D. Maia, R. P. de Souza, R. de Cicco, K. C. Saito, E. T. Kimura, M. B. de Carvalho, F. D. Nunes, E. H. Tajara, M. dos Santos, B. V. Nogueira, L. O. Trivilin, C. J. G. Pinheiro and A. M. Álvares-da-Silva, Sci. Rep., 2020, 10(1) DOI:10.1038/s41598-020-67270-5 .
  28. S. Genoud, M. W. M. Jones, B. G. Trist, J. J. Deng, S. Chen, D. J. Hare and K. L. Double, Chem. Sci., 2020, 11(33), 8919–8927 Search PubMed .
  29. K. Joppe, J. D. Nicolas, T. A. Grünewald, M. Eckermann, T. Salditt and P. Lingor, Biomed. Opt. Express, 2020, 11(7), 3423–3443 Search PubMed .
  30. L. Lemelle, A. Simionovici, P. Colin, G. Knott, S. Bohic, P. Cloetens and B. L. Schneider, Commun. Biol., 2020, 3(1) DOI:10.1038/s42003-020-1084-0 .
  31. C. Rumancev, T. Vöpel, S. Stuhr, A. von Gundlach, T. Senkbeil, S. Ebbinghaus, J. Garrevoet, G. Falkenberg, B. De Samber, L. Vincze, A. Rosenhahn and W. Schroeder, Biointerphases, 2021, 16(1), 011004,  DOI:10.1116/6.0000593 .
  32. G. Kourousias, F. Billè, R. Borghes, A. Alborini, S. Sala, R. Alberti and A. Gianoncelli, Sci. Rep., 2020, 10(1) DOI:10.1038/s41598-020-66435-6 .
  33. G. A. Adediran, J. R. M. Tuyishime, D. Vantelon, W. Klysubun and J. P. Gustafsson, Geoderma, 2020, 376, 114550,  DOI:10.1016/j.geoderma.2020.114550 .
  34. G. A. Adediran, D. Lundberg, G. Almkvist, A. E. P. del Real, W. Klysubun, S. Hillier, J. P. Gustafsson and M. Simonsson, Water Res., 2021, 189, 116585,  DOI:10.1016/j.watres.2020.116585 .
  35. A. V. Gamble, P. A. Northrup and D. L. Sparks, J. Environ. Qual., 2020, 49(1), 184–193 Search PubMed .
  36. W. Buss, A. Bogush, K. Ignatyev and O. Mašek, ACS Sustainable Chem. Eng., 2020, 8(32), 12295–12303 Search PubMed .
  37. S. Mitsunobu, T. Hiruta, J. Fukudo, Y. Narahashi, N. Hamamura, N. Matsue and Y. Takahashi, Geoderma, 2020, 373, 114415,  DOI:10.1016/j.geoderma.2020.114415 .
  38. T. Maruoka, Y. Nishio, T. Kogiso, K. Suzuki, T. Osawa, Y. Hatsukawa and Y. Terada, Geol. Soc. Am. Bull., 2020, 132(9–10), 2055–2066 Search PubMed .
  39. L. Lemelle, A. Bartolini, A. Simionovici, R. Tucoulou, W. De Nolf, F. Bassinot and T. de Garidel-Thoron, Sci. Rep., 2020, 10(1) DOI:10.1038/s41598-020-67481-w .
  40. M. Louvel, C. Sanchez-Valle, W. J. Malfait, G. S. Pokrovski, C. N. Borca and D. Grolimund, Solid Earth, 2020, 11(4), 1145–1161 Search PubMed .
  41. J. Segura-Ruiz, N. Pauc, J. Chrétien, L. Casiez, J. M. Hartmann, A. Chelnokov, V. Calvo, V. Reboud, S. Tardif, R. Tucoulou and J. Villanova, Mater. Sci. Eng., B, 2021, 264, 114899,  DOI:10.1016/j.mseb.2020.114899 .
  42. J. I. G. Dawkins, M. Z. Ghavidel, D. Chhin, I. Beaulieu, M. S. Hossain, R. F. Feng, J. Mauzeroll and S. B. Schougaard, Anal. Chem., 2020, 92(16), 10908–10912 Search PubMed .
  43. D. R. Wu, L. M. Housel, S. J. Kim, N. Sadique, C. D. Quilty, L. J. Wu, R. Tappero, S. L. Nicholas, S. Ehrlich, Y. M. Zhu, A. C. Marschilok, E. S. Takeuchi, D. C. Bock and K. J. Takeuchi, Energy Environ. Sci., 2020, 13(11), 4322–4333 Search PubMed .
  44. M. Ritzer, S. Schönherr, P. Schöppe, G. Larramona, C. Choné, G. Gurieva, A. Johannes, K. Ritter, G. Martínez-Criado, S. Schorr, C. Ronning and C. S. Schnohr, Phys. Status Solidi A, 2020, 217(21), 2000456,  DOI:10.1002/pssa.202000456 .
  45. R. G. Steinmann, G. Martinez-Criado, D. Salomon, H. Vitoux, R. Tucoulou, J. Villanova, S. Laboure, J. Eymery and J. Segura-Ruiz, J. Synchrotron Radiat., 2020, 27, 1074–1079 Search PubMed .
  46. I. de la Calle, V. Romero, I. Lavilla and C. Bendicho, Spectrochim. Acta, Part B, 2020, 168, 105843,  DOI:10.1016/j.sab.2020.105843 .
  47. Commision Directive, 2009/161/EU, Official Journal of the European Union, 2009, 32018, Search PubMed.
  48. Air Quality Guidelines for Europe, World Health Organisation, 2000, European series no. 91, 92 Search PubMed.
  49. S. Böttger, J. Kolny-Olesiak and U. E. A. Fittschen, Spectrochim. Acta, Part B, 2020, 170, 105903,  DOI:10.1016/j.sab.2020.105903 .
  50. L. Borgese, F. Bilo, A. Zacco, S. Federici, A. W. Mutahi, E. Bontempi, K. Trzepla, N. Hyslop, S. Yatkin, P. Wobrauschek, J. Prost, D. Ingerle and L. E. Depero, Spectrochim. Acta, Part B, 2020, 167, 105840,  DOI:10.1016/j.sab.2020.105840 .
  51. J. Osán, E. Börcsök, O. Czömpöly, C. Dian, V. Groma, L. Stabile and S. Török, Spectrochim. Acta, Part B, 2020, 167, 105852,  DOI:10.1016/j.sab.2020.105852 .
  52. S. Seeger, J. Osan, O. Czömpöly, A. Gross, H. Stosnach, L. Stabile, M. Ochsenkuehn-Petropoulou, L. A. Tsakanika, T. Lymperopoulou, S. Goddard, M. Fiebig, F. Gaie-Levrel, Y. Kayser and B. Beckhoff, Atmosphere, 2021, 12(3), 309,  DOI:10.3390/atmos12030309 .
  53. T. Matsuyama, H. Yamaguchi and K. Tsuji, J. Anal. At. Spectrom., 2021, 36(3), 570–575 Search PubMed .
  54. K. W. Fomba, N. Deabji, S. E. Barcha, I. Ouchen, E. Elbaramoussi, R. C. El Moursli, M. Harnafi, S. El Hajjaji, A. Mellouki and H. Herrmann, Atmos. Meas. Tech., 2020, 13(9), 4773–4790 Search PubMed .
  55. R. Fernández-Ruiz, M. L. Redrejo, K. E. J. Friedrich, N. Rodríguez and R. Amils, Spectrochim. Acta, Part B, 2020, 174, 105997,  DOI:10.1016/j.sab.2020.105997 .
  56. G. V. Pashkova, V. M. Chubarov, T. F. Akhmetzhanov, A. N. Zhilicheva, M. M. Mukhamedova, A. L. Finkelshtein and O. Y. Belozerova, Spectrochim. Acta, Part B, 2020, 168, 105856,  DOI:10.1016/j.sab.2020.105856 .
  57. A. S. Maltsev, G. V. Pashkova, R. Fernández-Ruiz, E. I. Demonterova, A. N. Shuliumova, N. N. Umarova, D. L. Shergin, M. M. Mukhamedova, V. M. Chubarov and E. A. Mikheeva, Spectrochim. Acta, Part B, 2021, 175, 106012,  DOI:10.1016/j.sab.2020.106012 .
  58. T. F. Akhmetzhanov, G. V. Pashkova, V. M. Chubarov, T. A. Labutin and A. M. Popov, J. Anal. At. Spectrom., 2021, 36(1), 224–232 Search PubMed .
  59. Y. Izumoto, K. Takamura, T. Matsuyama, H. Nagai, Y. Sakai, Y. Oguri and H. Yoshii, Spectrochim. Acta, Part B, 2020, 173, 105977,  DOI:10.1016/j.sab.2020.105977 .
  60. H. Nagai, H. Shiino and Y. Nakajima, Bunseki Kagaku, 2020, 69(9), 455–462 Search PubMed .
  61. K. Sanyal, S. Chappa, J. Bahadur, A. K. Pandey and N. L. Mishra, J. Anal. At. Spectrom., 2020, 35(11), 2770–2778 Search PubMed .
  62. J. F. Shao, W. B. Jia, X. L. Zhang, Y. Liu, X. R. Tang, G. C. Xiong and Q. Shan, J. Anal. At. Spectrom., 2020, 35(4), 746–753 Search PubMed .
  63. C. Horntrich, P. Kregsamer, P. Wobrauschek and C. Streli, Spectrochim. Acta, Part B, 2011, 66, 815–821 Search PubMed .
  64. M. Menzel, O. Scharf, S. H. Nowak, M. Radtke, U. Reinholz, P. Hischenhuber, G. Buzanich, A. Meyer, V. Lopez, K. McIntosh, C. Streli, G. J. Havrilla and U. A. E. Fittschen, J. Anal. At. Spectrom., 2015, 30(10), 2184–2193 Search PubMed .
  65. C. M. Sparks, U. A. E. Fittschen and G. J. Havrilla, Spectrochim. Acta, Part B, 2010, 65(9–10), 805–811 Search PubMed .
  66. T. Sugioka, H. Umeda and S. Kunimura, Anal. Sci., 2020, 36(4), 465–469 Search PubMed .
  67. A. von Bohlen and R. Fernández-Ruiz, Talanta, 2020, 209, 120562,  DOI:10.1016/j.talanta.2019.120562 .
  68. M. Czyzycki, M. Kokkoris and A. G. Karydas, J. Anal. At. Spectrom., 2020, 35(12), 2964–2973 Search PubMed .
  69. P. Hönicke, A. Andrle, Y. Kayser, K. V. Nikolaev, J. Probst, F. Scholze, V. Soltwisch, T. Weimann and B. Beckhoff, Nanotechnology, 2020, 31(50), 505709,  DOI:10.1088/1361-6528/abb557 .
  70. P. Hönicke, U. Waldschläger, T. Wiesner, M. Krämer and B. Beckhoff, Spectrochim. Acta, Part B, 2020, 174, 106009,  DOI:10.1016/j.sab.2020.106009 .
  71. R. Unterumsberger, P. Hönicke, Y. Kayser, B. Pollakowski-Herrmann, S. Gholhaki, Q. M. Guo, R. E. Palmer and B. Beckhoff, J. Anal. At. Spectrom., 2020, 35(5), 1022–1033 Search PubMed .
  72. K. V. Nikolaev, V. Soltwisch, P. Hönicke, F. Scholze, J. de la Rie, S. N. Yakunin, I. A. Makhotkin, R. W. E. van de Kruijs and F. Bijkerk, J. Synchrotron Radiat., 2020, 27, 386–395 Search PubMed .
  73. C. Vanhoof, J. R. Bacon, U. E. A. Fittschen and L. Vincze, J. Anal. At. Spectrom., 2020, 35(9), 1704–1719 Search PubMed .
  74. D. Ingerle, W. Artner, K. Hradil and C. Streli, Powder Diffr., 2020, 35, S29–S33 Search PubMed .
  75. E. Nolot, W. Pessoa, S. Torrengo, Y. Mazel, M. Bernard, P. Gergaud, Y. Ménesguen, M. C. Lépy and D. Eichert, Spectrochim. Acta, Part B, 2020, 168, 105864,  DOI:10.1016/j.sab.2020.105864 .
  76. S. Torrengo, D. Eichert, Y. Mazel, M. Bernard, Y. Ménesguen, M. C. Lépy and E. Nolot, Spectrochim. Acta, Part B, 2020, 171, 105926,  DOI:10.1016/j.sab.2020.105926 .
  77. C. Adams, C. Brand, M. Dentith, M. Fiorentini, S. Caruso and M. Mehta, Geochem.: Explor., Environ., Anal., 2020, 20(3), 366–380 Search PubMed .
  78. V. Laperche and B. Lemière, Minerals, 2020, 11(1), 33,  DOI:10.3390/min11010033 .
  79. S. B. Zhou, D. C. Weindorf, Q. M. Cheng, B. Y. Yang, Z. X. Yuan and S. Chakraborty, Spectrochim. Acta, Part B, 2020, 174, 105999,  DOI:10.1016/j.sab.2020.105999 .
  80. D. C. Weindorf and S. Chakraborty, Soil Sci. Soc. Am. J., 2020, 84(5), 1384–1392 Search PubMed .
  81. T. R. Tavares, A. M. Mouazen, E. E. N. Alves, F. R. dos Santos, F. L. Melquiades, H. W. P. de Carvalho and J. P. Molin, Agronomy, 2020, 10(6), 787,  DOI:10.3390/agronomy10060787 .
  82. E. K. Towett, L. B. Drake, G. E. Acquah, S. M. Haefele, S. P. McGrath and K. D. Shepherd, PLoS One, 2020, 15(12), e0242821,  DOI:10.1371/journal.pone.0242821 .
  83. Y. Sapkota, B. L. Drake, L. M. McDonald, T. C. Griggs and T. J. Basden, J. Environ. Qual., 2020, 49(2), 472–482 Search PubMed .
  84. P. Rai, M. Furger, I. El Haddad, V. Kumar, L. W. Wang, A. Singh, K. Dixit, D. Bhattu, J. E. Petit, D. Ganguly, N. Rastogi, U. Baltensperger, S. N. Tripathi, J. G. Slowik and A. S. H. Prévôt, Sci. Total Environ., 2020, 742, 140332,  DOI:10.1016/j.scitotenv.2020.140332 .
  85. Q. Y. Wang, F. S. Li, X. Y. Jiang, S. L. Wu and M. Q. Xu, Anal. Methods, 2020, 12(25), 3296–3307 Search PubMed .
  86. A. W. Beck, P. N. Peplowski and Z. W. Yokley, Planet. Space Sci., 2020, 190, 104990,  DOI:10.1016/j.pss.2020.104990 .
  87. L. R. Nittler, E. A. Frank, S. Z. Weider, E. Crapster-Pregont, A. Vorburger, R. D. Starr and S. C. Solomon, Icarus, 2020, 345, 113716,  DOI:10.1016/j.icarus.2020.113716 .
  88. S. A. B. Lins, B. Bremmers and G. E. Gigante, SoftwareX, 2020, 12 DOI:10.1016/j.softx.2020.100621 .
  89. Q. Q. Dai, H. Chopp, E. Pouyet, O. Cossairt, M. Walton and A. K. Katsaggelos, IEEE Trans. Multimed., 2020, 22(10), 2564–2578 Search PubMed .
  90. C. Caliri, M. Bicchieri, P. Biocca and F. P. Romano, X-Ray Spectrom., 2021, 50(4), 332–340 Search PubMed .
  91. M. Bicchieri, P. Biocca, C. Caliri and F. P. Romano, Microchem. J., 2020, 157, 104844,  DOI:10.1016/j.microc.2020.104844 .
  92. M. Nervo, F. P. Romano, C. Caliri, A. Piccirillo, P. Triolo, D. Demonte, A. Gatti, E. Vergallo, M. Cardinali and M. Ferrero, X-Ray Spectrom., 2021, 50(4), 253–262 Search PubMed .
  93. M. O. Pereira, V. S. Felix, A. L. Oliveira, D. S. Ferreira, A. R. Pimenta, C. S. Carvalho, F. L. Silva, C. A. Perez, D. Galante and R. P. Freitas, Spectrochim. Acta, Part A, 2021, 246, 118925,  DOI:10.1016/j.saa.2020.118925 .
  94. R. Molari, C. R. Appoloni and S. H. Rodriguez, Appl. Radiat. Isot., 2020, 165, 109346,  DOI:10.1016/j.apradiso.2020.109346 .
  95. M. Alfeld, K. Mösl and I. Reiche, X-Ray Spectrom., 2021, 50(4), 341–350 Search PubMed .
  96. G. Van der Snickt, K. A. Dooley, J. Sanyova, H. Dubois, J. K. Delaney, E. M. Gifford, S. Legrand, N. Laquiere and K. Janssens, Sci. Adv., 2020, 6(31), eabb3379,  DOI:10.1126/sciadv.abb3379 .
  97. M. Vadrucci, A. Mazzinghi, B. Sorrentino, S. Falzone, C. Gioia, P. Gioia, E. M. Loreti and M. Chiari, X-Ray Spectrom., 2020, 49(6), 668–678 Search PubMed .
  98. K. Laclavetine, D. Giovannacci, M. Radepont, A. Michelin, A. Tournié, O. Belhadj, C. Andraud and W. Nowik, X-Ray Spectrom., 2021, 50(4), 358–374 Search PubMed .
  99. A. Mazzinghi, C. Ruberto, L. Castelli, P. Ricciardi, C. Czelusniak, L. Giuntini, P. A. Mandò, M. Manetti, L. Palla and F. Taccetti, X-Ray Spectrom., 2021, 50(4), 272–278 Search PubMed .
  100. T. Christiansen, M. Cotte, W. de Nolf, E. Mouro, J. Reyes-Herrera, S. de Meyer, F. Vanmeert, N. Salvadó, V. Gonzalez, P. E. Lindelof, K. Mortensen, K. Ryholt, K. Janssens and S. Larsen, Proc. Natl. Acad. Sci. U. S. A., 2020, 117(45), 27825–27835 Search PubMed .
  101. M. Bicchieri, P. Biocca, C. Caliri, F. Vostal, L. Vostal and F. P. Romano, J. Cult. Heritage, 2020, 46, 140–147 Search PubMed .
  102. H. de la Codre, M. Radepont, J. P. Echard, O. Belhadj, S. Vaiedelich and V. Rouchon, X-Ray Spectrom., 2021, 50(4), 244–252 Search PubMed .
  103. R. Lins, S. Ridolfi, G. E. Gigante, R. Cesareo, M. Albini, C. Riccucci, G. di Carlo, A. Fabbri, P. Branchini and L. Tortora, Front. Chem., 2020, 8 DOI:10.3389/fchem.2020.00175 .
  104. S. Cagno, G. van der Snickt, S. Legrand, J. Caen, M. Patin, W. Meulebroeck, Y. Dirkx, M. Hillen, G. Steenackers, A. Rousaki, P. Vandenabeele and K. Janssens, X-Ray Spectrom., 2021, 50(4), 293–309 Search PubMed .
  105. T. Cavaleri, P. Buscaglia, C. Caliri, E. Ferraris, M. Nervo and F. P. Romano, X-Ray Spectrom., 2021, 50(4), 279–292 Search PubMed .
  106. A. Gianoncelli, S. Raneri, S. Schoeder, T. Okbinoglu, G. Barone, A. Santostefano and P. Mazzoleni, Microchem. J., 2020, 154, 104629,  DOI:10.1016/j.microc.2020.104629 .
  107. C. Sabbarese, J. Brocchieri and E. Scialla, X-Ray Spectrom., 2021 DOI:10.1002/xrs.3221 .
  108. J. Brocchieri, E. Scialla and C. Sabbarese, Nucl. Instrum. Methods Phys. Res., Sect. B, 2021, 486, 73–84 Search PubMed .
  109. M. Costa, A. Rousaki, S. Lycke, D. Saelens, P. Tack, A. Sánchez, J. Tuñón, B. Ceprián, P. Amate, M. Montejo, J. Mirão and P. Vandenabeele, Eur. Phys. J. Plus, 2020, 135(8) DOI:10.1140/epjp/s13360-020-00635-x .
  110. F. L. Melquiades, S. V. Santos, F. Lopes, J. Villanueva, J. L. Ticona and C. R. Appoloni, X-Ray Spectrom., 2021, 50(1), 53–67 Search PubMed .


Joint review coordinators.

This journal is © The Royal Society of Chemistry 2021