Atomic spectrometry update: review of advances in the analysis of clinical and biological materials, foods and beverages

Marina Patriarca *a, Nicola Barlow b, Alan Cross c, Sarah Hill d, Anna Robson e and Julian Tyson f
aIstituto Superiore di Sanita’, Viale Regina Elena 299, 00161 Rome, Italy. E-mail: marina.patriarca@iss.it; m.patriarca@outlook.it
bTrace Elements Laboratory, Black Country Pathology Services, Sandwell General Hospital, West Bromwich, West Midlands B71 4HJ, UK
cReading Scientific Services Ltd, The Reading Science Centre Whiteknights Campus, Pepper Lane, Reading, Berkshire RG6 6LA, UK
dLGC, Queens Road, Teddington, Middlesex TW11 0LY, UK
eDepartment of Biochemistry, Manchester University NHS Foundation Trust, Oxford Rd, Manchester M13 9WL, UK
fDepartment of Chemistry, University of Massachusetts Amherst, 710 North Pleasant St, Amherst, MA 01003, USA

Received 2nd February 2023

First published on 23rd February 2023


Abstract

This update covers publications from the second half of 2021 to the middle of 2022. Advances in the application of atomic spectrometry techniques to clinical and biological materials, foods and beverages are reviewed in the text, highlighting their key features. Technical details of sample collection and preparation, as well as progresses with analytical techniques are considered and three tables complement the text, summarising details of a larger spectrum of publications. During this period, the trend toward the application of multi-element techniques, such as EDXRFS, ICP-MS and LIBS continued, in particular for food authenticity studies. Triple quadrupole ICP-MS is becoming increasing popular, as it is less affected by interferences, as well as LIBS and XRF, that require minimal sample preparation. However, AAS is still considered a valid alternative for single or a limited number of elements: as in previous years, numerous pre-concentration techniques were presented, some of which explored “greener” reagents. The interest in NPs continued, both as a potential exposure risk and for their application as tags of biological materials, and led to a wider application of spICP-MS. Chromium speciation in food received more attention than usual during this period, providing evidence that the carcinogenic species CrVI was not present. A number of studies covered the application of atomic spectrometry techniques for the indirect determination of biological macromolecules, including an interesting application of LIBS for the rapid detection of the immune response to SARS-CoV-2.


1. Reviews

This latest Update follows on from last year’s1 and is a critical review of the relevant literature published during the approximately 12 month period from the second half of 2021 through the first half of 2022. It should be read in conjunction with the five other Atomic Spectrometry Updates from the previous year.2–6

The possibilities for supramolecular solvent-based microextraction techniques for sampling and pre-concentration of PTEs have been reviewed (113 references with titles).7 No time-frame was specified, but it appears as though most of the papers were published in the last 20 years. The reviewers explained what supramolecular solvents are and how, in general, they can be used in LLE procedures that are part of the sample preparation steps in methods for the determination of relevant elements. A rather limited number of examples were presented of applications to clinical and food samples as well as to environmental materials. Enrichment factors of a few tens seem to be typical, which does not compare all that favourably with the performance of other LLE procedures, for which enrichment factors of triple, or even quadruple, digits have been reported. The reviewers did not discuss what is needed in terms of sample digestion/extraction before the supramolecular solvent pre-concentration can be applied.

In an overview (41 references with titles) of metallic impurities in pharmaceuticals, the reviewers devoted much of the paper to the implementation of the 2013 ICH Q3D guideline.8 They also provided a tutorial introduction to the various sources of contamination, such as formulation ingredients, catalysts, process equipment, containers and closures, and a brief explanation of the operating principles of AAS, ICP-OES and ICP-MS, together with some examples of their application. The reviewers concluded that the quality risk assessment approach of the ICH Q3D guideline is an important milestone in efforts to harmonise control of elemental contaminants worldwide that allows manufacturers to provide vital information about the contribution of impurities in their drug products.

Concern over contamination by PTEs in medicinal plants, which have been a major source of drugs for hundreds, if not thousands, of years, means that concentrations of these elements must be monitored. The relevant literature over the past 25 years has been reviewed (206 references with titles),9 to provide an understanding of the current situation regarding the concentrations of PTEs in medicinal plants, but also to develop an in-depth awareness of the main methods of analysis. The review covered the important topic of sample preparation, including the possibilities for direct analyses of solids, before dealing with each instrumental technique. In addition to all the relevant atomic spectrometric techniques, NAA and ASV were also discussed. The reviewers highlighted the need for greater sample throughput capabilities and indicated that automation of sample preparation was a key area for future developments.

The applications of ICP-MS in clinical laboratories were reviewed (65 references with titles).10 The authors adopted a tutorial approach, covering the basic principles as well as recent technical advances in collision cells and analysers. The review covered a variety of applications that illustrated the role of ICP-MS in clinical and forensic toxicology laboratories, including determination of toxic elements, quantification of nutritional elements, monitoring of elemental deficiencies in metabolic diseases, and multi-element analyses of dried serum spots. This review is discussed further in Section 4.1.

A comprehensive, authoritative review (159 references with titles) of the determination of metallic nanoparticles in biological samples by spICP-MS11 started with a detailed description of how the relevant articles were retrieved from the Web of Science database, perhaps reflecting the co-authorship of a prominent member of the current ASU organisation. Also reflecting ASU practice, the review contained several lengthy tables that summarise protocols used to extract metal-based ENMs from (a) animal tissues, (b) plant tissues, and (c) biofluids. As well as a comprehensive section on sample preparation (aspects of which are discussed in Sections 3.1 and 3.2), the reviewers paid considerable attention to validation, with information summarised in a table entitled “Quality assurance checks and technical tips to aid the analysis of different types of control or reference samples”. The reviewers also discussed (a) progress towards the availability of CRMs and (b) practical considerations (at some length). They provided a helpful summary diagram of their key findings and recommendations, which included the most promising extraction procedures for biological matrices, and aspects of good practice, such as use of control samples and quality assurance protocols, and what particle parameters to report (particle size, size distributions, number and mass concentrations). They concluded that the next steps for the scientific community were (a) a consensus-based approach to developing standardised methods, based on identifying the most promising ones, (b) production of technical guidance documents on how best to apply the methods, with a decision tree to help select the most appropriate one for the type of sample or ENM, and (c) harmonisation of nomenclature and technical definitions.

Developments in XAS as an analytical tool for biological and biomedical applications over the past five years were reviewed (57 references, no titles) by Buzanich et al.12 Following a tutorial introduction, the authors discussed the application of μXAS and multiple energy μXRF combined with μXAS to obtain spatially resolved elemental and speciation information, which they described as among the most used analytical approaches for studying biological samples. They then discussed (a) high energy resolution fluorescence detection XAS and resonant X-ray emission spectroscopy (XES), and (b) time resolved XAS/XES (quick XAS, dispersive XAS, and ultrafast XAS). They concluded that developments in synchrotron- and X-ray free electron laser-based methods for XAS techniques, such as sub-μm sized X-ray beams together with fast measurement times, have opened a new window of opportunities for the characterisation of biological samples, which often contain low concentrations of analytes and are susceptible to radiation damage. They noted a trend in the use of theoretical calculations to verify the accuracy of experimental structural data, which they considered will become integrated with experimental investigations to describe fully both electronic properties and local structure of relevant materials.

The current status and prospects for the elemental analysis of non-invasive samples for human biomonitoring, as evidenced in the literature from the past ten years, were examined (150 references with titles).13 The reviewers discussed both the limitations of the information available from the analysis of blood and also how information from non-invasive samples might both correlate with and supplement this information. They comprehensively covered the relevant literature pertaining to the sampling and analysis of hair, nails, urine, meconium, breast milk, placenta, cord blood, saliva and teeth and summarised their findings in several useful tables dealing with (a) collection, storage, and pre-treatment, (b) sample preparation procedures for solution analysis, (c) selected speciation studies, (d) instrumental methods for solids analysis and (e) selected spatial analysis. They concluded with the hope that the advances reviewed will be beneficial not only in enhancing knowledge of the biochemistries of elements, but also in developing disease diagnoses, health evaluations and prognoses.

Two areas of clinical analysis in which atomic spectrometry is just one of many instrumental techniques applied were reviewed: boron analysis and imaging in boron neutron capture therapy (126 references with titles)14 and the characterisation and quantification of selenoprotein P (137 references with titles).15 Both reviews explained in some detail why the particular analytes are important and both included a limited number of applications featuring ICP-based methods with an emphasis on ICP-MS.

In combatting food fraud, knowledge of the multi-elemental composition is emerging as an important means of solving a number of food authentication and traceability challenges. In an overview (147 references with titles) of developments in the last ten years of multi-element analysis combined with chemometric data processing, Heredia et al.16 first provided a tutorial introduction to the basic principles of a comprehensive range of atomic spectrometry techniques before doing the same for various pattern recognition chemometric tools. The second half of the review was devoted to large numbers of examples of applications to the determination of geographical origin, botanical origin, and other classifications (e.g. based on type of fermentation or technological processing) as well as fraud detection. The reviewers concluded that ICP-MS appeared to be the best choice of instrumental technique because of its ability to determine a wide range of elements with good precision and accuracy. They did, however, indicate that (a) LIBS has the advantage of minimal sample preparation and (b) MIP-OES, with a plasma sustained in atmospheric nitrogen, is less costly than ICP-MS. They noted that the most frequently applied chemometric tools were PCA and LDA, though significant results have also been obtained with PLS-DA, SVM and ANN. A technique that does not feature at all prominently, XRF, was the subject of a separate review. Li et al. reviewed (111 references, no titles) the application of EDXRFS to the detection of elements in food over the past 20 years.17 The reviewers acknowledged the limitations of the technique with regard to food analysis (low and variable sensitivity in food matrices and hence poor detection capability and accuracy that is extremely dependent on the detector resolution) but indicated that recent improvements in detector performance and algorithm optimisation have allowed LODs of 0.1 μg g−1 to be reached. They also pointed out the inherent advantages of minimal sample preparation for the direct non-destructive analyses of solids. Much of the review was concerned with advances in instrument technology and data processing; however, the authors described the contents of 11 application papers that were mostly concerned with the analysis of grains (including rice) and tea. Comparisons with other techniques (WDXRF, ICP-OES, ICP-MS and Raman spectrometry) were also made. The review concluded with a list of five shortcomings of EDXRF that need to be addressed before the technique will be more widely applied.

In developing methods with ever decreasing LODs, pre-concentration is a major strategy to be employed, and methods based on LLE and SPE are continually being published, as may be readily seen from the contents of Table 1. It should be borne in mind that there are similar numbers of papers appearing describing pre-concentration in methods for the analysis of environmental materials. The recent interest in the possibilities of metal–organic frameworks for SPE has been summarised (157 references with titles).18 The reviewers pointed out the significant potential of these materials because of their porosity, enormous internal surface area, adsorption capacity and stability as well as the extensive range of possible functionalisation. The extent to which MOFs have already been used is clear from a useful summary table that lists work done over the past ten years for foods and the past five years for environmental materials. The authors introduced MOFs, explaining the naming system (based on the institution where the materials were first synthesised) and giving a brief overview of their synthesis. A significant part of the review was devoted to applications in the determination of specified elements, which included individual sections for As, Cd, Co, Cr, Cu, Hg, Ni, Pb and Zn and finally a section covering Ag, Au, Mn, Pd, Sb and Se. It is clear that several of the materials that have been prepared are capable of being separated magnetically from the liquid phase. It appears as though MOFs are not (yet) commercially available, although the reviewers speculated that increasing interest in applications will open the way to “large-scale production of the best performing adsorbents”. As it is not clear how easy they are to prepare for researchers that do not have expertise in organic synthesis, further developments may be relatively slow.

2. Metrology, interlaboratory studies, reference materials and reference ranges

In this section, we cover aspects of the applications of atomic spectrometry to clinical specimens, food and beverages, closely related to the improvement or assessment of the quality of measurement results and their comparability across space and time, that requires their traceability to common established references. Both the development of reference methods, the results of ILCs and the availability of appropriate RMs provide important information to improve or assess the quality of analytical results. Furthermore, in laboratory medicine, the results of analytical tests need to be compared to reference ranges covering both the general population, specific age ranges or physiological state, such as pregnancy, pathological conditions and environmental or occupational exposures. New or revised information in this area of research is also summarised here.

The group of Vogl et al.19 presented an application of atomic spectrometry, to give SI-traceable values to in-house protein standards, through the determination of their S content. The authors remarked on the growing importance of well-characterised protein standards, to support the growing field of proteomics, and lamented the present scarcity of such materials. They explained that characterisation of protein standards could in principle be achieved by other methods, but these generally required an already characterised standard or labelled protein. Traceable protein quantification could also be achieved, after their hydrolysis, using isotopically labelled amino acid standards; however, it was noted that accurate results strongly depended on careful control of the hydrolysis. Over the last twenty years, quantification of proteins through the determination of their S content by ICP-MS has emerged as an alternative, since S is present in 98% of all proteins, as part of the cysteine and methionine amino acids. The authors reviewed the methods of high accuracy already developed for this purpose, based on ID-ICP-MS/MS with S isotopes and sometimes coupled with chromatographic separation, and argued that their complexity limited their applicability. They aimed to develop a simplified procedure, that could be used by any laboratory to assign SI-traceable values to their in-house standards for proteins of known stoichiometry. To minimise S contamination, sample preparation and measurements were carried out in clean rooms or clean cabinets. The S content was quantified by ID-ICP-MS in both the protein solutions and in the corresponding fractions of coexisting small molecular mass S compounds, present as impurities, separated from a different aliquot of the sample by filtration on Amicon Ultra 0.5 centrifugal cellulose membrane filters (cut-off 3 kDa). Protein solutions between 0.3 and 1 mg kg−1 were prepared gravimetrically and spiked with appropriate amounts of a solution of 34S-enriched S to achieve a 32S[thin space (1/6-em)]:[thin space (1/6-em)]34S ratio of 1[thin space (1/6-em)]:[thin space (1/6-em)]1. After digestion in a microwave small vessel system, with 65% HNO3 (0.5–2 mL), the digests were transferred in PFA beakers, dried for 12–15 h at 150 °C on a hot plate, then dissolved in 2% (v/v) HNO3 prior to analysis. Additional sample aliquots (50–100 mg) were weighed onto the filter membrane and centrifuged for 15 min at 14[thin space (1/6-em)]000g and 18 °C. The filtrates were collected and the filters rinsed three times with 5 mmol L−1 NH4HCO3, each time followed by shaking, centrifugation and collection of the rinsing, to be added to the filtrate. After spiking with 34S and mixing overnight, the collected solution was analysed the following day. An alternative separation by gel filtration with a 25 mmol L−1 NaCl mobile phase required additional steps to remove the saline matrix from the eluates. Contamination of the procedural blanks for the two separation methods amounted to 11.1 ± 1.3 ng S for membrane filtration vs. 136–140 ng S for the other, leading to considerably lower LOD values (19.5 ng S vs. 347–393 ng S). The protein mass fractions were calculated from the measured S content, after correction for non-protein bound S, measured in the filtrates, taking into account the protein molecular mass and stoichiometry. The measurement uncertainty of each result was evaluated from the individual contributions of all the input quantities, including that of tabulated values for atomic and molecular weights and the 32S[thin space (1/6-em)]:[thin space (1/6-em)]34S ratio. The test performed on a high-purity protein CRM (NIST SRM 927e, bovine serum albumin solution) with a certified value (±expanded uncertainty, 95%) of 66.2 ± 1.4 g kg−1 yielded an average result of 66.1 ± 2.0 g kg−1 (n = 7), and both the measured value and its uncertainty agreed well with data reported using other high-accuracy methods. For commercial lyophilised avidin, the mass fraction (±expanded uncertainty, 95%) of 866 ± 86 g kg−1 was calculated from the data on purity (≥98%) and protein fraction (80–95%) provided by the manufacturer. Analysis by the proposed method yielded an average result (±expanded uncertainty) of 707 ± 65 g kg−1 (n = 3), thus supporting the need for the laboratory to assess its own protein standards. The authors also stressed the importance to check before hand, by molecular mass spectrometry or SDS-PAGE, the absence of contamination of the protein solution under test with impurities of other proteins. The correction for non-protein bound S amounted to 0.37%, for NIST SRM 927 and 0.29% for avidin. The procedure was applied to the characterisation of commercially available tau protein, to standardise measurements of this biomarker in experimental studies of a group of neurodegenerative diseases. The protein mass fraction (±expanded uncertainty, 95%) was determined as 0.328 ± 0.036 g kg−1vs. the corresponding value of 0.475 ± 0.039 g kg−1 calculated from the manufacturer’s data, comparing well with the value determined by amino acid analysis, via the tyrosine content, as 0.309 ± 0.040 g kg−1. These findings support the reliability of the proposed method, as well as its suitability for the purpose to determine SI-traceable values for in-house protein standards. The expanded uncertainty achieved on commercially available preparations (ca. 10% vs. 3.6% on the CRM) was attributed to the uncertainty associated with the value of the non-protein S fractions, where, due to the low levels measured, the uncertainties of the tabulated data for the isotope ratio and atomic and molecular weights had a higher impact.

The evaluation of laboratory performance in ILCs is an important educational tool and helps harmonise and improve the quality of analytical measurements in a given field of analysis. Within the framework of a EU project for human biomonitoring (HBM4EU), co-funded under Horizon2020, initiatives are in place to support the analytical quality of the measurements carried out toward the assessment of exposure of EU citizens to potentially harmful chemical substances and provide a sound basis for legislative measures. Göen and co-workers have previously reported about Interlaboratory Comparison Investigations (ICIs) among a network of selected European laboratories for the determination of Cd in urine, carried out as part of the quality assurance/quality control programme established under the HBM4EU.1 In a new paper20 they described the results of similar ICIs for biomarkers of exposure to Cr, as one of the substances in the priority list of HBM4EU. They noted that exposure to Cr may occur through the diet (mainly to non-toxic CrIII species), but people can be adversely affected by the presence of CrVI at the workplace, as part of industrial processes, such as electroplating and welding. The determination of total Cr in urine is the most common test used to monitor occupational exposure, although it cannot discriminate between exposure to either Cr species, due to the rapid interconversion of CrVI to CrIII within the human body. Assessment of other biomarkers, such as plasma and, especially, erythrocyte Cr, could provide additional information, as CrVI, but not CrIII, enters the red blood cells. The research group reported the cumulative assessment of four rounds of three ICIs for the determination of Cr in urine, plasma and whole blood, respectively, the last matrix chosen as a surrogate for human erythrocytes, that may better represent exposure to CrVI, but are more difficult to obtain. The organisation of these schemes followed the recommendations of international protocols and standards. Except for the first round, where test materials were prepared from lyophilised RMs (Trace Elements, Seronorm), human urine, bovine blood and bovine serum were spiked with known amounts of Cr to reach concentrations expected for exposed workers, then aliquoted and stored at −18 °C. Homogeneity and stability studies were carried out on all batches. Stability was assessed comparing samples kept at the recommended storage temperature (−18 °C) with those maintained at the reference temperature (−80 °C). Participants (from 8 to 25 in each round) were from different European countries. In each of the four rounds, they received two samples, shipped frozen, at different concentrations for each matrix, to be analysed using their standard analytical procedures. However, after the 2nd round, it was also recommended that the methods applied should achieve LOQs < 1 μg L−1. If a minimum of seven quantitative results were received, the organisers applied robust statistics to obtain the consensus values, their SDs and their uncertainty. Performance was evaluated as z scores, using a fixed value of 25%, as the standard deviation for proficiency assessment, based on an expert opinion of what was technically achievable in routine practices. Over the four rounds of the ICI for Cr in urine, consensus values ranged from 1.10 to 25.55 μg L−1 and the robust RSDs varied between 6 and 9%, except in one case (16%). Almost all participants (92–100%) achieved satisfactory z-scores. For Cr in plasma, with a lower concentration range for the consensus values (2.12–14.49 μg L−1), higher robust RSDs (7–12%, except an extreme value of 18%) were observed and the percentage of laboratories achieving satisfactory performances ranged from 86% to 100%. Finally, in the whole blood test materials, where even lower Cr concentrations were proposed (1.54–7.14 μg L−1), robust RSDs were comprised between 10% and 19%, with extreme values of 4% and 47% observed at the highest and lowest concentrations tested. The percentage of proficient laboratories varied from 70% to 100%. About 90% of the participants used ICP-MS, with the remaining ones applying ETAAS. Not surprisingly, most participants used an IS (from 78.5% to 87%, for the three matrices), whereas sample digestion was applied by 18.5%, 32% and 55% of the participants, respectively, for urine, plasma and whole blood samples, thus confirming the higher difficulties encountered in the analysis of whole blood. Comparisons of the z scores achieved by the participants according to analytical technique or use of an IS did not show significant differences for any of the three matrices, but the prevalence of one technique, that is also the one most commonly requiring an IS, may well hinder such assessment. On the other hand, participants digesting urine or plasma samples achieved significantly worse z scores (p < 0.05) than those who didn’t, but no significant difference was detected in the case of whole blood samples, an effect that may be due to the higher variability among results for this matrix. The researchers observed that the values of the RSDs compared well with those reported in other schemes for routine laboratories, were similar to those achieved by experienced reference laboratories, over similar ranges of concentrations, and were generally lower than the fixed value of 25% applied for proficiency assessment. They concluded that a more challenging approach may be considered to stimulate the continuous improvement of the network. Although this work provides a picture of the interlaboratory reproducibility of measurement results achieved within a network of expert laboratories, even if from limited sets of data, concern remains about whether these exercises may support the metrological comparability of results of the future biomonitoring activities undertaken by the participants, in particular for the lower levels and the more difficult matrices. Metrological comparability is demonstrated when measurement results are traceable to the same reference, e.g., an SI unit, and may be supported by participation in ILCs where the assigned value is itself traceable. However, this is not the case when assigned values are obtained from the consensus of the results provided by the participants. A further reason of concern is that the vast majority of the results are based on the same measurement principle. This may hinder biases associated with the method, for example, the effect of well known interferences associated with the determination of Cr in carbon-rich biological matrices. It may be worth remembering that the certified values for Cd concentration in the whole blood CRMs (BCR194, BCR195 and BCR196) obtained mainly by ETAAS had to be revised after the introduction of ICP-MS. The new technique highlighted uncorrected biases and allowed the application of a primary reference measurement method (ID-ICP-MS) with a lower LOD of 0.0005 ng L−1 Cd to obtain traceable certified values.21 The role of EQA schemes, based on metrologically traceable assigned values, for improving the quality of analytical results was further highlighted by the work of Shin et al.22 at the Chemical Metrology Laboratory of Singapore. This research group considered the importance of addressing the accuracy and variability of measurements of serum Cl, a key biomarker for the clinical assessment of acid–base and electrolyte status and a potential predictor of increased mortality risk, with a narrow reference range (96–106 mmol L−1), due to strict homeostatic control. Therefore, efforts to harmonise the results for serum Cl, put in place since 1963, led over the years to the development of both CRMs (e.g., NIST SRM 909 human serum) and reference methods. Coulometry and ID with TIMS or HR-ICP-MS were recognised as the primary reference measurement procedures for this test and were included in the BIPM JCTLM Database (https://www.jctlmdb.org/). Also, the IFCC developed guidelines for the application of ISE to Cl measurements in biological fluids.23 Quality specifications for analytical measurements in laboratory medicine, linking analytical quality (in terms of precision, bias and total error) to biological variation have been developed,24 and are now regularly updated at https://biologicalvariation.eu.25 The current estimates of within-subject and between-subject variability for serum Cl were 1.1% and 1.3% (as of 31.12.2022). From these values, minimum, desirable and optimal requirements for the allowable total error of measurement results (comparable to allowed expanded uncertainty) were calculated as 2.0%, 1.3% and 0.7%. Notwithstanding these efforts, the authors noted that there is still poor comparability among results obtained in routine laboratories using different analytical techniques (mercurimetric titration, spectrophotometry, coulometric–amperometric titration and ISE), leading to the need for each laboratory to establish its own reference ranges. To overcome this problem, they considered the need to improve laboratory performance through regularparticipation in EQA schemes with metrologically traceable assigned values, that would highlight biases and provide support for their correction. The primary reference measurement method used to assign values to the EQA samples was based on ID-SF-ICP-MS, using 37Cl and a complex sample preparation involving protein precipitation and separation of Cl from the remaining matrix by precipitation as AgCl. To avoid contamination, all experimental work was carried out in Class 100 low laminar flow fume hoods or in a Class 10000 clean room. Sample and calibration blends were prepared gravimetrically by spiking 0.3 g of either human serum or the Cl standard solution, prepared from NIST SRM 919b (sodium chloride), with appropriate amounts of the enriched 37Cl solution, to an approximate 37Cl[thin space (1/6-em)]:[thin space (1/6-em)]35Cl ratio of 1[thin space (1/6-em)]:[thin space (1/6-em)]1. After overnight equilibration, the serum blends were diluted to 10 mL with 1% HNO3 followed by addition of 1 mL ammonium molybdate (50 g L−1) and left to rest for 10 min, prior to two precipitation steps. The AgCl precipitate was re-dissolved with 1 mL of NH4OH (32% v/v), then diluted with 50 mL of ultrapure water. Both a reagent blank and a solution prepared from NIST SRM 975a (isotope standard for Cl) were analysed at the beginning and at the end of each batch in the analytical sequence, to check the blank contribution and allow for correction of mass bias. The effect of instrumental drift and imprecision was reduced by applying the bracketing technique for calibration. An LOQ of 2.2 mmol L−1 was achieved. Analysis of three CRMs for electrolytes in frozen human serum (NIST SRM 956c, levels 2 and 3; 956d, level 1) provided results within the expanded uncertainty of the certified values. Namely, the measured vs. certified values were: 122.2 ± 0.4 mmol L−1vs. 121.5 ± 2.5 mmol L−1; 137.3 ± 0.4 mmol L−1vs. 137.4 ± 1.8 mmol L−1; 94.3 ± 0.2 mmol L−1vs. 94.5 ± 0.21 mmol L−1. The relative expanded uncertainties (0.21–0.33%) were well below the optimal quality specification for total error (0.7%) for the measurements of Cl in serum. These findings were confirmed by the successful participation in a key-comparison (CCQM-K139 Elements in human serum), that is an international ILC for national metrology institutes. The eight participants used either IC, titrator, microcoulometry, ICP-MS, ICP-QQQ-MS or SF-ICP-MS. The result obtained with this method fell within the stated expanded uncertainty of the key-comparison reference value of 3871 ± 51 mg kg−1 (approx. 109.2 ± 1.4 mmol L−1). Between 2014 and 2018, the Singapore Health Sciences Authority carried out ten EQA exercises, with 24–33 participants, almost all of them applying ISE, on instruments from different manufacturers, but one using FAAS. Each EQA included one sample. In each year, both a low and a high concentration sample were distributed. The described method was applied to assign values to the EQA samples. Three bottles were randomly selected from each batch and two subsamples from each bottle were analysed. After testing for absence of outliers (Grubbs and Dixon tests), the mean of the six results was taken as the assigned value and the associated expanded uncertainty was evaluated according to the ISO/IEC Guide 98. The assigned values were in the range 88.84–92.24 mmol L−1 and 110.61–125.74 mmol L−1, respectively. The expanded uncertainties of the assigned values ranged from 0.20 to 1.13 mmol L−1, corresponding in relative terms to an average of 0.46%. These values satisfied the requirement for the uncertainty of the values assigned to EQA samples, that is less than one third of the standard deviation for proficiency assessment, set, in this EQA scheme, as ±3 mmol L−1 at levels ≤100 mmol L−1 and as ±3% for higher concentrations. For each exercise, robust statistics were applied to the participants’ results to obtain consensus values for comparison with those assigned by ID-SF-ICP-MS and robust SDs as an estimate of the interlaboratory reproducibility. Robust SDs (1.73–3.03 mmol L−1) were always within the standard deviation for proficiency assessment, for the 10 exercises, but relative deviations between consensus and traceable assigned values ranged from −3.44% to 1.15%. However, starting from the 8th cycle, the observed deviations started to decrease (0.35%, 0.25%, −0.22%) and the consensus values fell within the limits of the expanded uncertainty of the corresponding assigned values. The availability of independent assigned values also allowed identification of potential biases, via regression analysis of participants’ results against assigned values, by instrument manufacturer, although on limited sets of data, and this information, reported to participants, played a role to support corrective actions to reduce bias. Overall, these findings fully support the importance of traceable assigned values in EQA schemes to improve the comparability of measurement results.

Certified reference materials in suitable matrices are acknowledged as essential tools for the validation of analytical procedures, thus assuring the reliability and comparability of measurement results. Not surprisingly, over the period covered by this review, work continued to produce new CRMs, improve their features or further characterise existing ones. To this aim, two research groups implemented primary reference measurement procedures, “to obtain a measurement result without relation to a measurement standard for a quantity of the same kind”,26 thus linking directly the result to the SI. Baharom and Yim27 applied double ID-ICP-MS for the certification of the mass fractions of Ca, K and Mg in a new frozen human plasma CRM produced at the Korea Research Institute of Standards and Science (KRISS). The mechanism of homeostasis strictly controls the physiological levels of electrolytes in body fluids, thus leading to narrow reference ranges, commonly reported, for Ca, K and Mg, respectively, as 1.1–1.3 mmol L−1 (44–52 mg L−1), 3.5–5.0 mmol L−1 (137–196 mg L−1) and 0.6–1.07 mmol L−1 (15–26 mg L−1). Even small variations outside those levels are likely to be associated with a pathological condition, including malignant processes, parathyroid dysfunction, renal impairment or nutritional imbalances. Therefore, assuring the reliability of measurement results and limiting their uncertainty is of critical importance. Current desirable requirements for the allowable total error (comparable to allowed expanded uncertainty) are quoted in the EFLM Biological Variation Database25 as 2.3% (Ca), 4.8% (K) and 4.0% (Mg). In this work, twelve samples, randomly chosen from the batch of candidate RMs, were used for certification and homogeneity studies. Measurements were carried out by SF-ICP-MS. For the determination of the 42Ca[thin space (1/6-em)]:[thin space (1/6-em)]44Ca and 25Mg[thin space (1/6-em)]:[thin space (1/6-em)]24Mg isotope ratios, the instrument was operated at medium resolution, using a standard cyclonic spray chamber with a MicroMist concentric nebuliser, whereas the determination of the 41K[thin space (1/6-em)]:[thin space (1/6-em)]39K isotope ratio required an Aridus II membrane desolvator (CETAC Technologies, Omaha, USA), a microflow nebuliser and operation in HR mode. The sample blend was prepared gravimetrically, by spiking an amount of about 0.5 g of each CRM, and placed in a modified PTFE microwave digestion vessel with an appropriate amount of a solution of the enriched isotopes (42Ca, 41K and 25Mg), designed to obtain optimal isotope ratios for all three elements. After addition of 6 mL of sub-boiled HNO3, the sample blends and a procedural blank were left to equilibrate for 12 h at room temperature, then 2 mL H2O2 was added, and, after a further 30 min rest, MAD was performed at 180 °C and 990 W. The digests were diluted with deionised water, to obtain a HNO3 concentration of about 1 M. To obtain a calibration blend, a solution of the elements of interest was spiked gravimetrically with the same solution of the enriched isotopes used for the sample blend. Four calibration blends were prepared independently, using two different sets of standard solutions for Ca, K and Mg, traceable to the SI, prepared at KRISS as primary reference materials. After correction for mass discrimination and drift, the mass fraction of each element was calculated and the associated characterisation uncertainties were evaluated from the individual contributions of the input quantities, according to the uncertainty propagation law, as 0.32 mg kg−1 (Ca), 0.23 mg kg−1 (K) and 0.05 mg kg−1 (Mg). Uncertainty contributions due to inhomogeneity and calibration were included in the evaluation of the overall uncertainty of the CRM values, whereas the study of both short stability (at room temperature for 7 days) and long-term stability (at −70 °C for up to 21 months) indicated both contributions as negligible. The certified values and associated expanded uncertainties for the frozen plasma CRM were: 74.77 ± 0.90 (1.2%) mg kg−1 (Ca), 125.1 ± 5.4 (4.3%) mg kg−1 (K) and 15.95 ± 0.14 (0.9%) mg kg−1 (Mg), respectively. The authors highlighted the challenges associated with the determination of K isotope ratios, due to significant polyatomic interferences from 39Ar1H and 40Ar1H, whose residual effect determined a larger uncertainty as compared with those associated with Ca and Mg measurements. When compared to the desirable quality specifications for total error for these measurands, the CRM qualifies well as a tool to demonstrate the traceability of plasma Ca and Mg measurement results, in the processes of method validation and/or assessment, and provides a traceable reference also for plasma K determinations.

Over the years, the MODAS Consortium has been committed to producing CRMs of interest for the analysis of trace elements in food and environmental samples, according to the requirements of the ISO Guide 35 and ISO 17034, that were characterised in worldwide ILCs. Aiming to improve the traceability and reliability of the certified values, the research group28 decided to use results from reference measurement procedures (NAA for As, Cd, Co, Cr, Fe, Mo, Se and U; ID-ICP-MS for Hg) to obtain additional independent values for three new CRMs, based on herring, cormorant and cod tissue, respectively. The preparation of the candidate CRMs included freeze-drying of the tissues, followed by grinding and sieving through a 200 μm sieve. The resulting material was then homogenised in a polyethylene drum, allowed to rotate in three different directions for 16 h. Aliquots of 50 g were transferred to 100 cm3 amber glass bottles and sterilised by electron beam radiation energy (27 kGy). Aliquots of 10 g in 60 cm3 bottles were also prepared for the ILC. The homogeneity study was carried out determining the mass fractions of selected elements (Cd, Mg, Mn, Sr, V and Zn) on 11 bottles of each CRM, taking two subsamples from each. Samples (100 mg) were digested in PTFE vessels, with 6 mL conc. HNO3 and 2 mL 40% HF, for the complete dissolution of silica, using a multiwave 3000 high pressure microwave system (Anton Parr GmbH). After digestion, 6 mL H3BO3 was added and each vessel resealed, to allow complexation of the excess of F ions. The digests were diluted with 2% HNO3 and 5 ng mL−1 of In, as the IS, prior to analysis by ICP-MS. The standard uncertainty due to inhomogeneity, evaluated by means of ANOVA, was between 1.80% and 2.00% for the three CRMs. A long-term stability study was carried out over a period of 15 months at room temperature and resulted in an estimated shelf-life of 5 years, if stored protected from light in a tightly closed container, with an uncertainty contribution ranging from 1.07% to 1.13% for the three CRMs. The short-term stability study, carried out at 37 °C for 2 months, did not show significant variations of the mass fractions of the selected elements. The ILC involved 50 participants from 11 countries, using methods of their choice (AAS, AES, ICP-MS, NAA, XRF and others), simultaneously analysing CRMs of similar matrices along with the candidate ones, to document traceability to the SI. Results obtained with primary reference measurement procedures were not included in the evaluation of data, but used for independent assessment of the certified values. These were assigned as the overall mean of laboratory averages, after outliers’ rejection, provided that the dataset still met the quality criteria for analytical uncertainty, number of results and number of analytical techniques. Certified values were obtained for 6 major and minor elements (Cl, K, Mg, Na, P and S) and up to 24 trace elements (Ag, As, Ba, Br, Cd, Co, Cr, Cs, Cu, Fe, Hg, Li, Mn, Mo, Ni, Pb, Rb, Sb, Sc, Se, Sr, U, V and Zn) in at least one of the three CRMs. The associated uncertainties were evaluated combining the characterisation contribution, as the SD of the overall mean from the ILC, the inhomogeneity and long-term stability contributions and the standard uncertainty due to moisture determination. Relative expanded uncertainties, at 95% confidence level, ranged from 5.9% to 17.2% for the major and minor elements, from 5.4% to 12.8% for trace elements between 1 and 1000 mg kg−1 and from 7.1% to 17.9% for trace elements below 1 mg kg−1. Up to 18 traceable values were determined by primary reference measurement procedures for As, Cd, Co, Fe, Cr, Hg, Mo, Se and U, in at least one of the three CRMs, with relative expanded uncertainties between 2.2% and 12.3%. For all of them, the intervals value ± expanded uncertainty fell within the ILC certified ranges, except in only one case (Co in cormorant tissue), thus confirming the agreement between the two sets of values. The authors also explored the contribution of analytical techniques based on different measurement principles to certification campaigns carried out by MODAS over the years. They noted that, whereas ICP-MS was increasingly used, the application of NAA, the only technique for inorganic analysis essentially free from blank, was significantly reduced. Concern arises that both these trends may affect the reliability of values obtained from certification campaigns based only on ILCs, as the vast prevalence of only one measurement principle may hinder systematic errors associated with certain analyte/matrix combinations. Therefore, reference material producers should clearly address potential biases due to the certification methods in the information provided to their customers. Users also should consider such information to assess the suitability of the CRM for their specific intended use.

The production and maintenance of CRMs requires a lot of effort. Therefore, their supply is still limited and often costly. On the other hand, there is a growing demand from the analytical community for matrix-matched CRMs covering a larger number of properties. Three papers in this year’s review addressed work carried out on existing CRMs to extend the range of elements with certified values for their mass fractions or isotopic ratios. Safe drinking water is a necessity for human health. In most countries, criteria for its quality, including maximum allowable concentrations for a number of chemical elements, are set in legislation and regular monitoring of such parameters is required. Yeghicheyan and co-workers29 considered that, due to changes over the years, in both anthropogenic activities and climate, concern may arise about the levels of other elements (e.g., REEs) in drinking water and also that the determination of a large number of elements at low concentrations is a considerable challenge for any laboratory. They pointed out that few of the existing CRMs for drinking water are based on a real matrix, thus implying that, for some elements, especially new ones, results obtained on synthetic CRMs may not be representative of the actual analytical performance on real test samples, given the complexity and variability of real drinking water samples. Aiming to optimise the existing resources to support the traceability and reliability of laboratories’ measurement results, work was undertaken by the Isotrace CNRS workgroup, in collaboration with the NRCC (the Canada NMI), to propose reference values, for both mass fractions for additional trace elements and some isotope ratios, in the existing NRCC RM AQUA-1 (natural drinking water), released in 2017. Twelve laboratories participated in this exercise, using ICP-AES, ICP-MS and HR-ICP-MS for the determination of mass fractions for over 60 elements, with ICP-MS being the most frequently applied technique. Isotope ratios for Pb were determined by MC-ICP-MS, in one laboratory; those for Sr by two participants using MC-ICP-MS and TIMS, respectively. All participants analysed simultaneously another CRM (SLRS-6 river water) as quality control. The two materials had a very similar composition, as they were prepared from water from the same location, before (river water) and after (drinking water) the water treatment facility. The analytical performance of the participants was evaluated from the results (as laboratory means) obtained on both AQUA-1 and SRLS-6, in comparison with the certified values, with acceptability limits set as 2 SD of the overall mean. Results achieving z scores > |2| were removed, representing 30% and 15% of the datasets for AQUA-1 and SRLS-6, respectively. Robust statistics, weighted according to the number of measurements, were applied to determine the mass fractions for each element and its expanded uncertainty, only when valid results were reported by at least three laboratories. These values were found in agreement, within the respective uncertainties, with the certified ones, thus providing evidence of the participants’ capabilities. Notwithstanding the inherent difficulties of the reliable determination of low concentrations of trace elements, the authors were able to report new consensus values and associated expanded uncertainties, calculated with the same approach, for the mass fractions of REEs (except Eu) and seven other elements (B, Cs, Li, Rb, S, Si and Y) in AQUA-1, as well as indicative values for other elements. For REEs, the proposed values ranged from 0.43 ng L−1 (Lu) to 61 ng L−1 (Ce), with expanded uncertainty between 2.5% and 33%. For other elements, the assigned values varied from 0.0032 ± 0.0002 μg L−1 (Cs) to 8090 ± 220 μg L−1 (S). Isotope ratios were reported by only one (Pb) and two (Sr) laboratories and their values were 0.712845 for 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr, as the average of the two reported results) and 15.535 and 0.899 for 207Pb[thin space (1/6-em)]:[thin space (1/6-em)]204Pb and 207Pb[thin space (1/6-em)]:[thin space (1/6-em)]206Pb, respectively.

The natural isotopic composition of Li has been largely investigated in mineral materials, mainly linked to the assessment of geological and environmental processes. Although this element is widely used in various areas of human activities, including pharmaceutical applications, limited attention has been paid to its presence and isotopic variations in biological samples. To support further investigations in this area, Thibon et al.30 decided to investigate the Li isotopic composition of six RMs of biological origin: lobster hepatopancreas (TORT-2 and TORT-3), dogfish muscle (DORM-2 and DORM-4), mussel tissue (ERM-CE278k), and bone ash (SRM-1400). These complex matrices, together with low Li concentrations, posed an analytical challenge that was met by applying an extensive sample pretreatment, based on digestion and chemical separation, as well as measurement by high transmission MC-ICP-MS. To minimise procedural blanks, sample preparation was carried out in a pressurised clean laboratory, using ultrapure grade reagents. Samples, available as dried fine powder, were digested at room temperature with conc. HNO3–H2O2, followed, after 48 h, by addition of conc. HCl, then heated for 24 h before being evaporated and this second step repeated. After removal of the remaining NH4+, by leaving the samples at T > 100 °C for a few hours, 0.5 mL of 1.00 ± 0.05 mol L−1 HCl was added to dissolve the residue. Applying previously described procedures, chemical separation was performed by loading the samples on 2.7 mL of AG50-X12 (200–400 mesh) cation resin, preconditioned with 8 mL of 1.0 mol L−1 HCl, then eluting the Li with 6 mL of 1.0 mol L−1 HCl and removing the remaining matrix with H2O and 6 mol L−1 HCl. The process was repeated, then the eluted fraction was dried, before measurement of the isotope ratio. The MC-ICP-MS was operated in low resolution mode, using an Aridus II desolvating system (CETAC Technologies, Omaha, USA) to obtain dry plasma conditions, thus enhancing the sensitivity to about 0.8 V 7Li per 1 ppb Li and enabling the determination of isotope ratios at the low Li levels observed in biological materials. The intensities of 6Li and 7Li were measured simultaneously on two different collectors. Notwithstanding the complexity of the procedure, the authors reported procedural blanks ranging between 5 and 9 pg, thus amounting to a maximum blank contribution of 0.03%, for total Li, and 0.24‰ for the isotope ratios, that were considered acceptable. The measured level of Li (1.23 ± 0.06 μg g−1) closely matched the certified value in DORM-4 (1.21 ± 0.09 μg g−1), the only biological CRM for Li. Since no other biological CRM for the isotope composition is available, the authors assessed the long-term reproducibility of the 7Li reference solution in different analytical sessions over a year and the yield of the chemical separation in a Na-rich matrix by repeated analysis of a reference sea-water solution. They reported δ7Li values of 30.14 ± 0.18‰ (2 SD, n = 20) for the standard reference solution and of 31.32 ± 0.83‰ for the seawater reference solution, both in close agreement with the corresponding data reported in the literature (30.2 ± 0.3‰ and 30.31 ± 0.17‰; 31.2 ± 0.3‰ open ocean). Additionally, they further assessed the validity of the proposed procedure for the determination of the level of Li and Li isotope ratio by applying an alternative calibration approach (standard addition method) to DORM-2, for which results agreed well with those obtained with the standard approach. Lithium concentrations in the analysed RMs ranged from 0.05 ± 0.01 μg g−1 (TORT-2) to 1.23 ± 0.06 μg g−1 (DORM-4), whereas the δ7Li values observed in the six soft tissue RMs varied between 14.69 ± 0.35‰ (ERM-CE278k) and 29.17 ± 0.24‰ (DORM-4), but was −1.72 ± 0.48‰ for the bone ash RM (SRM 1400). Similarly, another group31 investigated the Cu, Pb and Zn isotope composition in commercially available RMs, to assist measurements aimed to contribute to the understanding of geological and biological processes. Among these RMs, 10 were of biological origin (mussel: SRM 2976 and ERM-CE 278; oyster: SRM1566b; fish protein; DORM-4; dogfish liver: DOLT-4; plankton: BCR-414; aquatic plant: BCR-670; sea lettuce: BCR-279; tomato leaves: SRM1573a; lobster: TORT-2). The presence of high concentrations of Ca, Mg and Na, typical of biological matrices, hinders the accurate determination of Cu and Zn isotopes, due to polyatomic interferences affecting the mass range from 63 to 70. For this reason, the authors applied an anion exchange resin (AG-MP1, 100–200 mesh), for Cu and Zn, and a Pb-specific resin (Eichrom Technologies, USA) to remove the matrix and isolate the analytes. Reference Material aliquots (0.1 g) were digested with HNO3–HF–HClO4 (4 + 5 + 1) in closed PTFE vessels on a hot plate and the residue was dissolved with 3% HNO3. The concentrations of Cu, Pb and Zn in the digested samples was measured by ICP-MS and aliquots containing approximately 600, 300 and 1200 ng of Cu, Pb and Zn, respectively, were evaporated at 90 °C on a hot plate. For Cu and Zn determination, the residue was re-dissolved in 7 mol L−1 HCl–0.001% H2O2 and 0.9 mL were loaded onto the column. After removal of the matrix, three fractions were eluted, with 19 mL of 7 mol L−1 HCl–0.001 H2O2%, 16 mL of 1 mol L−1 HCl–0.001% H2O2 and 10 mL of 0.5 mol L−1 HNO3, containing, respectively, Cu, Fe and Zn. To optimise the separation of Cu from interfering ions, the authors introduced an additional purification step: the Cu fraction was evaporated again at 90 °C for 48 h, re-dissolved as before and 0.9 mL were loaded on a second AG-MP1 column, to obtain the sample for analysis. The process allowed optimal recovery of both Cu (99.6%) and Zn (99.7%), whereas main interfering ions were not detected in the eluted fractions and there was no evidence of isotope fractionation occurring during the sample preparation. The authors highlighted the importance to collect the right fraction volumes, as the variable composition of the matrices may affect elution times and the recovery of the analytes of interest. On the other hand, the Pb fraction was obtained by loading 1.0 mL of the residue, re-dissolved in 2 mol L−1 HCl, onto the Eichrom column, followed by removal of the matrix with 2 mol L−1 HCl and elution of Pb with 6 mL of 7 mol L−1 HCl. Recovery after the separation process, evaluated from eight RMs, was 99.2%. Comparison of the 208Pb[thin space (1/6-em)]:[thin space (1/6-em)]206Pb and 207Pb[thin space (1/6-em)]:[thin space (1/6-em)]206Pb values of the reference solution NIST SRM 981 analysed with or without the chemical separation step did not show evidence of isotope fractionation. Measurements for all three elements were carried out by MC-ICP-MS, using standard solutions certified for their isotope composition (ERM-AE647, NIST SRM 981 and IRMM-3702) for mass discrimination correction. Bias and repeatability (2 SD) for Cu, determined over a period of 12 months using the CRM ERM-AE647 and CRM ERM-AE633, were δ65CuAE647: 0.00 ± 0.02‰ (n = 54) and δ65CuAE647: −0.21 ± 0.03‰ (n = 33). The corresponding values for Zn, determined over a period of 6 months using the CRM IRMM-3702 and CRM IRMM-651, were δ66Zn3702: 0.00 ± 0.02‰ (n = 15) and δ66Zn3702: −11.6 ± 0.03‰ (n = 8). Measurements of the 207Pb[thin space (1/6-em)]:[thin space (1/6-em)]206Pb and 208Pb[thin space (1/6-em)]:[thin space (1/6-em)]206Pb ratios in NIST SRM 981 yielded the values (±2 SD, n = 36) of 0.9145 ± 0.0003 and 2.1661 ± 0.0004 (2 SD, n = 36), respectively, reported as being in good agreement with other literature values. Copper, Pb and Zn isotope data in RMs were obtained from four replicate analysis of each material. In biological RMs, Pb levels were too low to allow reliable determinations of the isotope ratios, thus further improvement of this method may be necessary. To allow comparison of δ values for Cu and Zn with other literature data obtained with previously available reference standard solutions (NIST SRM 976 and JMC 3-0749), the researchers applied established conversion factors to the δ65CuAE647 and δ66Zn3702 values and reported them as δ65CuNIST976 and δ66ZnJMC. The δ65CuNIST976 values ranged from −0.19‰ to 0.70‰, with SDs between 0.01‰ and 0.04‰. Those for δ66ZnJMC varied between −0.96‰ and 0.78‰, with SDs ranging from 0.01‰ to 0.03‰. For some RMs isotope ratios were determined for the first time.

The drive towards more stratified reference ranges has continued this year, with three publications reporting on elemental concentrations in different age groups. A large national survey of 1099 Swedish adolescents, stratified into three age groups (mean age of 12, 15 and 18 years), demonstrated sex and age-related differences in whole blood toxic metal levels.32 Concentrations of Cd, total Hg and Pb in blood were measured by ICP-MS and interrogated in relation to food intake, sociodemographic characteristics and population-based toxicity reference values. Blood Pb levels above 12 μg L−1, the reference point for neurotoxicity set out by the European Food Safety Authority, were identified in 13% of the study participants. Luan et al.33 reported age-related reference intervals for nutritional elements in 589 apparently healthy children and adolescents from the Shandong Province, East China. Whole blood samples were collected and analysed for Ca, Cu, Fe, Mg, Mn, Se, Sr and Zn by ICP-MS and results were stratified into five age categories. Whilst the authors hoped the described ranges may be useful for clinical assessment, serum rather than whole blood analysis is more common for most of these elements and the study is limited by low sample numbers in some age groups. Lastly, toxic (As, Cd, Hg and Pb) and essential (Co, Cr, Mn, Mo, Ni and Se) metals were measured by ICP-MS in blood and urine samples collected from participants in a Healthy Aging and Biomarkers Cohort Study in China.34 This study has presented the largest documented dataset for the younger elderly aged 65–79 years (n = 932), octogenarians (n = 643), nonagenarians (n = 540) and centenarians (n = 386) and individual elements were given importance ratings for longevity.

Jayawardene et al.35 presented the first national human biomonitoring dataset for rare earth element concentrations in a nationally representative sample of Canadians. Whole blood samples (n = 5700) were obtained from the Canadian Health Measures Survey biobank and analysed for Al, Bi, Ce, Cr, Ge, La, Li, Nd, Pr, Te, Ti and Y using ICP-MS. Although this work has produced some baseline human biomonitoring data, the authors acknowledged limitations associated with the lack of CRMs for all the elements studied. They also suggested smaller targeted analysis for Al and Ti, due to the analytical challenges presented by environmental Al contamination and whole blood matrix interference for Ti measurements. Reference ranges for noble elements and REEs were also reported for the first time this year in umbilical cord blood and plasma samples from 125 Caucasian pregnant women in Belgrade.36 In contrast to the Canadian study, Stojsavljevic et al. used individual CRMs to validate measurements for each of the 38 elements measured by ICP-MS, using the standard addition method. Whilst the authors looked at age-related trends, conclusions are limited by the small cohort size.

3. Sample collection and preparation

3.1 Collection, storage and preliminary preparation

Strategies for the microsampling of biological fluids for elemental and isotopic analysis by ICP-MS have been reviewed.37 The reviewers divided samples into those requiring invasive procedures (blood, CSF, intraocular fluids, foetal urine, follicular, amniotic, peritoneal and synovial fluids) and those allowing non-invasive collection (urine, saliva, lacrimal fluid, nasal exudate, seminal fluid, breast milk and sweat). For each type of sample, the various (limited) options for collection were discussed together with the nature of the elemental information that can be obtained and how it may be related to various diseases. Some emphasis was given to the microsampling of blood: both dried blood spots and volumetric absorptive sampling were discussed. The review contains a useful table summarising isotopic analyses of human blood by MC-ICP-MS, to assess differences between diseased people and controls. The various options for microsample introduction to ICP-MS instruments, such as micro- and direct injection nebulisers, ETV and LA, were also discussed.

In a review of methods for the determination of metallic nanoparticles in biological samples by spICP-MS, the authors devoted a section to sample collection, storage and shelf-life.11 The reviewers cited sources showing that, in some cases, NPs could not be recovered after prolonged storage at −80 °C because of dissolution, chemical transformation and agglomeration/aggregation. The reviewers considered that storage conditions have not been systematically investigated for a range of ENMs and biological matrices, and so guidance on storing legal samples for forensic or regulatory purposes is some way off. Furthermore, as clinical samples may be stored for weeks or even years, depending on the type of study, it would be prudent to include spiked controls with stored samples.

In a study of the effects of two different anticoagulants (EDTA and heparin) and long-term frozen storage, metal concentrations in whole blood, blood cells and plasma from 22 healthy participants were measured by ICP-MS after 18 months storage at −80 °C.38 Of the 22 elements determined (Al, As, Ba, Ca, Cd, Co, Cr, Cu, Fe, Mg, Mn, Mo, Ni, Pb, Rb, Sb, Se, Sn, Sr, Ti, V and Zn), only the concentrations of Ba and Sb in all three sample types were significantly different for both anticoagulants. With EDTA, the Ni and Ti concentrations in blood cells changed significantly; and with heparin, the concentrations of Mo and Ni in blood cells, and of Sb in plasma, were also altered. The researchers concluded that Sb concentrations in peripheral blood should be preserved with EDTA, and heparin should be used in the determination of Ba. Additionally, heparin was more suitable in the determination of multiple metal concentrations in whole blood; whereas for blood cells and plasma, either EDTA or heparin could be used. Tanvir et al.39 also investigated the effects of storage temperature on the concentrations of trace elements (As, Bi, Br, Cd, Co, Cr, Cu, Hg, I, Mn, Mo, Ni, Pb, Sb, Se, Tl, V and Zn) in 0.1 mL samples of whole blood and plasma. They found that refrigeration (4 °C) and freezing (−20 °C) of both blood and plasma specimens were suitable storage conditions for many of the trace elements for periods up to six months, and that consistent recoveries were obtained by preserving specimens at −20 °C for up to one year. However, freezing at −80 °C did not improve stability but resulted in adsorption and/or precipitation of some elements. Samples frozen at this temperature required a significantly longer thawing time.

The potential for contamination of specimens by Zn leached from powder-free, single-use gloves has been clearly demonstrated.40 Release from three types of gloves—nitrile, vinyl and latex—were studied by (a) incubating a 1 cm square of fingertip material in serum for 10 min, (b) running serum from a pipette over the glove for 3 cm prior to collection and (c) touching a block of liver tissue once on one side producing an experimental fingerprint (latex gloves only). All the materials released significant amounts of Zn, as measured by ICP-MS, depending on the material, with release from latex being the highest. Increases in the Zn concentrations by double-, triple- and even quadruple-digit μg L−1 or kg−1 in the samples were measured. The authors concluded, in line with previously published advice, that researchers should use clean handling tools, rather than gloved hands, for all pre-analytical specimen-handling in veterinary and human medicine as well as for scientific studies, if this would be in accordance with hygiene requirements.

Ensuring the integrity of tissue samples during transportation to a laboratory for LA-ICP-MS imaging can be challenging. A new sample preparation protocol involving drying at room temperature followed by vacuum packaging has been developed.41 The researchers demonstrated that the approach overcame the disadvantages of commonly used procedures that either require elaborative storage and transport in frozen conditions or fixation in formalin or combined formalin fixation and paraffin-embedding. In particular, they showed that with the latter fixation protocols, significant losses of target elements occurred.

Hauser et al.42 showed that problems of representative sampling, which may be encountered in the determination of trace elements in the placenta, could be mitigated by slurry sampling for TXRF analysis. In the optimised protocol, 10 mg of dried, powdered placental tissue was mixed with 1 mL HNO3 by vortexing (30 s at 2500 rpm) and ultrasonication (30 min at 40 °C). The concentrations determined for major, minor, and trace elements were not significantly different from those determined (also by TXRF) following MAD with HNO3 and H2O2. Three elements, Ca, Fe and Zn, were also determined in the digests by ICP-OES as further validation. The researchers also studied the effects of fixation time and of sampling location (maternal vs. foetal side of the placenta) and found significant differences for foetal placenta tissue compared to maternal or intermediate tissue, showing accumulation of trace elements in the foetal side of the placenta. They also observed depletion up to 60% in concentrations with longer fixation times for foetal placenta tissue. They proposed that these effects were responsible for the large ranges of concentrations reported in the literature and indicated that further systematic investigation of non-homogeneous element distribution in placenta was needed.

García-Poyo et al.43 discussed the advantages and problems associated with the use of dried blood spots. They evaluated four commercially available systems, designed to produce dried blood spots of well-defined volumes, for the determination of Cu by HR-CS-ETAAS and showed, by comparison of the results with those obtained by venipuncture, that accurate results were obtained for all four devices (Mitra, HemaXis DB10, Capitainer qDBS and HemaPEN) with external calibration with aqueous standards. The devices differed in the contribution to the method blanks: HemaXis DB10 and HemaPEN would be preferred when low Cu concentrations (500 μg L−1 or lower), associated with some diseases, needed to be determined.

3.2 Digestion, extraction and pre-concentration

For the determination of As species in whole blood by HPLC-ICP-MS, Zhang et al.44 constructed a new sample preparation device based on magnet-assisted dispersive extraction and ultrasonic spray separation. In the first stage of the procedure, 100 μL of blood, 10 μL of 0.2% (v/v) Triton X-100 (cell disruption agent), and 100 μL of 12 g L−1N-ethylmaleimide (thiol masking agent) were added to a 2 mL centrifuge tube together with a magnetic stir bar and placed in a suitably modified fidget spinner (supplier not given) that had been mounted on a base to which several magnets had been strategically placed so that when flicked once with a finger, the magnetic stir bar rotated for about 5 s inside the centrifuge tube under the action of the “relatively changing magnetic force, forming a crushing mechanical movement”. After standing for 5 min, 60 μL of 15% (v/v) HClO4 (for protein precipitation) was added and the fidget spinner flicked a further five times (about 30 s). A second tube was mounted on top such that when the two tubes were inverted, the solution containing the arsenic species would pass through hydrophilic (F319-04, Whatman) and polyethersulfone (10 kDa, Pall Corporation) filters onto the spray sheet of a mesh-type piezoelectric ultrasonic transducer, consisting of a piezoelectric ceramic ring and a concentric stainless-steel substrate, and be nebulised into the other centrifuge tube by the high-frequency oscillation (110 kHz) of the ultrasonic spray sheet. The method was validated by spike additions to both human and rabbit blood and by the analysis of CRM GBW 09115 (freeze-dried human urine). The LODs were around 0.02 μg L−1.

To extract Ca, Mg, Na and Zn from cocoa butter, an emulsion breaking method was devised,45 for which the values of relevant parameters were optimised by response surface methodology with Box–Behnken design. Emulsions were formed by mixing 1.0 g sample (at 35 °C), 100 μL of xylene, and 1.0 mL of 0.7% (w/v) Triton X-114 in 1.8 mol L−1 HNO3. On heating to 96 ± 2 °C in a water bath (7 min) and centrifugation (5 min), phase separation occurred with deposition of the enriched aqueous phase at the bottom of the tube. The organic phase was removed, and 100 μL of the enriched aqueous phase taken for the determination of each element by HR-CS-FAAS (FAES in the case of Na). The LODs were 0.2, 0.1, 0.1, and 0.02, mg kg−1, respectively. The method was validated by spike recoveries (93 to 113%) from 3 real samples and by comparison of the results with those of a “calcination reference method” (a paired t-test showed no significant difference). All four elements were found in six real samples. A similar approach was taken46 for the determination of Ca, Fe, Mg, and Zn in edible oil samples, but this time the values of relevant parameters were optimised by constrained mixture and Doehlert designs. The emulsions were formed with the aid of ultrasound after the addition of 2.5 mL of 3.0 mol L−1 HNO3, and 2.0 mL of 1.0% Triton X-114 to 5.5 mL of sample. After heating to break the emulsion, the aqueous phase was diluted to 10 mL and analysed by HR-CS-FAAS. The method was validated by spike recoveries, though no details were given other than the range (87–113%). The slopes of the external calibration plots were not significantly different from those of standard additions plots. The LODs were 4, 8, 11 and 8 μg L−1, respectively and the procedure was applied to the analysis of soybean, canola, corn, sunflower, cottonseed, coconut, and olive oils; of the 28 samples analysed, only two coconut oils contained measurable concentrations of all four elements.

Several reviews of pre-concentration procedures have been published in the past 12 months. The possibilities for supramolecular solvent-based microextraction techniques were discussed7 in a review (113 references) that included examples of the determination of PTEs in a variety of matrices, including water, foodstuffs and several clinical materials (blood, urine and hair) by atomic spectrometry. The status of the applications of MOFs in the determination of PTE in foods and environmental materials was reviewed (157 references) by Gumus and Soylak.18 A summary table of all applications contained information from 57 references, about half of which were concerned with the analysis of foodstuffs. The reviewers considered that because of their porosity, enormous internal surface areas, adsorption capacities, functionalisation properties, and stability, MOFs are promising for fast and efficient extractions. It is not clear to what extent these materials are commercially available, can be reused and are cost effective. Mandal and Lahiri47 have reviewed (171 references) the applications of CPE to the extraction, pre-concentration and speciation of metal ions. Emphasis was placed on the determination of Al, As, Cd, Co, Cr, Cu, Fe, Hg, Ni, Pb, Sb and Se, though not all of the methods listed featured an atomic spectrometric measurement. Many of the applications surveyed were to environmental materials (mostly waters), but there were examples of the analysis of beverages and some foods and some biological materials. The reviewers also mentioned the determination of NPs and offered some critical evaluation of the advantages (low cost and semi-green) and limitations (difficult to automate and requires prolonged heating for phase separation) of methods based on CPE. In an overview (103 references) of sample preparation procedures for determining, by ICP-OES or ICP-MS, elemental impurities in medicines (active pharmaceutical ingredients, raw materials, pharmaceutical dosage forms, and dietary supplements), Pinheiro and Nóbrega48 identified the five procedures most commonly applied and summarised their characteristics in a useful table. The procedures, dissolution in aqueous media, dissolution in organic media, conventional heating-assisted digestion in a closed vessel, microwave-induced combustion and MAD, were also extensively discussed in the text. The reviewers also discussed what happens after the sample preparation in a section entitled “trends and challenges for the determination of elemental impurities using ICP-based techniques” in which they explained the roles of dissolved solids, acidity, spectral (and polyatomic) interferences, and calibration. They concluded that direct dissolution, microextraction or partial digestion could be promising alternatives for saving time, minimizing reagent consumption and decreasing instrumentation costs. However, they pointed out that the accuracy of such methods needs to be carefully evaluated, as there is an increased risk of interferences from the matrix components that remain.

As an alternative to conventional MAD with acids, Costa et al.49 developed a MAE with a DES for the determination of Cd, Cu, Fe, Mn, and Zn in medicinal herb samples by ICP-OES. Relevant parameters, optimised by a multivariate strategy (response surface methodology), were a DES of choline chloride–oxalic acid prepared in a microwave system, 35 s extraction time, 90% microwave power, and a 50 mg mL−1 sample–solvent ratio (150 mg of dried ground sample in 3 mL of extractant). The LOQs were 0.023 (Cd), 0.023 (Cu), 1.1 (Fe), 0.06 (Mn), and 0.67 (Zn) mg kg−1. The method was validated by the analysis of CRM NIST SRM 1515 (apple leaves) and SRM 1573e (tomato leaves), for which percent relative measurement errors ranging from −13% to +9% were obtained. These values are probably not significant because of the relatively high RSD values obtained for the MAE method, which ranged from 3.5 to 8.5%. The procedure was applied to the analysis of about 30 real samples, in which all five analytes were found. The method characteristics evaluated on a greenness scale (references provided) were deemed superior to three possible alternative methods. Curti et al.50 applied UAE to separate Cu, K, Mg, Mn, P, and Zn from sorghum flour, prior to determination by MIP-OES. To a sample mass of 0.5 g (milled to less than 0.5 mm), 5.0 mL of 7 mol L−1 HNO3 and 1 mL of H2O2 were added, and the mixture heated to 100 °C for 30 min followed by 30 min ultrasonication. The resulting suspension was diluted to 30 mL. The method was validated by (a) the analysis of a CRM (NCS ZC 73010, maize flour mealie) for which the percentage relative errors ranged from −18 to +10%, and (b) comparison of the results with those obtained by sealed vessel MAD with HNO3–H2O2. The method was applied to 20 real samples, in all of which all six analytes were found, and was also evaluated on a greenness scale involving a hexagon, for which an appropriate source was cited. A procedure based on digestion of biological samples with formic acid has been developed,51 in which 100 mg of sample was mixed with 5 mL conc. formic acid and heated to 90 °C. After 30 min, 0.5 mL of 30% H2O2 was added and the mixture heated for a further 30 min. Before analysis by ETAAS, the solution was diluted (factor not specified exactly). The method was evaluated by the accurate determination of Cd and Pb in CRM NRCC TORT-3 (lobster hepatopancreas) and NIST SRM1577b (bovine liver). The LODs were 3 and 18 μg kg−1 for Cd and Pb, respectively, and the procedure was applied to the analysis of molluscs and different fish tissues from coastal areas in Brazil, but only results for Pb, found in almost all samples, were given.

For the determination of Cd in milk by FAAS, two different kinds of magnetic NP were used for digestion, based on the Fenton process, and pre-concentration.52 In the Fenton process, ferrous salts and H2O2 react to form free hydroxyl radicals (˙OH), whose production can be enhanced by irradiation with UV light. The digestion procedure featured citric acid coated magnetic Fe3O4NP, 15 mg of which were mixed with 40 μL HNO3, 0.60 mL H2O2, 0.75 mL water and 0.20 mL milk sample. After vortexing, the mixture was UV irradiated for 2 h, and the NPs magnetically separated from the digest, to which 20 mg of polystyrene-coated magnetic Fe3O4NPs were added. After magnetic separation, the retained Cd was dissolved in 150 μL of HNO3. The LOD was 0.53 mg L−1 and the method was validated by spike recovery. Such a high LOD makes the method unsuitable for the analysis of many real milk samples, in which the concentration of Cd is typically single-digit μg L−1. A somewhat similar procedure has been devised by Mou et al.53 for the determination of Cd in rice by ICP-MS, in which cobalt reacted with H2O2 in a Fenton-like digestion followed by PVG for which cobalt acted as a catalyst. To 500 mg of powdered sample, 0.32 mL of H2O2 (30%, m/v), 80 mL of conc. HNO3, and 0.9 mL of 1000 mg L−1 cobalt were added, and the mixture diluted to 2 mL, heated to near dryness (25 min) and reconstituted in 20 mL of water. To remove the interference of molybdenum from the sample matrix, 1.0 g of anion-exchange resin was added and the solution shaken (30 s). After allowing the resin to settle, the supernatant was diluted to 10 mL with formic acid (the PVG reagent). The LOD was 1.6 μg kg−1. The method was validated by the analysis of two rice CRMs (NRCCRM GBW(E)100351 and GBW(E)100357) and applied to the analysis of 7 real samples, in all of which Cd was detected. The procedure is discussed further in Section 4.4.

Several research groups have evaluated alkaline digestions, despite the fact that many metal hydroxides are insoluble. For the determination of clinical trace elements in whole blood and plasma by ICP-MS, Tanvir et al.39 treated 0.1 mL of freshly thawed and homogenised (roller mixer) samples with 0.5 mL of a solution prepared by mixing 50.0 mL water, 1.0 g of EDTA, 10 mL NH4OH, 10.0 mL 2-propanol and 1.0 mL of 1% (w/v) Triton X-100 and diluting to 100.0 mL. After vortexing for 5 s, the solutions were made up to 5.0 mL and analysed. In reviewing the determination of metallic NPs in biological samples by spICP-MS, Laycock et al.11 found that alkaline-based extraction methods have been quite widely applied, particularly in the preparation of animal tissue samples. The review contains a useful summary table in which papers are grouped by sample preparation method. The table also contained references to a substantial number of enzyme-assisted extraction methods, which have been applied to plant as well as animal tissues. In the determination of AgNPs in seaweed by spICP-MS,54 the analyte was liberated from the sample matrix by ultrasound-assisted enzymatic hydrolysis optimised in terms of type of sonication (bath vs. ultrasonic probe), ultrasound amplitude, sonication time, sonication mode (pulsed vs. continuous sonication), concentration of the enzymes mixture (Macerozyme R-10), and enzymatic hydrolysis time. The stability of the NPs during extraction was evaluated by TEM for a 15 nm polyvinylpyrrolidone coated AgNP standard. The LODs were 14 nm (size) and 4.34 × 107 part per g (number), and the method was applied to 10 real samples, in six of which AgNP were detected. On the other hand, to extract As species (AsIII, AsV, AB, DMA, MMA) from carrageenan-producing seaweed (for which previously published methods were deemed unsuitable) for determination by HPLC-ICP-MS, a method based on heating (at 90 °C) with dilute HNO3 (0.2%) for 60 min has been devised.55 Samples were first washed, frozen (−20 °C) for 24 h, freeze dried and ground to a particle size <0.12 mm. The parameters were optimised for both maximum extraction and minimal conversion of AsIII to AsV. The LODs were between 8 and 10 μg kg−1. The method was validated by spike recoveries and applied to the analysis of four varieties of K. alvarezii; all analytes, except AB, were detected in all samples, together with an unknown compound. The researchers did not compare the sum of species concentrations with total As concentrations. In an investigation into the effects of the interference of C on the determination of Br, Cl and I in medical plants following MAD, it was found56 that C concentrations (measured by ICP-OES) in the extracts (H2O, 100 mmol L−1 NH4OH or 100 mmol L−1 TMAH) ranged from 980 to 9800 mg L−1, with the highest concentrations obtained for TMAH. However only TMAH (0.5 g sample, 6 mL, 240 °C, 20 min) gave acceptable extraction efficiencies, and so dilution was necessary, as although no interferences were observed on Cl and Br up to 1000 mg L−1 of C, signal enhancement for I was observed above 500 mg L−1 of C. The accuracy of the method was evaluated by the analysis of two CRMs (NIST SRM 1572 (citrus leaves) and NIST 1547 (peach leaves)), spike recoveries and comparison of the results with those of a method involving microwave-induced combustion. The LOQs were 0.125, 125, and 0.045 μg g−1 for Br, Cl, and I, respectively. On this basis, Br was found in all but one of the nine real samples, Cl was found in all, whereas I was only detected in two of them.

To avoid the need for concentrated acids (and H2O2), the possibilities of a device featuring a microwave-assisted single reaction chamber containing pressurised oxygen has been evaluated.57 The figure of merit was the residual carbon content, defined as 100-times the ratio of the DOC in the final digest to the TOC of the sample. Samples included bovine liver (NIST SRM 1577a), linseed, parsley and whole milk powder (NIST SRM 8435) in which the TOC in the dried samples were 49%, 60%, 41%, and 51%, respectively. The researchers showed that satisfactory digestions of 250 mg of samples for analyses by ICP-OES were obtained (residual carbon contents of 4%, corresponding to 200 mg L−1) with 1 mol L−1 HNO3, when the reactor was pressurised to 50 bar with O2. The method was further validated by the analysis of CRM NIST SRM 8433 (corn bran) and by comparison of the results with those obtained by a “conventional” procedure with conc. HNO3 under 50 bar N2. The researchers attributed the destruction of C-containing compounds to the regeneration of HNO3 by the reaction of O2 with NO, a hypothesis that was supported by the residual acidity, which was about 70%, compared with the 7.5% remaining after digestion under N2. The researchers duly noted the alignment with green chemistry recommendations.

As has been the case for the past several years, there is no diminution in the number of papers reporting the results for yet another pre-concentration methodology. The majority of such procedures involve SPE, with possibly a growing interest in the synthesis of magnetic NPs functionalised with a suitable ligand. These and methods involving LLE are summarized in Table 1. In the case of solid samples, deciding whether a method involves pre-concentration is not so straightforward. As the initial sample dissolution step always involves considerable dilution, the solution introduced to the instrument for measurement may still be more diluted than the original sample even after a pre-concentration step. A paper is described in Table 1, only if the method clearly involves a step in which the concentration of the analyte species in a solution (either the original sample or the sample digest or extract) was increased. Procedures in which a solution is produced that is more concentrated than that from a “standard” procedure, such as a MAD in acid, are not included in Table 1. Other pre-concentration protocols are dealt with elsewhere in this ASU. For example, procedures in which the analyte species are converted into volatile derivatives, collected (typically on a solid), and then released as a concentrated slug are discussed in Section 4.4. Often much of the material presented in the reports concerns (a) the synthesis and characterisation of the novel solid extractant, and (b) the optimisation of the extraction procedure. Sometimes crucial information about the performance of the resulting analytical method is relegated to the electronic supplementary information (ESI), a trend that is to be discouraged. Authors of such studies are encouraged to give greater prominence to the analytical performance and put the information about the synthesis and characterisation into the “ESI”. Many reports do not contain a description of the optimised method in sufficient detail that would enable it to be followed by another researcher, a trend that is also to be discouraged. A variety of optimisation strategies have been adopted, often with little or no explanation of the basis for the choice, and often with no clear statements about the figure of merit or the boundary conditions. A few reports are noteworthy because of the extent of the factor space that was searched. Sun et al.,58 in devising a method for the determination of Cd in rice by ultrasound-assisted DLLME and FAAS, investigated the effects of 10 ligands, four disperser solvents and three extraction solvents. In a similar procedure, ultrasound-assisted DLLME involving DESs, for the pre-concentration of Se prior to determination by HG-AAS, Altunay et al.59 investigated the possibilities of six different DESs as well as the usual parameters of pH, amount of ligand, volume of DES, sonication time, and sample amount.

Table 1 Pre-concentration by liquid- or solid-phase (micro) extraction
Element Matrix Technique Extraction mode/reagents Procedure/comments LOD in μg L−1 (unless stated otherwise) Validation Ref.
Am, Pu, 90Sr, U Urine, water (lake, sea) ICP-MS/MS SPE on DGA-branched resin and Sr resin Sample (10 mL) was loaded onto the DGA-B resin, rinsed and Sr eluted with 6 mL of 8 mol L−1 HNO3 and the eluate loaded on the Sr resin. U and Am were then eluted from the DGA-B resin, and Sr from Sr-resin with 0.01 mol L−1 HNO3. System was fully automated and rapid compared with traditional radiometric counting methods. Analytes not found in samples 0.65 pg L−1 (241Am), 0.56 pg L−1 (239Pu), 1.48 pg L−1 (90Sr), 1.75 pg L−1 (234U) Spike recoveries 78
As Water (bottled, well, waste), honey, rice (white, brown) ETAAS LLME of complex with ethylenediamine-N,N′-disuccinic acid into DES, benzyltriphenylphosphonium chloride + ethylene glycol Samples (0.5 g) MAD with HNO3 and H2O2 made up to 25 mL, adjusted to pH 7, then AsV reduced to AsIII by KI in acid. To 5 mL solution adjusted to pH 6, complexant 600 μL DES and 300 μL THF (as emulsifier) were added. Vortexed (3 min) and centrifuged (5 min at 4000 rpm). Organic phase too viscous and was diluted to 2 mL with acidic EtOH. As not found in three of five honey samples or the bottled water 6.5 ng L−1 CRM NIST CRM 1643e (simulated fresh water-trace elements), 1568a (rice flour) 196
As Food (rice and flour) ICP-MS SPME with alpha-amylase modified magnetic carbon nanotubes (alpha-amylase-Fe3O4/MWCNTs) To 15 mg of sample, 200 μL of pH 7 buffer and 5 mg of alpha-amylase-Fe3O4/MWCNTs at 37 °C were added. Following 10 min ultrasonication, 800 μL of water was added and phases magnetically separated. Analyte found in five rice samples but not in any of four flour samples, including rice flour 14 μg kg−1 NIST SRM 1568a (rice flour) 197
As Rice and water (spring and sea) PVG-ICP-MS SPE with kinetic discrimination on CdS/MIL-100(Fe) composites acting both as adsorbent and photocatalyst 1.0 g of ground rice was extracted with 20 mL of 0.15 mol L−1 HNO3 at 90 °C for 2.5 h. The extract was cooled, centrifuged, and filtered. Then 16 mg of CdS/MIL-100(Fe)-50% was added and dispersed. After incubation (5 min for AsV and 105 min for AsIII) the composites were collected on microfiltration membranes (0.45 μm, Nylon 66) and then ultrasonically re-suspended in 10 mL of diluted formic acid and pumped to a thin-film, flow-through photochemical reactor. The procedure was applied to the determination of total As in several food samples (digested to convert all species to AsV) and to the speciation analysis of the rice CRM and three real samples; both species were detected in all samples 0.11 ng L−1 Total As in CRM: NRCCRM, GBW(E)080684 (rice), GBW08517 (kelp), GBW10017 (milk powder), GBW10015 (spinach), NRCTORT-3 (lobster hepatopancreas), DOLT-5 (dogfish liver), DORM-4 (fish protein), CASS-6 (seawater). Comparison of rice results with those of an HPLC-ICP-MS procedure (no details given) 120
AsIII Water (tap, river), urine, fruit juice HG-ETAAS with lanthanum and iridium modifiers DSPME on magnetic Fe3O4NPs modified with chitosan Into 20.0 mL of filtered sample (pH 8.1) solution was injected 1 mL of a suspension of 30 mg sorbent and 20 mg CTAB. The mixture was sonicated (5 min at room temperature) and the sorbent magnetically separated. 148 μL of HNO3 (0.5 mol L−1) was added, the mixture was sonicated (8 min) to desorb the AsIII and magnetically separated. The interferences studied included AsV, MMA and DMA. AsIII was found in all samples except urine 0.003 Spike recovery 65
AsIII and AsV Dairy (peanut milk, soy milk, milk tea, and yogurt) ETAAS Magnetic DSPME of AsV on ZnFe2O4 nanotubes (ZFONTs) was coupled with DLLME Sample (6.0 mL) was extracted with 40 mL of artificial gastric juice (formulation given) for 3 h at 37 °C, then filtered (0.45 μm) and diluted to 60 mL with 0.01 mol L−1 HNO3. To 20 mL (pH 5.0) was added 20 mg of ZFONTs and the mixture ultrasonicated for 2 min at room temperature. After magnetic separation, the AsV retained was desorbed with 0.5 mL of 0.5 mol L−1 HNO3 by sonication for 1.5 min. The separated solution (20 mL) containing AsIII was mixed with 100 μL APDC, 0.4 mL MeOH (disperser), and 50 μL CCl4 (extractant). After sonication (2 min) and centrifugation (5000 rpm for 4 min), the organic phase was withdrawn and diluted to 100 μL with EtOH. Both species were found in all three real samples. There was no evidence of species interconversion 0.001 (AsIII), 0.002 (AsV) CRM: National Research Centre for Geology (Beijing, China) GBW 10017 (milk powder) and spike recoveries 64
iAs Wild shrimp ICP-MS SPE on magnetic iron NP functionalised with dimethyl triamine-pentamethylene phosphonic acid, Fe3O4@DTPMP Sample (500 mg) plus 10 mL 2% HNO3, heated for 1 h at 90 °C and diluted to 20 mL, centrifuged, and filtered (0.45 μm). For extraction of iAs, to 10 mL was added 4 mL Britton–Robinson (BR) buffer solution (pH 4.0), diluted to 40 mL, and 20 mg of NP (Fe3O4@DTPMP) added, stirred for 15 min, and magnetically separated. To desorb the iAs from the NPs 0.5 mL of 10% HNO3 was added, vortexed (3 min) magnetically separated and the resulting solution diluted to 10 mL. The procedure was applied to two real samples 16 ng kg−1 CRMs DOLT-5 (dogfish liver) and DORM-4 (fish protein). Comparison of results with those of an HPLC-ICP-MS method 63
Bi Urine SQT-FAAS DSPME with salicylic acid modified magnetic iron nanoparticles (SA-MNPs) Sample (0.50 g) subjected to MAD with 8 mL HNO3–H2O2 then urine was “diluted 15 times” with deionised water. To 30 mL of digest (pH adjusted to 5), 20 mg of SA-MNPs were added, and after vortexing (15 s) and magnetic separation, the Bi was eluted with 100 μL HNO3. One sample was analyzed, but Bi was not detected. Spike recoveries of 60, 80, 100, and 150 μg L−1 were between 95 and 101% 6 Spike recovery 198
Cd Milk (whole and low-fat) FAAS SPE onto polystyrene-coated magnetic Fe3O4 NPs (PS-MNPs) with elution by conc. HNO3 In a Fenton digestion process, to 200 μL of sample, 15 mg citric acid-coated magnetic Fe3O4NPs, 40 μL HNO3, 600 μL H2O2 and 750 μL water were added. The mixture was UV irradiated for 2.0 h. After magnetic separation, the sample solution was transferred into a 15 mL tube containing 20 mg of PS-MNPs and the pH adjusted to 6.0. After vortexing (15 s) and magnetic separation, the adsorbed cadmium was eluted with 150 μL conc. HNO3. It is not clear whether any Cd was found in any samples. Spike recoveries of 5–25 mg L−1 were between 87 and 111% 530 Spike recoveries 52
Cd Rice (polished, brown, glutinous) FAAS DSPME of the DDTC complex into 1-heptanol, with EtOH as the disperser To 0.1 g dried, powdered and sifted sample was added 10 mL of HNO3–HClO4 (9 + 1, v/v). After soaking for 24 h and heating to dryness, the residue was dissolved in 0.1 mol L−1 HNO3 and diluted to 100 mL. To 8 mL of this solution (adjusted to pH 7), 0.8 mL 0.001 mol L−1 DDTC solution, 0.5 mL of 1-heptanol and 0.1 mL of EtOH were added. After sonication and centrifugation at 40 °C, 0.3 mL organic phase was removed and mixed with 0.4 mL MeOH. Cd was found in all three samples 0.69 Spike recovery 58
Cd Infant food and water (tap, mineral, well) FAAS Co-precipitation with zero-valent iron produced by reaction of ironII with BH Infant formula: samples (0.5 g) were digested with 5 mL HNO3 and 3 mL H2O2 and evaporated to near dryness. The residue was dissolved in 25 mL 0.1 mol L−1 HCl solution, filtered and diluted to 100 mL. The digest (or water sample) was adjusted to pH 4, heated to 55 °C and 0.1 g of KHP and 2.5 mL of 500 mg L−1 ironII added followed by 0.06 g of BH. After magnetic separation, the precipitate was dissolved in 0.2 mL of 6 mol L−1 HCl. Cd was found in all four samples 0.1 CRM. NIST SRM 1643e (trace elements in water), spike recoveries and comparison of results with those of a standard method, for which no details were given 61
Cd Water (bottled, mineral, well, waste), food (rice, black tea, wheat, honey, corn, leek, eggplant, spinach, tomato, grilled meat, grilled chicken) FAAS DLLME of cysteine complex into magnetic ionic liquid (MIL) trihexyl(tetradecyl)phosphonium chloride + CoCl4 Samples (1 g) MAD with HNO3 and H2O2 made up to 10 mL. To 5 mL of solution adjusted to pH 4.5, 680 μL of 50 mmol L−1L-cysteine, and 560 μL of the MIL were added, sonicated (2 min), magnetically separated and the MIL phase dissolved in 1 mL of acidic EtOH. Cd was not found in bottled or mineral water, but was found in all other samples 0.6 CRM INCT TL-1 (tea leaves), NIST SRM 1573a (tomato leaves), CRM 7502-a (trace elements in white rice flour), spike recoveries 199
Cd Water (industrial waste), foods (radish, spinach, bread (white and brown)), cigarette tobacco FAAS SPE dispersive, magnetic on glycidyl methacrylate ion imprinted polymer (Fe3O4@GMA@IIP) Impossible to find a description of the method. Even information about sample preparation is in a prior paper. It appears as though perchloric acid might be involved. Cd was detected in all samples 3 CRM, rice flour NIES 10c (rice flour), spike recoveries 200
Cd Water (tap, mineral, lake, physiological saline solution) FAAS with thermospray introduction SPE on metal–organic framework (MOF) UiO-66 Sample (10 mL buffered at pH 8) passed at 10 mL min−1 though a mini-column (1.5 cm × 0.5 cm) containing 20 mg of UiO-66 MOF. Eluted with 1 mol L−1 HCl at 1 mL min−1. Analyte only found in the mineral water 0.03 Spike recoveries 201
Cd Water (tap, ground, mineral and reuse) HR-CS-FAAS CPE of PAN complex into Triton X-114 surfactant-rich phase, which was separated by SPE on hydrophobic cotton and eluted with dilute HNO3 Additions to and removal from the reaction vessel were by computer-controlled peristaltic pumps. The following were added: sample (4 mL), PAN, Tris buffer, Triton X-114. After addition of Tris/HTris+ to induce phase separation, the lower surfactant-rich phase was pumped onto a mini-column containing 35 mg of hydrophobic cotton and eluted with 500 μL of 0.5 mol L−1 HNO3. It is not clear if Cd was found in any of the samples. Spike recoveries of 5 μg L−1 were between 95 and 105% 1.3 CRM: SCI science EP-H (drinking water) and spike recoveries 60
Cd Water (drinking) and food (rice wheat and watermelon) ETAAS DLLME with a hydrophobic natural deep eutectic solvent prepared from salicylic acid, and L-menthol Sample (500 mg) MAD with HNO3–H2O2 and diluted to 25 mL, adjusted to pH 6. Then 200 μL of the L-menthol–salicylic acid extraction solvent was rapidly injected, vortexed (5 min), ultrasonicated (40 °C for 5 min), and the NADES aggregates were gradually fragmented. Then, 20 μL was taken and, together with 10 μL of Pd(NO3)2 modifier solution, injected into the ET atomiser. Analyte was found in all samples 0.37 × 10−4 Spike recoveries 202
Cd Beverages (bottled mineral water, pear juice, green tea, and distilled spirits) ETAAS Successive (first pyrethroids, then Cd) homogeneous LLME based on switchable hydrophilicity solvents and ionic liquids. For Cd hydrophobic ILs were generated in situ as extraction solvents by hydrophilic ILs and sodium fluoroborate with APDC as complexing agent In the second extraction, 100 μL of APDC (100 mg L−1), 132 mg of sodium fluoroborate, and 126 mg of sodium carbonate were added whereby 1-ethyl-3-methylimidazolium[thin space (1/6-em)]tetrafluoroborate, (as extraction solvent, freezing point of 15 °C) and CO2 bubbles were generated in situ. The mixture was centrifuged at 1370g for 3 min and left in an ice bath for 5 min to solidify the extraction solvent (at the bottom). Which was separated and dissolved in 100 μL of 0.1 mol L−1 HNO3 in EtOH. Cd was not detected in any of the samples 0.006 Spike recoveries 203
Cd, Co, Hg, Ni, Pb and V Oral and parenteral drugs ICP-OES DLLME of 8-hydroxyquinoline complexes into a DES (a mixture of 2[thin space (1/6-em)]:[thin space (1/6-em)]1 molar ratio of DL-menthol and decanoic acid) To 8 mL of sample solution, at pH 3.4 containing 0.1% 8-hydroxyquinoline, was added 70 μL of DES, vortexed (3 min), and centrifuged (4 min 3000 rpm). 50 μL was introduced directly to the spectrometer. The researchers wrote, (a) “all analytes were below their respective LOQ values for all oral and parenteral drug samples” and (b) “LOQ values… are suitable to meet USP requirements even using oral liquid drugs” 0.05 (Cd), 0.6 (Co), 0.8 (Hg), 0.9 (Ni), 1.2 (Pb), 0.8 (V) Spike recoveries 204
Cd, Co, Cu, Pb Lake water and urine FI-FAAS SPE on sol–gel thiocyanatopropyl functionalised silica Acid digestion in an autoclave of urine (HNO3) and mussel tissue (HClO4). 120 mg of extractant packed into a microcolumn. Samples (pH 5) were loaded at 10 mL min−1 for 2 min and retained analytes eluted with 1 mol L−1 HNO3. Cd and Pb were not detected in any of the samples, Co was not detected in one lake sample or the urine sample, but Cu was detected in all samples. No loss of functionality after 700 load-elute cycles 15 (Cd), 0.5 (Co), 0.5 (Cu), 2 (Pb) CRM: NIST SRM 1643e (trace elements in water), BCR 278-R (trace elements in mussel tissue). Spike recoveries 205
Cd, Cr Herbs (mint, basil, pennyroyal, parsley, dill) ETAAS Magnetic SPE of 8-dithiocarbamate complexes with multi-walled carbon nanotubes (MWCNTs) Samples (0.4 g) MAD with HNO3 and H2O2 and made up to 100 mL. To 25 mL adjusted to pH 5, complexant and 0.12 g (Cd) or 0.02 g (Cr) magnetic CNTs were added. After magnetic separation, Cd dissolved in 2 mL ACN–HNO3 and Cr in 2 mL MeOH–HNO3. Both Cd and Cr were found in samples 0.3 Spike recoveries 206
Cd, Cu, Ni Pharmaceuticals (eye drop, anesthetic, serum) and water (tap mineral and spring) ICP-OES SPE with N,N′-bis(5-methoxsalicylidene)-2-hydroxy-1,3-propanediamine modified silica gel All samples were aqueous solutions; 35 mL (adjusted to pH 4) was passed through a column (10 × 1 cm containing 750 mg of extractant) at 3.0 mL min−1 and the retained analytes eluted with 4.0 mL of 0.5 mol L−1 HNO3 at 2.6 mL min−1. Cu and Cd were found in all six samples, whereas Ni was not found in any. The possible interference by EDTA or thiourea was investigated 0.03 (Cd), 0.06 (Cu), 0.04 (Ni) CRM Certipur ICP multi-element standard solution IV and spike recovery 207
Cd, Cu, Fe, Mn, and Zn Medicinal herbs ICP-OES LLE (MAE) into DES choline chloride–oxalic acid Sample (150 mg) plus 3 mL solvent ChCl[thin space (1/6-em)]:[thin space (1/6-em)]Ox[thin space (1/6-em)]:[thin space (1/6-em)]water (1[thin space (1/6-em)]:[thin space (1/6-em)]1[thin space (1/6-em)]:[thin space (1/6-em)]1 molar ratio) was vortexed (5 s). Microwave irradiated (90%, 630 W, 35 s), vortexed, diluted to 6 mL, and centrifuged (8000g 5 min). 31 real samples were analysed; all analytes found in all samples. The procedure scored highly on the analytical greenness assessment scale LOQ 23 (Cd), 23 (Cu), 1100 (Fe), 60 (Mn), and 670 (Zn) CRM NIST SRM 1515 (apple leaves), NIST SRM 1573a (tomato leaves) 49
Cd, Cu, Pb Shellfish (abalones, scallops, oysters, mussels, clams and razor clams) ICP-MS Magnetic DSPME of ionic liquid-coated amino silanized magnetic graphene oxide (MGO@SiO2-APTES-IL) MAD of 0.2 g of freeze dried sample with H2O2–HNO3 diluted to 50 mL. To 30 mL (adjusted to pH 5.6) was added 25 mg of MGO@SiO2-APTES-IL. After extraction and separation, analytes dissolved in 3 mL 5% HNO3. Analytes found in all samples 0.004 (Cd), 0.003 (Cu), 0.002 (Pb) Spike recoveries 208
Cd, Fe, Pb Water (drinking and dental) FAAS CPE of complexes with 2,6-diamino-4-phenyl-1,3,5-triazine and 3-amino-7-dimethylamino-2-methylphenazine and Triton X-114 Methods were summarized in a table. Slightly different conditions for each element. Sample volume was maybe 100 mL and final volume was 5 mL. It is not clear whether any of the analytes were found in the samples 10 (Cd), 300 (Fe), 10 (Pb) Spike recoveries 209
Cd, Fe, Pb Drinking water FAAS CPE with 2,4-diamino-6-phenyl-1,3,5-triazine and 3-amino-7-dimethylamino-2-methylphenazine as chelating agents, and Triton X-114 The optimum conditions were different for each element. The analytes in 100 mL of sample were transferred into 5 mL of diluted extractant. The procedure takes an hour. No results for real samples were given 5 (Cd), 25 (Fe), 5 (Pb) Spike recoveries 210
Cd, Hg, Mn, Pb, Sb Rice and rice products (crackers, noodles, infant cereal) ICP-MS SPE on oxidized multi-walled carbon nanotubes (ox-MXCNTs) A mixture of 3.0 mg of ox-MWCNTs and 30 mg of Epolene® was packed in a microcolumn (25 × 2.3 mm). Epolene® is a low-density polyethylene wax that offers good high-temperature stability, low-temperature flexibility, and very good compatibility with the use of mineral acids. Sample (5 mL adjusted to pH 7) pumped through column at 1 mL min−1 then eluted with 10% HNO3 at 1.9 mL min−1. 16 samples were analyzed: Cd was determined in 14; Hg in 3, Mn in 14; Pb in 6 and Sb in 8 Ranged from 0.8 ng L−1 (Mn) to 0.09 μg L−1 (Hg) CRM NIST SRM 1568b (rice flour) and NIST 1643f (trace elements in water) 211
Cd, Hg, Pb Medicines (drugs in solid dosage forms) ICP-OES DLLME of DDTC complexes into toluene. No disperser Sample (500 mg) MAD final volume 25 mL. To 8 mL, pH adjusted to 6, DDTC was added followed by 100 μL of toluene. Shaken, centrifuged, 80 μL introduced directly to ICP-OE spectrometer via a multi-nebuliser device that accommodates two independent liquid inlets. All analytes were below their respective LOQ values for all samples analysed. The improvement in LOQs compared to those without pre-concentration was 40-fold LOQs 0.3 (Cd), 1.8 (Hg), 1.6 (Pb) Spike recoveries 212
Cd, Pb Vegetables (celery, basil, fenugreek and savory) FAAS SPE on dual template imprinted polymer-modified mesoporous silica coated with Fe3O4 magnetic nanoparticles Samples (300 mg) digested with HNO3 and H2O2, final volume maybe 50 mL. To this (adjusted to pH 6.12) was added 40.62 mg of the extractant and after 17.38 min and magnetic separation, the retained metals were eluted in 5 mL of 0.1 mol L−1 HNO3–MeOH (9 + 1). Neither analyte was found in any of the four samples 0.15 (Cd), 0.35 (Pb) Spike recoveries 213
Cd, Pb Food (vegetables, and barbecued meat), water (tap, river, mineral and well) FAAS DSPME with poly-3-hydroxy butyrate-polyvinyl triethyl ammonium chloride comb-type amphiphilic cationic block copolymer (PHBvbNCl) Acidic MAD of 1 g adjusted to pH 7 (final volume not given), extracted with 55 mg of PHBvbNCl, and analytes dissolved in 1.5 mL acidic EtOH. Analytes found in all samples except tap and river waters 0.15 (Cd), 0.03 (Pb) CRMs: NIST SRM 1643e. (trace elements in water), INCT TL-1 (tea leaves) 214
Co, Cu, Ni Blood, urine, tap water FAAS CPE of complexes with (E)-2-(2,4-dihydroxybenzylidene)-N-phenylhydrazine-1-carbothioamide (DHBPHC) into a surfactant-rich phase of Triton X-114 Blood (0.5 mL) and urine (2.0 mL) were subject to MAD with HNO3–H2O2 and diluted to 10.0 mL. To 50 mL sample (pH 6) was added an unspecified volume of 0.08% v/v Triton X-114 and 10−4 mol L−1 of DHBPHC. After separation, the volume of the surfactant-rich phase was made up to 1.0 mL with EtOH–conc. HNO3 (5 + 1, v/v). None of the analytes were detected in tap water and Co was not detected in blood 0.3 (Co), 0.8 (Cu), 0.9 (Ni) Spike recoveries 215
Cr, Cu, Fe, Mn, Pb, Zn Coffee (instant coffee, filter coffee), tea, (tea bag (black), Turkish tea (black), green tea), broad beans, beans, chickpea, red lentils and green lentils FAAS SPE on synthesized ternary polymer composite, polystyrene/polyacrylonitrile/polyindole elution with HNO3 Extractant (300 mg) in column (10 cm × 1 cm). Samples (0.5 g) digested with HNO3 and H2O2 and made up to 20 mL after adjusting to pH 7 and loaded onto the column then eluted with 10 mL of 0.1 mol L−1 HNO3. Cu, Fe, Mn and Zn were found in all samples, Cr in about half of them and Pb in only 2 (broad bean and green lentils) 0.9 (Cr), 1.2 (Cu), 2.0 (Fe), 1.4 (Mn), 0.9 (Pb) 2.0 (Zn) CRM INCT-TL-1 (tea leaves) and spike recovery 216
Cr, Cu, Pb Water (tap, river, lake, urine, hair) ETAAS Magnetic SPE on melamine-based triazine polymers on the surface of magnetic Fe3O4@SiO2 NPs Water filtered and acidified. Hair (0.2 g wet ashed with HNO3 and made up 50 mL), urine (5 mL wet ashed with HNO3 and made up 50 mL). To 50 mL sample (pH 6) 5 mg of extractant added, shaken (10 min) and magnetically separated. Analytes eluted into 0.4 mL of 1 mol L−1 HNO3 with vortexing (3 min). Analytes found in all samples except Cr and Pb were not found in tap water 2.9 (Cu), 1.4 (Cr) and 6.1 (Pb) ng L−1 CRM GSB07-3186-2014 (water quality standard), GBW091027 (human urine), spike recoveries 217
CrIII Tap water, green tea infusion ETAAS SPE with ion imprinted polymer based on CrIII–1,10-phenanthroline complex and styrene Sample (adjusted to pH 4.5), volume not given (maybe 20 mL), loaded through column containing 0.1 g extractant at 1.0 mL min and eluted with 1 mL of 0.1 mol L−1 EDTA. The method showed good specificity for CrIII over CrVI. Matrix effects were observed for both samples 0.35 CRM NIST SRM 1643e (surface water) 66
Cu Water (tap) FAAS SPE on activated carbon-based ion-imprinted sorbent (Cu(II)-IAC) Column: 100 mg adsorbent in 15 cm × 0.8 mm column. Up to 750 mL of sample (at pH 5) passed. Eluted with 5 mL 4 mol L−1 HNO3. Batch: 25 mL sample (pH 5), 15 mg adsorbent shaken for 2 h at 500 rpm then filtered. Analyte not found in the one real sample 0.038 CRMs ERML-CA021e (soft drinking water), NIST 1643e (simulated freshwater) 218
Cu, Mn Coconut water FAAS DLLME of (2-(5-bromo-2-pyridylazo)-5-(diethylamino)-phenol, 5-Br-PADAP complex into CHCl3 with ACN as disperser (according to the abstract) Samples (30 mL) subject to MAD with HNO3 and H2O2. NaCl added. pH 4 for Cu and pH 9.5 for Mn. In contrast to the abstract, the text indicated EtOH (0.5 mL) was the disperser and ACN (3.8 mL) was the extractant. In another location, the text suggested that CHCl3 was the disperser. Both analytes were found in 5 real samples LOQ 4.8 (Cu) and 3.3 (Mn) Spike recoveries 219
Cu, Mn Plants (sesame, peanut, eggplant, maize and cucumber), barbecue (meat doner, chicken doner, meatballs, grilled chicken, fish) FAAS SPE on dithizone@PAA (details in previous papers) Solid samples (0.2 to 0.50 g), MAD with HNO3 and H2O2. Final volume not given. To 30 mL sample (pH 4.0) was added 55 mg adsorbent, shaken (20 min), centrifuged (4000 rpm, 2 min), separated and analytes dissolved in 2.0 mL of acidic EtOH. Both analytes found in all plants and all barbecue samples, except Cu in meat doner and Mn in grilled chicken and meat doner 0.06 (Cu), 0.2 (Mn) CRM NIST SRM 1568 (rice flour), 1515 (apple leaves), INCT-TL-1 (tea leaves), INCT-MPH-2 (mixed Polish herbs), spike recoveries 220
Cu, Mn Food (potato, carrot, olive, cabbage, cherry, green and black teas, tomato, spinach, apple juices, walnut, rice, honey, pepper), tap water and mineral water ICP-OES SPE on magnetic γ-Fe2O3 NPs modified with Bacillus cereus To 1.0 g was added 5.0 mL of HNO3–HCl (1 + 1, v/v), and after evaporation to dryness and addition of 6 mL HNO3–HCl–H2O2 (1 + 1 + 0.2, v/v/v) the mixture was subjected to MAD. The final volume was 50 mL, which after pH adjustment (6) were loaded onto the 100 mg column (1 × 10 cm). The retained analytes were eluted with 5 mL 1.0 mol L−1 HCl. Both analytes were found in all samples. The column could be used 32 times 0.09 (Cu), 0.08 (Mn) NCS DC73350 (poplar leaves), NRCC, DORM2 (dogfish muscle), and NWTM-15 (fortified water) and NCS ZC73014 (tea leaves) 221
Cu, Ni, Pb Water (tap, mineral), food (black tea, walnut, cucumber, potato, tomato, onion, rice, corn, baby milk, biscuit) ICP-OES SPE with E. profundum loaded onto Amberlite XAD-4 resin A lot of information in previous publications or supplementary materials. MAD but sample masses and final volumes not given. 100 mg of absorbent in column, up to 200 mL (pH 4–6) loaded at 3 mL min−1. Elution with 5 mL of HCl. None of the elements were found in tap water, Ni was not found in the mineral water, and Pb not found in walnut, cucumber, tomato, baby milk and biscuit 0.031 (Cu), 0.042 (Ni), 0.043 (Pb) CRM NCSZC 73014 (tea leaves), NWTM-15 (fortified water), and NCSZC73350 (poplar leaves), CWW-TM-D 222
Cu, Pb, Zn Food (fish, chicken and spinach) and water (river and sea) FAAS SPE on MCM-41 silica rice husk modified with DTZ (RH@MCM-41@DTZ) Samples (200–500 mg) subjected to MAD with HNO3–H2O2, evaporated to near dryness and diluted to 50.0 mL and mixed with 50 mg of RH@MCM-41@DTZ. After pH adjustment (6.0) and shaking (20 min), the sorbent material was separated by filtration and shaken with. 5 mL of 0.5 mol L−1 HNO3 to desorb the loaded metal ions. All analytes were found in all samples, except for Pb in spinach 0.3 (Cu), 0.5 (Pb) and 0.2 (Zn) CRM: NIST SRM1643f (trace elements in water) and SRM 1570a (spinach leaves). Spike recoveries 223
Hg Water (tap, sea) CV-ETAAS Magnetic DSPME of the methyl thiosalicylate complex into a ferrofluid based on Fe3O4@graphene oxide nanospheres together with an ionic liquid, 1-butyl-3-methyl-imidazolium-tetrafluoroborate To sample (volume not specified), NaCl and methyl thiosalicylate were added, and pH adjusted to 1. Then 220 μL of ferrofluid injected, ultrasonicated, magnetically separated and Hg eluted into 2 mL of 5% HNO3 and 5% urea. The Hg was trapped on the iridium-modified platform of the transversely heated graphite atomizer. No Hg was found in the tap water 0.25 ng L−1 CRM NIST SRM 2976 (mussel tissue), spike recoveries 224
Hg Food (fish muscle) ICP-OES SPE on electrochemically produced aluminium oxide (AO) membrane modified with polyaminophosphonic acid A PTFE column (10 × 0.8 cm) was packed with 300 mg of functionalised AO. 50 mL of sample solution (pH 6.0) was passed at 6.0 mL−1 and the sorbed Hg eluted with 3 mL 0.5 mol L−1 H2SO4. One sample of fish muscle was analyzed; the concentration was 0.28 μg g−1 0.02 NIST SRM 1641d (mercury in water) and spike recovery 177
Hg Water ICP-MS SPE on carbon fibres functionalised with (3-mercaptopropyl)trimethoxysilane Sample volume not given. Mass of fibres not given. Elution with 2 mL 2% HNO3 containing 200 μg L−1 Au NP to eliminate memory effects. No real samples analyzed 0.002 None apart from interference studies 225
Hg, Zn Soil, water (tap, mineral) food (rice, wheat flour, meat carrot, potato, garlic peanut, milk, sugar, corn, cheese, olive, chocolate) ICP-OES SPE with B. licheniformis loaded onto Amberlite XAD-4 resin A lot of information in previous publications or supplementary materials. MAD but sample masses and final volumes not given. Column loaded at 3 mL min−1, and eluted with 5 mL of 0.1 mol L−1 HCl. Zn found in all samples, but Hg only found in soil 0.06 (Hg), 0.03 (Zn) CWW-TM-D, EU-L-2, NCS ZC73O14 (tea leaves), NCS ZC73350 (poplar leaves), NWTM-15 (fortified water) 226
iHg, MeHg Rice RP-HPLC-HG-AFS SPE on C-18 microcolumn modified with DDTC Solid sample (1.0 g) shaken with 10 mL 5 mol L−1 HCl for 60 min, centrifuged (8000 rpm for 15 min) and filtered (0.22 μm); 2.0 mL neutralized with 6 mol L−1 NaOH and diluted to 10 mL. Sample solution (10 mL) loaded for 3 min and eluted in 0.1 mL. Eluted with 0.25% 2-mercaptoethanol–60 mmol L−1 CH3COONH4–4% acetonitrile, which was also the mobile phase. Both analytes found in all seven rice samples at single- or low double-digit μg kg−1 concentrations 0.3 ng L−1 (iHg), 0.2 ng L−1 (MeHg) CRM GBW(E)100348 (rice), GBW(E)100361 (rice), META-DJTZK-024 (brown rice), spike recoveries, comparison of total Hg (MAD and AFS) with sum of species 227
iHg and MeHg Fish HPLC-ICP-MS Magnetic SPE on metal–organic framework. Fe3O4 nanospheres coated with sulfur-functionalised UiO-66 Sample (500 mg) plus 2 mL 5 mol L−1 HCl sonicated (10 min), centrifuged (5000 rpm for 5 min). Two extracts filtered (0.22 μm), diluted to 50 mL, and pH adjusted to 4.0. Then 20 mg of Fe3O4 @UiO66-SH added, sonicated (15 min) to adsorb the mercury species, and magnetically separated. Elution (vortexed 2 min) into 1 mL 0.5% HNO3 and 2% thiourea. Neither species was found in the two water samples and only iHg was found in the fish sample, though MeHg was accurately determined in the CRM 1.4 (iHg) and 2.6 (MeHg) ng L−1 CRM GBW10029 (fish tissue) and spike recoveries 228
Hg species (iHg, MeHg, EtHg) Cereals (corn, rice, and wheat flour), Hg-methylating bacteria (Geobacter sulfurreducens) and periphyton HPLC-ICP-MS SPE on DTZ-modified C18 Sample preparation not given. The C18 SPE column (5 μm, 4.6 mm i.d. × 12.5 mm) located in the loop of the HPLC injection valve was functionalised with 5 mL 2.5 mg L−1 dithizone at 5 mL min−1 for 84 s. Then, 5 mL sample was loaded, and the retained analytes eluted with 1% 2-mercaptoethanol directly into the HPLC column. The concentration of MeHg (found in all samples) in cereal ranged from 0.15 ± 0.01 to 1.42 ± 0.02 ng g−1. The procedure was also applied to the analysis of natural waters For 5 mL sample, 0.15 (iHg), 0.07 (MeHg), and 0.04 (EtHg) ng L−1 Comparison of results with those of a purge and trap GC-AFS method 229
Hg, Sn Food (gluten-free biscuit, flour, rice, tuna fish, meat, chicken meat, potato, chocolate, coffee), tap water, energy drink and mineral water ICP-OES SPE on Amberlite XAD-4 modified with Geobacillus galactosidasius sp. nov. To 1.0 g was added 5.0 mL of HNO3–HCl (1 + 1, v/v), and after evaporation to dryness and addition of 6 mL HNO3–HCl–H2O2 (1 + 1 + 0.2, v/v/v) the mixture was subjected to MAD. The final volume was 50 mL which, after pH adjustment (5–7), was loaded onto the column (1 × 10 cm). The retained analytes were eluted with 5 mL 1.0 mol L−1 HCl. Of the 10 samples analyzed, Sn was found in three and Hg in six. The column could be used 30 times 0.53 (Hg), 0.27 (Sn) CRM NRCC DORM-2 (dogfish muscle) 230
In Water (distilled, drinking, mineral) ETAAS SPE of oxalate complex on silica gel modified with azolium groups (details in previous papers) Sample (pH 3 with oxalic acid) passed though 50 mg of sorbent packed in a column. Removed, dried and slurried with 1 mL 0.1% HNO3. Analyte not found in any samples 5.5 ng L−1 Spike recoveries 231
MnII and MnVII Beverages (tap water, ice tea, energy drink, mineral water, sprite drink and carbonated drink) ETAAS Magnetic DSPME of MnVII on ZnFe2O4 nanotubes (ZFONTs), then solidified floating organic drop microextraction of MnII with 1-phenyl-3-methyl-4-benzoyl-5-pyrazone (PMBP) in 1-dodecanol To 20 mL solution (adjusted to pH 4) was added 40 mg of ZFONTs. MnVII retained on ZFONTs was desorbed with 0.5 mL of 0.4 mol L−1 NaOH. To 10 mL upper solution was added 50 μL of 0.8 mmol L−1 PMBP in 1-undecanol. After extraction and freezing (ice bath) the solidified organic solvent was transferred and diluted to 100 μL with EtOH. MnII was found in all samples, but MnVII was not found in any samples 0.005 (MnII) and 0.007 (MnVII) CRM Institute of Reference Materials of State of Environment Protection Agency (Beijing) GSBZ 50019-90 (water) and spike recoveries 67
Ni, Zn Food (hydrogenated edible oils, fish, and milk) FAAS Ultrasound-assisted LLME of 1,10-phenanthroline complexes into a hydrophobic DES tetrabutylammonium chloride–decanoic acid (1[thin space (1/6-em)]:[thin space (1/6-em)]2) Different digestion procedure for each sample. Oil (5 g to 25 mL), fish (5 g to 50 mL), milk (dried 2 g in 10 mL). To sample (10 mL) at pH 6, 0.6 mL 1% 1,10-phenanthroline, and 0.8 mL DES were added, sonication (3 min 25 °C), and centrifuged (2 min 4000 rpm). DES phase made up to 5 mL with MeOH. Zn was found in all samples, but Ni only in milk. The text contains a number of inconsistencies 0.029 μg kg−1 (Ni), and 1.5 μg kg−1 (Zn) Spike recoveries 232
Pb Edible oils (sesame, olive, and canola) FAAS Silver nanoparticles impregnated with a DES of choline chloride and phenol (1[thin space (1/6-em)]:[thin space (1/6-em)]2 mol L−1 ratio) + THF Samples (2.0 g) were digested by heating with 2 mL HNO3 and 1 mL H2O2 and evaporated to near dryness when a further 2 mL HNO3 and 1 mL H2O2 were added. Finally, the solution was diluted to 10 mL and, after adjustment to pH 2, 1 mL of Ag-NPs (0.1 mmol L−1) were added followed by 500 μL of DES. Following vortexing, THF (500 μL) was added and the mixture ultrasonicated (1 min) and centrifuged (3500 rpm for 2 min). The separated DES layer was diluted to 5 mL with EtOH and then HNO3 was added to give a final concentration of 0.1 mol L−1. The final volume was not given. Pb was found in all three samples 0.28 Spike recovery 233
Pb Water (spring, well, tap, river, mineral) ETAAS DSPME with in situ synthesis of magnetic Fe3O4–PbS nanocomposites To sample (20 mL) containing 1 mol L−1 NaOH and 0.8 mmol L−1L-cysteine was added 0.3 mL of the concentrated Fe3O4NP slurry, and sonicated for 1 min. Magnetic separation (5 min) and resuspension in 250 μL of water; 12 μL + 5 μL palladium nitrate modifier injected into furnace. Analyte detected in all samples but <LOQ (0.07 μg L−1) 0.02 Spike recoveries 234
Pb Water (tap, lake), urine, blood serum ETAAS SPE on carboxylate-functionalised magnetic adsorbent was synthesized by calcination from waste toner Biological samples diluted 100-times. All adjusted to pH 6. To 50 mL was added 5 mg WT-500, dispersed and ultrasonicated (5.0 min). Magnetically separated and retained Pb dissolved in 0.5 mL 0.2 mol L−1 HNO3. Analyte was found in all four real samples 0.003 CRM GSB 07-1183-2000 (environmental water) and spike recoveries 235
Pb Bovine liver and mussel tissue, kidney, muscle, lungs ICP-MS SPE on lemon peel modified with magnetic Fe3O4 Samples (25–100 mg) MAD with HNO3 made up to 10 mL. Adjusted to pH 5, 50 mg of Fe3O4 – lemon peel added and shaken (1 min). Magnetic separation and Pb dissolved in 2.5 mL of 3% HNO3. Lead found in all samples except lungs 39 ng L−1 CRM BCR 185 R (bovine liver), NIST SRM 2976a (mussel tissue) 236
REE (Lu, Pr, Sm, Tb, Tm, Yb) Food RM (scallop, green tea, corn) ETV-ICP-MS SPME on fibrous graphitic-C3N4@TiO2 nanocomposites (FGCTNCs) coupled with solidified floating organic drop LLME Sample (1.0 g) subjected to MAD with HNO3–H2O2, final volume not given. 30 mL was repeatedly passed through 30 mg of FGCTNCs in the needle hub of a syringe. The retained REEs were eluted with four 2.0 mL 0.5 mol L−1 HNO3. The REEs in 5 mL of this solution were extracted (50 μL pivaloyltrifluoroacetone in 1-dodecanol) by shaking at 45 °C in a 10 mL syringe. After cooling in an ice bath, the aqueous phase was expelled and the viscous organic phase diluted to 80 μL with THF. Only reference materials were analyzed 0.002 to 0.065 ng L−1 CRM: Institute of Geophysical and Geochemical Prospecting including GBW 10024 (scallops), GBW 10052 (green tea) and GBW 10012 (corn) 237
SbIII and SbV Rice wine ETAAS SPE with ZnFe2O4 nanotubes (ZFONTs) and solidified floating organic drop microextraction Samples were degassed, diluted (5-fold) and filtered. 40 mg of ZFONTs were packed into the needle hub of a syringe and the SbV in 20 mL of sample (pH 5) was extracted by repeated passage (7 times in 4 min) through the extractant and the dissolved in four 300 μL portions of 0.5 mol L−1 HNO3. The SbIII in 5 mL was extracted (50 μL DDTC in 1-dodecanol) by shaking at 40 °C in a 10 mL syringe. After cooling in an ice bath, the aqueous phase was expelled and the viscous organic phase diluted to 100 μL with EtOH. Both species were found in three real samples 0.005 (SbIII) and 0.003 (SbV) CRM GSB-07-1376-2001 (water) and spike recoveries 68
Se Fish (mullet, Atlantic bigeye) HG-AAS SPE on magnetic NP of graphene oxide GO γ-Fe2O3 Samples (600 mg) MAD with HNO3 adjusted to pH 2, made up to 40 mL, then 60 mg of magnetic NPs added and shaken for 30 min. Magnetic separation and Se eluted with 500 μL of 0.1 mol L−1 EDTA (pH 12). Se found in all samples 70 ng g−1 CRM NRCC TORT-2 (lobster hepatopancreas), ERM-BB422 (fish muscle), NIST SRM 1577a (bovine liver), SRM 1577b (bovine liver) 238
SeIV Food (rice flour, mushroom, soya, corn flour, broccoli, pumpkin, buckwheat flour, oat flour, egg, tomato, brown rice, green tea, canned tuna, canned shrimp and chicken liver) and water (tap, river, mineral, and well) HG-AAS LLME of 2-(3,4-dihydroxyphenyl)-3,5,7-trihydroxychromen-4-one (quercetin) complex into a DES Six different DES were investigated. Samples (1 g) subject to MAD with HNO3–H2O2 and diluted to 50.0 mL. To 10 mL of sample solution (pH 5.5), 10 μL of 100 mmol L−1 of quercetin and 400 μL of a mixture of menthol and lauric acid were added. After sonication and centrifugation, the aqueous phase was separated by decantation. The remaining phase was diluted to 2 mL with acidic MeOH. Analyte was found in all samples except the tap and river waters 0.0003 CRM: NIST 1567a (wheat flour) and 1548a (typical diet) and spike recovery 59
SeIV and SeVI Food (cow milk, soy milk, milk tea and yoghurt) ETAAS Magnetic DSPME of SeVI on ZnFe2O4 nanotubes (ZFONTs) and DLLME of SeIV with APDC into CHCl3 with EtOH disperser To 6 mL of sample was added 40 mL of artificial gastric juice. Finally diluted to 50 mL with dilute HNO3. To 20 mL of this solution (pH 2), were added 15 mg of ZFONTs after separation, the SeVI was dissolved in 0.5 mL of NaOH. To the supernatant, 100 μL APDC, 0.3 mL EtOH (as disperser), and 50 μL CHCl3 were added. Both species were found in all samples. SeVI was not detected in the CRM 0.001 CRM GBW 10017 (milk powder) and spike recoveries 69
VV and total V Water (mineral river), food (cabbage, carrots, mint, tomato) ICP-OES CPE with back-extraction. Extraction of VV as complex with bis(3,4-dihydroxybenzylidene)isophthalohydrazide (DHBIP) in Triton X-114. Back-extracted with 1.0 mL of 1.0 mol L−1 HNO3 Sample (0.5–1.0 g) MAD with HNO3 and H2O2. Made up to 25 mL with pH adjustment to 7.0. Sample (5 mL) mixed with 500 μL of 1 mmol L−1 of DHBIP, 500 μL of 2.5% (v/v) Triton X-114 and 2.0 mL of hexamine buffer (pH 7.0) and diluted to 50 mL. Incubated at 45 °C for 15 min, centrifuged (4000 rpm, 3 min). Transferred to ice bath and phases separated. Surfactant-rich phase volume made up to 1.0 mL with 1 mol L−1 HNO3 followed by heating to 45 °C and centrifuging. Analyte found in all samples 0.12 Spike recoveries 70
Zn Milk (liquid and powder) and water (tap, river, ground and spring) FAAS Magnetic DSPME with a nickel oxide/nickel ferrite (NiO/NiFe2O4) nanocomposite In the case of the milk, 0.5 mL was heated with 5.0 mL of HNO3 and 3.0 mL of H2O2. Then, the volume was reduced to 0.5 mL and diluted with water (volume not given). The pH was adjusted to 6. In the case of milk powder and the baby formula, 0.05 g was mixed with 4.0 mL of HNO3 and 10.0 mL of H2SO4 and was heated until the volume was about 0.5 mL. The digest was diluted (volume not given) and the pH adjusted to 6. For pre-concentration, 100 mg of the magnetic nanocomposite was added to 150 mL of sample and the mixture agitated for 15 min. After magnetic separation, the adsorbed Zn was dissolved in 1 mL of 2.0 mol L−1 acetic acid and filtered. The material could be reused 100 times. Zn was found in all six samples 0.04 CRM NIST SRM 1549 (non-fat milk powder) 239
Various (20) Oil (vegetable, mineral) ICP-OES LLE with DES of choline chloride, lactic acid and water Sample (5 g) mixed with 0.05 g DES, 60 °C for 20 min. Centrifuged (5 min at 6000g) and 0.1 g lower DES phase dissolved in 9.9 mL of water. Twelve analytes <LOD in sunflower oil; 14 analytes <LOD in motor oil. Not really pre-concentration, but demonstration that MAD is not necessary 0.02 to 17 mg kg−1 CRM 10066-2012 (motor oil). Comparison of results with those of a MAD method 240


One of the limitations of SPE and LLE procedures is that they are time-consuming, requiring considerable operator involvement in a series of manual manipulations, and thus the report of an automated CPE procedure is noteworthy. Akiba et al.60 devised a computer-controlled multi-commuted flow-batch system for the pre-concentration of Cd from water (tap, ground, mineral and ‘reuse’) samples with determination by HR-CS-FAAS. The procedure was based on the extraction of the pyridyl-azo-naphthol complex into a Triton X-100 enriched phase, which was stacked in a short plug in a small column of hydrophobic cotton functioning as the interface between the extraction and detection stages. The throughput was 6 h−1 and the LOD was 1.3 μg L−1 The method was validated by the analysis of a drinking water CRM (EP-H, supplier not given) and spike recoveries.

A co-precipitation procedure has been developed for the determination of Cd in water (well tap and mineral) and in infant formula by FAAS.61 The latter was first subjected to MAD: 0.5 g of sample in a final volume of 100 mL of diluted HCl. After pH adjustment and heating to 50 °C, KHP and iron(II) (500 mg L−1) were added followed by 0.6 g of NaBH4. The mixture was stirred for 4 min at 100 rpm. The black suspension of zero-valent iron particles containing co-precipitated Cd were magnetically separated and dissolved in 0.2 mL of 6 mol L−1 HCl. The method was validated by the analysis of CRM NIST SRM 1643e (trace elements in water) and spike recoveries. Infant formula was analysed by a second procedure (EN 13805:2014), though it is not clear to what extent this differed from the co-precipitation method. The LOD was 0.1 μg L−1, 880 times lower than the LOD without pre-concentration (even though the enrichment factor was only 430).

A pre-concentration procedure, based on tandem electromembrane extractions has been devised for the determination of CrVI in food (milk powder, fish tissue and basil) by ETAAS.62 Samples (1.0 g) were first dry-ashed (550 °C for 4–5 h), dissolved in 100 mL of diluted HCl and the pH adjusted to 5.0. An 8 mL subsample was subject to the first electromembrane procedure, in which species migrated across a supported liquid membrane (1-octanol containing Aliquat-336 as carrier) into 40 μL of acceptor solution that acted, following acidification, as the donor solution for the second procedure, in which species migrated across 1 μL of the same organic solvent, acting a free liquid membrane, into 10 μL of acceptor solution. Separation from CrIII was assured as the two Cr species bore opposite charges in the separation stages. The LOD was 0.003 μg L−1. The method was validated by the analysis of CRM NIST SRM 2700 (hexavalent chromium in contaminated soil), which contains 14.9 ± 1.2 mg kg−1 and for which a relative measurement error of −11% was reported that is not significant, and by spike recoveries. The concentration in the basil was below the method LOD, which is presumably 0.3 μg kg−1, though this was not calculated by the authors.

Several researchers have devised non-chromatographic speciation procedures based on separation and pre-concentration of target species by SPE. Although this approach is limited in the number of species that can be determined, it has the advantages of (a) avoiding the inherent dilution of an HPLC procedure, and (b) consenting the possible use of measurement techniques other than ICP-MS. Such procedures include the determination of iAs species in shrimp,63 iAs in dairy products,64 AsIII in water, urine and fruit juice,65 CrIII in tap water,66 Mn species in beverages,67 iSb in rice wine,68 iSe species in dairy products,69 SeIV in foods,59 and VV in water.70

4. Progress with analytical techniques

4.1 Mass spectrometry

The large number of publications describing ICP-MS-based methods over the review period demonstrates the wide use of this technique in elemental analysis for applications covered by this review. It was noteworthy that ICP-QQQ-MS, with the advantages that it provides for interference removal, is now utilised in many routine applications. The main themes encountered this year were reduction of matrix effects through choice of sample introduction system and calibration/internal standardisation strategies, simplification of highly accurate methods to make them accessible to laboratories with standard ICP-MS equipment and automation of pre-analytical steps to attain faster and less labour-intensive methods.

There was interest in mitigating matrix effects in ethanol-containing samples this year with two papers reporting different approaches. The first study utilised a high temperature torch integrated sample introduction system.71 A spray chamber temperature of 300 °C, extraction lens voltage of 5 V and torch injector inner diameter of 2.5 mm removed all aerosol transport interferences and achieved 100% transport efficiency, allowing use of 25% EtOH calibration standards. However, it was necessary to optimise the plasma sampling zone to reduce interferences at the expense of sensitivity to achieve accurate results. Under optimised conditions, spiked recoveries for a number of elements, including As, Cd, Co and Ni, in bioethanol, wine and spirit beverages, ranged from 80 to 115%, although RSDs were relatively high at approximately 10%. In the second work, matrix effects due to EtOH and variability between individual samples were reduced through application of a multi-isotope calibration strategy to undiluted wine samples.72 The calibration strategy is based on the multi-energy calibration developed originally for ICP-OES and utilises measurements of at least two isotopes of an element in a pair of samples; one diluted 1 + 1 with a spiked analyte solution of a known concentration and the other diluted 1 + 1 with a blank solution. Similar LODs were obtained to a conventional internal standardisation method except in the case of Pb, for which the LOD was 20-fold lower. When applied to real wine samples, concentrations of Cu, Fe, Pb, Sn and Zn compared well with the internal standardisation method while Cr, Ni and Sr were slightly overestimated.

Sample introduction strategies for sampling small volumes of biological fluids for ICP-MS analysis were the focus of a review paper this year and encompassed microflow nebulisers coupled with small spray chambers, total consumption systems, flow-injection, ETV and sampling of dried matrix spots by LA.37 Representative applications, including in blood, CSF, intraocular fluids, urine, saliva, lacrimal fluid and sweat, were also presented.

Sample introduction is a key area of development in time resolved ICP-MS analysis, which includes scICP-MS and spICP-MS. In a study aiming to quantify the amount of Pt from chemotherapy drugs incorporated into single human cells, along with several endogenous elements (Cu, Fe, P, Se and Zn), a commercial sample introduction system consisting of a narrow internal diameter nebuliser (250 μm id) and a total consumption spray chamber was assessed. With introduction of a cell solution of 5 × 104 cells per mL at a flow rate of 10 μL min−1, a transport efficiency of 15% for intact cells was achieved.73 Tandem mass spectrometry was utilised to eliminate interferences, using O2, as the reaction gas, for P and S, and NH3 for Fe, respectively, to generate product ions that were monitored “off-mass”. Interestingly, when monitoring the reaction product ions, there was an approximate 4-fold increase in signal duration compared with that for the equivalent atomic ions necessitating careful consideration of dilution factors to avoid double events. A LOD of 0.004 fg per cell was reported for Pt and a range of 0.02 to 0.4 fg per cell for the endogenous elements. The results obtained were comparable with those from bulk digestion analysis. Wu et al.74 described a high throughput two-dimensional cytometry platform for single cell analysis consisting of a Dean flow-assisted vortex capillary cell sampling unit, to pre-focus and monodisperse the single cell stream, coupled to both a LIF detector for traditional flow cytometry mLIF measurements and scICP-MS for mass detection via a 107Ag label. A throughput of 43.13 cells per s for LIF and 12.65 cells per s for scICP-MS was achieved with the proportion of effective signals being greater than 96.2%. In a novel application of scICP-MS and spICP-MS, discussed further in Section 6.5.1, TiNP release from in vitro degradation of dental implants as well as incorporation of the NPs into single osteoblasts was investigated.75 A relatively long dwell time of 5 ms was employed but experiments using different concentrations of TiO2 suspensions demonstrated that only single particles were detected. For spICP-MS, a transport efficiency of 4.25% was achieved with a flow rate of 0.32 mL min−1 while for scICP-MS, the commercial sample introduction system utilised with a cell suspension of 5 × 104 cells per mL and a flow rate of 10 μL min−1 effectively functioned as a total consumption system and transport efficiencies were in the range of 50% to 60%.

Two approaches to traceable characterisation of protein and peptide standards via determination of heteroatom, S, were described this year. This is an important issue because commercially available standards for proteins/peptides often have variable purity and are not always adequately characterised to enable use as calibration standards. A highly accurate method for measuring peptide content of β-amyloid was reported by Schaier et al.76 The method involved oxidation and hydrolysis of the peptide and then separation of S-containing amino acids, methionine and cysteine, using a strong anion-exchange column before determination of S using ICP-QQQ-MS in O2 mode. An 34S-labelled yeast hydrolysate was used for species specific ID to account for systematic errors, sample loss and any transformation of species during the sample preparation process. Recoveries of almost 100% were reported for an amino acid SRM (NIST 2389a) and RSD values were less than 4%. Meanwhile, another report described a simplified and accessible method for characterising protein standards that would be applicable to any S-containing protein and which avoided the need for a hyphenated chromatography system or a specific 34S labelled protein standard.19 Quantification of S was performed using non-species specific ID (inorganic 34S spike) and SF-ICP-MS. To allow correction for the presence of S contaminants in the protein solutions, non-protein bound S was isolated using membrane filtration and subsequently quantified. The analysis required only 200 μg of protein sample. Validation of the method was conducted using SRM 927e (bovine albumin), where results were comparable to other published methods, as well as comparison of measured Tau protein concentrations with those obtained by aromatic amino acid analysis.

A strategy to simplify a reference ICP-MS method for the determination of K and Se in human serum to make it more accessible for laboratories with standard ICP-QMS equipment followed a two-step ID procedure, which in contrast to conventional ID obviates the need to accurately pre-determine the concentration of the enriched isotope in the ID spike by SF-ICP-MS.77 Interferences were removed using the less common reaction gas CH4. The method achieved LODs of 0.8 mg kg−1 and 2.7 μg kg−1 for K and Se, respectively, RSDs were <0.2 and <0.7% and deviations from certified values were <0.9% in human serum SRMs.

An interest in developing methods for rapid determination of radionucleotides in urine using ICP-QQQ-MS was again evident this year. A fully automated system for separation, pre-concentration and measurement of multiple radionucleotides (90Sr and isotopes of U, Am and Pu) required less than 10 mL of urine and had a rapid analysis time of 46 min per sample.78 An automated flow injection set-up with an automated two-column ion chromatography system (DGA-branched resin and Sr resin) was used. Two sample introduction systems were compared: APEX-Ω, which increased sensitivity, and a double pass spray chamber, which although less sensitive, was taken forward because it facilitated faster analysis times. Different gases (O2–He, He, O2–H2) were used sequentially in the DRC of the ICP-QQQ-MS to successfully eliminate polyatomic interferences. Following optimisation of instrument sensitivity, LODs were comparable to, or lower than, those reported in other studies, ranging from 0.56 pg L−1 for 239Pu to 1.75 pg L−1 for 234U. Urine spiked recoveries ranged from 95.9 to 109.3% for all analytes with a notable improvement when U and Pu results were corrected for using 233U and 242Pu tracer recovery. In contrast, Wu et al.79 focussed on determination of ultra-trace Pu in 1 L urine samples, a volume that would require a 24 h collection rather than a spot sample. A relatively involved sample preparation procedure with respect to the previous paper taking 8 h per sample was described, which comprised co-precipitation with HTiO2 and digestion in HNO3–H2O2 before separation of Pu using an anion-exchange (AGMP-1 M) and extraction chromatographic resin (TEVA). The high-efficiency APEX-Ω sample introduction system yielded an 8-fold improvement in sensitivity with respect to a spray chamber. A mixture of gases (NH3–He) were used to mitigate tailing of 258U and interference from U hydrides. Excellent LODs of 0.873 fg L−1 for 239Pu and 0.14 fg L−1 for 240Pu were achieved. The method was validated through analysis of urine spiked with 239Pu and 240Pu.

The use of matrix matched calibrators to improve accuracy of LA-ICP-MS measurements in solid materials was the focus of two papers over the review period. Novo et al.80 reported a well-conducted study in which in-house Br and I-spiked hair strands were used as calibrators to determine concentrations and spatial distributions of these halogens in human hair. Highly linear calibration curves were obtained and longitudinal and cross-sectional scanning studies proved that the calibration material was adequately homogeneous. Isotope, 34S, which is naturally present in hair, was used as an IS to correct for temporal differences in ablation rate and instrumental drift. A low-dispersion ablation cell and aerosol rapid transport system provided a single pulse response duration of some 1 ms to allow spatially resolved LA-ICP-MS analysis. The results obtained agreed well with those from pneumatic nebulisation ICP-MS following digestion by microwave-induced combustion. The LODs achieved were 0.36 μg L−1 and 0.14 μg L−1 for Br and I, respectively. Application of the method identified higher concentrations of I in hair from individuals on synthetic thyroid hormone replacement versus controls. The second paper was concerned with the matrix of multivitamin supplements and strived to develop a rapid and automated method that didn’t require acid digestion.81 Matrix matched calibrators prepared using a multivitamin reference material together with a multi-element fortified cellulose powder were compared to those composed of the multi-element cellulose powder alone. Normalisation of the results to an IS was essential to correct for different ablation yields arising from differences in focus on the surface of the samples and the best choice of IS varied between the two calibration strategies. Although overall improved recoveries were obtained using the matrix matched calibrators for S and Fe and similar recoveries were observed for the other elements studied, the high concentrations of some of the target elements present in the multivitamin matrix meant that only a single point calibration could be used.

There were a substantial number of papers describing speciation methods within the scope of this review. Of note were two reports exploring different calibration strategies in speciation analysis to avoid the impracticalities of gold standard species specific ID, e.g., cost or availability of isotopically enriched species or measuring monoisotopic elements. Petrov et al.82 employed a novel on-column IS strategy, which involved chromatographic injection of the analyte species as an IS following the sample injection. Analysis of a standard containing MMA and AsV demonstrated that the approach may be used for multiple species as long as there is adequate resolution of the species. The strategy was successfully applied to iAs in rice for which an expanded uncertainty of <10% was achieved for mass fractions of 60 to 300 μg kg−1 iAs. In other work, the compound independent calibration of five Se species in rice using ion-pairing reversed phase chromatography coupled to ICP-MS was found to be a satisfactory alternative to species specific ID.83 The approach relies on an identical response for a particular element regardless of its chemical form. Optimal enzymatic extraction efficiency whilst maintaining the integrity of the original species was crucial. Spiked recoveries in rice were satisfactory (96.1 to 102.9%) for four species (SeIV, SeVI, MeSeCys and SeMet) but low for SeCys2 (66.1–77.1%), which was explained by the lack of stability of this species in the rice matrix. The method achieved RSDs of less than 7.0%. The approach was applied to measuring the five Se species in Chinese rice and validated against external calibration as well as total Se concentrations.

Other developments in the area of speciation included a comparison of three MS techniques for determination of MeHg, EtHg and iHg in biological CRMs by triple spike species-specific ID-MS.84 Perhaps unsurprisingly, GC-ICP-MS demonstrated superior accuracy, precision, LODs and matrix tolerance over molecular techniques, GC-EI-MS and GC-EI-MS/MS, particularly in a blood matrix. Low sensitivity was observed for GC-EI-MS/MS, despite its higher selectivity, and this method could not be recommended for routine use. Two rapid automated arsenic speciation methods in urine were developed by Quarles et al.85 One separated and detected six As species using a single column with an impressive separation time of around 2 min, while the second method utilised an additional column to also separate TMAO, which increased the separation time to approximately 4.5 min. In urine, LODs ranged from 2.8 to 9.1 ng L−1 and recoveries for the As species were between 94% and 107%. Accuracy was verified through PT schemes. Another interesting study, this time involving simultaneous speciation of As and Hg in fish by HPLC-ICP-MS with a run time of less than 4.5 min, reported impressive LODs particularly for the As species (0.005 to 0.007 ng As per mL and Hg species: 0.013 to 0.015 ng Hg per mL).86 Use of O2 reaction gas improved sensitivity both for As, detected as reaction product, [75As16O]+, and for Hg, detected “on-mass”, the latter presumably through collision damping. Spiked recoveries were between 97% and 103%, while RSDs were less than 7%. Following enzymatic digestion, the method was applied to CRM, NRCC DORM-3 fish protein, for which the sum of the species was in concordance with the total As and Hg certified values, and to local fish.

Isotope ratio determination in clinical samples was also the subject of several papers with most applications making use of MC-ICP-MS due to its high throughput and ionisation power. The current interest in the capabilities of high precision isotope measurement as a diagnostic and prognostic tool was captured in a review by Vanhaecke and Costas-Rodríguez87 although as yet, the technique has little in the way of routine clinical applications. The main areas of potential application discussed were disorders affecting iron homeostasis, bone diseases, liver disease, cancer, neurodegenerative disorders and diabetes mellitus. As well as a need for larger studies to ascertain findings, simpler and more automated sample preparation procedures are necessary to render the technique more accessible to routine laboratories. Meanwhile, the performance of a new pseudo high sensitivity MC-ICP-MS instrument (Thermo Scientific Neoma), which incorporates 11 independent Faraday cups connected to 20 amplifier slots, for measuring K isotope ratios (41K[thin space (1/6-em)]:[thin space (1/6-em)]39K) in clinical samples under hot plasma conditions was assessed.88 A relative insensitivity to matrix effects was accomplished through the newly designed sample introduction system integrating an iCAP Qnova Series ICP-MS torch and injector assembly. Intermediate precision was reported as 0.07‰. Following a one-step chemical purification of K by ion exchange chromatography, the method was applied to determine δ41K in biological CRMs with good agreement reported, as well as to blood from patients with acute myeloid leukaemia and controls. Finally the gold standard method for isotope ratio measurement, MC-TIMS, featured in a 42Ca–48Ca double spike method to determine Ca isotope ratios in small tissue samples.89 A low Ca blank of ≤10 ng, achieved through optimisation of the separation of Ca from the matrix and isobaric interferences using DGA resin, reduced the minimum mass of Ca required to 670 ng, which is equivalent to 2.3 μg bone material or 10 μL blood. The reported expanded uncertainty for the procedure for δ44/40Ca was 0.08‰. The δ values for bone and human hair SRMs were in good agreement with published data and δ values for three hair CRMs were reported for the first time.

4.2 Atomic absorption and atomic emission spectrometry

As an alternative to ICP-MS, ETAAS offers some advantages in situations where (a) sample amount is limited, (b) only a limited number of elements are to be determined and (c) matrix interferences can be overcome. In particular, HR-CS-AA spectrometers are capable of simultaneous determinations of more than one analyte and offer spectral interference correction modes not available with line-source instruments. Nakadi et al. devised a time-absorbance profile ratio correction of spectral overlaps in HR-CS-ETAAS.90 They considered that the time–absorbance profile (TAP) of a species, measured under the same instrumental conditions, should be the same at every wavelength measured. Therefore, subtracting a TAP normalised spectrum of the interfering species from the normalised absorbance of the atomic line, would be leaving only the analytical signal of the analyte. They explained in detail how the TAP method could be deployed without the need to perform any additional measurements or know in advance the nature of the overlapping species. They accurately determined Pb in NIST SRM 1570a (spinach leaves) and NIST SRM 1577b (bovine liver), Ni in NIST SRM 1573 (tomato leaves), and Cu in dried plasma spots produced from Seronorm whole blood.

There have been several reports of determinations by HR-CS-AAS, including that of Cd in waters60 by FAAS following automated CPE (see Table 1 and Section 3.2) and of Cd and Pb in canned food91 (see Table 1) also by FAAS. In addition, there have been several reports of determinations by HR-CS-ETAAS. A method for the determination of Pt in human pleural effusions has been devised92 in which the diluted sample was ashed in the presence of O2 in the atomizer. The LOD was 1 μg L−1, and the method was validated by comparison of the results with those obtained by an ICP-MS method. The researchers showed that the simultaneous determination of Fe at concentrations around 1 mg L−1 (the usual serum value) was also possible. A method for the direct determination of Cr, Fe and Ni in powdered edible seed (pumpkin, amaranth, chia, sunflower and quinoa) samples has been devised.93 Samples (0.35–0.65 mg) were weighed directly onto the solid-sampling graphite platforms and thermally pretreated in the presence of a combination of Pd–Mg and H2O2. Then, Cd was determined at 228.802 nm, using 400 °C and 1600 °C as pyrolysis and atomisation temperatures, respectively, followed by in sequence, the measurement of Ni (232.003 nm) and Fe (232.036 nm), using 2500 °C as the atomisation temperature. The LODs were 3.1 μg kg−1 (Cd), 3.4 mg kg−1 (Fe) and 0.062 mg kg−1 (Ni), and the method was validated by the analysis of NIST SRM 1573a (tomato leaves). Ten samples were analysed: Fe and Ni were found in all of them, but Cd was below the LOQ (10 μg kg−1) for three of them. A similar procedure was devised by Bustos et al. for the determination of Cd and Ni in powdered cocoa (although the title of the paper indicates that the samples were chocolate),94 though in this case the samples were handled as slurries (100 mg particle size < 45 μm plus 10 mL of 2% HNO3 and 0.1% Triton X-114) of which 20 μL was injected onto the platform, which had been pretreated with 2.0 μg of palladium. A slow heating ramp of 400 °C s−1 was applied, allowing the simultaneous measurement of Cd at the principal line (228.8018 nm) and of Ni (at a secondary line 228.9984 nm). The LODs were 0.027 and 0.22 μg g−1 for Cd and Ni, respectively and the method was validated by the analysis of the same CRM, NIST SRM 1573a (tomato leaves), and by spike recoveries from real samples. The method was applied to the analysis of five cocoa samples, in all of which both analytes were found.

Two research groups have reported on the determination of Cd, as part of a multi-element analysis, by line-source ETAAS. Akbaba et al. determined Cd and Se in a range of food samples (eggs, white cheese, milk powder, pollen, thyme, green pepper, mint, green beans and Turkish coffee).95 Samples (1.0 g) were dissolved by MAD with HNO3 and H2O2 and made up to a final volume of 10.0 mL. The researchers investigated a number of modifiers, settling on 20 μg of nickel + 4 μg of platinum as the optimum, allowing pyrolysis temperatures of 500 °C for Cd and 900 °C for Se, with atomization at 1700 and 2300 °C, respectively. The LODs were 0.5 μg L−1 (Cd) and 0.4 μg L−1 (Se), and the method was validated by the analysis of CRMs, IAEA-155 (whey powder) and IRMM BCR-150 (skim milk powder). Only Se was determined in the BCR-150 material. Both analytes were found in all samples. For the determination of As, Cd and Se in medicinal plant raw materials, Nikulin et al.96 considered the “universal” modifier of Pd(NO3)2 and its mixtures with Mg(NO3)2 to be “extremely expensive” and, furthermore, may not always be available to laboratories involved in the analysis of natural substances. They proposed some simpler alternatives: a 1% ascorbic acid solution, as a modifier for Pb, to reduced it to Pb0, and a 1% solution of Ni(NO3)2 which forms intermetallic compounds with volatile elements. They showed that no modifier was needed for the determination of Cd. Samples (0.4 g) were dissolved by MAD with HNO3–H2O2 and made up to a final volume of 100 mL. Some investigation of the digestions conditions was carried out, to ensure the maximum extent of degradation of the matrix. The pyrolysis temperatures were 1400 °C, 350 °C and 400 °C for As, Cd and Pb, respectively with atomisation at 2500 °C, 2100 °C and 1900 °C. No details of the instrument were given, and so it is not clear if atomisation was from a platform or from the wall of the tube. The procedure was applied to 32 plant samples. Cadmium was found in most of them, Pb in some, and As only in one (sea kale, Thalli laminariae) at a concentration of 8.5 mg kg−1.

A method for the determination of Au NPs in biological tissue by ETAAS first converted all of the Au to dissolved species via MAD.97 The researchers devoted some effort to the optimisation of the digestion procedure, in which 200 mg of sample was digested with 1.5 mL of 65% HNO3 and 9.0 mL 37% HCl with a final volume of 20 mL. The modifier was 5 μL of a solution containing 5 μg of palladium and 3 μg of magnesium nitrate, with pyrolysis and atomization at 800 and 1800 °C, respectively. They showed that the same calibration was obtained for both AuNP and dissolved Au and validated the procedure by spike additions of NP. The LOD was 0.6 μg L−1 and the method was applied for the determination of AuNPs in six biological tissues (liver, small intestine, heart, lungs, brain and kidneys). The results were not given in detail but summarized as showing that Au accumulated at “low level in liver”, while for the other organs, the Au concentration was below the LOQ of 13 μg kg−1.

Parsons and co-workers have updated their 1993 method for the determination of Pb in blood by ETAAS in the light of the decrease in Pb concentrations in children’s blood in the USA, so as to be fit for purpose to support measurements at the 2021 CDC reference value of 3.5 μg dL−1.98 Their goal was an appropriately lower LOD and applicability to a lower concentration range of 1–40 μg dL−1 that better reflects current population blood Pb concentrations. They used two different instruments (both with transversely heated atomizers with integral platforms and Zeeman background correction), for which slightly different optimum operating conditions and performance characteristics were obtained. A major feature of the updated method is the use of a tungsten–rhodium permanent modifier: the coating procedure consisted of depositing 250 μg of tungsten and 200 μg of rhodium on a pyrolytically-coated graphite tube in a series of drying steps, which takes approximately 45 min, but which is then suitable for approximately 350 firings (equivalent to an 8 h workday). The researchers pointed out that tubes could be recoated to extend lifetime up to 1000 firings, which may be compared to the 350 firings for the established GFAAS method in which a phosphate modifier was used. They also evaluated sample volume and sample dilution, settling on 16 μL and 1 + 9, respectively. They noted that with the permanent modifier, double peaks could not be avoided with aqueous standards and so they considered matrix-matched standards to be mandatory. These were prepared daily as a 1 + 9 dilution of 50 μL of a Pb calibration working stock solution with 50 μL of “base” blood, and 400 μL of Triton X-100–HNO3 diluent. They considered that a “base” blood should be confirmed as having no more than 0.5 μg dL−1 endogenous Pb, and for their study, the base blood was obtained from a caprine source available to the New York State Department of Health, though they considered that any whole blood source (human, caprine or bovine) should suffice. They adopted a two-stage drying procedure (110–130 °C, followed by 150–170 °C) with pyrolysis at 600–700 °C and atomisation at 1400 °C and obtained LODs of 0.16 and 0.21 μg dL−1, depending on the instrument, corresponding to LOQs of 0.5 and 0.7 μg dL−1. They validated the method by (a) analysis of several CRMs (NIST SRM 955c, toxic metals in caprine blood, levels 1, 2 and 3, and SRM 955d, toxic metals and metabolites in frozen human blood, levels 1 and 2); (b) analysis of seven archived PT samples, and (c) comparison of the results for 30 real samples with those obtained by an established ICP-MS method. The researchers concluded that the improved method could benefit hospitals and smaller laboratories in terms of keeping pace with the goals of detecting and quantifying lower blood Pb concentrations by ETAAS instrumentation, especially if an ICP-MS instrument is not available.

Several research groups have described OES determinations based on APGDs. Gorska and Pohl99 determined Ca, K, Mg, and Na in fruit juices (apple, banana, blackcurrant, lemon, lime, pomegranate, quince, and tomato). After a simple 1000-fold dilution of the samples and their acidification to 0.1 mol L−1 with HNO3, by measuring the light emitted when this solution formed the cathode in an APGD sustained between the solution, flowing up and over a tungsten tube, and a solid tungsten anode, by an applied dc potential of 800–1200 V. Matrix effects were encountered for Ca and Mg, for which standards additions calibration was required. The researchers investigated the effects of a number of operating parameters, not all of which were amenable to interpretation: the effect of flow rate, which caused a decrease in signal-to-background ratio at all currents requires further investigation. The LODs were 0.6 (Ca), 0.14 (K), 0.63 (Mg), and 0.02 (Na) μg L−1, and the method was validated by the comparison of the results for real samples with those obtained by an ICP-OES method in which samples (0.5 g) were subject to a lengthy digestion procedure in sealed vessels in a heating block: 5.0 mL conc. HNO3 for 3 h at 120 °C followed by 5.0 mL of 30% H2O2 and further heating for 1 h. Although the final mass was given as 50 g, further dilutions were applied. No statistical evaluation of the results (such as a paired t-test) was presented, so it is not possible to evaluate whether the results were significantly different or not; visual inspection indicates that they are not. All four analytes were found in all 8 samples. The researchers did mention the advantage of omitting a time-consuming sample preparation step, but did not quantify the improvement in throughput so achieved. They also noted that the relevant reagents are not needed. A somewhat similar system was used by Chen and Wang100 to generate Hg vapour that was then transported into a second APGD for determination by AES. The two sources were optimised independently by single-cycle univariate searches with LOD as the figure of merit, which makes interpretation of the results (or comparison with those of Gorska and Pohl) somewhat challenging. The optimised LOD was 0.25 μg L−1. The method was validated by the analysis of the CRM GBW10029 (total Hg and MeHg in fish tissue), issued by the Chinese NMI, and by the analysis of human hair, both of which were also analysed by an ICP-MS method. Samples (500 mg) were digested with HNO3–H2O2 in sealed vessels and eventually diluted to 30 mL. Standard additions were needed for the fish material. Cai et al.101 determined Pb in hair with sample introduction via an induction heating furnace (see Section 4.4). For the determination of Hg species in fish oil, an integrated CV microplasma OES device has been constructed by 3D printing.102 Mercury species contained in fish oil (mainly iHg and MeHg) were extracted into an aqueous medium containing 1.5% (m/v) L-cysteine and 1% (m/v) CH3COONH4 by vortexing for 3 min. Only iHg reacted with BH to release Hg0 and so MeHg was determined as the difference between the iHg and the total Hg concentrations, the latter obtained from the CV reaction, following conversion of the Hg species by reaction with KMnO4 (and destruction of the excess MnO4 by reaction with hydroxylamine hydrochloride). The generated Hg0 was separated with an Ar carrier gas in a GLS and, after a passage through a CaCl2 drier, was introduced into the microplasma, sustained in a 2 mm gap between two tapered tungsten electrodes by a potential difference of 100 V. The emission at 253.6 nm was monitored. The LOD for both species was 0.1 μg L−1. The method was validated by the analysis of CRMs (NRCC DORM-4, fish muscle and DOLT-5, dogfish liver), obtaining relative measurement errors, for MeHg, of −10% and −9%, respectively. The method was applied to the analysis of eight fish oil capsules, none of which contained either species. Spike recoveries at 5 μg L−1 of both species ranged from 84% to 114%. The researchers concluded that their device and methodology would be suitable for on-site Hg species determinations with applications in food safety assessment and customs inspection.

Two research groups have reported on determinations by arc emission spectrometry. Savinov and Drobyshev103 placed a few tens of μL of biological fluid sample (saliva or serum) on the end of one electrode, together with 0.15 mg of NaCl, as spectroscopic buffer, which was then dried under an IR lamp before striking the arc. They quoted LODs for 150 μL of saliva of 1 μg L−1 for Ag, Al, Cd, Cu, Fe, Mn, and Zn and several μg L−1 for Cr, Pb, and Ti. As only 50 μL of blood could be handled, the LODs in this matrix were correspondingly 3-times poorer. They compared the results obtained for saliva with those of an ICP-OES method, which involved MAD with HNO3 and H2O2. For blood, they compared the results with those of ICP-OES, ETAAS and TXRF methods and concluded that the concentrations of Ag, Al, Cr, and Mn could not be reliably determined. To analyse medicinal plant materials, Otmakhov et al.104 dried the samples and reduced them to ashes, then mixed them with graphite powder (ten-fold dilution) and a potassium salt as spectroscopic buffer. The method was validated by the analysis of a CRM (SRS 8923-2007 LB-1, birch leaves, A. P. Vernadsky Institute of Geochemistry, SB, RAS, Irkutsk, Russia), certified for the Ca, K, Mg, Na, P and Si contents. The LOQs were around 0.01 μg g−1. The method was applied to the analysis of five real samples, in which some 24 elements could be determined.

4.3 Laser induced breakdown spectroscopy

The application of LIBS for disease diagnosis has been noted in previous ASU Reviews. Within this period, two review articles have brought together and summarised recent research in this subject. Zhang et al.105 provided a broad overview covering multiple specimen types (e.g. tumours, tissue sections, blood products, etc.) and different disorders from common diseases through to cancer. Quantification and imaging were also discussed. Khan et al.106 focussed specifically on cancer diagnosis and classification. The review examined instrumentation, signal enhancements and quantitative analytical approaches, whilst noting the importance of chemometric methods to support data interpretation. Both highlight the potential of LIBS as a clinical tool. An interesting application of LIBS was described by Berlo et al.107 for the rapid detection of SARS-CoV-2 immune response. The researchers analysed 97 blood plasma samples (50 controls and 47 positive RT-PCR) and compared several LIBS instruments as well as LIBS coupled with ICP-MS. Utilising the combined data sets and chemometric models, it was ascertained that the two groups could be distinguished with 95% accuracy. Furthermore, Ba and Zn were depleted in the SARS-CoV-2 positive samples compared to the controls which was inversely correlated to CN lines from the LIBS spectra. The data provided a curious insight but further work with larger datasets is required. The analysis of hair and nails using a low-cost LIBS system was evaluated by Zhang and co-workers108 for the detection of Ca, Na and Mg. By applying a calibration free approach using standard reference lines, the content ratios of Ca[thin space (1/6-em)]:[thin space (1/6-em)]Na and Mg[thin space (1/6-em)]:[thin space (1/6-em)]Na were determined. The accuracy of this method was compared to that of ICP-OES demonstrating agreement within 10% for both matrices.

There has been continued interest in LIBS as a rapid measurement tool for food and beverage analysis due to the minimal sample preparation required. Tea leaves were the focus in a publication by Rehan et al.109 to determine Al, Ca, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, Pb and Zn. The results were verified by ICP-OES, achieving good agreement. In a similar approach, Ali et al.110 implemented LIBS with the iterative discrete wavelet transform algorithm for the quantification of Ca, K and Mg in oat flour. The results demonstrated linear behaviour (R2 > 0.98) with LODs of 20 ppm. Furthermore, the concentrations determined by LIBS were in good agreement with those from ICP-OES.

This past year has shown the continued interest in utilising LIBS for provenance and authentication investigations. Following on from previous work, Stefas et al. published two papers111,112 concerning the analysis of honey. In the first study,111 LIBS was used to investigate the impact of bee food on the final honey product to determine if additional feeding occurred or to establish adulteration. Inverted syrup (fructose, glucose, maltose), sugar water (sucrose) and candy paste (sucrose) were compared to control feedstocks and the honeys were analysed by LIBS as well as HPLC refractive index detection as the standard method for comparison. The LIBS spectral features were analysed using LDA and random forest classifiers, achieving >90% accuracy for discrimination of the different feedstocks. It was also found that Ca, K, Mg and Na were the most important lines for discrimination. In the second work,112 glucose syrup was used as an adulterant and mixed with different genuine floral honeys followed by LIBS analysis. The use of LDA and randomised tree chemometrics provided a classification accuracy >90%, also showing that Ca, K and Na were important spectral lines. Both papers demonstrated the potential of LIBS for fraud detection in this valuable food commodity. Ginseng is a herb where geographical origin and age play a significant role in its value and is subject to fraud and adulteration. Zhao and co-workers113 applied LIBS in combination with hyperspectral imaging as rapid analytical tools to investigate plant species, geographical origin and age. In the study, samples of American, Chinese, Japanese and Korean ginseng were compared alongside samples from two geographically close regions and of different ages (2 and 4 years). For data processing, PCA was first applied to reduce the dimensionality followed by PLS-DA. The LIBS and hyperspectral imaging datasets were treated separately and combined to form a fusion set. With further optimisation, it was established that the fusion data set provided the highest accuracies of 99%, 99% and 100% for plant species, geographical origin and age, respectively. Additionally, the spectral lines Al(I) (394.424 nm), Al(I) (396.152 nm), Ca(I) (422.672 nm) and P(I) (253.561 nm) were highly ranked and contributed significantly to the model. In a similar approach, Park et al.114 analysed sea salt from Japan, Korea and France, establishing Ca, K and Mg as the key features for classification. It was observed that those with higher levels of Mg also tended to have higher signals for H(I) at 656 nm due to the absorption of water linked with Mg salts such as MgSO4 and MgCl2. The data was subjected to PLS-DA modelling and a two-step modelling approach was implemented to improve the classification accuracy to 98.9%. Finally, in a more unusual application of LIBS, Sezer et al.115 focussed on the classification of cheese and processing technique. Kashar cheese (a Turkish specialty), prepared by heating and stretching the curds, was compared against processed cheese. The LIBS data was combined with ICP-OES and AAS analyses, followed by PCA and PLS-DA modelling. Both approaches achieved excellent discrimination accuracy but PLS-DA prediction ability was found to be 100% whereas PCA was 97%. It was an interesting example of the methodology applied to an unusual foodstuff.

4.4 Vapour generation procedures and atomic fluorescence spectrometry

As for previous review periods, there is continued interest in the application of VG procedures with not only AFS, but also with AAS, OES and MS. Work that is discussed in detail in this section involves some novelty of the VG method and/or the combination of VG with the mode of detection. If the novelty concerns sample pretreatment, the paper is discussed in Section 3.2 or, if pre-concentration by LLE or SPE was involved, in Table 1. In addition to CVG, with either SnCl2 (specifically for Hg) or BH, a further number of reports of PVG have appeared. There were also reports of VG in liquid electrode GDs. Several methods to increase the sensitivity, by trapping the vapour phase species followed by their rapid release, were described, including procedures involving a DBD device, a tungsten coil, as well as the well-known gold amalgam trapping of Hg. Several of the procedures described in this year’s Update were devised with the goal of bypassing the conventional MAD sample preparation step.

Determinations based on the atomic emission from APGDs require that samples are introduced into the discharge as a vapour. In addition to the liquid electrode procedures discussed in Section 4.2,99,100 a procedure using electromagnetic induction heating vaporisation was described by Cai et al.101 for the determination of Pb in a single human hair. The LOD was 31 μg kg−1. The method was validated by the analysis of the CRM GBW09101b (human hair) and applied to the analysis of three real samples, whose results (all well above the LOD) were compared to those obtained by an ICP-MS method, that involved prior digestion of the samples. The researchers pointed out the increased throughput associated with the absence of such a digestion step.

Researchers have used APGDs as vapour generation devices as well as sources for AES. Pan et al.116 generated volatile derivatives of Cd, Hg and Zn in an APGD in argon with a liquid electrode. They referred to their procedure as microplasma-induced VG. The volatile species generated in the plasma were swept into an AF spectrometer for quantification. They showed that significant increases in sensitivity could be obtained by enhanced VG efficiencies (up to 90% in the case of Hg) obtained on the addition of MeOH. The LODs were 0.05 μg L−1, 0.007 μg L−1 and 0.5 μg L−1 for Cd, Hg and Zn, respectively. The method was validated by the analysis of a CRM (NRCC TORT-3, lobster hepatopancreas) and applied to crayfish samples collected from three different provinces of China. All three analytes were detected in all samples, and the results (quantified by standard additions) were compared with those obtained by an ICP-MS method. For both methods, the samples were subjected to MAD. The researchers devoted a significant effort to elucidating the nature of the reaction products and of the mechanism of reactions, concluding that the species of Cd and Zn generated in the presence of MeOH included their hydrides, dimethyl Cd, dimethyl Zn, as well as the free atomic vapours of Cd0 and Zn0. Cai et al.117 generated volatile Hg in a solution cathode GD, delivered, via a GLS and drying tube, to an ICP optical emission spectrometer for quantification. The LOD was 0.22 μg L−1 for measurement at 194.1 nm. The method was applied to the analysis of a CRM, GBW10029 (fish), and to samples of shrimp, crawfish, soil and human hair. No details of sample preparation were given, and Hg was only found in the CRM and two of the three hair samples. Recoveries of spike additions, at single digit μg kg−1 concentrations, to the other samples ranged from 92% to 104%.

Further developments in element determinations in which gas-phase pre-concentration in a DBD device have been reported. To determine As in edible seaweed by DBD-AES with HG sample introduction, Zhang et al. dissolved the dried, ground samples (200 mg) by MAD with HNO3H2O2 in a final volume of 250 mL, of which 10 mL was transferred to a batch hydride generator and the arsine generated by reaction with BH was swept by air and trapped on the interior surface of the DBD device.118 The trapped species were released on the passage of an Ar/H2 mixture, producing a transient atomic emission signal. A modification to previous versions of the instrumentation allowed signals to be captured as peak volumes (intensity × wavelength × time) leading to a four-fold increase in sensitivity and a solution LOD of 0.2 μg L−1, corresponding to 0.25 mg kg−1 in the solid samples. The method was validated by the analysis of CRMs GBW08521 (As, Cd and Pb in laver) and GBW10023 (laver) and by comparison of the results obtained for three real samples with those of an ICP-MS method. Recoveries of 20 mg kg−1 spikes ranged from 103% to 114%. Gu et al.119 trapped the As species, generated by decomposition of slurried seafood samples with a tungsten coil vaporiser, on a DBD device, but then released the species into an AF spectrometer for quantification. The LOD was an unimpressive 40 μg kg−1 (probably because of the extensive dilution of the slurry sampling step) and the method was validated by the analysis of three CRMs GBW08521 (laver) GBW10023 (laver) and GBW10024 (scallop) and applied to the analysis of three real samples that were also analysed by ICP-MS, for which samples were pretreated by MAD.

Applications of PVG continue to be reported. This year, researchers have devised methods for the determination of iAs in rice (and water), halogens in water (by ICP-MS), Cd in rice (by ICP-MS), thiomersal in vaccines (by AAS), and selenium species in mineral water (by HPLC-ICP-MS). For the determination of iAs Wang et al. employed pre-concentration by SPE (see Table 1) on a metal oxide framework, MIL-100(Fe), doped with CdSNPs. This material acted both as adsorbent and photocatalyst.120 The researchers showed that the As species could be separated on the basis of the kinetics of the reaction with the solid extractant: AsV was absorbed and separated after 5 min, whereas AsIII was absorbed only after 105 min. The separated solid was resuspended in formic acid and the arsine generated on flow-through UV irradiation, swept via a GLS into an ICP mass spectrometer. The LOD was 0.11 ng L−1 and the method was validated by the analysis of eight CRMs. Hu et al. showed that volatile derivates of Br and Cl could be generated from solutions containing metal acetates but no organic acids.121 In particular, copper acetate was the most effective, forming the basis of a procedure with ICP-MS LODs of 0.03 and 3 μg L−1 for Br and Cl, respectively. The researchers devoted considerable effort to the optimisation of the vapour generation reaction, the identification of reaction intermediates and elucidation of reaction mechanism, concluding that methyl halides were generated in organic-acid-free media as a result of ligand-to-metal charge transfer between acetate and copper and the charge transfer to solvent excitation of halides. The method was applied to the analysis of bottled water and seawater, for which spike recoveries between 92 and 101% were obtained. For the determination of Cd in rice, Mou et al. devised a procedure in which cobalt played a dual role, both in the sample digestion via a Fenton-like reaction (see Section 3.2) and also as a catalyst in the PVG of volatile Cd derivatives by reaction with formic acid. The researchers found that interferences from coexisting ions were greatly reduced in comparison with those found using either ferric ion assisted PVG or direct solution nebulisation with ICP-MS measurement. Under the optimised experimental conditions, the LOD, for a 0.05 g sample, was 1.6 μg kg−1. The method was validated by the analysis of two rice CRMs GBW(E)100351 and GBW(E)100357 and by spike recoveries from 7 real samples, in all of which Cd was detected at double- or low treble-digit μg kg−1 concentrations. No Cd was detected in a wheat flour sample. Thiomersal, an organomercuric compound used to prevent bacterial proliferation thereby guaranteeing safety and stability in cosmetics pharmaceuticals and vaccines, was determined by PVG of Hg0 from formic acid solutions to which a graphene quantum dot–TiO2 nanocomposite had been added.122 The reaction was carried out in a batch reactor and the Hg generated swept into an AA spectrometer with a multi-pass cell for quantification. The LOD was 15 ng L−1 and the top of the linear working range was 1000 ng L−1. The researchers showed that different Hg species were photo-degraded at different times under UV, with maximum evolution of Hg0 from Hg2+ at 5 min, from EtHg at 9 min, MeHg at 13 min and thiomersal at 16 min. The sequential photo-degradation of mercurial species resulted in temporally resolved profiles, as the evolution of Hg0 from one specific species reached maximum and returned to baseline before the Hg0 from the next mercurial species was released. The method was successfully applied to the analysis of two vaccines containing notionally 100 μg mL−1 thiomersal. The method also allowed the estimation of residual concentrations of Hg2+ and EtHg (due to natural degradation of thiomersal). In a hepatitis B vaccine, the concentrations were 2.7 and 3.2 μg L−1, and in a diphtheria-tetanus vaccine, the concentrations were 2.3 and 3.8 μg L−1, respectively. For conventional CV-AFS determination of thiomersal (aka thimerosal), digestion is needed to convert the organoHg to iHg. de Oliveira et al.123 investigated four possible digestion systems, showing that either KBr/KBrO3 or KMnO4 was suitable as an alternative to conventional MAD with HNO3–H2O2. The LOD was 20 ng L−1 and the method was applied to five vaccine samples with differing thimerosal concentrations, for which the results were in agreement with those of a MAD method (reference method). Inorganic Se species were determined by HPLC-ICP-MS with post column PVG by reaction with formic acid in the presence of a metal–organic framework MIL-125-NH2 as catalyst.124 Extensive optimisation of all the parameters was carried out, resulting in a mobile phase composition of 13.5 mmol L−1 Na2CO3 and 4.5 mmol L−1 NaHCO3, and a column temperature of 40 °C. The LOD for both species was 0.8 μg L−1. The method was applied to the analysis of two samples of bottled natural mineral water, one of which contained no detectable Se species and the other only SeVI at about 20 μg L−1. Recoveries of spike additions of both species to this sample at concentrations of 10–50 μg L−1 ranged from 92% to 107%. The abstract and introduction indicated that the method was validated by the analysis of CRMs, but no results for any such materials were presented.

To improve the sensitivity and hence LOD for the determination of As by HG-AAS, Yildiz et al.125 trapped the arsine, originally generated by reaction with BH in acid solution, at 60 °C on a platinum-coated tungsten coil. The As was then released, by heating to 950 °C, and transported to a flame-heated quartz tube atom cell for quantification by AAS. The LOD was 16 ng L−1 (a 16-fold improvement over the LOD without trapping) and the method was validated by the analysis of CRMs NIST SRM 1640a (trace elements in natural water), EnviroMAT (drinking water), and CRM-023 (Sandy Loam 7), 0.1 g of which was subject to MAD with concentrated acids (6 mL HCl–2 mL HNO3–0.5 mL HF), followed by dilution to 50 mL. The AsV in all samples was reduced to AsIII by reaction with KI. The method was applied to the analysis of four drinking water samples, but in all of them the As concentrations were <LOD. Spike recoveries at 0.25 and 0.50 μg L−1 ranged from 102% to 105%. The tungsten coils could be reused 350 times with no loss in sensitivity.

For the direct determination of Sb in fruit juices Lima et al.126 constructed a flow-batch HG system in which the various reagents and sample could be pumped, under independent control, into a batch reactor, from which the generated stibine was swept via a dryer to an atomic fluorescence spectrometer for quantification. The relevant parameters were optimised through a Box–Behnken design, and HCl was selected as the best acid, leading to an LOD of 20 ng L−1. The method was applied to the determination of total Sb in six commercial samples of grape juice and the results were compared with those of a batch HG-AFS method based on dry mineralisation with MgO–Mg (NO3)2. No significant differences were found based on a paired t-test. The concentration of total Sb in the grape juices ranged from 1.23 to 4.58 μg L−1. Spike recoveries at 0.5, 2, and 4 μg L−1 ranged from 89% to 116%. The researchers argued that with a throughput of 87 h−1, and a total waste per determination of 1.15 mL the method is a fast and ecofriendly tool for determining Sb in grape juices.

There have been relatively few reports of post-column HG in speciation studies. In an extensive study of the determination of Hg species in blood and hair by a variety of methods Petry-Podgórska et al.127 devised and evaluated an HPLC-HG-ICP-MS method in which volatile derivatives were generated post-column by merging streams of acid and of BH. The volatile species were introduced into the plasma make-up gas, rather than via the nebuliser, which was used to introduce an IS. Although a 30–40 fold increase in sensitivity was obtained compared with HPLC-ICP-MS, the LODs were only improved by a factor of 5, to 3–6 ng L−1 for iHg, MeHg, EtHg and PhHg, though the coupling via CVG allowed a greater range of mobile phase compositions to be investigated compared with that which could be introduced directly into the ICP and diminished the effect of the gradient needed to elute PhHg. They developed a simple and short (7 min) two-stage extraction procedure for simultaneous determination of iHg, MeHg and EtHg in human blood that was validated by analysis of human blood CRMs (Seronorm Trace Elements Whole Blood L-1 and L-2). They also developed a method for the selective extraction of MeHg from hair samples with 2 mol L−1 HCl, which was validated by the analysis of CRM IAEA-086 (human hair). They combined this with the total Hg determination in a second aliquot by solid sampling AAS with the dedicated mercury analyser AMA-254. As it is known that hair contains only iHg and MeHg, the iHg content was obtained by difference. They also compared the results obtained by the AMA-254 instrument for the MeHg content of IAEA-086 material (MeHg 258 ± 22 μg kg−1) with those obtained by HG-AAS, ICP-MS and HPLC-ICP-MS obtaining relative measurement errors that ranged from −10 to +10%. Chinese researchers128 devised a procedure for the determination of five Se species (SeCys2, MeSeCys, SeMet, SeIV and SeVI) in livestock and poultry meat by ion-pair RP-HPLC-HG-AFS. A post-column UV reactor was used to convert species to BH active precursors. The LODs ranged from 0.6 to 0.9 μg L−1. Sample preparation involved both enzymatic extraction with trypsin, protease XIV, and pronase, as well as the extraction of the inorganic species with iodoacetamide.

It has been suggested that bees and their products could be useful bioindicators of anthropogenic activities and could overcome some of the deficiencies of air quality networks. To this end, researchers129 have developed a suitable sample preparation method for the determination of Hg in bees and beehive products (pollen, propolis, royal jelly, beeswax, honey, honeydew) by CV-AFS. Results were obtained from small sample masses (20–100 mg) heated with a mixture of 0.5 mL HCl, 0.2 mL HNO3, and 0.1 mL H2O2 in a water bath at 95 °C, which was made up to a final volume of 5.0 mL. The solution LOD was 0.01 μg L−1, corresponding to 3, 1, 0.5 μg kg−1 for sample masses of 20, 50 and 100 mg, respectively. The method was applied to the bees and bee products collected in six areas of the Lazio region in Italy.

Researchers130 at the US FDA have evaluated the suitability of a thermal decomposition amalgamation AAS procedure as an alternative to XRFS for the determination of Hg in certifiable colour additives. They concluded that the method has a sensitivity much improved over that of XRF allowing an LOD of 30 μg kg−1, which makes it suitable for the analysis of samples for which the Code of Federal Regulations specification limit is often 1 mg kg−1. The researchers noted that analyses of certain matrixes, notably those that released nitrogen or sulfur oxides or halogens upon combustion, meant more frequent replacement of the catalyst and recalibration. However, additives containing BaSO4, in colour additive lakes, that are difficult to analyse by other techniques, were, the researchers concluded, well suited for the thermal desorption amalgamation procedure.

4.5 X-ray spectrometry

A comprehensive review of recent advances in X-ray spectrometry5 complements the applications with clinical and biological materials, foods and beverages covered within this Update. Imaging applications of X-ray spectrometry are described separately in Section 6.2.

Within this review period, the development and use of EDXRF spectrometry has continued to grow within the food and beverage field. A review by Li et al.17 summarised the advances in EDXRF spectrometers and algorithm optimisation, which have enabled significant improvements in detection levels, reduced sample preparation protocols and improved sensitivity for low mass elements. It also covered a number of food-related applications. By combining DSPME pre-concentration with XRF detection, Hg was quantified in the parts-per-trillion range for a variety of waters, beverages, foods and tissues in two studies by Musielak and co-workers. In the first,131 thiosemicarbazide-functionalised carbon nanotubes were used to selectively adsorb HgII from the sample within 10 min, followed by analysis with TXRF spectrometry. The approach achieved an LOD of 2.6 pg mL−1 for 600 s measurement time with a sample enrichment factor of 200. The method was tested with spike recoveries in various water samples and seawater RMs, as well as biological tissues, which were digested prior to pre-concentration (lobster hepatopancreas, herring, cormorant and cod tissues). Furthermore, two quantification techniques were compared, namely classical external calibration with internal standardisation (Y in this case) and only IS using the instrumental software. The results between the two calibration approaches were extremely comparable with recoveries for the water samples and biological tissues of 99–103% and 96–105%, with external calibration, vs. those of 99–106% and 97–108%, with the IS only method. The concentrations ranged from 0.05 ng mL−1 to 2200 μg kg−1. The internal standardisation method using the instrumental software has the significant advantage of not requiring separate Hg calibration standards, further increasing throughput. The work demonstrated that DSPME with TXRF spectrometry was a powerful combination for ultra-sensitive Hg detection in various samples across a range of concentrations. In the second study,132 applied the DSPME method with thiosemicarbazide modified graphene oxide for selective HgII separation, followed by measurement by both EDXRF and TXRF spectrometry. Following a similar approach to the first paper, the pre-concentration was optimised then validated with spiked samples (water, apple juice, beer, wine) and various reference materials (seawater, groundwater, wastewater, herring, cormorant and cod tissues, pig kidney, lobster, tobacco, scallion, celery, and spinach). The LODs for TXRF spectrometry were 2.1 pg mL−1 for liquids and 1.8 ng g−1 for solid samples. As expected, the LODs for EDXRF spectrometry were higher due to the poorer sensitivity of the technique, 60 pg mL−1 and 73 ng g−1 for liquid and solid samples, respectively, however, they represent a significant improvement if pre-concentration was not applied.

Two papers have focussed on the use of EDXRF spectrometry as a rapid analytical tool for product quality and safety. Milinovic et al.133 utilised EDXRF spectrometry for the direct and non-destructive measurement of Ca, K and P in edible seaweeds. Green, red, and brown seaweeds from Portugal were collected and analysed. The workers used conventional ICP-OES analysis to determine the concentrations which were used to build the statistical calibration model. It was found that cross validation PLS with the nonlinear iterative partial least squares algorithm was appropriate and the model was validated using the CRM SRM3232 Thallus laminariae (kelp). For the analytes of interest, the correlation between ICP-OES and EDXRF spectrometry achieved R2 between 0.88 and 0.95, with an RMSE < 1.02 and the bias was low. It demonstrated a fit for purpose rapid screening method for the edible seaweed industry. Ma et al.134 described a development with EDXRF instrumentation to significantly improve the LODs for As, Cd and Pb in Chinese herbal medicines. A monochromatic excitation source was used by incorporating a doubly curved LiF crystal which reduces the scattering background noise and improving the S/N. Quantification was achieved through fundamental parameters with a software algorithm for fast integration and spectral fitting. The approach was tested with 99 herbal samples by comparison with ICP-MS which included a quality control material. Good correlations were found for all three elements with R2 ranging between 0.9059 and 0.9631 for those samples above the LOD with EDXRF spectrometry. The LODs achieved were respectable for the X-ray technique: 0.093, 0.114 and 0.232 mg kg−1 for As, Cd and Pb, respectively. The authors also noted the relative low cost and ease of use for rapid analysis (<30 min in total per sample) for regulatory applications, such as customs, or at remote production sites.

The use of a portable XRF system for rapid biomonitoring analysis using human toenails as a non-invasive specimen was investigated by Bhatia and co-workers.135 Samples were collected from 3 populations in Chennai, India, totalling 97 individuals. The researchers quantitatively determined As, Ni, Pb, Se and Zn, by using in-house manufactured nail phantoms. Additionally, the samples were acid digested and analysed by ICP-MS. The Passing–Bablok regression analysis from both techniques showed there was no bias for Ni, Pb and Zn but a strong proportional bias was observed for As and Se at concentrations close to or below the XRF LOD. However the portable XRF spectrometer showed good agreement with ICP-MS when the levels were significantly above the LOD, demonstrating fit for purpose applicability to in-field population studies.

A study by Hauser et al.42 investigated the application of an easy sample preparation process for the analysis of placenta tissue with TXRF spectrometry. Only 10 mg of powdered tissue was suspended in 1 mL of HNO3, then 10 μL of the suspension were dropped and dried on to quartz plates. This was compared with classical MAD. Quantification was achieved by the addition of V as an IS. The samples were screened and those above the LOD were quantified, namely Ca, Cr, Cu, Fe, K, Ni, Rb, Se, Sr and Zn. Three elements were selectively compared against ICP-OES as a reference technique using the acid digested samples. The recoveries were 102 ± 13% for Ca, 112 ± 5% for Fe and 120 ± 11% for Zn, demonstrating acceptable agreement. When comparing the sample preparation methods by TXRF spectrometry, excellent correlation was observed for all elements except Ni. The Ni values in the suspension were slightly positively biased compared with digested samples, likely due to the Ni levels being close to the LOD and a possible interference from Fe. However, the approach was satisfactory for the purpose of examining the impact of sampling location on placental tissue, where a number of trends were found.

5. Nanomaterials

This section covers the developments for the detection and biological impact of NPs. The use of NPs for sample preparation/pre-concentration is covered within Section 3.2. The majority of papers within this review period cover the identification and quantification of various NPs whilst considering the impact on biological systems e.g. dietary exposure, tagging for spatial distribution. It is also worth highlighting a review article by Laycock et al.,11 who systematically evaluated the analytical process for NP detection in biological samples, from sample preparation to analysis.

A number of publications have demonstrated the application of spICP-MS for the detection and assessment of exposure risks from NPs. Grasso et al.136 analysed canned seafood products for ZnONPs to examine potential dietary exposure. The homogenised samples were prepared using TMAH and sonication for NP extraction, in addition to total Zn determination by MAD followed by standard ICP-MS detection. The spICP-MS established the presence of ZnONPs with a mean size diameter of 61.3–78.6 nm at a number concentration from 5.51 × 106 to 9.97 × 106. With respect to the samples, there was no significant trend considering trophic level and size of the seafood species, however the ZnONPs were slightly larger in diameter in the mollusc species. From the data, the authors estimated the dietary intake of the NPs between 0.010 and 0.031 μg kg−1 bw for adults and from 0.022 to 0.067 μg kg−1 bw in children. From the spICP-MS profile, the dissolved Zn was estimated and was comparable to the total Zn from the acid digestion, confirming the efficiency of the alkaline extraction method. Overall, the study demonstrated the existence of trace amounts of ZnONPs in seafood products, resulting in low level human exposure, however, the in vivo fate via ingestion requires further research to better understand the risks. Also focussed on seafood, Suzuki et al.137 utilised spICP-MS to investigate the risk from dietary exposure from HgNPs. Samples of fish and shellfish (n = 90) were enzymatically digested with pancreatin and lipase for the NP analysis. Additionally, total Hg was determined by a thermal decomposition mercury analyser and Se by ICP-MS following MAD. Furthermore, MeHg was determined by HPLC-ICP-MS. The results showed the presence of HgNPs in almost all samples, with the highest particle number concentrations observed in tuna and swordfish (17.7 × 107) compared to shellfish (1.2 × 106). When combining the data with the total Hg, MeHg and Se, some trends were observed indicating that HgNPs were formed immediately from the demethylation of Hg. There was also evidence to suggest the formation of HgSeNPs. Based on the results, the dietary exposure was calculated as 0.067, 5.75 and 5.32 μg per person per day for HgNPs, total Hg and MeHg, respectively. Compared to the recommended daily maximum dose of 0.57 μg per kg bw per day for Hg, the level for HgNPs was estimated as 1.2 ng per kg bw per day, indicating the risk from redissolved iHg from HgNPs was very small. However, the authors concluded that toxicity risks from the particles still remains unclear, requiring further investigation. Li and co-workers138 described the method development for assessing AgNP migration from packaging, specifically storage bags for breast milk. The researchers optimised and validated two testing methods, namely A4F-ICP-MS and spICP-MS. The procedures were rigorously evaluated for accuracy, repeatability, LOD, LOQ and recovery. A number of breast milk storage bags were tested under three incubation conditions: 37 °C for 1 h, 4 °C for 5 days and −20 °C for 10 days. The milk samples were enzymatically treated with proteinase K in a buffer of 10 mmol L−1 Tris–HCl and 1% w/v Triton X-100 at pH 7.5. The results from both techniques showed no migration of AgNPs. Following this, three sets of harsher conditions were tested: heating at 70 °C for 1 h, boiling at 100 °C for 30 min and microwave heating for 1 min. Again, no migration of AgNPs occurred, demonstrating the applicability of the techniques for nanomaterial safety assessments.

Another key area of NP research, is their use for labelling and tagging of proteins, cells, etc. for diagnostic and quantitation purposes. Li et al.139 applied this principle as a new novel approach to quantify enzyme activity. In this work, uracil-DNA glycosylase was used as a model due its importance for DNA repair mechanisms and prevention of mutagenesis. The procedure utilised two DNA probes with partially complementary sequences, in combination with AuNPs and functionalised magnetic beads. Each enzymatic reaction was proportional to the number AuNPs released therefore by counting the particles with spICP-MS, the enzyme activity could be inferred. The LOD achieved was 0.0003 U mL−1 which was better than or comparable to other methods and proof of concept was demonstrated. Sun and co-workers140 utilised lanthanide NPs as tags for two applications: to quantify alpha-fetoprotein and for counting cancer cells. The NPs were synthesised by the hydrothermal method to form NaYF4:Yb,Er with amino modifications to allow the conjugation of the anti-alpha-fetoprotein antibody. The method was tested with human serum achieving acceptable results. The tags were also applied to cancer cell detection which was demonstrated with MCF7 breast cancer cells. After incubation, LA-ICP-MS was employed to check the reaction of the tags with the cells, whilst also determining spatial distribution information. The samples were then analysed by time resolved ICP-MS to count the cancerous cells. The paper demonstrated the potential of NPs with ICP-MS for clinical research. In a paper employing scICP-MS, Oliveira et al.141 quantified the uptake of AuNPs in individual cells to support improved understanding of nano-based drug delivery mechanisms. Prostate cancer cells (PC-3) were used as the model and were incubated with the AuNPs at various concentrations. It was found that increasing the concentration of particles (from 2 to 32 μg L−1) increased the number of AuNP containing cells somewhat proportionally (from 1.83% to 38.36%, respectively). Additionally, it was possible to distinguish the number of NPs per cell. The work demonstrated the power of the application and potential benefits for nanotherapeutics research.

6. Applications: clinical and biological materials

In this section, developments in various areas of work involving clinical and biological materials are discussed. In addition technical details of selected procedures are summarised in Table 2.
Table 2 Clinical and biological materials
Analyte Matrix Technique Study aim, procedure and comments Reference
Al Dextran 40 glucose injection ICP-MS Validation of a method, comprising MAD in HNO3–HCl with detection of Al in He mode. Figures of merit were an LOD of 0.482 ng mL−1, an RSD of 8.8% and spiked recoveries between 102.2 and 107.4%. The method was applied to 44 glucose injection batches, produced by nine pharmaceutical companies; three exceeded the regulatory limit for Al of 25 ng mL−1 241
As species (DMA, iAs, MMA) Granulocytes, leukocytes HPLC-ICP-MS Evaluation of As species in 21 patients with acute promyelocytic leukaemia at different stages of treatment with As2O3. The LOQ of the As method was 0.5 ng mL−1. In both cell types, iAs was the predominant species, whereas MMA and DMA were only detected in leukocytes (MMA > DMA) 242
As species (AB, AC, AsIII, AsV, DMA, MMA) Urine IC-ICP-MS A rapid method for determination of six As species in urine with a 3 min cycle time and 12 h calibration stability. Roxarsone was used as the IS. The LOQs for the species were 1 μg L−1. At species concentrations of 2.2 to 4.6 μg L−1, RSDs ranged from 4 to 7% and spiked urine recoveries were 99 to 104%. Further method validation was performed using samples from different proficiency testing schemes 142
Au Prostate cancer cells (PC-3) scICP-MS Development of a novel method to study the uptake of AuNPs into individual cancer cells, which achieved transport efficiencies for intact cells of 19.8 to 29.4% and an LOQ for Au of 230 ag per cell. Average Au masses per cell obtained by scICP-MS agreed well with those determined by bulk digestion followed by ICP-MS analysis. The proportion of individual cells identified as containing AuNPs increased from 1.83 to 38.36% as the AuNP exposure concentration increased from 2 to 32 μg L−1. At higher AuNP exposure concentrations, more cells contained multiple AuNPs, e.g., at 32 μg L−1, 22.79, 12.83 and 1.79% of cells contained one, two and three AuNPs, respectively 141
Ca, K, Mg, Na Bile LIBS The feasibility of direct bile analysis to discriminate gallbladder cancer (n = 7) from gallbladder polyps (n = 17) was investigated. A laser patterned silicon wafer substrate allowed the non-pretreated bile to be spread homogeneously for LIBS analysis. The IS used was LiCl. Lower peak areas were observed in gallbladder cancer samples, in particular for Mg and K. Optimal discrimination between the clinical scenarios was obtained using the ratio of the K and Na peak areas 243
Cd Blood ICP-MS This work aimed to test the hypothesis that blood Cd concentration is positively associated with coronary artery calcification (assessed by computed tomography), as a measure of coronary artery atherosclerosis, in a large (n = 5627) Swedish population study. The median blood Cd concentration was 0.24 μg L−1. Blood Cd in the highest quartile (median 0.63 μg L−1) was positively associated with coronary artery Ca score 244
Cd Blood, cord blood ICP-MS An investigation into the association of prenatal Cd exposure and neurodevelopment in 2 year old Japanese children, involving 3545 mother–child pairs. Median (IQR) blood Cd concentrations in maternal blood, collected mid/late pregnancy, were 0.7 (0.52 to 0.95) μg L−1 while those in cord blood were 0.04 (0.03 to 0.06) μg L−1. The findings indicated that prenatal Cd exposure was negatively associated with neurodevelopment in boys and in children of mothers who smoked or had gestational diabetes during pregnancy 245
Cd Urine ICP-MS The aim of the study was to assess the effect of maternal Cd exposure during pregnancy on foetal and infant size from 24 weeks gestation to 6 months of age. Mean urinary creatinine-corrected Cd concentrations were 1.00 and 0.98 μg g−1 creatinine in the first and third trimesters, respectively. An association between increased maternal urinary Cd and both decreased head circumference in boys and decreased infant weight amongst girls was observed 246
Cu Serum ICP-MS This work focussed on establishing reference ranges for total and exchangeable Cu in children aged 1 to 5 (n = 122), 6 to 12 (n = 125) and 13 to 18 years old (n = 120). The exchangeable Cu method involved ultrafiltration (30 kDa cut-off) of serum followed by determination of Cu by ICP-MS. The figures of merit of the Cu method were LOQ, 0.1 μg dL−1, RSDs < 2% at 185.42 μg dL−1 and recoveries from 99.1 to 108.0% 247
Cu, Fe, Zn Plasma, serum TXRF A method to simultaneously determine Cu, Fe and Zn in plasma/serum. Evaluation of sample preparation parameters using response surface methodology identified direct analysis of plasma/serum without dilution to be the best strategy. Good agreement with reference values for SRMs was obtained as well as comparable results to those measured by ICP-MS and ICP-OES for patient samples 248
Cu, Fe, Mn, Zn CSF, serum SEC-ICP-MS Concentrations of transition metals and their species were compared in paired CSF and serum samples from 24 healthy controls. The concentrations of the total elements were lower in CSF compared with serum. Speciation analysis suggested that high molecular weight metalloproteins are diluted in the CSF, whilst low molecular weight fractions are enriched 249
I Urine ICP-MS A study investigating the association between maternal urine I and adverse birth outcomes, which included 870 subjects. The median urinary I concentration was 172 ± 135 μg L−1. Higher urine I, ≥250 μg L−1, was associated with an increased risk of macrosomia and multivitamin supplements containing I and frequent milk consumption were identified as dietary risk factors 250
I Urine ICP-MS This work aimed to evaluate the association between I status of pregnant Portuguese women and their milk and dairy product consumption. The study cohort consisted of 468 women, of which 269 were on I supplements. Although milk intake (but not yoghurt/cheese) was positively associated with urinary I concentrations in the whole group and in those not taking supplements, the results suggested that frequent milk intake alone may not ensure adequate I status 251
I Urine ICP-MS Urinary I concentrations and thyroid goitre were assessed following 20 years of universal salt iodisation through an epidemiological study involving adults (n = 78[thin space (1/6-em)]470) and 9 to 11 year olds (n = 1860) from 31 Chinese provinces. In the adult and child cohorts, median (IQR) urine I was 177.89 (117.89 to 263.90) μg L−1 and 199.75 (128.41 to 303.37) μg L−1, respectively, while the incidence of goitre was 1.2 and 3.5%, respectively. Only 3.4% of individuals had urine I <50 μg L−1 demonstrating the success of the salt iodisation programme in eliminating I deficiency in this population 252
I Urine ICP-MS A study to establish dietary I intake and I status in Finnish pregnant females and their infants. Urine I was determined in early (n = 174) and late (n = 186) pregnancy as well as in the mothers (n = 197) and infants (n = 123) 3 months postpartum. Over the various time points, 59 to 70% of mothers and 29% of infants had urinary I concentrations consistent with insufficiency. A lower dietary intake was recorded in the I insufficient women, supporting the findings 253
I, Se Breast milk, erythrocytes, plasma, urine ICP-MS An investigation into I and Se status of infants in relation to maternal status during pregnancy and lactation in a Swedish birth cohort (n = 604). Urine I and erythrocyte/plasma Se were measured in the mothers at 29 weeks gestation while breast milk I and Se, infant urine I and infant erythrocyte Se were determined 4 months postpartum. The median urinary I (113 μg L−1) and plasma Se (65 μg L−1) concentrations in the mothers were consistent with insufficient intake. Infant urinary I concentrations were similar to maternal concentrations and strongly correlated with those in breast milk, while infant erythrocyte Se, although similar to maternal concentrations, did not correlate with breast milk Se concentrations 254
K, Mg, Na Bile LIBS, NIRS The work aimed to differentiate gallbladder cancer from gallstones and gallbladder polyps by way of direct bile analysis using two techniques. A combination of the LIBS Na to K peak area ratios with second NIR principal component scores enhanced the ability to differentiate gallbladder cancer from the benign conditions with respect to using either technique alone 255
Li (Cl, K, Mg, Na) SKOV3 cells LIBS, XRF A novel method (FROZEN!) to determine intracellular Li concentrations in single cells. Rapid cleaning with deionised water removed extracellular electrolytes, followed by flash freezing to preserve the cell contents before removal of H2O in a freeze drier. Positively charged ions were determined by LIBS and negatively charged ions by XRF. Cells that had been incubated in a medium containing 1.0 mmol L−1 Li were found to have intracellular Li concentrations of 0.5 mmol L−1 and to contain lower concentrations of electrolytes, K, Mg and Na, vs. non-incubated cells 256
Li Serum ICP-MS The validation of a standard addition method for determination of serum Li. Following dilution of serum in 0.3% HNO3 and spiking with Ge as the IS, the Li/Ge ratio was detected in He mode. Reported RSDs were ≤1.11% and mean biases from target values for SRM 956d were −0.71% for level 1, −0.17% for level 2 and 2.20% for level 3. External quality assurance results were satisfactory 257
Mn (Cu, Fe) Liver biopsies SF-ICP-MS A study to establish liver Mn concentrations in hepatic steatosis and non-alcoholic fatty liver disease in 76 patients with chronically raised liver function tests. Biopsy samples underwent both conventional histological analysis and trace element determination (which also included Cu and Fe). Lower Mn concentrations were observed in patients with steatosis compared with those without (mean 3.8 vs. 6.4 μg g−1, p < 0.001) and a strong inverse correlation was apparent between steatosis grade and hepatic Mn content 258
Mo Urine ICP-MS A retrospective study across eight industrial sites in France evaluating pre- and post-shift urinary Mo concentrations over a period of two years. An average of six samples was collected from each of 77 workers. Post-shift urinary Mo concentrations were significantly higher than pre-shift concentrations (median (95th percentile): 60.6 (190.0) vs. 37.9 (91.1) μg g−1 creatinine) although no general increase in urine Mo over time was observed 259
Pb Dried blood spots TXRF The aim of the study was to develop a method to measure Pb in dried blood spot samples for use in human biomonitoring. Blood SRMs were used to validate the final method and the figures of merit were an LOD of 0.28 μg dL−1, RSD of 15% and accuracy of 111%. When the method was applied to a non-exposed population (n = 41) and to occupationally exposed electronic waste workers (n = 40), good agreement was obtained with Pb concentrations in whole blood as measured by ICP-MS 260
Pb Cord blood GFAAS A nested cohort cross-sectional study examining the association between prenatal factors and cord blood Pb concentrations, involving 121 mother–child pairs in China. Mean cord blood Pb concentrations were 21.9 μg L−1 and concentrations correlated significantly with maternal blood Pb concentrations measured in the third trimester of pregnancy. Passive smoking was found to be a risk factor for cord blood Pb >20 μg L−1, while taking docosahexaenoic acid and consuming more whole grains appeared to decrease the risk 261
Pb Blood AAS An investigation into the association between blood Pb concentrations and markers of Fe deficiency in women in the second trimester of pregnancy (n = 396). In the control group, Fe deficiency and Fe deficiency anaemia groups, the mean blood Pb concentrations were 3.25 ± 0.407, 7.96 ± 0.502 and 22.12 ± 1.02 μg dL−1, respectively. A significant negative correlation between blood Pb and markers, haemoglobin, total Fe-binding capacity, ferritin and serum Fe, was observed 262
Pt Ovarian tissue LA-ICP-MS The Pt distribution in ovarian cancer tissue following neoadjuvant chemotherapy in 27 patients with advanced high-grade serous ovarian cancer was investigated. Of the two distribution patterns observed, that in which there was little Pt accumulation in the tumour tissue was associated with more rapid recurrence of the cancer after Pt therapy and worse prognosis. The findings implied that the Pt distribution pattern could be useful in predicting Pt resistance prior to tumour recurrence 263
Ru Serum ICP-MS An ultra-sensitive label-free method for detecting hepatitis B virus DNA, which utilised a Ru-containing dual function probe to allow both ICP-MS and fluorescence detection. For the ICP-MS method, the linear range was 3.5 to 200 amol L−1 and the LOD was 1 amol L−1. The results obtained were verified against RT-PCR 264
Se Blood, plasma ICP-MS A reference range study in Chinese child-bearing age females (n = 187). The age-specific reference ranges established (2.5th to 97.5th percentiles) were 73.81 to 140.75 μg L−1 and 81.06 to 164.75 μg L−1 for plasma and whole blood, respectively. When the reference ranges were applied to 1950 subjects from a nutrition surveillance study, 24.1% were found to have low Se concentrations and 5.1%, high 265
Se Urine ICP-MS A birth cohort study investigating the association between maternal urine Se and newborn telomere length in 746 Chinese mother–newborn pairs. Median creatinine-corrected urinary Se concentrations in the first, second and third trimesters were 16.29, 18.08 and 18.35 μg g−1 creatinine, respectively. A doubling of the maternal urine Se in all trimesters was associated with a 6.02 to 6.44% increase in the telomere length at birth 266
Si Lung biopsies XRF Quantitation of Si content in biopsies of lung-transplanted silicosis patients (n = 17) vs. idiopathic pulmonary fibrosis controls (n = 17). The Si content measured in the silicosis biopsies was higher than in controls (7284 ± 4694 vs. 899 ± 366 ppm) and there was a negative correlation between Si content and age, body mass index and pulmonary function test results. A Si cut-off of 1128 ppm was proposed to predict artificial stone-induced silicosis and this had a sensitivity of 100% and specificity of 94% 267
Tl Urine ICP-MS A study into the relationship between Tl exposure in early pregnancy and preterm birth, involving 2104 pregnant women from a Chinese Maternal and Child Health Cohort. The median urinary Tl concentration at 20 weeks of pregnancy was 0.35 μg L−1 (0.47 μg g−1 creatinine). The results suggested an increased risk of preterm birth in males whose mothers had urine Tl >0.57 μg g−1 creatinine 268
Y Blood, breast cancer cells (MCF-7), serum ICP-MS, scICP-MS Development of an indirect method to measure tumour marker, alpha fetoprotein, in serum using synthesised lanthanide NPs (NaYF4:Yb,Er) as immunolabelling probes. Elemental tag, 89Y, was detected by ICP-MS. The method had a linear range of 2 to 180 ng mL−1 and an LOD of 0.44 ng mL−1. Further work involved labelling AFP-positive cancer cells with the Y-containing NPs to allow them to be counted in blood via scICP-MS 140
Various (6) Artificial saliva ICP-MS, SEM/EDS A comparison of metal ion (Cr, Co, Fe, Mo, Ni and Ti) release from orthodontic appliances in both an in vitro (90 day incubation of devices in artificial saliva and analysis by ICP-MS) and in vivo setting (analysis of appliance debris composition using SEM/EDS). For all metals except for Fe, the metal content in the debris was higher than that released into artificial saliva. The released metal ion concentrations did not exceed recommended upper limits for daily intake 269
Various (8) Blood ICP-MS A reference range study focussing on Ca, Cu, Fe, Mg, Mn, Se, Sr and Zn in Chinese children and adolescents (n = 589) in which the influence of age, gender and season were assessed. Gender and age-specific reference intervals for whole blood were established 33
Various (11) Serum ICP-MS A reference range study for nutritional and toxic metals involving 2217 children, aged <16 years, in China. Sex, age and season were reported to influence the concentrations measured 270
Various (23) Serum ICP-MS A prospective study to investigate the relationship between exposure to multiple metals and all-cause and cardiovascular mortality in Chinese adults (n = 6155). The mean duration of follow-up was 9.8 years. After adjusting for confounding factors, plasma Cu, Mo and V were positively associated with both all-cause and cardiovascular mortality whereas Mn, Se and Tl were negatively associated with all-cause mortality only and Se and Tl, with cardiovascular mortality 271
Various (17) Blood and urine ICP-MS Element concentrations, including those of REEs, were compared in electronic waste recycling workers (n = 100) and controls (n = 51). Mean concentrations of blood Pb, Sr, TI, and urinary Eu, La, Pb, Tb, and TI were significantly higher in the workers compared with controls 272
Various (18) Urine ICP-MS A study looking at the association between urine metal/metalloid concentrations and factors affecting male fertility in 796 students. The highest concentration quartiles of urinary Ni and V were associated with decreases of 10.84 and 11.53% in normal sperm morphology, respectively, while urine Ag concentrations were found to have a positive correlation with number of sperm, sperm concentration and semen volume. Lower testosterone concentrations were observed in a sub-group with overall higher urine metal concentrations and testosterone displayed a negative association with urine Li 273
Various (24) Urine ICP-MS The work aimed to establish baseline pre-exposure concentrations of 17 REEs as well as As, Cr, Co, Ni, Sb, Tl and U in pooled urine samples from the Inuit population in Quebec. Of the REEs, only Ce and La were detected 274
Various (20) Erythrocytes, lymphocytes and plasma TXRF Concentrations of various elements were measured in isolated blood components, collected from 36 healthy individuals (15 male, 21 female), for use as reference values. Significant concentration differences were found between males and females for P in lymphocytes, Se and Rb in erythrocytes and V in plasma 275
Various (15) Nails TXRF Determination of various elements, including As, Br, Ca, Cr, Cu, Fe, Mn, K, Ni, P, S, Se, Ti, V and Zn, in nails digested in HNO3–HCl, with the aim of differentiating men with colon cancer (n = 31) from healthy controls (n = 31). Significant differences were observed between the groups for As, Ca, Cr, Cu, Fe, Se and Zn (p < 0.05) 276
Various (9) Thyroid tissue LIBS An investigation into the feasibility of differentiating papillary thyroid cancer tissue (n = 14) from normal tissue (n = 13) using spectral emissions corresponding to C, Ca, molecular CN, H, K, Mg, N, Na and O. The Ca and Mg content of the cancerous tissue was significantly higher than that in normal tissue. Using the chemometric method, SVM classifier, and all 10 spectral emissions, cancerous tissue was discriminated from normal tissue with 92.6% accuracy, 92.9% sensitivity and 92.3% specificity 277


6.1 Metallomics

A comprehensive review of recent advances in speciation and related applications complements the work with clinical and biological materials, foods and beverages covered in this Update.6 Discussion of work relating to NPs is included in Section 5. Extraction techniques, including those applied to species selection, are considered in more detail in Section 3.2 and Table 1. In this section, we specifically address advances in speciation techniques, over the period covered by this Update. Once again, it is noted the interest in the identification and quantification of As and Hg species, both in clinical and food samples. Selenium speciation is also a rather regular feature in this section, whereas the accurate discrimination of CrIII and CrVI in a variety of foods received more attention than usual during this period and produced substantial evidence that the diet is not a source of the carcinogenic form of Cr (CrVI).

The speciation of As in biological fluids was the subject of three interesting papers in this year’s Update. Zhang and co-workers44 explored a novel extraction procedure, based on magnetic-assisted dispersive extraction, followed by ultrasonic spray separation, to assist the identification and quantification of As species (AsIII, AsV, DMA, and MMA) in blood, as an aid to investigate As metabolism and biomarkers of exposure. Blood (100 μL) was mixed with 10 μL of 0.2% (v/v) Triton X-100 and 100 μL of 12 g L−1N-ethylmaleimide with the aid of a magnetic stirrer. The extract was separated from the remaining matrix using an ultrasonic spray sheer and an ultrafiltration membrane, then was presented to IC-ICP-MS for the determination of the As species. The extraction efficiency for total As was 96% and no interconversion of As species was observed, as assessed from spiking experiments. The LODs for the four As species in whole blood ranged from 0.017 to 0.023 μg L−1, and RSDs were between 2.3% and 4.2%.

Another group of researchers142 described a method for the quantification of As species (AsIII, AsV, AB, AC, DMA, and MMA) in urine by IC-ICP-MS. Samples were simply diluted with a 50 mM NH4HCO3 solution, containing Roxarsone as IS. The chromatographic separation was achieved using a Dionex 5000 instrument and gradient elution with the following mobile phases: (A) 1 mM NH4HCO3 in 2% MeOH at pH 8.2; (B) 200 mM NH4HCO3–2% NH4OH in 2% MeOH at pH 9.5 and (C) 200 mM NH4HCO3–1% HNO3–1% NH4OH in 2% MeOH at pH 8.2. The ICP mass spectrometer was operated in KED mode using He as the collision gas. The LOD for each species was less than 0.1 μg L−1. For As species concentrations within the ranges 0.4–1.0 μg L−1, 1.1–4.6 μg L−1 and 8.0–96.5 μg L−1, the observed inter-day RSDs were <33%, <15% and <4%, respectively. The accuracy of the procedure was evaluated by spiking donor urine samples with the As species, yielding recoveries in the range 99–104%, successful participation in PT schemes for total As, iAs and As species as well as comparison of the results with a previously validated internal procedure to determine iAs in urine by ICP-MS after liquid–liquid extraction. However the authors highlighted that the proposed procedure did not resolve AB from TMAO.

Procedures for the fast separation of As species in urine by HPLC-ICP-MS were further investigated by Quarles et al.85 This group proposed a method (A) based on anion-exchange chromatography, that would allow the separation of AsIII, AsV, AB, AC, DMA and MMA, within 2 min, as well as an alternative procedure (B), using an additional C18 column, for the determination of TMAO in addition to the other species, within a slightly longer time frame (4.5 min). In both cases, an automated system (prepFAST IC, Elemental Scientific) was used, that included an autosampler, the ability to dilute standards and samples inline to the ICP-MS and to switch the sample introduction between different options (no column, one column, two columns). For total As determination, 2% (v/v) HNO3 at a flow rate of 300 μL min−1 was used as the carrier, whereas elution of As species was achieved with 0.5 mmol L−1 (NH4)2CO3, at pH 9.5 (for method A) and followed by a mobile phase consisting of 80 mmol L−1 (NH4)2CO3, at pH 9.2 (for method B). The ICP mass spectrometer was operated with a collision cell, at a helium gas flow of 4.0 mL min−1. The LODs in a urine matrix were within the ranges 2.8–6.0 ng L−1 As (method A) and 4.1–9.1 ng L−1 As (method B), respectively. Recoveries for the two methods, assessed on a urine sample spiked with 10 μg L−1 of each of the 7 As species, ranged from 94% to 107% (A) and from 97% to 105% (B), whereas RSDs (n = 3) varied between 0.9% and 9.9% (A) and between 2.1% and 8.0% (B), respectively. Successful participation in 16 PT exercises further supported the validation of these methods.

The accurate determination of Hg species in biological fluids still poses challenges, particularly at low levels and given the possibility of species interconversion. Rodríguez-González and co-workers84 have explored the potential of analytical techniques based on different principles, to reliably quantify Hg species in biological matrices. This group assessed the performances of three analytical techniques (GC-ICP-MS, GC-EI-MS and GC-EI-MS/MS) for the determination of iHg, EtHg and MeHg, using ID with a triple isotope spike. The assessment was carried out on CRMs of different biological matrices (NRCC DOLT-4, dogfish liver; IAEA-085, human hair; IAEA-08, human hair and NIST SRM-955c, caprine blood). Samples (0.1–0.5 g) were spiked with appropriate amounts of isotopically enriched compounds (201MeHg, 198EtHg, and 199HgII) to obtain an endogenous vs. labelled compound ratio between 0.1 and 10. After the addition of 3 mL 25% TMAH, samples underwent focused MAE, followed by addition of 4 mL acetate buffer and adjustment to pH 4, with 10% sub-boiled HCl. Derivatisation was carried out by adding 0.4 mL of 2% w/v sodium(tetra-n-propyl)borate and 1 mL hexane. After manually shaking for 5 min, the hexane phase was cleaned-up on a Florisil® home-made column, filtered and stored at −18 °C. Prior to analysis, the samples were pre-concentrated under an N2 flow, to a final volume of approximately 30 μL. The authors stated that the major challenge of the work was the optimisation of the instrumental conditions for the three techniques and the evaluation of species interconversion rates and concluded that GC-ICP-MS showed better performances and matrix tolerance. The GC-ICP-MS values for LODs and LOQs, evaluated from the measurements of procedural blanks, were 0.11 and 0.37 ng g−1 (iHg) and 0.02 and 0.06 ng g−1 (for both EtHg and MeHg). In comparison, GC-EI-MS and GC-EI-MS/MS achieved LOQs of 1.12 and 1.71 ng g−1 (iHg) 0.23 and 0.86 ng g−1 (EtHg) and 0.28 and 1.01 ng g−1 (MeHg).

Another group of researchers127 investigated a variety of procedures, involving the generation of volatile Hg species and/or RP-HPLC, combined with AAS or ICP-MS, to determine Hg2+, MeHg+, EtHg+ and PhHg+ in blood and hair samples. In one of these approaches, volatile Hg derivatives were generated with NaBH4 after the HPLC separation of the Hg species on a C18 column with isocratic elution (mobile phase: 95% 0.5 g L−1L-cysteine–0.5 g L−1L-cysteine·HCl, pH 2.3; 5% MeOH). This achieved a 30–40 times enhancement of sensitivity for the determination of Hg species by ICP-MS, with LODs between and 6 ng L−1, as compared to those obtained by HPLC-ICP-MS only (15–26 ng L−1). The authors noted that the VG step significantly reduced the effect of the organic mobile phase on the plasma. To further improve the performance of the method and limit the matrix effects, the authors devised an additional extraction step for blood samples, carried out by pipetting in a 10 mL glass vial 300 μL of blood to which 430 μL of cysteine buffer and 270 μL of conc. HNO3 were added. The mixture was heated at 85 °C for 2 min. After addition of 3.8 mL of cysteine buffer, the extraction mixture was placed in an ultrasonic water bath at 40 °C for 5 min, then filtered prior to analysis by either HPLC-ICP-MS or HPLC-VG-ICP-MS. The accuracy of the procedures was confirmed by analysis of CRMs (Seronorm™ Trace Elements Whole Blood L-1 and L-2) and also by spiking experiments with the species of interest, that yielded recoveries better than 90% (Hg2+, MeHg+) and 80% for EtHg+, respectively. In addition, MeHg+ was selectively extracted from the solid hair samples by 2 mol L−1 HCl and the extract analysed by either a mercury analyser (AMA-254), VG-AAS or ICP-MS. The respective LODs were 35 μg kg−1 (AMA-254), 41 μg kg−1 (VG-AAS) and 10 μg kg−1 (ICP-MS). Recoveries of the certified value for MeHg+ in the CRM IAEA-086, human hair, were between 90% and 110% for all three detectors.

Given the increasing interest in the levels of As species in food and beverages, a number of papers continue to explore methods for the speciation of As in these matrices. Among these, Nawrocka et al.143 developed a procedure to determine six As species (AsIII, AsV, AB, AC, DMA and MMA) in different types of seafood (Atlantic jackknife clam (Ensis directus), blue mussel (Mytilus edulis), Pacific oyster (Crassostrea gigas), common cockle (Cardium edule), tuna (Thunnus sp.) and Atlantic salmon (Salmo salar)). Samples (0.1 g) underwent MAE with 5 mL of MeOH–H2O (3 + 1 v/v). The separation of the extracted species was carried out by anion-exchange HPLC-ICP-MS, using isocratic elution, at a flow rate of 0.8 mL min−1, with a mobile phase of 50 mmol L−1 NH4(CO3)2–1% (v/v) MetOH–0.2 mmol L−1 EDTA, at pH 8.5. The LODs and LOQs for the As species ranged from 0.27 to 0.52 μg kg−1 and between 0.38 and 0.92 μg kg−1, respectively. Recoveries, evaluated on samples spiked with As species at three levels of concentration, ranged from 88.8% to 116.2% and similar figures were obtained on three CRMs certified for AB and DMA.

The speciation of CrIII and CrVI in food was the topic of the work of Saraiva et al.,144 who reported the development of a method based on specific ID-HPLC-ICP-MS in meat and dairy products. Subsequently, they went on to apply this method to assess the presence and fate of Cr species in raw and cooked milk and meat samples,145 rice146 and bread and breakfast cereals.147 The procedure was based on the formation of a CrIII–EDTA complex, negatively charged, followed by the reduction of CrO42− to CrIII with 1,5-diphenylcarbazide and subsequent complexation with its oxidised form (DPC) to form a positively charged CrIII–DPC complex. For the sequential complexation of CrIII and CrVI, 0.3 g of sample were diluted with 16 mL of EDTA and heated at 70 °C for 25 min in a heating block. After cooling down the mixture, 200 μL of 1,5-diphenylcarbazide was added and the final volume was made up to 20 mL with ultrapure water. After heating again at 70 °C for 25 min, the extracts were cooled down and filtered, then analysed the same day. To perform the ID, known amounts of isotopically enriched solutions (50CrIII and 53CrVI) were added to each standard and sample, that were then left to equilibrate. The HPLC separation of the complexed species was carried out in a single analytical run of less than 3 min, using anion-exchange HPLC and isocratic elution with 0.01 mol L−1 HNO3–2.5% (v/v) MeOH–0.30 mol L−1 EDTA at pH 2. The ICP mass spectrometer was operated in KED mode, using He as the collision gas. The LOQs were 0.013 μg kg−1 for CrIII and 0.049 μg kg−1 for CrVI, respectively. The intermediate reproducibility RSDs, evaluated over a period of two months, ranged from 6.8 to 13% for CrIII and from 6.8% to 25.9% for CrVI, respectively. The measurement bias was estimated to be within the range 0.01–0.11%. The method’s application to a variety of food did not reveal the presence of CrVI as a potential risk for the general population. Marković and co-workers148 also applied an ID procedure, based on spikes of 50CrIII and 53CrVI, coupled with anion-exchange HPLC-ICP-MS, for the speciation of Cr in alcoholic beverages and concluded that, due to the presence of antioxidants, no CrVI was detectable in beer or wine.

Selenium is an important nutrient for human beings and several efforts are in place to assure that adequate Se amounts are present in the diet, including Se fortification of some crops or assumption of food supplements rich in Se. In addition, the chemical form of Se present in food is also important, since not all of them are absorbed by the human body to the same effect. Although several food categories have been evaluated for their contribution to the presence of Se in the diet, including cereals, meat and vegetables, eggs have received less attention. Aiming to fill this gap, a group of researchers149 undertook a thorough investigation of analytical procedures suitable to determine the total Se content as well as those of selected Se species (SeIV, SeVI, SeCys2, SeMet and MeSeCys), in raw and cooked whole eggs. Four extractants (H2O; 0.1 mol L−1 HCl; 0.1 mol L−1 NaOH; 25 mmol L−1 CH3COONH4 buffer–5% MeOH (v/v)), as well as enzymatic hydrolysis with protease XIV, were studied, of which the last gave the best results. Samples of liquid whole eggs (2.5 g) were diluted in 10 mL H2O and sonicated for 30 min, after addition of 50 mg protease XIV and shaking, the mixture was placed in a water bath shaker, at 37 °C for 18 h. Finally the extracts were centrifuged and filtered. Two chromatographic separations (ion-pairing C18 RP chromatography, with a mobile phase of 0.5 mmol L−1 TBAH–10 mmol L−1 CH3COONH4–2% MeOH (v/v), at pH 5.5; AEC with 5 mmol L−1 citric acid, at pH 5.0, as the mobile phase) were evaluated. Both gave satisfactory results, but the first one provided better sensitivities due to the presence of MeOH. LOQs were 1.86 μg L−1 (SeCys2), 1.62 μg L−1 (MeSeCys), 1.82 μg L−1 (SeMet), 2.37 μg L−1 (SeIV), and 1.98 μg L−1 (SeVI). Recoveries of spiked amounts of SeCys2, SeMet and MeSeCys at three levels (0.1, 0.2 and 0.4 μg g−1), varied from 85.5% to 98.2% (external calibration) and from 86.9% to 96.7% (standard addition method).

6.2 Imaging with MS and X-rays

Within this review period, a number studies have focused on the use of LA-ICP-MS as a diagnostic tool for disease detection. Paul and co-workers150 described the development of a digital platform for the identification of colorectal tumours in tissue sections from LA-ICP-MS imaging data. The training data combined haematoxylin and eosin staining, Ki67 immunolabelling and complementary elemental images of C, Ca, Cu, Fe, Mg, P and S. Tissue features were grouped into four major anatomical categories, namely smooth muscle, mucosa, adenomatous polyp and primary tumour (where present), and the elemental data was pooled for each region. This created a binary image where tumour pixels were assigned a value of 1 and other areas were 0. Thresholds were established for each element and a Monte Carlo simulation was utilised, based on previous work. It was found that primary tumour cells were correctly identified, including the ability to differentiate the malignancy. For example, mucinous carcinoma is prone to errors from subjective assessment and can have a poor prognosis and treatment response but it was clearly distinguished. However, false positive pixels were flagged in areas of high cell proliferation. Therefore, additional tissue sections were analysed and the researchers developed a random decision forest approach which significantly improved the differentiation and prevented the false positives. The method offers opportunities to improve cancer diagnostics, however the authors did note the current limitations, in particular, the slow speed of LA-ICP-MS with sequential ICP-MS systems and the need for clinical specialists to supervise the results. Mello et al.151 utilised LA-ICP-MS to assess the potential of multiplexing antibody immunolabelling with metal tags. The workers selected three muscle proteins as models, namely dystrophin, myosin and sarcospan, which are important to Duchenne’s muscular dystrophy research. The antibodies were prepared individually in murine muscle tissue sections as well as multiplexed (n = 8 in both cases) to determine the impact of label combination. The lanthanide tags (Dy, Gd and Nd) were imaged by LA-ICP-MS to highlight the specific location of the protein in the tissue and were also quantified using in-house prepared gelatine calibration standards. The results demonstrated there was no statistical difference from the single conjugate versus the multiplex, opening the possibility to extend the method to additional antibody–metal conjugates. Furthermore, the technique enabled protein localisation as well as a quantitative assessment, which can provide crucial data in disease research.

Novo et al.80 published an investigation into the quantification and distribution of Br and I in hair. Matrix-matched calibration standards were prepared using hair strands soaked in various concentrations of Br and I for 24 h, before rinsing and drying. The total levels were characterised by MIC and ICP-MS analysis. Samples were collected from hypothyroidism patients (n = 7) and controls (n = 18) which were split into two portions. One was subjected to total analysis and the second was measured directly by LA-ICP-MS using the in-house prepared hair calibration materials. As part of the method development, the researchers utilised 34S as an IS to successfully normalise signal variabilities and instrumental drift. Correlation coefficients were >0.999 with LODs of 0.36 μg g−1 for Br and 0.14 μg g−1 for I, which is impressive for these poorly ionised elements by direct solid analysis. The results showed that the hypothyroidism patients receiving treatment with synthetic thyroid hormones had significantly higher I concentration than the healthy volunteers at the 95% CI, 5.77 ± 2.92 μg g−1vs. 1.95 ± 0.74 μg g−1, respectively. However, no significant difference was observed for Br; 1.29 ± 0.91 μg g−1 for the patients vs. 1.18 ± 0.53 μg g−1 for the controls. Additionally, depth profiling and longitudinal scanning were performed, which enabled spatial profiling. This revealed variation in I levels through the strand which can provide temporal information, based on hair growth of 1 cm per month. The work demonstrated the possibility of the approach for non-invasive biomonitoring and metabolism studies.

A new instrumental development was described by Wu and co-workers74 for a two-dimensional cytometry platform by coupling laser induced fluorescence and ICP-MS. The set up enabled the analysis of single particles and single cells at a high throughput rate. The concept was demonstrated with HepG2 cells incubated with Ag but has the potential to expand to multiple elements. The technique could provide useful data for life science research.

The use of SR-XRF spectrometry for biochemical research has received some interest in this review period. He et al.152 investigated the application of non-targeted metallomics for cancer screening in blood serum using SR-XRF spectrometry and machine learning. Samples from cancer patients (n = 100) and controls (n = 100) were qualitatively and rapidly analysed by SR-XRF spectrometry (5 s acquisition time) and the data was used to build machine learning algorithms. The model achieved accuracy of >96% for cancer screening and identified Ca, Ti and Zn as elements that varied between the patients and controls. The combination of fast analysis and elemental trend identification highlighted the potential of the approach. The authors also noted the possibility to extend the method to other diseases such as COVID-19 and neurodegenerative disorders. Suarez and co-workers153 reported the use of SXRF microscopy (the combination of SR-XRF spectrometry and TEM) for the imaging of elements within cell organelles. HepG2 cells were used as a model and were exposed to various AgNPs. A nanoprobe beamline was utilised for the SR-XRF measurements with rasters of 50 × 50 nm2. Quantification was achieved by calibration with a thin film reference sample (AXO RF8-200-S2453). The same section was analysed with TEM and the resulting images were overlaid. It was possible to identify Ag localisation in the mitochondria due to the AgNPs, at a level 10-fold higher than the surrounding cytosol. From this information, methods of transportation within the cell could be hypothesised. The authors also noted that since these experiments were completed, the synchrotron has been upgraded leading to significant improvements in photon flux resulting in faster acquisition times. This enables the opportunity for further research utilising this approach and achieving increased data in less time, which is always a critical factor with beam time.

6.3 Trace elements in pregnancy

This year saw the first published data on the use of arsenic trioxide therapy during pregnancy in human patients.154 While As2O3 is a successful front line therapy for acute promyelocytic leukaemia, administration in pregnancy is challenging due to potential foetal exposure to As.154 In this study trough level blood samples were collected from three pregnant and five non-pregnant patients on As2O3 therapy. Blood and amniotic fluid samples were also collected from two pregnant patients on the day of delivery. Concentrations of iAs, MMA and DMA were determined in blood and amniotic fluid samples using HPLC-HG-AFS. Despite the small sample size, this work demonstrated differential permeation of the placental barrier (iAs > DMA > MMA) resulting in high foetal exposure and potential teratogenicity.

Two articles investigated prenatal elemental exposure and the risk of neural tube defects. Liu et al.155 demonstrated a correlation between prenatal Al exposure and increased risk of neural tube defects (NTDs), particularly the anencephaly subtype. Levels of Al in serum and placental samples from 144 case and 256 control mothers from a Chinese birth defect monitoring study were quantified by ICP-MS following acid digestion. This is the first study to measure Al in paired serum and placental samples in human subjects and builds on evidence previously reported from animal studies. Of note, no correlation was observed between serum and placental Al levels. A further case control study from China reported on prenatal As levels as a risk factor for NTDs.156 Concentrations of As in placental samples from 408 NTD affected pregnancies and 593 healthy pregnancies from a population-based birth defects program were measured by ICP-MS. Elevated As levels demonstrated an increased risk of NTDs with an OR of 1.68 (95% CI 1.09–2.58) which was not evident in women taking folic acid supplementation (OR 1.30, 95% CI 0.58–2.92).

Growing interest in associations between toxic and nutritional elements and in vitro fertilization (IVF) outcomes is evident.157–159 Of the various articles published, the largest study (n = 1184) measured Cd, Co, Pb and Se by ICP-MS in serum samples collected pre-oocyte retrieval from subfertile patients undergoing their first cycle of IVF.157 Using Bayesian Kernel machine regressions (BKMR), negative associations were observed between Pb and Cd levels and rates of high quality embryos. Higher Pb concentrations were also associated with lower rates of oocyte fertilization. Despite the large dataset, the authors accepted that further work was required to confirm findings as limitations included measurement at one single time-point and lack of consideration for metal exposure in the male partner. A much smaller case-control study used BKMR to identify a link between higher Ba exposure and early embryonic arrest in subfertile women undergoing assisted reproduction.158 Fasting blood samples from 228 infertile women, including 74 with embryonic arrest, were analysed by ICP-MS for As, Ba, Hg and Pb. Of the elements measured, a statistically significant difference between case and control group geometric mean concentrations was only observed for Ba (p = 0.009, case group 34.15 μg L−1, control group 23.00 μg L−1). Zhou et al.159 investigated metal exposure in couples (n = 195) by multi-element ICP-MS analysis of female serum and follicular fluid and male semen. Exposure to Ba, Cr and Pb was associated with total sperm concentration, total motile sperm count and MII oocytes. Fertilisation parameters also correlated with serum As, Ba, Cr, Hg and Tl levels and seminal plasma Mo concentrations.

6.4 Elements as tags for indirect determinations

A couple of articles of note this year described developments in protein targeting using ICP-MS readout methods. The determination of immunoglobulin E (IgE) and apolipoprotein E (APOE) in aqueous humour was carried out by a competitive ELISA, using iridium nanocluster (IrNCs)-labelled immunoprobes and ICP-MS.160 The method achieved superior sensitivity to other NP based methods (LOD 0.02 ng mL−1 for IgE in aqueous humour) and the authors claimed acceptable comparability with a commercial ELISA for IgE in human sera, based on data from just three subjects. Asensio et al.161 reported an scICP-MS method using Lu-labelled antibodies to detect HER2 positive cells from biopsies, complementing the tag-LIBS approach discussed in our last review.1 Analysis of mixtures containing BT-474 (HER2 positive) and MDA-MD-231 (HER2 negative) cells gave a correlation coefficient of 0.995 for median concentration (fgLu per cell) against % BT-474. However, the applicability of this method for discriminating between different populations is limited given the large interquartile ranges observed.

The quantification of global DNA demethylation intermediates 5-hydroxymethylcytosine (5hmC) and 5-formylcytosine (5fC) using a Cu-based metal–organic framework with ICP-AES detection was described.162 This strategy circumvents the use of bisulfite treatments, requirement for complex enzymes and primers and long analytical times associated with MS, PCR and fluorescence based methods. In brief, extracted genomic DNA was fragmented and denatured into ssDNA, on which 5hmC and 5fC loci were carboxylated to anchor Cu ions. Copper-based MOF were grown at these locations forming pearl necklace-like ssDNA-MOF composites which can be cross-linked with aldehydic magnetic beads. Following magnetic separation and acid digestion, ICP-AES determination of Cu content gave an indication of the amount of 5hmC and 5fC. Applying 5hmC and 5fC calibration curves, obtained by using DNA extracted from human brain and newborn mouse brain tissue, respectively, levels of 5hmC and 5fC detected in various murine tissues were comparable with conventional methods.

6.5 Multi-element applications

6.5.1 Specimens analysed to investigate metallic implants and biomaterials. Although the number of studies involving metal hip and joint implants has declined over recent years, during this review period, Schoon et al.163 published an interesting report describing the characterisation of CoCrMo particles in human bone marrow by spatially resolved and synchrotron-based nano-XRF. In all eight patients, who had either total hip arthroplasty, hip resurfacing or total knee arthroplasty implants, CoCrMo particles were detected at concentrations of up to 1 × 1011 particles per mL bone marrow or 2 × 104 particles bone marrow per cell. A wide range of particle sizes was observed, ranging from nano- to microparticles with a size distribution that differed between implant types. In samples from patients with total knee arthroplasty and hip resurfacing implants, smaller particles had significantly lower Co content than larger particles, suggesting release of Co within the periprosthetic bone marrow. The results are useful for guiding clinically relevant dosing and particle composition in studies investigating the carcinogenicity risk of CoCrMo particles.

A second paper of note evaluated the release of NPs from Ti-based dental implants.75 The implants were incubated in a 50 + 50 mixture of artificial saliva and bacterial culture medium, both under sterile conditions and in the presence of oral bacteria. Single particle-ICP-MS and TEM, revealed that the released Ti-containing NPs were heterogeneous in diameter (10 to 160 nm) and that the number of NPs was diminished in the presence of the bacteria, which was explained by the formation of a biofilm on the surface of the implants. A discrepancy in the measured particle sizes of some 50% between spICP-MS and TEM was attributed to impurities within the TiO2NPs leading to underestimation of particle size by spICP-MS. In sequential runs, release of Al- and V-containing NPs from the implants into the artificial saliva was confirmed. The second part of the work, using scICP-MS, established significant uptake of Ti into single osteoblast cells.

6.6 Progress for individual elements

6.6.1 Arsenic. An HPLC-ICP-MS method was developed to quantify different As species in urine from patients with arsenism.164 Arsenic species (AsIII, AsV, AB, AC, DMA and MMA) were measured in urine from patients with As poisoning following realgar powder administration.164 Whilst AB and AC were not detected in urine, concentrations of other As species were elevated in urine from patients with arsenism and decreased following detoxification treatment with sodium dimercaptopropane sulfonate. The authors did not state how many urine samples were analysed, nor did they give concentrations for healthy controls, making it difficult to attribute a level of confidence to the findings.
6.6.2 Cadmium. An environmentally friendly procedure for determining Cd in saliva used simultaneous spray assisted droplet formation-liquid phase microextraction with slotted quartz tube-flame atomic absorption spectrometry.165 This approach achieved an LOD of 0.65 ng mL−1 and a linear range of 2.0–50 ng mL−1 which is comparable with the analytical performance of other methodologies. Primary advantages of the proposed technique are minimization of sample preparation steps and the environmental benefits associated with reduced external dispersive solvent usage.
6.6.3 Chloride. Metrologically traceable assigned target values for serum Cl in external quality assessment schemes highlighted method biases of clinical analysers. 22 Accuracy-based Cl target values for the Health Services Authority EQA scheme were obtained using an SF-ICP-IDMS primary reference measurement procedure, of which further details are discussed in Section 2, validated with CRMs (NIST SRM 956c and 956d, electrolytes in frozen human serum). The relative deviation from the assigned values of the results obtained with each analytical brand of analysers was assessed. For Roche and Siemens analysers, the observed relative deviations showed significant correlations with Cl concentration (R2 0.8536 and 0.5136, respectively). For Roche analysers, relative deviations rendered results at lower levels outside of allowable limits defined by the Royal College of Pathologists of Australasia, but performance improved over time with EQA participation. This article strengthens the argument for applying traceable assigned target values, which are currently lacking in many EQA schemes.
6.6.4 Copper. Two papers reported on the use of Cu as potential biomarker in cancer diagnosis and prognosis.

A comprehensive study of Cu in ascites demonstrated Cu induced angiogenic effects and consolidated the link between high ascite Cu content and malignant ovarian tumour progression.166 Primary screening of trace elements in 27 ascitic fluid samples by ICP-MS indicated associations of As, Co, Cu, Mo, Ni, Se and Zn concentrations with malignant tumour progression. Multivariate linear regression analysis of Cu concentrations, measured by AAS, in ascites from patients with benign (n = 88), borderline (n = 11) and malignant (n = 25) ovarian cancer, as a follow-up, showed increased Cu concentrations in malignant compared with benign tumours (p < 0.001). Microarray analysis revealed upregulation of angiogenesis biological processes in Cu-treated ovarian cancer cell lines OVCAR3 and A2780 compared with the mesothelial cell line Met-5A. Finally, vascular epithelial growth factor (VEGF) were increased in all cell lines following incubation with Cu and measured Cu concentrations in ascitic fluid correlated with VEGF concentrations.

Marković et al.167 documented a novel analytical method for serum Cu speciation using conjoint liquid chromatography on short bed connective interaction media monolithic disks with UV detection of protein elution profiles and ID-ICP-MS quantitation of Cu species. Separation of Cu bound to caeruloplasmin (Cp), albumin (HSA) and lower molecular mass (LMM) species was achieved, of which the latter was not determined in previously reported methods. Speciation was accomplished by retention of Cu–HSA on a α-HSA disk, followed by separation of Cu–Cp and Cu–LMM on a DEAE disk with elution by NH4Cl at pH 7.4. Post-column quantification by ID-ICP-MS yielded LODs for Cu–Cp, Cu–HSA and Cu–LMM of 6.1, 5.3 and 3.3 ng mL−1 Cu, respectively. Analysis of human serum from healthy individuals (n = 4), kidney transplant patients (n = 4) and cancer patients (n = 6) found higher concentrations of total Cu and Cu–Cp in the latter, however sample numbers were too small to support reliable conclusions and the cancer type was not stated.

García-Poyo et al. described Cu isotopic analysis in serum using ETV-MC-ICP-MS168and fs-LA-MC-ICP-MS169 both following acid MAD. Accuracy and precision were assessed using NIST SRM 3114 (copper solution with a certified isotope Cu value), as both standard and sample, at various concentrations, and results were expressed as δ values. Both methods achieved comparable precision for NIST SRM 3114 at 1 mg L−1 Cu with average δ65Cu values of 0.00 ± 0.17‰ Cu (5 μL sample volume) and −0.01 ± −0.19‰ Cu (1 μL sample volume), respectively. Examination of patient samples with fs-LA-MC-ICP-MS169 demonstrated negative δ65Cu values (median −1.11‰, range from −0.31‰ to −2.29‰) for Wilson’s disease patients on chelation therapy (n = 8), compared with zero or positive values for other patient groups, including Wilson’s disease patients at diagnosis or on Zn therapy (n = 22). Adoption of these strategies into routine clinical use is likely precluded by simpler in-use ICP-MS methods achieving comparable LODs.

6.6.5 Chromium. Occupational exposure to CrVI is a known risk factor for cancer and respiratory disorders. Zhang et al.170 reported on an 8 year longitudinal cohort study investigating the relationship between Cr and lung function in 515 workers from chromate production and application plants in China between 2010 and 2017. Blood Cr levels and lung function were determined by ICP-MS and spirometry, respectively, and exposure–response curves demonstrated a non-linear correlation between lung function and ln-transformed blood Cr. A reference value of 6 μg L−1 was proposed for blood Cr, above which all lung function parameters showed downwards trends.
6.6.6 Iodine. Dekker et al.171 presented an association between salivary and urinary I levels measured by ICP-MS. Paired saliva and urine samples were collected from 20 healthy volunteers, 10 differentiated thyroid carcinoma patients with low dietary I intake, determined as urinary I excretion (UIE < 150 μg 24 h−1) and 10 patients on amiodarone with high iodine status (UIE ≥ 150 μg 24 h−1). Urine was collected in eight 2 h sub collections, with an additional overnight collection, and saliva was collected at the start of each urine sub collection. Salivary I content, with dilution corrected for by salivary protein and urea, was significantly different between participant groups (p < 0.01) and correlated with 24 h UIE. The study was limited by small sample size and correlation data was expressed as medians and interquartile ranges of different subject groups with no linear regression statistics. Data were insufficient to ascertain an optimal sampling time and large intra-individual variations were observed (37.7% CV for protein corrected salivary I and 26.9% CV for urea corrected urine I, respectively).
6.6.7 Iron. Iron accumulation in post-mortem liver and bone marrow samples from haemodialysis patients was documented in liver and bone marrow samples were obtained from 21 patients. Liver Fe content (LIC) was quantified by ETAAS and expressed as μmol g−1 dw. Paraffin embedded liver sections were stained with hematoxylin–eosin and Perl’s Prussian blue and bone marrow aspirate slides were stained with Prussian blue. More than half of the patients (57%) had an increased LIC (>36 mmol g−1), while bone marrow iron content (BMIC) was raised (>Gale’s Fe grade 3) in 45%. Resistance to erythropoiesis stimulating agents (ESA) was found in patients with higher LIC (p = 0.002). This contrasts with previous beliefs that ESA hypo responsiveness results from Fe deficiency and suggests a blocking mechanism of Fe utilisation. Measured LIC was also positively correlated with ferritin (r = 0.86, p < 0.001) and AUC–ROC curve analysis demonstrated diagnostic accuracy for Fe overload (LIC > 36 μmol g−1) with a ferritin concentration cut-off of 422 ng mL−1 (sensitivity 83.3%, specificity 77.8%). These observations reinforce the limit of 500 ng mL−1 for administering intravenous Fe to haemodialysis patients, suggested by a non-profit international organisation (KDIGO – Kidney Disease: Improving Global Outcomes).
6.6.8 Lithium. Salivary Li analysis was proposed by Parkin et al. as a non-invasive sample type for monitoring patients on Li treatment.172 Measurement of Li was performed using ICP-OES in 169 drool samples collected from 75 patients with psychiatric disorders across two recruitment sites in an 18 month longitudinal study. Salivary Li showed good correlation with serum Li (r = 0.77) when adjusted for covariates such as Li dose, smoking and diabetes using multivariate linear regression. The group proposed intrasubject saliva/serum ratios, calculated across multiple visits, to predict serum results from saliva samples at subsequent visits without blood collection. Robust predictions (predicted vs. observed r = 0.9) were achieved employing the mean saliva/serum ratio from three prior observations. Much larger studies are warranted before this approach could be considered for Li monitoring in clinical practice.
6.6.9 Mercury. Speciation analysis of Hg and Se in human brain tissue samples using HERFD-XAS was used to evaluate the differences between chronic low-level and short-term high level methyl mercury exposure from fish consumption.173 Samples were collected from subjects representing chronic long-term low-level Hg exposure (n = 4), short-term high-level exposure with long-term (n = 1) and shorter-term (n = 2) survival as well as healthy controls with no known exposure to Hg (n = 2). Brain tissue from long-term low-level exposure showed MeHg coordinated to an aliphatic thiolate as previously seen in marine fish. Mercuric selenide (HgSe) deposits and Hg(II)–bis-thiolate complexes were identified in samples derived from acute exposures, the latter of which may be a signature of intoxication and result in neuropathological changes. The authors concluded that this study demonstrated inadequacies of using acute exposure mechanistic animal studies to model chronic low-level Hg exposure in humans.
6.6.10 Titanium. Luders et al.174 investigated potential associations between plasma Ti concentrations and metal abrasion from magnetically controlled growing rods used in the treatment of early-onset scoliosis. This is the first study to evaluate material loss from the retrieved rods (n = 44 in 23 patients) in relation to plasma Ti quantified by ICP-MS, but results were relatively inconclusive. Metal wear was observed for most implants, however plasma Ti levels were not associated with any variables, such as implantation time or number of surgical lengthening procedures carried out to preserve spinal growth. The reported correlation with material loss was weak (R2 = 0.3743) and the authors also acknowledged that blood Ti levels may not reflect total Ti debris as deposition in other tissues and organs had not been included.
6.6.11 Zinc. Zinc concentrations and isotope ratios (δ66Zn) were determined in 68 breast tissue samples, comprising malignant and benign tumours, histologically normal tissue adjacent to each tumour type and healthy controls, using a double-spike (64Zn and 67Zn) MC-ICP-MS method.175 While findings of higher Zn concentrations and lighter δ66Zn in malignant tumour tissue with respect to tissue from healthy controls confirmed previous reports, the difference between δ66Zn in malignant tumour vs. adjacent normal tissue did not reach statistical significance. Two interesting and perhaps unexpected findings are noted. First, Zn concentrations and isotope ratios are almost indistinguishable between malignant and benign tumour tissue (15.2 ± 16.2 vs. 15.4 ± 16.2 μg g−1 and −0.37 ± 0.17 vs. −0.32 ± 0.16‰, respectively), meaning that this approach may not be suitable for differentiating malignant and benign disease. Second, significantly higher Zn concentrations were measured in normal tissue adjacent to benign tumours (7.4 ± 4.4 μg g−1) compared with both healthy control tissue (2.3 ± 1.7 μg g−1) and normal tissue adjacent to malignant tumours (1.9 ± 1.6 μg g−1). Zinc isotope ratios, however, were similar in all “normal” tissue groups (−0.17 ± 0.15 vs. −0.20 ± 0.13 and −0.25 ± 0.23, respectively). Various mechanisms were suggested to explain the results, including a specific immune response to a benign tumour. It is noted that there was no consideration as to whether the finding of higher Zn concentrations in normal tissue adjacent to benign tumours could have resulted from contamination during collection of this subset of samples.

7. Applications: drugs and pharmaceuticals, traditional medicines and supplements

The number of papers on the subject of elemental analysis in medicines seems to have decreased during this Update period, likely due to ICH Q3D regulation being well explored and established analytically. One interesting paper from Nascimento et al.56 investigated Br, Cl and I concentrations in herbal medications utilising optimised MAE in alkaline conditions. As there is little information on the effects of carbon on signal enhancement for these elements, the team devised to investigate them by dosing samples with known amounts of citric acid. The addition of carbon was found to cause signal enhancement in both aqueous solutions of citric acid as well as in herbal samples after alkaline MAE, with the signal of I being the most impacted, followed by that of Br, but with almost negligible effect on the Cl signal. A range of extraction conditions were tested, including different temperatures (90 °C or 240 °C), times (5 or 20 min) and extraction solutions (water, 100 mmol L−1 NH4OH or 100 mmol L−1 TMAH). Whereas Cl and Br were extracted (100%) under any of these conditions, effective recovery of I was only achieved under the stronger conditions. Extraction of the herbal samples with 100 mmol L−1. TMAH, heated to 240 °C for 5 min gave the optimal conditions to achieve good recoveries (95–105%) for CRMs (NIST SRM 1572 citrus leaves and NIST SRM 1547 peach leaves).

8. Applications: foods and beverages

In this section, prominent papers are discussed to highlight the most relevant progresses in the area of food and beverages analysis. In addition, technical details and most relevant findings of a number of other papers in this area are summarised in Table 3, complementing those reported in Table 1. The major focus, given also the increasing availability of instrumental techniques with capabilities for multi-element detection, and advanced chemometric tools, was on applications to a more and more wide range of foods and beverages to determine their authenticity and/or provenance. This trend may also be driven by the increasing attention paid by the general public to buy food and beverages produced locally or, otherwise, of trusted origin, especially for those goods that claim high prices.
Table 3 Food and beverages
Analyte Matrix Technique Study aim, procedure and comments Ref
Al Food ICP-MS The exposure to Al through the diet in North China was assessed by determining the Al content of 1232 food samples included in the traditional Chinese diet. After drying at 85 °C for 4 h, 0.1–0.5 g sample underwent MAD with 3 mL 60% HNO3–2 mL 30% H2O2, then the digests were diluted to 50 mL with deionised H2O. Scandium was used as the IS. LOD and LOQ were 0.5 and 1.5 mg kg−1, respectively, and recovery ranged from 90.5 to 96.7%. The average Al intake was estimated as 1.82 mg per kg bw per week and the P95 as 2.3 mg per kg bw per week, in this last case exceeding the PTWI of 2.0 mg per kg bw per week proposed by JEFCA. The main contributors were deep-fried dough sticks, starch products and steam bread, with average Al content of 219 mg kg−1, 84.5 mg kg−1 and 28.6 mg kg−1, respectively 278
As species Rice ICP-MS, HPLC-ICP-MS, LA-ICP-MS As levels in 53 pairs of paddy rice and parboiled paddy rice were determined to assess the role of rice treatment to reduce the risk of dietary exposure to As. Samples were dehusked, then part of each sample was milled to obtain a corresponding polished sample. Twenty-two pairs of rice bran were also obtained from the milling process. A total of 256 samples were finely ground, then digested with 6 mL 65% HNO3–1 mL 30% H2O2. Total As was determined by ICP-MS and As species (AsIII, AsV, MMA and DMA) by anion-exchange HPLC-ICP-MS. Analysis of rice flour CRMs for both total As and As species (NIST SRM 1598 and ERM BC211) gave recoveries ranging from 95% to 115%. LA-ICP-MS was applied to investigate As distribution within the rice grains. Polishing reduced the content of iAs in rice, but not in parboiled rice. DMA levels were low in both types of rice and MMA was not detected. Total As contents were 159 ± 46 μg kg−1 and 108 ± 33 μg kg−1 in unpolished and polished rice, whereas the corresponding values for parboiled rice were 145 ± 44 μg kg−1 and 145 ± 42 μg kg−1 279
As species Rice ICP-MS, HPLC-ICP-MS To assess As bioavailability from rice and the factors influencing it, 11 white rice grain samples underwent a modified physiologically-based extraction test (PBET). As species (AsIII, AsV, DMA and MMA) were separated by anion-exchange chromatography with 8.0 mmol L−1 (NH4)2HPO4 and 8.0 mmol L−1 NH4NO3 at pH 6.2 as the mobile phase. The results showed similar As bioavailability from both raw and cooked rice in the gastric phase (44–88% vs. 42–73%) and in the intestinal one (47–102% vs. 43–99%) 280
As species Seafood ICP-QMS, HPLC-ICP-MS A method was developed to determine As species (AsIII, AsV, AB, AC, DMA and MMA) in different types of seafood (Atlantic jackknife clam (Ensis directus), blue mussel (Mytilus edulis), Pacific oyster (Crassostrea gigas), common cockle (Cardium edule), tuna (Thunnus sp.) and Atlantic salmon (Salmo salar)), by anion-exchange HPLC-ICP-MS, after MAE with 5 mL of MeOH–H2O (3 + 1, v/v). LODs and LOQs for the As species ranged from 0.27 to 0.52 μg kg−1 and between 0.38 and 0.92 μg kg−1, respectively. Recoveries, evaluated on samples spiked with As species at three levels of concentration, ranged from 88.8% to 116.2% and similar figures were obtained on three CRMs certified for AB and DMA. 143
As species Water (deionised, mineral natural river water) LC-HG-AFS The stability of As species (AsIII, AsV, DMA and MMA) in water samples was investigated in different storage conditions. Species separation was carried out using a PRP-X100 anion exchange column under gradient elution with (NH4)2CO3, followed by on-line production of volatile arsenic hydrides with 1.2 mol L−1 HCl as carrier and 1.4% NaBH4 in 0.1 mol L−1 NaOH, as the reductant. LOQs ranged from 0.83 to 1.9 μg L−1 and RSDs from 3.4 to 6.4%. As species were found to be stable in samples acidified with HCl or HNO3, stored at +4 °C 281
As, Pb, Sn Soft drinks ICP-MS (Cd, Pb), ICP-AES (Sn) To simplify sample preparation, MAD with 2 mL HCl–5 mL HNO3 was applied to 0.5 g sample, followed by reduction of the digest to 0.5 mL, addition of 10 mL H2O and 1 mL (As, Pb) or 5 mL (Sn), HNO3, heating at 100 °C for 5 min and dilution to 50 mL with H2O. Addition of 1 mL H2SO4 after MAD, improved the performance for Sn. Recovery of spiked amounts of As (0.2 mg kg−1), Pb (0.2 mg kg−1) and Sn (150 mg kg−1) was 95%, 100% and 93%, respectively, and the corresponding intralaboratory RSDs were 9%, 8.6% and 1.8%. Analysis of six types of soft drinks (fermented beverage with plant extracts, Cola, reconstituted orange juice, vegetable juice cocktail, soy milk, liquid protein drink) spiked at the same levels as before yielded average recoveries of 95.8% (As), 97.6% (Pb) and 72.4% (Sn), thus supporting the applicability of the method to a variety of beverages 282
B, Pb and Sr Wine (“Lambrusco”) ICP-MS, MC-ICP-MS, GFAAS The concentrations of B and Pb (by ICP-MS) as well as Sr (by Zeeman-corrected GFAAS) were determined in samples of Lambrusco wine of different PDOs. In addition, isotope ratios (δ11B, δ18O, 206Pb[thin space (1/6-em)]:[thin space (1/6-em)]204Pb, 207Pb[thin space (1/6-em)]:[thin space (1/6-em)]204Pb, 207Pb[thin space (1/6-em)]:[thin space (1/6-em)]206Pb, 208Pb[thin space (1/6-em)]:[thin space (1/6-em)]204Pb, 208Pb[thin space (1/6-em)]:[thin space (1/6-em)]206Pb, and 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr) were measured by MC-ICP-MS, after acid digestion and SPE. These datasets, analysed by ANOVA, post hoc Tukey–Kramer test and PCA, allowed to identify the origin of the wine samples 283
Ca Seeds (rice, wheat, pearl millet, bean, maize, and cowpea) ICP-MS, EDXRF Human Ca deficiency worldwide could be addressed by biofortification of staple crops. EDXRF was assessed as a technique able to provide a fast response. Calcium levels in a set of cereal samples from biofortification breeding programs were determined by ICP-MS and used as reference. EDXRF results were comparable to those obtained by ICP-MS (average difference < ± 5 mg kg−1; RSD for duplicate analysis <10%) 284
Ca, K and P Edible (green, red, and brown) seaweeds EDXRF, ICP-OES To assess the quality of seaweeds currently used in gastronomy, a non-destructive method (EDXRF) was applied to determine Ca, K, and P levels in edible green, red, and brown seaweeds and the results compared with those obtained by ICP-OES, using a variety of statistical techniques as well as analysis of a CRM (NIST SRM 3232, kelp powder), with satisfactory results 133
Cd Rice LIBS A rice film sample was prepared after scraping the rice suspension onto a glass slide and compared to the conventional sample preparation method for Cd determination by LIBS. The intensity of the Cd-spectrum increased by 5–7 times, thus achieving a lower LOD 285
Cd, Pb Rice ICP-MS Soaking rice for 3 h with 200 ppm of potassium tartrate or potassium citrate, followed by cooking for 15 min with the same amount of chelating agent, was found more effective in reducing Cd (93–95%) and Pb levels (99%), than each process taken on its own, without affecting the taste of the food. The rice samples from these experiments, each replicated 10 times, were acid-digested (500 mg dry-sample) with 9 mL HNO3–3 mL HCl 286
Co (vitamin B12) Human milk HPLC-ICP-MS A method was developed for the determination of vitamin B12 in human milk via the measurement of Co by ICP-MS, to overcome difficulties experienced with this matrix by the microbiological and chemiluminescence enzyme immunoassay methods. To 1.0 mL sample, 0.5 mL of 1 mg mL−1 KCN, 4.0 mL of 0.4 mol L−1 CH3COONa buffer at pH 4.0 and 4.5 mL of water were added, followed by heating at 120 °C for 30 min. After cooling in a ice bath and centrifuging to remove the protein precipitate, 9.0 mL of the supernatant were loaded onto an immunoaffinity cartridge. The eluted fraction was evaporated to dryness then the residue was dissolved in 200 μL H2O for HPLC-ICP-MS analysis. The LOQ was 54 ng L−1, recovery of spiked amounts and indicative values from milk and infant formula CRMs were between 80% and 120% and within- and between-day RSDs were <10% and <15%, respectively 287
Cr species Beer, wine ICP-MS, HPLC-ICP-MS The speciation of Cr in beer and wine samples, within Cr ranges of 1.8–5 ng mL−1 and 6 to 22 ng mL−1, respectively, revealed concentrations <0.06 ng mL−1 CrVI in all samples. The separation of Cr species was achieved by anion-exchange chromatography. The addition of stable isotopes (50CrVI and 53CrIII) to beer and wine demonstrated that, due to the presence of antioxidants, CrVI was reduced to CrIII 148
Cr species Rice ID-HPLC-ICP-MS A procedure based on species-specific ID-HPLC-ICP-MS was developed to assess the presence of CrVI in rice, with LOD and LOQ of 0.004 and 0.014 μg kg−1 for CrIII and of 0.014 and 0.047 μg kg−1 for CrVI. In 10 rice samples, CrVI levels were <LOQ, while those of CrIII ranged from 0.59 and 104 μg kg−1, comparing well with the results obtained for total Cr determination. Spiking experiments confirmed that CrVI could not be detected, as it was reduced to CrIII 146
Cu, Mn Glycyrrhiza LIBS Cu and Mn mass fractions were determined in glycyrrhiza, a traditional Chinese medicine. To improve the accuracy of LIBS, in this work both an IS and calibration by standard addition were introduced. This approach led to both faster analyses (0.3 h) and improvement of accuracy and stability from 3 to 25 times. The results agreed well with those obtained by ICP-OES (p 0.05) 288
Fe, Mn, Zn Fortified infant foods (lactea cereal flour, infant cereal, milk powder, and chocolate powder) FAAS After measuring the levels of Fe, Mn and Zn in fortified infant foods (lactea cereal flour, infant cereal, milk powder, and chocolate powder), their in vitro bioaccessibility was evaluated, in the presence of potential inhibitors/promoters such as total dietary fiber, phytic acid, ascorbic acid and calcium, in three steps simulating salivary, gastric and intestinal digestion 289
Hg Fruit and vegetables (lemon, orange, tomato) VG-AFS A procedure based on microplasma-induced VG AFS was developed for the determination of Hg in fruits and vegetables without sample preparation. Hg0 CV was generated in a portion of sample juice uploaded in a stainless steel capillary, by applying a high voltage between the capillary and a tungsten electrode, then the vapour was moved to an AF spectrometer for rapid screening with LODs < 0.5 μg L−1. In the positive samples, Hg was quantified using the standard addition method 290
Hg, Pb Human milk ICP-MS The levels of Hg and Pb in 70 colostrum samples collected in Morocco as part of the CONTAMILK study were reported in two studies. Hg concentrations ranged from 1.64 to 124 μg L−1 (median 3.56 μg L−1) and those of Pb were between 1.38 and 515.39 μg L−1 (median 9.08 μg L−1) in both cases exceeding the “normal” values indicated by the WHO (Hg: 1.7 μg L−1; Pb: 2–5 μg L−1) 291,292
Se species Eggs HPLC-ICP-MS To assess the availability of Se species for human intake, analytical procedures to determine SeIV, SeVI, SeCys2, SeMet and MeSeCys, as well as total Se, in raw and cooked whole eggs were thoroughly investigated. Four extractants (H2O; 0.1 mol L−1 HCl; 0.1 mol L−1 NaOH; 25 mol L−1 CH3COONH4 buffer–5% MeOH (v/v)), as well as enzymatic hydrolysis with protease XIV, were studied, of which the last gave the best results. For the separation of Se species in the filtered extracts, two chromatographic separations (ion-pairing C18 RP chromatography, mobile phase 0.5 mol L−1 TBAH–10 mmol L−1 CH3COONH4–2% MeOH (v/v), at pH 5.5; anion-exchange chromatography with 5 mol L−1 citric acid, at pH 5.0, as the mobile phase) were evaluated. Both gave satisfactory results, but the first one provided better sensitivities due to the presence of MeOH. LOQs were 1.86 μg L−1 (SeCys2), 1.62 μg L−1 (MeSeCys), 1.82 μg L−1 (SeMet), 2.37 μg L−1 (SeIV), and 1.98 μg L−1 (SeVI). Recoveries of spiked amounts of SeCys2, SeMet and MeSeCys at three levels (0.1, 0.2 and 0.4 μg g−1), varied from 85.5% to 98.2% (external calibration) and from 86.9% to 96.7% (standard addition method) 149
Se species Garlic SEC-ICP-MS, HPLC-ICP-MS The presence and distribution of Se amino acids, that may play a role in cancer prevention, was investigated in four garlic clones, fortified with Se, at different stages of growth. Analysis by SEC-ICP-MS established that Se was associated with two molecular weight fractions (7–5 kDa and 2–4 kDa). In addition, Se amino acids were determined by anion-exchange chromatography coupled with ICP-MS, after peak identification by ESI-MS/MS. MeSeCys was the major species present 293
Se species Kefir grains RP-HPLC-ICP-OES SeO32− and seleno-DL-Met were determined simultaneously in kefir grains, with LODs and LOQs (as Se) of 0.52 and 1.73 mg kg−1 and 0.26 and 0.87 mg kg−1, respectively, for the two species. Recoveries from spiked kefir grains were 98.4 ± 0.8% and 93.6 ± 1.0%, respectively. Experiments indicated that supplementation of kefir grains with SeO32− did not induce conversion to seleno-DL-Met after 4 days at room temperature 294
Se species Z0206 selenium-enriched polysaccharide HPLC-ICP-MS To determine Se distribution of different Se species (Se, SeCys2, SeMet and MeSeCys) in Se-enriched food, such as Se-enriched polysaccharides, incubation with proteinase K for 18 h, followed by centrifugation, proved to be an efficient sample pretreatment. Se species could then be determined with LODs from 0.44 to 2.35 μg L−1 and LOQs in the range of 1.54–7.21 μg L−1, RSDs < 10% and recovery > 80%, except for SeCys2 295
Sr Wine TIMS, ICP-MS A lower isotopic ratio for 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr (<0.7072) was observed in a study of 101 wines from Tenerife and attributed to the volcanic geology of the Canary Islands. These values may help discriminate the local wines from those with other provenance. The influence of other factors, such as rainfall, on the variations of the 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr was discussed 296
Tl Food (agricultural, fishery, livestock, and processed foods), CRM (white cabbage) ICP-MS Potential interferences from different types of food on the determination of Tl were evaluated, using six representative items (apple, beef, orange juice, rice, sea salt and sesame oil), spiked with Tl at three concentration levels. After adjusting the dilution of the sea salt samples (0.1 g to 80 mL), recoveries ranged from 92.05% to 110.44% and RSDs from 1.09% to 9.79%. The analysis of a CRM (BCR-679, white cabbage) gave a recovery of 101% ± 2.94%. The method LOQs in the 6 food types ranged from 0.0222 μg kg−1 (beef) to 0.01585 μg kg−1 (apple). Thallium, measured in 304 food samples, was detected in 148 of them (<10 μg kg−1 in 98% of the samples) 297
V species Apples (V-enriched) ICP-OES V species (VIV and VV) in V-enriched apples were separated by RP-LC prior to quantification. V species could be preserved by addition of EDTA. VIV accumulated exclusively in the apple peel 298
Various Algae-based supplements EDXRF The elemental composition of 15 algae-based supplements (including Kelp, Sea spaghetti, Arame, Hijiki and Wakame) were evaluated by μEDXRF. Kelp algae were found to be a good source of I, although a high consumption may cause a health risk. Also, As or Pb levels in some of these products, may be of concern, in case of ingestion of a high daily dose of these supplements, taking into account the beneficial effect of the other nutrients they provide 299
Various Almonds (Prunus dulcis MILL.) ICP-MS, LA-ICP-MS The application of chemometric methods (SVM) to profiles of chemical elements obtained from 250 samples of almonds from more than 30 varieties, 6 countries and 4 crop years, allowed the country of origin of the samples to be distinguished, without influence of neither the type of almond or the crop year 300,301
Various Human milk ICP-MS The determination of macro and trace elements in human milk samples, collected before and after each feed, in a period of 24 h, from 11 mothers, allowed the assessment of variations in Ca, Cu, Fe, I, Mn, Mo, P, Se and Zn concentrations, associated with the time of sampling, that may affect the evaluation of dietary intake of these minerals in lactating infants 302
Various Milk HR-ICP-MS, IRMS The stable isotope composition (C, H, N and O) and element fingerprints were assessed to determine the origin of milk produced in a tropical country. 250 mL samples were collected on 3 consecutive days, during both the wet and dry periods, then stored at −20 °C. For ICP-MS analysis, MAD (200 °C for 10 min) was applied to 1 mL sample with 5 mL supra-pure >69% HNO3–1 mL 35% H2O2, followed by dilution to 25 mL with deionised H2O. Recoveries of certified elements from the EC IRMM CRM BCR-063R (skim milk powder) ranged from 85–115%, with RSDs <10%. LDA applied to the isotope signatures and the concentrations of Al, Ba, Co, Cr, Fe, Li, Mn, and Sr achieved a 100% discrimination power for the agroclimatic origin of the milk samples 303
Various Milk powder LIBS An assessment carried out on 25 milk powder samples mixed with four different types of exogenous proteins, identified an approach based on chemometric methods (LDA, kNN, RF and SVM as well as CNN) applied to datasets of chemical elements determined by LIBS, as a rapid and reliable procedure to identify milk powder adulteration 304
Various Octopus (Octopus vulgaris) ink TXRF Octopus ink was selected as a potentially useful sample for authenticity studies, based on elemental profiles. The application of chemometric techniques successfully discriminated four collection areas along the mainland Portuguese coast. Arsenic, Na, Pb, Rb and Sr gave the most important contribution to the identification of provenance 305
Various Rice EDXRF EDXRF was applied to commercially available rice samples (17 Basmati, 11 Thai, and 7 Long Grain rice) to obtain elemental profiles as a tool to authenticate different types of aromatic rice. Analysis of the data by SIMCA and PLS allowed the correct classification of 94.1, 85.6, and 100% of the Basmati, Long Grain, and Thai rice, respectively 306
Various Rice LIBS Fusion models based on LIBS and NIRS data were successfully applied to the identification of parents and progenies of hybrid rice, with accuracy >95% 307
Various Tea TXRF An assessment of sample preparation methods for TXRF analysis, based on tea powder (5 independent replicates, particle size > 180 mesh), highlighted a significant contribution (>60%) of the sample preparation to the global precision of the measurement results, regardless of the element and concentration range. An evaluation of dispersant, sample quantity and particle size showed that better RSDs and accuracy could be obtained by dispersing tea powder samples in deionised water rather than 1% Triton X-100, in a ratio of 20 mg in 5 mL, and with particle size between 200 and 300 mesh 308
Various Wheat flour EDXRF PCA and CARS were applied to elemental databases obtained from 68 wheat flour samples from three different origins to characterise their geographical features 309
Various (4) Beverages (beer, juice, soft drinks and wine) TXRF, ICP-MS The effect of sample preparation (direct analysis and acid digestion) on the interference of carbohydrate-rich matrices on the determination of Ca, K, P and S in alcoholic and nonalcoholic beverages by TXRF was investigated, by comparison of the results with ICP-MS. 310
Various (4) Milk powder TXRF The determination of elements in solid suspensions of milk powder by TXRF gave acceptable results for Ca, Fe, K and Zn, after careful evaluation of sample preparation and measurement conditions, as well as of the quantification approaches (internal standardisation vs. empirical calibration), and may be suitable for the screening of other elements 311
Various (5) Malatya cheese ICP-OES The variation of the concentrations of 5 elements (Ca, K, Mg, Na and P) depending on the process of production and temperature of storage of Malatya cheese were investigated 312
Various (5) Salt (refined, rock, and sea salt) ICP-MS, HG-AAS A survey of levels of As, Cu, Hg, Pb and Zn in edible salt samples from Iran (n = 60) was carried out to assess the contribution to dietary intake of these elements. Arsenic, Cu, Pb and Zn levels were measured by ICP-MS, after their co-precipitation from a 10% (w/v) salt solution with Dy(OH)3. Mercury was determined by HG-AAS, after reduction with NaBH4 and SnCl2. Recoveries between 96.1% and 101.7% were reported, but LODs and LOQs (between 0.01 and 3) were given in μg L−1, without further specification. Median levels of Cu (3.3–7.1 μg g−1) and Pb (9.96–27.2 μg g−1) were much higher than national and Codex Alimentarius limits (2 mg kg−1 and 1 mg kg−1, respectively) 313
Various (6) Drinking water, CRM ICP-QMS, SF-ICP-MS The performance of Q- and SF-ICP-MS for the determination of As, Cd, Cu, Pb, U and Zn in drinking water were compared, based on the respective performance in the international ILC IAEA 2015. Recoveries of the assigned values were 90–110% and 92–120%, respectively, except for Cu by SF-ICP-MS (64%). RSDs were better for SF-ICP-MS (<2%) than for Q-ICP-MS (<5%). LODs for both techniques were <ng L−1, except for Zn. Both methods were applied to measure the concentrations of these elements in drinking water 314
Various (6) Human milk FAAS, ETAAS Levels of essential elements (Cu, Fe and Zn) were determined by FAAS and those of PTEs (Cd, Cr and Pb) by ETAAS in 95 samples of milk, collected from mothers, two months after birth, at five health care centres in Kenitra, Morocco. All samples were digested prior to analysis. The mean concentrations were reported as 0.60 μg L−1 (Cd); 8.84 μg L−1 (Cr); 0.26 mg L−1 (Cu); 0.43 mg L−1 (Fe); 3.20 μg L−1 (Pb) and 6.11 mg L−1 (Zn). The WHO recommended limits were exceeded in 6.46% (Cd) and 27.37% (Pb) of samples and the mean values observed for Cr and Zn were also above the WHO indications 315
Various (6) Lamb meat ICP-OES An assessment of the bioaccessibility of Cu, Fe, K, Mg, P, and Zn from lamb meat samples, depending on the cooking method (grill, microwave oven, air fryer, pressure cooker, and electric oven), was carried out by measuring the element levels in the raw and cooked meat (recoveries from CRMs: 87–101%) and by an in vitro gastrointestinal simulation (recoveries of spiked amounts: 87–115%). The bioaccessibility of the studied elements ranged from 1 to 76%, with K being the most preserved element (68–76%) and Fe and Zn the less ones (4–19% and 1–21%, respectively) 316
Various (6) Seaweeds ICP-MS, HPLC-ICP-MS The total concentrations of As, Cd, Hg, I, Ni and Pb were determined in 33 dried seaweed samples after MAD (0.5 g sample digested with 3 mL HNO3–2 mL H2O–0.5 mL H2O2–2 mL HCl). An aliquot of the digest was diluted with an alkaline solution (40 g L−1 NH4OH, 10 g L−1, 1-butanol and 1 g L−1 EDTA) for I analysis. iAs was quantified in 0.2 g sample, after MAE with 10 mL of 0.2% HNO3–1% H2O2, at 90 °C for 20 min, followed by centrifugation (4500g, 10 min, 2 °C) and analysis by anion-exchange HPLC-ICP-MS, with a mobile phase of 3 mM NaH2PO4–0.3 mol L−1 EDTA–0.3% MeOH at pH 8.6. Seaweed samples included dulse (Palmaria palmata), hijiki (Hizikia fusiforme), kelp (Laminaria spp.), kombu (Laminaria japonica), nori (Porphyra tenera), sea lettuce (Ulva lactuca), sea spaghetti (Himanthalia elongata), wakame (Undaria pinnatifida) and agar–agar (Gelidium spp.). Iodine levels were high in most samples and iAs was high in the hijiki samples 317
Various (7) Human milk ICP-MS A study was carried out in Japan to assess the current levels of relevant trace elements in human milk and their correlation with other factors. 79 healthy mothers of single newborns were recruited to collect milk samples at 1 and 3 months after delivery as well as provide information on food consumption frequency. The observed median concentrations (at 1 and 3 months postpartum) were: Cr: 0.8 and 0.6 μg dL−1, Cu: 50 and 33 μg dL−1, Fe: 98 and 55 μg dL−1, Mn: 0.8 and 0.7 μg dL−1, Mo: 0.7 and 0.7 μg dL−1, Se: 2.2 and 2.1 μg dL−1, Zn: 272 and 177 μg dL−1 318
Various (8) Cheese ICP-OES Three typical Italian Pecorino cheeses, two of them with PDO, were classified with total accuracy of 93.3% and a predictive accuracy of 91.3%, by applying PCA and PLS-DA to datasets of 8 major elements (Ba, Ca, Fe, K, Mg, Na, P, and Zn) obtained on a total of 53 cheese samples 319
Various (9) Wild mushrooms (Craterellus cornucopioides, Boletus aereus, and Cantharellus cibarius) ICP-OES Samples of wild mushrooms were analysed to determine their concentrations of Al, B, Ca, Cr, Cu, Fe, Mg, Ni, and Zn. The method's LOQs ranged from 0.08 mg kg−1 (Mn) to 4.57 mg kg−1 (Al). Recoveries were between 80.9% and 119.7%, and RSDs were >8.8%. PCA, ANOVA and PLS-DA were applied to classify the samples according to their species 320
Various (10) Chickpeas (Cicer arietinum L.) ICP-OES Chickpea samples were ground with a mortar, the peel discarded and 0.25 g of the flour underwent MAD with 5 mL HNO3–5 mL H2O2, at constant power (1000 W), a ramp time to 180 °C of 10 min and a hold time of 40 min. The digests were diluted to 25 mL with ultrapure H2O. Statistical and chemometric tools (ANOVA, LDA and SIMCA) were applied to data obtained for 10 elements (Ca, Cu, Fe, K, Mg, Mn, Mo, P, Sr and Zn) on 60 samples of chickpeas harvested in different areas of Italy, to identify patterns related to their provenance 190
Various (12) Milks and plant-based drinks ICP-MS The content of 12 elements (Ca, Co, Hg, K, Mg, Mn, Na, Ni, P, Pb, U and V) was determined in cow and goat milk (including lactose-free, fresh and UHT types) and plant-based drinks (soy, almond, rice and oat) as well as in infant formulas, from organic and conventional production systems, using previously described methods, with reported LODs of 250 μg g−1 (K, Na and P), 50 μg g−1 (Ca), 25 μg g−1 (Mg), 0.05 μg g−1 (Hg, Mn, Ni, Pb and V) and 0.025 μg g−1 (Co and U). Recoveries were assessed on CRMs (NIST SRM 1549a whole milk powder, for Ca, K, Mg, Na and P; NIST SRM 1570a Spinach leaves, for Co, Hg, Mn, Ni, Pb, V and U) and ranged from 75 to 110%, except for Ni (55%) 321
Various (13) Human milk FAAS, ICP-MS The levels of As, Be, Ca, Cd, Co, Cr, Cu, Fe, Mn, Pb, Sr, U and Zn were determined in colostrum and mature milk from 150 women, divided in three groups of about 50 subjects, according to their smoking/nonsmoking habit and exposure to second hand smoke. Samples underwent MAD with 1.5 mL 65% HNO3–0.5 mL ≥ 30% H2O2. Cd and Pb concentrations were higher in both the smoking and exposed groups compared to controls 322
Various (13) Wine TXRF PCA and cluster analysis, applied to datasets of 13 elements, allowed identification of a panel of elements (Ba, K, Mn, Ni, Rb and Sr) that allowed discrimination between red and white wines of different origin. Furthermore, LDA achieved 100% accuracy and 96.43% prediction capability 323
Various (15) American ginseng (Panax quinquefolius L.) ICP-MS Fifteen elements (Al, Ca, Cd, Co, Cu, Fe, K, Mg, Mn, Na, Ni, Rb, Sr, Ti, and Zn) were determined in 44 American ginseng samples from three countries (Canada, America and China), as a potential panel for geographical authentication. Both PCA and LDA were applied to the datasets. LDA classified the ginger samples according to their country of origin with satisfactory predictive ability (93.2%) 324
Various (15) Kiwi fruit (organic and conventionally grown) ICP-OES, IRMS Classification methods (SIMCA, PLS-DA and LS-SVM) were applied to datasets of 15 elements (Al, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Na, Pb, Se, Sr, and Zn) and three stable isotopes (δ13C, δ15N, and δ18O) obtained from samples of organically (n = 78) and conventionally grown (n = 85) fresh kiwi fruit. This approach proved successful for the discrimination between the two types of kiwifruits, with a classification accuracy of 0.950 and 0.983 for organic and non-organic kiwifruits, respectively 325
Various (16) As species, Fe species Yerba mate (Ilex paraguariensis) HPLC-HG-ICP-OES, IC-ICP-OES, ICP-OES The concentrations of sixteen elements (Al, As, Cd, Co, Cr, Cu, Fe, Hg, Li, Mn, Mo, Ni, Pb, Sb, Se, Zn) were determined in 58 acid digested samples of yerba mate of different origins as well as in the corresponding extracts, with recoveries within the range 80–120% and LOQs ranging from 0.006 mg kg−1 (Li) to 0.360 mg kg −1 (As). Furthermore, As (AsIII, AsV, DMA) and Fe (FeII and FeIII) species were determined in the same samples. As speciation was achieved by HPLC-HG-ICP-OES and Fe species were separated by IC-ICP-OES. 326
Various (17) Edible insects ICP-MS, FAAS Edible insects, harvested in the natural environment, are regularly consumed in tropical countries. The content of trace elements (Ag, As, Be, Cd, Co, Cr, Cu, Ga, Hg, Mn, Ni, Pb, Rb, Se, Sr, V, Zn) was determined in 12 species of insects, caught in the wild and stored frozen. After thawing, drying at 65 °C and crushing into flour, 1 g of each species sample was digested in PTFE vessels with 8 mL HNO3–1 mL H2O2, by heating in a sand bath at 150 °C for 2 h. The digests were diluted to 100 mL with deionised H2O, prior to analysis. Unfortunately, no detail of analytical quality are provided. The reported mass fractions were (mg kg−1): Ag: 0.050–7.636; As: 1.534–6.36; Be: ND; Cd: 0.016–0.504; Cr: 1.01–5.64; Co: 0.932–11.59; Cu: 3.023–5.832; Ga: 0.164–0.475; Hg: 0.010–0.315; Mn: 2.224–9.927; Ni: 0.136–8.637; Pb: 0.041–0.501; Rb: 0.125–0.471; Se: 0.410–3.094; Sr: 0.074–0.493; V: 0.895–3.778 and Zn: 4.117–12.041 327
Various (17) Olive oil (extra virgin) Q-ICP-MS The concentrations of 17 trace elements (As, Ba, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Ni, Rb, Pb, Sr, V, and Zn) was determined in 42 olive oil and 9 soil samples from three different Mediterranean countries. Oil samples were centrifuged, then 0.5 g of the supernatant was mixed with 0.5 mL H2O2 and left overnight at room temperature. MAD was performed after the addition of 5 mL 69% sub-boiled HNO3. The digests were evaporated to dryness, then re-dissolved in 5 mL of 2% sub-boiled HNO3 prior to analysis. LODs were between 0.0002 and 0.313 μg kg−1 and RSDs between 2% and 15%. Recoveries for the certified elements (Ca, Cu, Fe, K, Mg, Mn, Zn) in a CRM (NIST SRM 2387 peanut butter) ranged between 86% and 102%. PCA, applied to these datasets, allowed to discriminate the oil origin 328
Various (18) Teas XRF 75 tea (black, green, white, herbal) samples from different countries were classified into 5 major geographical region of origin, by applying PCA, SIMCA and PLS-DA to datasets of 18 chemical elements 329
Various (18) Wine (red and white) ICP-MS, IRMS LDA was applied to datasets of 18 elements (Al, As, B, Ca, Co, Cu, Fe, K, Li, Mg, Mn, Na, Ni, Pb, Rb, Sr, V and Zn) and two stable isotopes (δ13C and δ18O), determined in white and red wines from Austria and Argentina. The wine origin was correctly identified in 100% of the cases, on the basis of 12 elements (As, B, Co, Li, Mg, Mn, Na, Pb, Rb, Sr, V and Zn) and δ18O 330
Various (19) Legumes (common bean grain and pod, lupin, faba bean, lentil, chickpea, soybean, field pea and runner bean) ICP-MS, EDXRF Two instrumental methods were applied to the determination of 20 homogenised and powdered legume samples. For ICP-MS analysis, 250 mg of each sample underwent MAD with 6 mL 65% HNO3–2 mL 30% H2O2, then were diluted to 50 mL. For EDXRF, pellets were prepared from 0.5 g to 1.0 g of sample. Based on the comparison, EDXRF was preferred for the determination of Br, Ca, Cl, K, Fe, Mo, P, Rb, S, Sr, Ti and Zn, being cheaper, simpler and more environmentally friendly, whereas ICP-MS was the method of choice for Co, Cr, Cu, Mg, Mn, Na and V, owing to better sensitivity and accuracy. Results obtained by each technique for its panel of elements were in good agreement with the certified values of a CRM (NIST SRM 1573a tomato leaves), although this matrix may not fully represent that of legumes. Nickel, that is present in most legumes, was not included in the assessment 331
Various (19) Saffron, food and environmental CRMs EDXRF The possibility to reduce the sample size to only 1 g was evaluated by comparing two sampling methods (small sample holders and the double pellet method, consisting of a thin sample layer on a wax pellet) using 17 different organic matrix RMs (lichen, wheat flour, rice flour, bran, brown bread, cabbage, vegetable feed, pine needles, spinach, tomato and tobacco leaves). Both methods had advantages and drawbacks, the double pellets providing 100% accurate results for the lighter elements and the small sample holders giving recoveries >80% for most of the tested elements, except the lightest (Al, Mg and Si). These methods allowed the analysis of saffron samples, available in limited amounts 332
Various (23) Bee honey ICP-MS The levels of trace elements (Ag, As, Ba, Be, Bi, Cd, Co, Cr, Cu, Fe, Hg, Li, Mn, Mo, Ni, Pb, Sb, Sn, Sr, Te, Tl, V, and Zn) in four unifloral (asphodel, eucalyptus, strawberry tree, and thistle) bee honeys from Sardinia, Italy, were investigated, to support geographical origin. After samples were heated to 40 °C and homogenised, 0.7 g of the liquified sample were digested with 0.5 cm3 HNO3–3 cm3 H2O2–4 cm3 H2O (type I) in PTFE vessels on a Single Reaction Chamber (SRC) microwave digestion system, followed by dilution of the digests to 15 cm3, filtering and storing in the dark at 4 °C until analysis. Given the large variability of the analyte concentrations in the samples, calibrations were aimed to cover, for each element, only the relevant range of concentration. No CRM was available, hence trueness was evaluated by recovery of spiked amounts and ranged from 85% (Be) to 130% (Hg). LDA, applied to the elemental datasets, allowed correct identification of their botanical origin 333
Various (23) Sea cucumber (Apostichopus japonicas) ICP-OES, ICP-MS Techniques for data analysis (XGBoost and SHAP) were applied to datasets of 23 elements determined in 167 sea cucumber samples to identify their geographical origin. Selenium was identified as the most important marker of provenance 334
Various (24) Seafood (Pollicipes pollicipes) TXRF Chemometric tools (CAP, LDA, RF, S-LDA and VIP-PLS-DA) were applied to datasets of 24 elements, measured in 90 samples of stalked barnacles, from 6 sites along the Portuguese western coast, to evaluate the ability of each approach to trace the geographical origin of the animals collected 335
Various (25) Soybeans ICP-MS An approach for the identification of the geographical origin of soybeans from 4 regions of the same Chinese province, based on the measurement of 25 chemical elements, was developed. A panel of techniques (difference, correlation, cluster heat map, and OPLS-DA) was applied to the datasets. This study indicated that the relative distribution of the elements reflected differences in the soil composition in the growing areas. In addition, classification of the soybeans according to their origin could be achieved by means of a discriminant model using the information from 17 elements (Al, Ca, Cr, Cu, Fe, Ga, K, Mg, Mn, Mo, Ni, P, Pd, Pb, Rb, Se and Zn), with a correct discrimination rate of 93.33% 336
Various (26) Edible seaweed ICP-MS Elemental profiles were determined in 46 edible seaweed samples, divided in 13 subgroups/species based on DNA barcoding analysis, after acid MAD. Relationship with taxonomy of the species were explored. High interspecies variation was also observed for some elements 337
Various (30) Bee honey (monofloral) ICP-OES, ICP-MS, FAAS To determine the botanical origin of monofloral bee honey, an approach based on the statistical method of self-organizing maps was developed and successfully applied to datasets of both 30 chemical elements and 9 physicochemical parameters obtained on 62 monofloral bee honey samples (31 linden, 14 rapeseed, 13 sunflower, and 4 acacia), in comparison with results of melissopalynological analysis 338
Various (30) Rice ICP-MS/MS The elemental profile of rice samples was determined to investigate their provenance. Interferences were removed by applying a collision cell, with H2, O2 or NH3–He. LODs ranged from 0.0000662 to 0.144 mg kg−1 and LOQs between 0.000221 and 0.479 mg kg−1. Recovery varied between 80.6% and 110.5% at spike levels from 0.10 to 100.00 mg kg−1. RSDs were <10% 339
Various (31) Whisky ICP-OES, ICP-MS, CVAAS The elemental profiles of 170 samples of whisky from 11 countries were determined and the resulting datasets were examined in relation to country of origin, type of whisky (single malt and blended) and age of products. Single malt Scotch whisky was characterised by significantly higher levels of Cu, that increased, together with Mn content, with aging in barrels 340
Various (40) Insects (Prionoplus reticularis) ICP-MS The content of 40 chemical elements was investigated in larvae of Prionoplus reticularis, a traditional food in New Zealand. Measurements were carried out on freeze-dried samples of harvested insects, pooled according to the stage of development. A CRM (NRCC DORM-4 fish protein) was used as a control material, however, most of the certified concentrations and maybe even the matrix did not closely match the samples. A total of 28 elements were detected 341
Various (45) Olive oils (extra virgin), CRM ICP-MS 45 elements were quantified in 0.5 g of extra virgin olive oils. Wet digestion in a water bath with 2[thin space (1/6-em)]:[thin space (1/6-em)]1 (v/v 10% (v/v) HNO3–H2O2, at 95 °C for 40 min, applied to an oil-based CRM (Conostan S-21) and three real samples, gave the best results, as compared with other reagent mixtures and UAE. LOQs ranged from 0.004 μg kg−1 (Be) to 270 μg kg−1 (Si). Recoveries of spiked amounts varied from 85% to 120%, except for Ba, Ca and Fe (<14%) 342
Various (46) Hazelnuts ICP-OES, ICP-MS Samples (20 g) of three different hazelnut cultivars and of five processed products (roasted hazelnuts, hazelnut paste, hazelnut cream, Gianduja paste and Gianduiotto paste) obtained from each group, were dry-ashed with the support of microwave irradiation prior to analysis by ICP-OES and ICP-MS. This sample treatment maximised the sample size for presentation to the instrument, hence the LOQs, although some elements were lost by volatisation (As, Cd, Hg and Pb). LDA and PCA, applied to the elemental profiles, allowed discrimination of the cultivars and the respective products, with Mo and Ni being the most prominent markers 343
Various (49) REEs Wine ICP-MS The levels of 49 elements, including REEs, in organic and conventional wines, produced in an area of volcanic origin (Canary Islands), was investigated. No significant differences were found between the two production methods, whereas the element profile could be linked to the island of origin 344
Various (52) Anchovies (salt-ripened), cuttlefish ICP-MS, DMA (Hg) The elemental profile of transformed anchovies was determined in 180 products from three different fishing areas before and after packaging. Data were analysed with four decision trees-based algorithms (C5.0, CART, CHAID, and QUEST). Six elements (As, Cd, K, Li, P and Sr) supported the identification of provenance of bulk anchovies, whereas the concentrations of As, B, Cd, K, and Pd allowed classification of packaged products according to their origin. In a second paper, elemental profiles from 68 cuttlefish samples were studied. Among various chemometric techniques, VIP-PLS-DA allowed classification of samples according to their origin with 100% sensitivity, specificity and accuracy 345,346
Various (53) Wild Astragali Radix ICP-MS, LIBS The authenticity of Wild Astragali Radix, used both as food and medicine, was assessed by applying PCA and PLS-DA to datasets of 53 elements, achieving accuracy up to 93.8% 347
Samples from three provinces in China were analysed after MAD
Various (63) Cheese (Emmer type) ICP-OES Samples of Emmer cheese, produced in three different locations in Italy, were analysed, after MAD, to determine their elemental profile. The method LODs and LOQs ranged from 0.02 to 1.03 μg g−1 dw and from 0.07 to 3.44 μg g−1 dw, respectively. For spiked amounts, analytical recoveries of 83–107% were achieved. Statistical (ANOVA) and chemometric (PLS-DA) approaches allowed the identification of the samples according to the geographical origin, with a 100% accuracy 348


8.1 Progress for individual elements

8.1.1 Arsenic. A useful non-chromatographic method for iAs determination was proposed by Wang et al. 120 The team utilised CdS/MIL-100(Fe) as a sorbent and photocatalyst which allowed determination by PVG-ICP-MS. They synthesised the CdS/MIL-100(Fe) using previously described methods modified to establish optimal conditions. It was found that pH had a small effect on the rate of adsorption, minimising the need for pH adjustment. Furthermore, the rate of adsorption of AsIII was considerably greater than that for AsV, allowing for effective speciation. AsIII was entirely adsorbed within 5 min, whereas AsV took 105 min for complete adsorption. The LOQ for the method was established to be 1.7 ng mL−1 with direct analysis and 0.11 ng mL−1 with a 20-fold pre-concentration. Good recoveries were obtained from a variety of water and food matrices and equivalent results were obtained on samples analysed by chromatography and this non-chromatographic method, demonstrating the method to be suitable for As speciation with sufficient sensitivity for As speciation in food and water samples.

The current desire to reduce animal products in diets has led to a huge increase in the number of plant-based milk alternatives available on the market. Traditional almond and soy milks have been joined by rice, millet, oat and coconut milks. To this end Ruzik et al.176 examined the As content in a variety of plant based milks available on the Polish market. Total As was determined using MAD in HNO3 followed by ICP-MS. Unsurprisingly it was found that levels of total As were highest in rice-based milks (2.3 μg L−1), followed by those in milks made from other plants, with almond milk showing the lowest levels of As (0.016 μg L−1). The authors noted that no organic forms of As were observed in the HPLC-ICP-MS part of the study with only iAsIII and iAsV being detected.

8.1.2 Mercury. By utilising modified AlO substrates Ahmad et al.177 were able to determine HgII species in water and fish samples by ICP-OES with a low LOD of 0.02 ng mL−1. Aluminium oxide substrates were functionalised with polyaminophosphonic acid giving a high flow rate SPE membrane. By controlling flow rate and pH, the membranes were found to be selective at extracting HgII ions from other elements, such as alkaline earth metals. The SPE was found to be effective in the pH range of 3–7, with an optimal range of pH 6–7 linked to the zeta potential of the membrane surface. The membrane was shown to have a potential maximum loading of 361 mg g−1 allowing a pre-concentration factor of 933 under the preparation conditions used. Recovery from RMs and spiked RMs was found to be 98–100% with RSD < 1% across all samples matrices, demonstrating that the material may be a useful tool for determination of low level HgII in food and water matrices.
8.1.3 Iodine. The most commonly used preparation method for I analysis by ICP-MS is alkaline digestion. In some cases, the high pH generated in this procedure can cause precipitation of Ca and Mg in water samples leading to nebuliser blockages. Rosen et al.178 looked at a simple dilution method using 2% NH4OH only or 2% NH4OH–0.1% EDTA, It was found that in bottled and tap waters precipitation of Ca and Mg occurred in all samples when stored with just NH4OH whereas only one sample was found to have precipitate with the EDTA solution. Analysis of the samples showed that the I concentration was unaffected by either treatment. This indicated that this preparation was suitable for I determination in waters and the authors proposed this procedure as a green replacement for dilution with TMAH which is commonly used.

8.2 Single and multi-element applications in food and beverages

8.2.1 Cereals. In a study carried out by Sadiq et al.,179saliva, gastric and intestinal leachates of a variety of rice based infant foods, obtained from batch leaching as well as from a novel on-line leaching method, were analysed for their element content. Arsenic, Cd, Cr, Cu, Fe, Pb, Se, and Zn concentrations were studied using both leaching methods and recoveries established by comparison with acid extracted aliquots of the samples. In the on-line leaching method, ∼0.1 g of sample was exposed to saliva, gastric and intestinal solutions for 5 min at the rate of 1 mL min and elements were measured in time resolved mode. For the batch leach method ∼0.2 g aliquots of sample were leached in subsequent aliquots of the saliva, gastric and intestinal solutions over a period of 4 h to mimic in vivo conditions. Although some differences were seen in extraction profiles for the batch and on-line methods, the total bioavailable concentrations were found to be similar suggesting that the on-line method may be a suitable rapid assessment tool for bioavailability of elements in rice based infant foods. Speciation analysis was also carried out for As, Cr and Hg using HPLC-ICP-MS on samples that had been extracted using the on-line extraction method. In saliva and gastric solutions. As and Se were found to exist entirely as AsV and SeVI forms. For Cr, unusually CrVI was detected in the extracts for all samples, particularly in the saliva models.

Chromium speciation was also the subject of a study carried out by Saraiva et al.147 The team used SSID-HPLC-ICP-MS to ascertain CrIII and CrVI levels in breads and breakfast cereals. Total Cr was also measured by ICP-MS after MAD. A sequential extraction procedure was optimised, using EDTA to complex CrIII species, followed by reduction of CrVI to CrIII with 1,5-diphenylcarbazide which was further complexed by its oxidised form (DPC). The EDTA and DPC forms were then separated by anion-exchange HPLC in a 3 min run. For the SSID, 50CrIII and 53CrVI was spiked into all samples. The method was validated and the efficacy of the speciation was assessed by spiked recovery studies using CrVI, that returned good recoveries between 97% and 113%. The method LOQ was found to be 0.014 μg kg for CrIII and 0.047 μg kg for CrVI. The CrVI species was not found above the LOQ in any of the samples, and the authors suggested that previous reports of CrVI being present in bread samples may be likely due to analytical artifacts.

8.2.2 Vegetables, fruit, mushrooms and nuts. Tree nuts are widely consumed globally and are prized for their high nutritional value, but due to the complex composition, and presence of high levels of fats, elemental analysis can be complex and time consuming. Rovasi Adolfo and team180 utilised solid–oil–water emulsion breaking to investigate levels of Ba, Co, Fe and Ni in Brazil nuts, cashews, hazelnuts, pecans and peanuts. Approximately 0.8 g underwent emulsification in an optimised Triton X-100–HNO3 solution, and, after briefly heating to 90 °C, the mixture centrifuged to allow for extraction of the aqueous layer. The same samples were also prepared by MAD in HNO3. The extraction procedure was considerably shorter, taking 14 min compared to 45 min for the acid digestion. All samples were analysed using HR-CS-GFAAS, using secondary wavelengths close in spectral range, less than 0.4 nm between all of them, allowing simultaneous analysis of the four elements, a feature which is uncommon for this technique. As secondary wavelengths were utilised sensitivities were reduced, but LOQs of 1.146 mg L for Ba, 7.421 μg L for Co, 0.285 mg L for Fe, and 6.614 μg L for Ni were still achieved. Recoveries of analytes from the extraction procedure were comparable to those obtained by MAD, and ranged from 93.5% for Ni to 104.5% for Co, thus demonstrating the suitability of the method as a rapid, green extraction for Ba, Co, Fe and Ni in nuts.
8.2.3 Authenticity. Studies of authenticity and origin formed a significant proportion of this Update in the area of food and beverage analysis. This field has grown in significance in a short space of time with a wide variety of techniques and matrices being explored.

Panax notoginseng, or Chinese ginseng is a traditional herbal medicine widely used globally with many accepted and proposed health benefits. Ji et al.181 measured 49 elements by ICP-QQQ-MS in 89 samples to build discriminant models to identify material from different regions and cultivation types. Chemometric tools, such as LDA, LR, kNNs, NB, NNs, PLS-DA, RF and SVMs, were used to build prediction models. For both origin and cultivation discrimination 200Hg, 202Hg, 97Mo, 98Mo, 23Na, 60Ni, 118Sn, 205Tl, and 232Th were found to have no significant impact on the models. For assessment of cultivation only 107Ag, 137Ba, 43Ca, 44Ca, 24Mg, and 66Zn were required to discriminate between field and forest grown samples. For geographical origins, 35 elements were used to discriminate between the five regions studied.

Avocados were the subject of a study by Muñoz-Redondo182 to identify those from a Spanish origin. Commercially available avocados (131 samples) from eight countries (Spain, Brazil, Chile, Colombia, Kenya, Mexico, Peru and South Africa) were used for the modelling. Samples underwent ICP-MS analysis for 46 elements. Furthermore, the determination of stable isotope ratios for C, H, N, O and S, using IRMS, was performed on the lipid and protein fractions (denoted with the subscript letters “L” and “P”, respectively) of the avocado, to allow for variations of lipid concentrations, which have previously been reported to show lower δ13C concentrations. The PLS-DA model yielded a high classification ratio of 98%, revealing the potential of stable isotopes (δ13CP, δ2HL, δ2HP, δ15NPδ18OL and δ18OP) and elemental profiles of Ba, Ca, Eu, Fe, Mg, Mn, Rb and Sr to trace the provenance of Spanish avocados.

As a staple commodity for half the world’s population, rice adulteration is common and adulteration and mislabelling is a global issue. Origin of rice from India, China and Vietnam was assessed by Quinn et al.183 using ICP-MS followed by chemometric analysis, using DD-SIMCA, kNN and PLS-DA models. Fifty-nine Chinese, 71 Indian and 29 Vietnamese rice samples were analysed. Samples underwent MAD with equal volumes of HNO3 and H2O2, followed by the determination of 40 target elements by ICP-MS. Modelling with PLS-DA was able to discriminate between the three origins. Chinese rice samples showed higher concentrations of Al, Ca and Mg, whereas those from Vietnam contained high levels of Zn and low concentrations of Ge, and rice samples from India had higher concentration of B, Co, Cu, Fe, Mo, Se, Sr, Ti and W. This model showed a sensitivity of 99% and specificity of 100%. Alternative models (DD-SIMCA and kNN) returned similar levels of sensitivity and specificity as seen with the PLS-DA model.

Virgin coconut oil is a versatile commodity with a demand from the food, pharmaceutical and cosmetic industries, which brings in the requirement to be able to assess the origin of these oils. Amit et al.184 used HCA, LDA and PCA, to build the models from the data obtained by ICP-MS on 18 elements. Both Principal Component Regression and PLSR were able to predict the origins of coconut oils with a R2 of 0.999 and low RMSEP values (0.074% and 0.075%, respectively).

To ascertain the region of production, artisanal cheeses from Brazil were analysed for a small group of elements (Ca, Cu, K, Mg, Mn, Na and K) by ICP-OES, and the resulting data explored using a variety of statistical models (ANN, kNN, LVQ, RF and SVM).185 The models provided by SVM and RF were the most efficient for classifying cheese types, with no significant difference in performance. Copper was the primary element of influence in the models. An accuracy of 100% was achieved for both models, demonstrating the usefulness of chemometric tools applied to the results of the determination of these elements as a tool for identifying the cheese production regions in Brazil and supporting PDO claims.

In a study by Zhang et al.,186the possibility of discriminate the origin of Pu’er teas from a small geographical region of ten adjacent sites was explored by modelling element (Ca, K, Mn, Ni, P and S) profiles, determined by low-power TXRF, and the results of quantification of caffeine and polyphenols by spectrophotometry, obtained on brewed extracts. Whereas PCA showed high variance, proving too unreliable for modelling of samples of close geographical origin, LDA gave lower variance and a predictive ability of 100%, hence proving to be a suitable tool for discriminating the origin of Pu’er teas, even from similar geographical locations.

The demand for sustainable food is ever increasing, and the demand for sustainably grown chicken is no exception. As with any food that demands a premium price, adulteration is a possibility. Islam et al.187 measured 21 elements in chicken meat by ICP-MS (As, Ba, Cd, Co, Cr, Cu, Ga, Li, Mn, Pb, Rb, Se, Sr, V and Zn) and ICP-OES (Ca, K, Mg, Na, P and S), then utilised PCA and LDA modelling to assess whether these tools could discriminate chicken from sustainable sources vs. conventionally grown ones. With analytical methods achieving recoveries of spiked amounts between 95% and 110% for all 21 elements, chemometric analysis by PCA and LDA successfully differentiated the two sample types with a 100% accuracy.

The content of eighteen elements in grape juices from a single vineyard was explored by LDA and PCA analysis, with the aim of identifying different harvests as well as discriminating between Japanese and European grape varieties.188 Fruits from seven European and three Japanese grape varieties were collected across three growing seasons between 2017 and 2019. These samples were prepared by acid digestion prior to analysis by ICP-MS and ICP-OES, to accumulate data to build the models. After applying PCA to the datasets, it was found that Ba, Ca, Ga and Sr could discriminate between European and Japanese varieties, whereas vintage could be discriminated by the concentrations of Mg, Mn, P and S. The authors noted that, since relatively small variance was seen in mineral content from samples coming from the same vineyard, the model would be able to appreciate larger differences in elemental composition deriving from different geographies.

Lemons from Sorrento (Italy) are protected by PGI status, dependent on both the variety and where they are grown. Ruggiero et al.189 investigated the use of chemometrics applied to element levels to establish a model for identifying lemons from PGI groves. Samples were analysed by ICP-MS for 14 elements, following MAD. The resulting datasets underwent PCA and Subspace LDA (SLDA) statistical analysis. Modelling using eight key elements (Ba, Ca, Co, Fe, Mg, Mo, Rb and Sr) could discriminate lemons from the Sorrento PGI region from those grown in other nearby areas, with no PGI status, with 98% correct classification, 98% accuracy and 94% external validation. The models could not identify different varieties, due to the strong influence of the soil on the elemental composition of the fruit, that would require further sampling points to improve the reliability of the model.

Italian chickpeas (Cicer arietinum L.) were the subject of origin studies by Di Donato et al. 190 Sixty chickpea samples, harvested in three different areas of Italy, were analysed by ICP-OES. Using a relatively small number of elements (Ca, Cu, Fe, K, Mg, Mn, Mo, P, Sr and Zn), successful modelling of the geographical origin of the chickpeas was established. All test samples were correctly identified following training by the LDA model. Three SIMCA class models, built for the Cicerale, Valentano and Navelli varieties, exhibited good sensitivity (88%, 90% and 100%, respectively), despite the samples coming from relatively close geographical regions, thus demonstrating a simple, yet specific, technique for determining chickpea origin by ICP-OES.

Elemental profiles of Greek wines from different harvests were obtained by Pasvanka et al. 191 using ICP-MS to measure the contents of 44 elements. The resulting data were processed using a variety of chemometric models. Hierarchical cluster analysis (HCA) allowed harvest years to be discriminated, whereas PLS-DA modelling showed potential links between variety and region of production. The authors noted that, due to a small dataset, further samples from additional harvests and regions would be needed to consolidate the methods applied in this paper.

Elemental profiles of olive oils from different cultivars, obtained using LIBS, coupled with additional information from UV absorption spectroscopy, were explored with the aim of origin classification.192 A total of 41 monovarietal oils were analysed and the results combined for chemometric analysis using LDA with a gradient boosting model. When using the LDA algorithm, the training and testing accuracy were 96.0% and 82.5%, respectively. With the gradient boosted model, accuracies of 100% (training) and 86.3% (testing) were obtained. This initial study showed the potential of LIBS for rapid testing of olive oils to determine their origin.

Investigation of Sr isotope ratios in food samples were carried out in two separate studies. In one of them,193 milks and dairy products underwent Sr isotope ratio analysis, followed by discriminant analysis to determine provenance. It was found that 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr ratios showed sufficient precision and accuracy to identify milks from geological areas with different soil and rock compositions, but full region discrimination was not possible using this model. Traditionally made Aceto Balsamico Tradizionale di Modena (ABTM) is a highly prized foodstuff, therefore mislabelling with the industrially product (Aceto Balsamico di Modena – ABM) is a prevalent food fraud. Durante et al.194 investigated the 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr ratios in ABTM and ABM, as well as in the ingredients used for industrial production, to assess whether a profile for monitoring could be established. They observed that the ABTM samples had very similar Sr ratios, due to the controlled conditions used for this production and this feature enabled to differentiate between ABM and ABTM samples. Their results demonstrated that Sr isotope ratio analysis of balsamic vinegar can be a useful tool against food fraud.

8.2.4 Honey and bee products. Propolis samples from Hungary were digested by MAD and analysed by ICP-MS and ICP-OES for 36 elements to build a model to for geographical origin determination. The team195 measured 36 elements in 252 samples of propolis from around Hungary for the models. The key elements selected for the modelling were Al, B, Ca, Ce, Cs, Dy, Er, Fe, Gd, Ho, K, Lu, Mg, Mn, Na, Nd, P, Pr, S, Sm, Sr, Tb, Tm and Yb. Linear discriminant analysis was used for the modelling and found to be 77% efficient in identification, which was preferable to the 62% effectiveness obtained with cross validated analysis.

Adulteration of honey with glucose syrup was investigated by applying chemometric analysis to data obtained by LIBS.112 Fifteen samples of honey from a variety of different plant origins were adulterated with between 10 and 90% glucose syrup. These samples, along with unadulterated and pure glucose samples, were split into training and classification sets. Data obtained by LIBS on the training set were used to build the model. The results of measurements of Ca, K and Na were sufficient to establish models, with a clear decrease in the concentrations of these elements with increasing adulteration. The subsequent analysis, by LDA and ERT, of the samples in the classification set showed that both chemometric tools gave an accurate prediction of adulteration. The authors stated the simplicity of using three spectral lines and the LDA model could provide a simple tool for investigating honey adulteration.

9. Abbreviations

AAatomic absorption
AASatomic absorption spectrometry
ABarsenobetaine
ACarsenocholine
AECanion-exchange chromatography
AESatomic emission spectrometry
AF4asymmetric flow-field flow fractionation
AFPalpha fetoprotein
AFSatomic fluorescence spectrometry
ANNartificial neural network
ANOVAanalysis of variance
APDCammonium pyrrolidine dithiocarbamate
APGDatmospheric pressure glow discharge
ASUAtomic Spectrometry Update
ASVanodic stripping voltammetry
ATRattenuated total reflectance
AUC–ROCarea under the curve–receiver operating characteristic
BHborohydride
BIPMBureau Internationale des Poids et Mesures
bwbody weight
CAPcanonical analysis of principal components
CARScompetitive adaptive reweighted sampling
CARTclassification and regression trees
CCQMConsultative Committee for Amount of Substance – Metrology in Chemistry and Biology
CDCCentres for Disease Control and Prevention
CHAIDchi-square automatic interaction detection
CIconfidence interval
CNNconvolutional neural network
CPEcloud point extraction
CRMcertified reference material
CScontinuum source
CSFcerebrospinal fluid
CTABcetyl trimethylammonium bromide
CVcold vapour
CVGchemical vapour generation
DBDdielectric barrier discharge
dcdirect current
DD-SIMCAdata driven soft independent modelling of class analogy
DDTCdiethyldithiocarbamate
DEAEdiethylamine
DESdeep eutectic solvent
DGAdiglycolamide
DLLMEdispersive liquid–liquid microextraction
DMAdimethylarsenic
DOCdissolved organic carbon
DPCdiphenylcarbazone
DRCdynamic reaction cell
DSPMEdispersive solid phase micro extraction
dwdry weight
DTZdithizone
EDenergy dispersive
EDSenergy dispersive spectroscopy
EDTAethylenediaminetetraacetic acid
EDXRFenergy dispersive X-ray fluorescence
EDXRFSenergy dispersive X-ray fluorescence spectroscopy
EFLMEuropean Federation of Clinical Chemistry and Laboratory Medicine
EIelectron ionisation
ELISAenzyme-linked immunosorbent assay
ENMengineered nanomaterials
EQAexternal quality assessment
ERMEuropean Reference Material
ERTextremely randomised trees
ESIelectrospray ionisation
ETelectrothermal
ETAASelectrothermal atomic absorption spectrometry
EtHgethylmercury
EtOHethanol
ETVelectrothermal vaporisation
EUEuropean Union
FAASflame atomic absorption spectrometry
FAESflame atomic emission spectrometry
FIflow-injection
fsfemtosecond
FTIRFourier-transform infrared spectroscopy
GCgas chromatography
GDglow discharge
GFAASgraphite furnace atomic absorption spectrometry
GLSgas liquid separator
HCAhierarchical cluster analysis
HERFDhigh-energy resolution fluorescence detection
HGhydride generation
HPLChigh performance liquid chromatography
HRhigh resolution
IAEAInternational Atomic Energy Agency
iAsinorganic arsenic
ICion chromatography
ICHInternational Council on Harmonization (of Technical Requirements for Pharmaceuticals for Human Use)
ICPinductively coupled plasma
ICP-QMSICP-quadrupole MS
IDisotope dilution
idinternal diameter
IDMSisotope dilution mass spectrometry
IFCCInternational Federation of Clinical Chemistry and Laboratory Medicine
iHginorganic mercury
ILionic liquid
ILCinterlaboratory comparison
IQRinterquartile range
IRinfrared
ISinternal standard
iSeinorganic Se
ISEion selective electrode
ISOInternational Organization for Standardisation
KEDkinetic energy discrimination
KHPpotassium hydrogen phthalate
kNNk-nearest neighbour
JCTLMJoint Committee for Traceability in Laboratory Medicine
LSleast squares
LAlaser ablation
LCliquid chromatography
LDAlinear discriminant analysis
LIBSlaser induced breakdown spectroscopy
LIFlaser-induced fluorescence
LLEliquid–liquid extraction
LLMEliquid–liquid microextraction
LODlimit of detection
LOQlimit of quantification
LRlogistic regression
LVQlearning vector quantization
MADmicrowave-assisted digestion
MAEmicrowave-assisted extraction
MCmulticollector
MeHgmethylmercury
MeOHmethanol
MeSeCysmethylselenocysteine
MICmicrowave-induced combustion
MILmagnetic ionic liquid
MIPmicrowave induced plasma
MMAmonomethylarsenic
MOFmetal–organic framework
MSmass spectrometry
MS/MStandem MS
NAAneutron activation analysis
NBNaive Bayes algorithm
NADESnatural deep eutectic solvent
NIRnear-infrared
NIRSnear-infrared Spectroscopy
NISTNational Institute of Standards and Technology
NMINational Metrology Institute
NNneural network
NPnanoparticle
NRCCNational Research Council of Canada
NRCCRMNational Research Centre for Certified Reference Materials (China)
OESoptical emission spectrometry
OPLS-DAorthogonal partial least squares discriminant analyses
ORodds ratio
PAGEpolyacrylamide gel electrophoresis
PCAprincipal component analysis
PCRpolymerase chain reaction
PDOprotected designation of origin
PGIprotected geographical indication
PLSpartial least squares
PLSRpartial least squares regression
PLS-DApartial least squares discriminant analysis
PTproficiency testing
PTWIprovisional tolerable daily intake
PTFEpoly(tetrafluoroethylene)
PTEpotentially toxic element
PVGphotochemical vapour generation
Qquadrupole
QQQtriple quadrupole
QUESTquick unbiased efficient statistical tree
REErare earth element
RFrandom forest
RMreference material
RMSEroot mean square error
RMSEProot mean square error of prediction
RPreversed phase
RSDrelative standard deviation
RT-PCRreal time polymerase chain reaction
S/Nsignal-to-noise ratio
S-LDAstepwise linear discriminant analysis
scsingle cell
SDstandard deviation
SDSsodium dodecylsulfate
SECsize exclusion chromatography
SeCys2selenocystine
SeMetselenomethionine
SFsector field
SHAPSHapley Additive exPlanations
SISystème International d’unités – International System of Units
SIMCAsoft independent modelling of class analogy
spsingle particle
SPEsolid phase extraction
SPMEsolid phase microextraction
SQTslotted quartz tube
SRsynchrotron radiation
SRMstandard reference material
SSIDspecies specific isotope dilution
SVMsupport vector machine
TBAHtetrabutyl ammonium hydroxide
TEMtransmission electron microscopy
THFtetrahydrofuran
TIMSthermal ionisation mass spectrometry
TMAHtetramethylammonium hydroxide
TMAOtrimethylarsine oxide
TOCtotal organic carbon
TXRFtotal reflection XRF
UAEultrasound-assisted extraction
US FDAUnited States Food and Drug Administration
USPUnited States Pharmacopeia
UVultraviolet
VIPvariable in importance
VGvapour generation
vs. versus
WDwavelength dispersive
WHOWorld Health Organisation
XGBoostExtreme Gradient Boosting
XRFX-ray fluorescence
XRFSX-ray fluorescence spectroscopy

Conflicts of interest

There is no conflict to declare.

References

  1. M. Patriarca, N. Barlow, A. Cross, S. Hill, A. Robson, A. Taylor and J. Tyson, J. Anal. At. Spectrom., 2022, 37(3), 410–473 RSC.
  2. 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., 2022, 37(1), 9–49 RSC.
  3. S. Carter, R. Clough, A. Fisher, B. Gibson and B. Russell, J. Anal. At. Spectrom., 2021, 36(11), 2241–2305 RSC.
  4. E. H. Evans, J. Pisonero, C. M. M. Smith and R. N. Taylor, J. Anal. At. Spectrom., 2021, 36(5), 868–891 RSC.
  5. C. Vanhoof, J. R. Bacon, U. E. A. Fittschen and L. Vincze, J. Anal. At. Spectrom., 2021, 36(9), 1797–1812 RSC.
  6. R. Clough, C. F. Harrington, S. J. Hill, Y. Madrid and J. F. Tyson, J. Anal. At. Spectrom., 2021, 36(7), 1326–1373 RSC.
  7. V. Jalili, R. Zendehdel and A. Barkhordari, Rev. Anal. Chem., 2021, 40(1), 93–107 CrossRef CAS.
  8. P. Kushwaha, Curr. Pharm. Anal., 2021, 17(8), 960–968 CrossRef CAS.
  9. C. Guo, L. Lv, Y. Liu, M. Ji, E. Zang, Q. Liu, M. Zhang and M. Li, Crit Rev Anal Chem., 2023, 53(2), 339–359 CrossRef CAS PubMed.
  10. H. S. Yang, D. R. LaFrance and Y. Hao, Am. J. Clin. Pathol., 2021, 156(2), 167–175 CrossRef CAS PubMed.
  11. A. Laycock, N. J. Clark, R. Clough, R. Smith and R. D. Handy, Environ. Sci.: Nano, 2022, 9(2), 420–453 RSC.
  12. A. Guilherme Buzanich, X-Ray Spectrom., 2022, 51(3), 294–303 CrossRef CAS.
  13. J. T. Lum, Y. N. Chan and K. S. Leung, Talanta, 2021, 234, 122683 CrossRef CAS PubMed.
  14. Q. Dai, Q. Y. Yang, X. Y. Bao, J. J. Chen, M. Han and Q. C. Wei, Mol. Pharm., 2022, 19(2), 363–377 CrossRef CAS PubMed.
  15. J. Lamarche, L. Ronga, J. Szpunar and R. Lobinski, Int. J. Mol. Sci., 2021, 22(12), 6283 CrossRef CAS PubMed.
  16. J. Zaldarriaga Heredia, M. Wagner, F. C. Jofre, M. Savio, S. M. Azcarate and J. M. Camina, Crit. Rev. Food Sci. Nutr., 2022, 1–21 CrossRef PubMed.
  17. F. Li, L. Meng, W. Ding, J. Wang and L. Ge, X-Ray Spectrom., 2022, 51(4), 346–364 CrossRef CAS.
  18. Z. Pinar Gumus and M. Soylak, TrAC, Trends Anal. Chem., 2021, 143, 116417 CrossRef CAS.
  19. N. Lemke, A. H. El-Khatib, T. Tchipilov, N. Jakubowski, M. G. Weller and J. Vogl, Anal. Bioanal. Chem., 2022, 414(15), 4441–4455 CrossRef CAS PubMed.
  20. S. Nubler, M. Schafer, K. Haji-Abbas-Zarrabi, S. Markovic, K. Markovic, M. Esteban Lopez, A. Castano, H. Mol, H. M. Koch, J. P. Antignac, J. Hajslova, C. Thomsen, K. Vorkamp and T. Göen, J. Trace Elem. Med. Biol., 2022, 70, 126912 CrossRef PubMed.
  21. J. Diemer, J. Vogl, C. R. Quétel, T. Lisinger, P. D. P. Taylor, A. Lamberty and J. Pauwels, Fresenius’ J. Anal. Chem., 2001, 370(5), 492–498 CrossRef CAS PubMed.
  22. W. Z. Yu, Q. Liu, H. W. Leung, B. M. K. Tong, G. Chew, T. K. Lee, R. Y. C. Shin, T. L. Teo and S. K. Sethi, J. Trace Elem. Med. Biol., 2021, 68, 126825 CrossRef CAS PubMed.
  23. M. C. B. Rayana, R. W. Burnett, A. K. Covington, P. D’Orazio, N. Fogh-Andersen, E. Jacobs, R. Kataky, W. R. Külpmann, K. Kuwa, L. Larsson, A. Lewenstam, A. H. J. Maas, G. Mager, J. W. Naskalski, A. O. Okorodudu, C. Ritter and A. St John, Clin. Chem. Lab. Med., 2006, 44, 346–352 Search PubMed.
  24. C. Ricos, V. Alvarez, F. Cava, J. V. Garcia-Lario, A. Hernandez, C. V. Jimenez, J. Minchinela, C. Perich and M. Simon, Scand. J. Clin. Lab. Invest., 1999, 59, 491–500 CrossRef CAS PubMed.
  25. A. K. Aarsand, P. Fernandez-Calle, C. Webster, A. Coskun, E. Gonzales-Lao, J. Diaz-Garzon, N. Jonker, M. Simon, F. Braga, C. Perich, B. Boned, F. Marques-Garcia, A. Carobene, B. Aslan, E. Sezer, W. A. Bartlett and S. Sandberg, The EFLM Biological Variation Database. https://biologicalvariation.eu/, 31.12.2022 Search PubMed.
  26. JCGM 200:2012 International Vocabulary of Metrology – Basic and General Concepts and Associated Terms (VIM), 3rd edn, 2008 version with minor corrections, https://www.bipm.org/en/committees/jc/jcgm/publications Search PubMed.
  27. N. Baharom and Y. H. Yim, Bull. Korean Chem. Soc., 2021, 42(12), 1593–1601 CrossRef CAS.
  28. H. Polkowska-Motrenko, Z. Samczyński, R. S. Dybczyński, E. Chajduk, B. Danko, P. Kalbarczyk, A. A. Krata, M. Pyszynska and I. Zuba, Food Anal. Methods, 2022, 15(2), 377–390 CrossRef.
  29. D. Yeghicheyan, P. Grinberg, L. Y. Alleman, M. Belhadj, L. Causse, J. M. Chmeleff, L. Cordier, I. Djouraev, D. Dumoulin, J. Dumont, R. Freydier, H. Mariot, C. Cloquet, P. Kumkrong, B. Malet, C. Jeandel, A. Marquet, J. Riotte, M. Tharaud, G. Billon, G. Trommetter, F. Seby, A. Guihou, P. Deschamps and Z. Mester, Anal. Bioanal. Chem., 2021, 413(20), 4959–4978 CrossRef CAS PubMed.
  30. F. Thibon, L. Weppe, M. Montanes, P. Telouk and N. Vigier, J. Anal. At. Spectrom., 2021, 36(7), 1381–1388 RSC.
  31. H. Jeong, K. Ra and J. Y. Choi, Geostand. Geoanal. Res., 2021, 45(3), 551–563 CrossRef CAS.
  32. P. Almerud, G. Zamaratskaia, A. K. Lindroos, H. Bjermo, E. M. Andersson, T. Lundh, E. H. Ankarberg and S. Lignell, Environ. Res., 2021, 197, 110991 CrossRef CAS PubMed.
  33. F. Luan, B. Liu, S. Sun, Y. Chen, Y. Xu, X. Jiang, X. Guo, S. Cheng and Y. Wang, Biol. Trace Elem. Res., 2022, 200(7), 3078–3087 CrossRef CAS PubMed.
  34. Y. Lv, Y. Wei, J. Zhou, K. Xue, Y. Guo, Y. Liu, A. Ju, B. Wu, F. Zhao, C. Chen, J. Xiong, C. Li, H. Gu, Z. Cao, J. S. Ji and X. Shi, Environ. Int., 2021, 156, 106717 CrossRef CAS PubMed.
  35. I. Jayawardene, J. F. Paradis, S. Belisle, D. Poddalgoda and K. Macey, J. Trace Elem. Med. Biol., 2021, 68, 126830 CrossRef CAS PubMed.
  36. A. Stojsavljević, M. Rovčanin, J. Jagodić, Ž. Miković, A. Jeremić, M. Perović and D. Manojlović, Exposure Health, 2021, 14(3), 623–633 CrossRef.
  37. M. Aranaz, E. Valencia-Agudo, L. Lobo and R. Pereiro, J. Anal. At. Spectrom., 2022, 37(1), 50–68 RSC.
  38. Y. Bao, X. Ge, L. Li, J. He, S. Huang, X. Luo, X. Chen, P. Chen and X. Yang, BioMetals, 2021, 34(5), 1191–1205 CrossRef CAS PubMed.
  39. E. M. Tanvir, T. Komarova, E. Comino, R. Sumner, K. M. Whitfield and P. N. Shaw, J. Trace Elem. Med. Biol., 2021, 68, 126804 CrossRef CAS PubMed.
  40. E. Humann-Ziehank, C. Wolf, S. Schuler and K. Gerst, J. Trace Elem. Med. Biol., 2021, 68, 126871 CrossRef CAS PubMed.
  41. R. Buchholz, S. Krossa, M. K. Andersen, M. Holtkamp, M. Sperling, U. Karst and M. B. Tessem, Metallomics, 2022, 14(3), mfac013 CrossRef PubMed.
  42. S. Hauser, S. Andres and K. Leopold, Anal. Bioanal. Chem., 2022, 414(15), 4519–4529 CrossRef CAS PubMed.
  43. M. C. García-Poyo, C. Pécheyran, L. Rello, E. García-González, S. Alonso Rodríguez, F. V. Nakadi, M. Aramendía and M. Resano, J. Anal. At. Spectrom., 2021, 36(8), 1666–1677 RSC.
  44. X. Zhang, S. Liu, X. Wei, Y. L. Yu and J. H. Wang, Anal. Chem., 2021, 93(30), 10577–10583 CrossRef CAS PubMed.
  45. L. B. Guimarães, J. C. Romero, S. M. S. Queiroz, E. G. P. da Silva, D. C. Lima, J. P. Coutinho, F. S. Dias, M. A. Bezerra and F. A. C. Amorim, Food Anal. Methods, 2022, 15(2), 458–467 CrossRef.
  46. U. M. F. da Mata Cerqueira, G. S. Valasques, C. T. de Souza, S. A. Araújo, M. A. Bezerra and C. G. Novaes, Food Anal. Methods, 2022, 15(4), 1098–1106 CrossRef.
  47. S. Mandal and S. Lahiri, Microchem. J., 2022, 175, 107150 CrossRef CAS.
  48. F. C. Pinheiro and J. A. Nóbrega, Microchem. J., 2022, 175, 107189 CrossRef CAS.
  49. F. S. Costa, L. S. Moreira, A. M. Silva, R. J. Silva, M. P. dos Santos, E. G. P. da Silva, M. T. Grassi, M. H. Gonzalez and C. D. B. Amaral, J. Food Compos. Anal., 2022, 109, 104510 CrossRef CAS.
  50. M. I. Curti, F. Cora Jofre, S. M. Azcarate, J. M. Camina, P. D. Ribotta and M. Savio, J. Anal. Methods Chem., 2021, 2021, 9201094 Search PubMed.
  51. L. S. Moreira, F. S. Costa, R. d. C. Lidorio, L. W. S. Toledo, A. Oliveira, M. H. Gonzalez, E. G. P. da Silva and C. D. B. Amaral, Food Anal. Methods, 2022, 15(6), 1687–1694 CrossRef.
  52. M. F. Ayyildiz, M. Saylan, G. D. Bozyigit, D. S. Chormey, O. K. Ozdemir, C. Buyukpinar and S. Bakirdere, Anal. Biochem., 2021, 629, 114309 CrossRef CAS PubMed.
  53. Q. Mou, L. Dong, L. Xu, Z. L. Song, Y. Yu, E. H. Wang, Y. H. Zhao and Y. Gao, J. Anal. At. Spectrom., 2021, 36(7), 1422–1430 RSC.
  54. J. J. Lopez-Mayan, B. Alvarez-Fernandez, E. Pena-Vazquez, M. C. Barciela-Alonso, A. Moreda-Pineiro and P. Bermejo-Barrera, Talanta, 2022, 247, 123556 CrossRef CAS PubMed.
  55. G. Subramaniam, J. Bakar, N. Surugau and K. Muhammad, Food Anal. Methods, 2022, 15(10), 2858–2878 CrossRef.
  56. M. S. Nascimento, G. T. Druzian, L. S. F. Pereira, M. F. Mesko, R. S. Picoloto, P. A. Mello and E. M. M. Flores, J. Anal. At. Spectrom., 2022, 37(3), 535–543 RSC.
  57. C. A. Bizzi, M. F. Pedrotti, D. M. Betiolo, M. S. Nascimento, E. I. Muller, G. Cravotto and E. M. M. Flores, Anal. Methods, 2021, 13(46), 5555–5563 RSC.
  58. Q. Sun, X. Cui, Y. Wang, P. Zhang and W. Lu, Molecules, 2022, 27(3), 590 CrossRef CAS PubMed.
  59. N. Altunay and M. Tuzen, Food Chem., 2021, 364, 130371 CrossRef CAS PubMed.
  60. N. Akiba, R. S. N. Mancini, A. Z. Carvalho and I. Gaubeur, Anal. Lett., 2022, 55(5), 741–754 CrossRef CAS.
  61. N. Alvand, M. Baghdadi, M. Alimoradi, A. Marjani and T. M. Isfahani, Food Anal. Methods, 2022, 15(7), 2048–2058 CrossRef.
  62. L. Goodarzi, M. R. Bayatloo, S. Chalavi, S. Nojavan, T. Rahmani and S. B. Azimi, Food Chem., 2022, 373(Pt A), 131442 CrossRef CAS PubMed.
  63. F. L. F. da Silva, D. M. A. Neto, F. L. de Menezes, I. P. Sa, J. M. de Higuera, P. B. A. Fechine, L. S. da Costa, A. R. A. Nogueira, G. S. Lopes and W. O. Matos, Food Chem., 2021, 345, 128781 CrossRef CAS PubMed.
  64. S. Z. Chen, Y. X. Liu, C. L. Wang, J. T. Yan and D. B. Lu, Anal. Lett., 2022, 55(2), 269–280 CrossRef CAS.
  65. F. Fallah Tafti and M. Masrournia, J. Iran. Chem. Soc., 2022, 19(6), 2383–2394 CrossRef.
  66. L. Trzonkowska, B. Leśniewska and B. Godlewska-Żyłkiewicz, Water, 2022, 14(4), 529 CrossRef CAS.
  67. S. Chen, J. Yan, Y. Liu, C. Wang and D. Lu, Food Chem., 2021, 359, 129958 CrossRef CAS PubMed.
  68. S. Chen, J. Liu, J. Yan, C. Wang and D. Lu, Food Addit. Contam., Part A, 2022, 39(3), 499–507 CrossRef CAS PubMed.
  69. S. Chen, Y. Liu, C. Wang, J. Yan and D. Lu, Food Addit. Contam., Part A, 2021, 38(9), 1539–1550 CrossRef CAS PubMed.
  70. W. I. Mortada, M. M. El-Defrawy, E. Erfan and H. A. El-Asmy, J. Food Compos. Anal., 2022, 108, 104445 CrossRef CAS.
  71. C. Sanchez, R. Sanchez, C. P. Lienemann and J. L. Todoli, J. Anal. At. Spectrom., 2021, 36(10), 2085–2096 RSC.
  72. M. S. Wheal and E. N. Wilkes, J. Anal. At. Spectrom., 2021, 36(11), 2383–2390 RSC.
  73. T. Liu, E. Bolea-Fernandez, C. Mangodt, O. De Wever and F. Vanhaecke, Anal. Chim. Acta, 2021, 1177, 338797 CrossRef CAS PubMed.
  74. C. X. Wu, X. Wei, X. Men, X. Zhang, Y. L. Yu, Z. R. Xu, M. L. Chen and J. H. Wang, Anal. Chem., 2021, 93(23), 8203–8209 CrossRef CAS PubMed.
  75. M. Cosmi, N. Gonzalez-Quinonez, P. T. Diaz, A. Manteca, E. Blanco-Gonzalez, J. Bettmer, M. Montes-Bayon and M. Corte-Rodriguez, J. Anal. At. Spectrom., 2021, 36(9), 2007–2016 RSC.
  76. M. Schaier, G. Hermann, G. Koellensperger and S. Theiner, Anal. Bioanal. Chem., 2022, 414(1), 639–648 CrossRef CAS PubMed.
  77. J. Yang, L. Chi and S. Li, J. Anal. Methods Chem., 2021, 2021, 5533631 Search PubMed.
  78. W. Wang, R. D. Evans and H. E. Evans, Talanta, 2021, 233, 122507 CrossRef CAS PubMed.
  79. Y. Wu, Y. Xu, S. Xing, X. Dai, N. Yuan and M. Luo, Spectrochim. Acta, Part B, 2021, 184, 106259 CrossRef CAS.
  80. D. L. Novo, T. Van Acker, J. Belza, F. Vanhaecke and M. F. Mesko, J. Anal. At. Spectrom., 2022, 37(4), 775–782 RSC.
  81. C. Martinez Lopez, J. A. Carter, R. Corzo and T. I. Todorov, Food Anal. Methods, 2022, 15(11), 3079–3091 CrossRef.
  82. P. Petrov, S. Cowen and H. Goenaga-Infante, Anal. Methods, 2021, 13(33), 3641–3648 RSC.
  83. Q. Ma, Q. Zhang, X. Li, Y. Gao, C. Wei, H. Li and H. Jiao, J. Chromatogr. A, 2022, 1674, 463134 CrossRef CAS PubMed.
  84. L. Suárez-Criado, S. Queipo-Abad, A. Rodríguez-Cea, P. Rodríguez-González and J. I. García Alonso, J. Anal. At. Spectrom., 2022, 37(7), 1462–1470 RSC.
  85. C. D. Quarles, P. Sullivan, N. Bohlim and N. Saetveit, J. Anal. At. Spectrom., 2022, 37(6), 1240–1246 RSC.
  86. Y. C. Chen and S. J. Jiang, J. Anal. At. Spectrom., 2021, 36(5), 938–945 RSC.
  87. F. Vanhaecke and M. Costas-Rodríguez, View, 2021, 2(1), 20200094 CrossRef CAS.
  88. P. Télouk, E. Albalat, T. Tacail, F. Arnaud-Godet and V. Balter, J. Anal. At. Spectrom., 2022, 37(6), 1259–1264 RSC.
  89. A. Retzmann, D. Walls, K. A. Miller, J. Irrgeher, T. Prohaska and M. E. Wieser, Anal. Bioanal. Chem., 2022, 414(1), 675–689 CrossRef CAS PubMed.
  90. F. V. Nakadi, M. C. García-Poyo, C. Pécheyran and M. Resano, J. Anal. At. Spectrom., 2021, 36(11), 2370–2382 RSC.
  91. P. Chaikhan, Y. Udnan, R. J. Ampiah-Bonney and W. C. Chaiyasith, Food Chem., 2022, 375, 131857 CrossRef CAS PubMed.
  92. O. Zverina, O. Venclicek, J. Kuta, P. Coufalik, I. Hagarova and K. Brat, J. Trace Elem. Med. Biol., 2021, 68, 126869 CrossRef CAS PubMed.
  93. N. K. V. dos Santos, L. dos Santos, I. C. F. Damin, M. G. R. Vale and M. B. Dessuy, J. Food Compos. Anal., 2022, 111, 104625 CrossRef CAS.
  94. D. E. Bustos, J. A. Toro, M. Briceno and R. E. Rivas, Talanta, 2022, 247, 123547 CrossRef CAS PubMed.
  95. T. N. Akbaba and O. Aacar, J. Anal. Chem., 2022, 77(5), 555–562 CrossRef.
  96. A. Nikulin, O. Potanina, M. Alyussef, V. Vasil’ev, R. Abramovich, O. Novikov, N. Boyko, A. Khromov and E. Platonov, Farmacia, 2021, 69(3), 566–575 CrossRef CAS.
  97. O. Cadar, T. Mocan, C. Roman and M. Senila, Nanomaterials, 2021, 11(12), 3370 CrossRef CAS PubMed.
  98. E. J. Pacer, C. D. Palmer and P. J. Parsons, Spectrochim. Acta, Part B, 2022, 190, 106324 CrossRef CAS.
  99. M. Gorska and P. Pohl, J. Anal. At. Spectrom., 2021, 36(7), 1455–1465 RSC.
  100. H. X. Chen and Z. Wang, J. Anal. At. Spectrom., 2021, 36(10), 2185–2191 RSC.
  101. Z. Q. Cai, L. Qian, X. X. Peng and Z. Wang, Anal. Chem., 2021, 93(44), 14701–14707 CrossRef CAS PubMed.
  102. J. Yang, Y. Luo, Y. Su, Y. Li, Y. Lin and C. Zheng, Microchem. J., 2022, 179, 107569 CrossRef CAS.
  103. S. S. Savinov and A. I. Drobyshev, J. Anal. Chem., 2022, 77(3), 328–333 CrossRef CAS.
  104. V. I. Otmakhov, E. S. Rabchevich, E. V. Petrova, I. V. Shilova, T. M. Gindullina, E. S. Sheleg and D. E. Babenkov, Pharm. Chem. J., 2022, 56(2), 277–282 CrossRef CAS.
  105. K. Zhang, Z. W. Xu, C. S. Chen and F. Z. Fang, Spectrosc. Spectral Anal., 2021, 41(6), 1961–1965 CAS.
  106. M. N. Khan, Q. Wang, B. S. Idrees, W. Xiangli, G. Teng, X. Cui, Z. Zhao, K. Wei and M. Abrar, Front. Phys., 2022, 10, 821057 CrossRef.
  107. K. Berlo, W. Xia, F. Zwillich, E. Gibbons, R. Gaudiuso, E. Ewusi-Annan, G. R. Chiklis and N. Melikechi, Sci. Rep., 2022, 12(1), 1614 CrossRef CAS PubMed.
  108. S. Y. Zhang, Z. L. Hu, Z. F. Zhao, F. Chen, Y. Tang, Z. Q. Sheng, D. Zhang, Z. J. Zhang, H. L. Jin, H. S. Pu and L. B. Guo, Optik, 2021, 242, 167067 CrossRef CAS.
  109. I. Rehan, M. A. Gondal, R. K. Aldakheel, M. A. Almessiere, K. Rehan, S. Khan, S. Sultana and M. Z. Khan, Arabian J. Sci. Eng., 2022, 47(6), 7531–7539 CrossRef CAS.
  110. N. Ali, Z. X. Huang, J. Zong, Y. W. Ma, Y. B. Xiao, L. Wang, P. F. Zhang and D. Chen, J. Food Saf. Food Qual., 2021, 72(4), 131–136 Search PubMed.
  111. D. Stefas, N. Gyftokostas, P. Kourelias, E. Nanou, C. Tananaki, D. Kanelis, V. Liolios, V. Kokkinos, C. Bouras and S. Couris, Food Control, 2022, 134, 108770 CrossRef CAS.
  112. D. Stefas, N. Gyftokostas, P. Kourelias, E. Nanou, V. Kokkinos, C. Bouras and S. Couris, Appl. Sci., 2021, 11(14), 6435 CrossRef CAS.
  113. S. Y. Zhao, W. R. Song, Z. Y. Hou and Z. Wang, J. Anal. At. Spectrom., 2021, 36(8), 1704–1711 RSC.
  114. J. Park, S. Kumar, S. H. Han, V. K. Singh, S. H. Nam and Y. Lee, Appl. Spectrosc., 2022, 76(9), 1042–1050 CrossRef CAS PubMed.
  115. B. Sezer, M. Ozturk, H. Ayvaz, H. Apaydin and I. H. Boyaci, Food Chem., 2022, 390, 132946 CrossRef CAS PubMed.
  116. X. M. Pan, Y. Lin, Y. B. Su, J. H. Yang, L. B. He, Y. R. Deng, X. D. Hou and C. B. Zheng, Anal. Chem., 2021, 93(23), 8257–8264 CrossRef CAS PubMed.
  117. Z. Q. Cai, H. J. Zou, Y. R. Chen and Z. Wang, Chin. Chem. Lett., 2022, 33(5), 2692–2696 CrossRef CAS.
  118. Y. Zhang, X. Mao, D. Tian, J. Liu and C. Li, Anal. Methods, 2021, 13(36), 4079–4089 RSC.
  119. S. Y. Gu, X. D. Huang, M. L. Chen, J. X. Liu, X. F. Mao, X. Na, G. Y. Chen and Y. B. Shao, Anal. Chem., 2021, 93(45), 15063–15071 CrossRef CAS PubMed.
  120. M. Wang, J. He, J. Luo, J. Hu and X. Hou, Food Chem., 2022, 375, 131841 CrossRef CAS PubMed.
  121. J. Hu, H. J. Chen, X. M. Jiang and X. D. Hou, Anal. Chem., 2021, 93(32), 11151–11158 CrossRef CAS PubMed.
  122. J. R. Miranda-Andrades, M. J. Pedrozo-Penãfiel, J. A. Mendes and R. Q. Aucelio, Spectrochim. Acta, Part B, 2022, 194, 106477 CrossRef CAS.
  123. M. J. de Oliveira, F. A. S. Cunha and J. C. C. Santos, J. Anal. At. Spectrom., 2021, 36(4), 740–746 RSC.
  124. J. Luo, Z. Hu, F. Xu, D. Geng and X. Tang, Microchem. J., 2022, 174, 107053 CrossRef CAS.
  125. D. Yildiz, M. Atasoy and A. Demirak, Anal. Lett., 2022, 55(17), 2776–2788 CrossRef CAS.
  126. E. A. Lima, F. A. S. Cunha, M. J. Oliveira, W. S. Lyra, M. M. S. Junior, J. C. C. Santos, S. L. C. Ferreira, M. C. U. Araujo and L. F. Almeida, Food Chem., 2022, 381, 132194 CrossRef CAS PubMed.
  127. I. Petry-Podgórska, V. Schrenková, M. Migašová, T. Matoušek and J. Kratzer, Microchem. J., 2021, 170, 106606 CrossRef.
  128. Y. H. Wei, Q. Q. Huang, J. Y. Zhang, S. Y. Qiu, T. H. Tu, L. F. Yuan, T. C. Dai, B. J. Zhang, W. H. Li and H. Yan, Spectrosc. Spectral Anal., 2021, 41(12), 3822–3827 CAS.
  129. M. L. Astolfi, M. E. Conti, M. Ristorini, M. A. Frezzini, M. Papi, L. Massimi and S. Canepari, Molecules, 2021, 26(16), 4878 CrossRef CAS PubMed.
  130. K. Phan, N. Richardson and N. M. Hepp, J. AOAC Int., 2022, 105(1), 69–73 CrossRef PubMed.
  131. M. Musielak, M. Serda, E. Talik, A. Gagor, J. Korzuch and R. Sitko, J. Anal. At. Spectrom., 2021, 36(7), 1533–1543 RSC.
  132. M. Musielak, M. Serda and R. Sitko, Food Chem., 2022, 390, 133136 CrossRef CAS PubMed.
  133. J. Milinovic, F. M. J. Figueiredo, J. P. Noronha and J. Sardinha, J. Food Compos. Anal., 2022, 110, 104559 CrossRef CAS.
  134. X. Ma, M. Z. Hua, C. Ji, J. Zhang, R. Shi, Y. Xiao, X. Liu, X. He, W. Zheng and X. Lu, Analyst, 2022, 147(16), 3628–3633 RSC.
  135. M. Bhatia, A. J. Specht, V. Ramya, D. Sulaiman, M. Konda, P. Balcom, E. M. Sunderland and A. Qureshi, Environ. Sci. Technol., 2021, 55(19), 13113–13121 CAS.
  136. A. Grasso, M. Ferrante, A. Moreda-Pineiro, G. Arena, R. Magarini, G. Oliveri Conti, A. Cristaldi and C. Copat, Ecotoxicol. Environ. Saf., 2022, 231, 113217 CrossRef CAS PubMed.
  137. Y. Suzuki, M. Kondo, H. Akiyama and Y. Ogra, Environ. Pollut., 2022, 307, 119555 CrossRef CAS PubMed.
  138. B. Li, S. L. Chua, D. Yu, S. H. Chan and A. Li, Molecules, 2022, 27(8), 2539 CrossRef CAS PubMed.
  139. B. R. Li, H. Tang, R. Q. Yu and J. H. Jiang, Anal. Chem., 2021, 93(24), 8381–8385 CrossRef CAS PubMed.
  140. X. Sun, X. Wei, X. Liu, X. Zhang, N. Wu, J. Liu, Y. Wang, M. Chen and J. Wang, Anal. Chim. Acta, 2022, 1201, 339639 CrossRef CAS PubMed.
  141. S. Ruella Oliveira, K. Tuttis, A. Rita Thomazela Machado, C. Cristina de Souza Rocha, L. Maria Greggi Antunes and F. Barbosa, Microchem. J., 2022, 177, 107275 CrossRef CAS.
  142. P. J. Wegwerth, S. A. Erdahl, M. L. Wermers, M. M. Hanley, S. J. Eckdahl and P. J. Jannetto, J. Appl. Lab. Med., 2021, 6(4), 846–857 CrossRef PubMed.
  143. A. Nawrocka, M. Durkalec, M. Michalski and A. Posyniak, Food Chem., 2022, 379, 132045 CrossRef CAS PubMed.
  144. M. Saraiva, R. Chekri, A. Leufroy, T. Guerin, J. J. Sloth and P. Jitaru, Talanta, 2021, 222, 121538 CrossRef CAS PubMed.
  145. M. Saraiva, R. Chekri, T. Guerin, J. J. Sloth and P. Jitaru, Food Addit. Contam., Part A, 2021, 38(2), 304–314 CrossRef CAS PubMed.
  146. M. Saraiva, T. Guerin, P. Jitaru and J. J. Sloth, Food Addit. Contam., Part A, 2021, 38(10), 1735–1742 CrossRef CAS PubMed.
  147. M. Saraiva, P. Jitaru and J. J. Sloth, J. Food Compos. Anal., 2021, 102, 103991 CrossRef CAS.
  148. S. Marković, M. Gabrič, M. Islamčević Razboršek, R. Milačič and J. Ščančar, J. Food Compos. Anal., 2022, 108, 104422 CrossRef.
  149. Y. Zhao, M. Wang, M. Yang, J. Zhou and T. Wang, Separations, 2022, 9(2), 21 CrossRef CAS.
  150. B. Paul, K. Kysenius, J. B. Hilton, M. W. M. Jones, R. W. Hutchinson, D. D. Buchanan, C. Rosty, F. Fryer, A. I. Bush, J. M. Hergt, J. D. Woodhead, D. P. Bishop, P. A. Doble, M. M. Hill, P. J. Crouch and D. J. Hare, Chem. Sci., 2021, 12(30), 10321–10333 RSC.
  151. M. G. Mello, M. T. Westerhausen, P. Singh, P. A. Doble, J. Wanagat and D. P. Bishop, Anal. Bioanal. Chem., 2021, 413(21), 5509–5516 CrossRef CAS PubMed.
  152. L. He, Y. Lu, C. Li, H. Xie, J. Zhao, Y. Wang, L. Wang, X. Wang, W. Wang, D. Chen, Y. Gao, B. Li and Y. F. Li, Talanta, 2022, 245, 123486 CrossRef CAS PubMed.
  153. V. Tardillo Suarez, B. Gallet, M. Chevallet, P. H. Jouneau, R. Tucoulou, G. Veronesi and A. Deniaud, J. Struct. Biol., 2021, 213(3), 107766 CrossRef CAS PubMed.
  154. M. Guo, J. Lv, X. Chen, M. Wu, Q. Zhao and X. Hai, Front. Oncol., 2022, 12, 887026 CrossRef PubMed.
  155. M. Liu, D. Wang, C. Wang, S. Yin, X. Pi, Z. Li, L. Wang, J. Liu, C. Yin, L. Jin and A. Ren, Chemosphere, 2021, 284, 131387 CrossRef CAS PubMed.
  156. X. Pi, C. Wang, S. Yin, L. Jin, Z. Li, L. Wang, J. Liu, Y. Zhang and A. Ren, Exposure Health, 2022 DOI:10.1007/s12403-022-00487-z.
  157. D. Li, C. Liang, Y. Cao, D. Zhu, L. Shen, Z. Zhang, T. Jiang, Z. Zhang, K. Zong, Y. Liu, D. Liang, Y. Cao, D. Ji and X. Xu, Ecotoxicol. Environ. Saf., 2022, 233, 113309 CrossRef CAS PubMed.
  158. T. Jiang, Y. Hu, S. He, R. Jiang, Y. Yao, Z. Jin, J. Shen, F. Tao, Y. Ji and C. Liang, Environ. Res., 2022, 211, 113072 CrossRef CAS PubMed.
  159. L. Zhou, K. Liang, M. Li, C. Rong, J. Zheng and J. Li, J. Trace Elem. Med. Biol., 2021, 68, 126810 CrossRef CAS PubMed.
  160. P. Menero-Valdes, A. Lores-Padin, B. Fernandez, H. Gonzalez-Iglesias and R. Pereiro, Talanta, 2022, 244, 123424 CrossRef CAS PubMed.
  161. A. F. Asensio, M. Corte-Rodriguez, J. Bettmer, L. M. Sierra, M. Montes-Bayon and E. Blanco-Gonzalez, Talanta, 2021, 235, 122773 CrossRef CAS PubMed.
  162. Z. N. Yu, S. Y. Liu, Y. F. Zhang, Y. D. Li, Y. Z. Xu, D. P. Chen, Z. Dai and X. Y. Zou, Anal. Chem., 2021, 93(22), 7787–7791 CrossRef CAS PubMed.
  163. J. Schoon, B. Hesse, R. Tucoulou, S. Geissler, M. Ort, G. N. Duda, C. Perka, G. I. Wassilew, G. Perino and A. Rakow, J. Mater. Sci.: Mater. Med., 2022, 33(6), 54 CrossRef CAS PubMed.
  164. X. H. Song, H. L. Li, C. Ma, W. Y. Duan, Y. B. Sun, Y. Q. Li, T. H. Huang and B. Zhou, At. Spectrosc., 2021, 42(5), 278–281 CAS.
  165. S. Topal, M. Saylan, B. T. Zaman and S. Bakirdere, J. Trace Elem. Med. Biol., 2021, 68, 126859 CrossRef CAS PubMed.
  166. T. Onuma, T. Mizutani, Y. Fujita, S. Yamada and Y. Yoshida, J. Trace Elem. Med. Biol., 2021, 68, 126865 CrossRef CAS PubMed.
  167. K. Marković, M. Cemazar, G. Sersa, R. Milačič and J. Ščančar, J. Anal. At. Spectrom., 2022, 37(8), 1675–1686 RSC.
  168. M. C. García-Poyo, A. L. Ronzani, J. Frayret, S. Bérail, L. Rello, E. García-González, B. Lelièvre, F. V. Nakadi, M. Aramendía, M. Resano and C. Pécheyran, Spectrochim. Acta, Part B, 2021, 185, 106306 CrossRef.
  169. M. C. García-Poyo, S. Bérail, A. L. Ronzani, L. Rello, E. García-González, B. Leliévre, P. Cales, F. V. Nakadi, M. Aramendía, M. Resano and C. Pécheyran, J. Anal. At. Spectrom., 2021, 36(5), 968–980 RSC.
  170. Y. Zhang, Z. Su, G. Hu, S. Hong, C. Long, Q. Zhang, P. Zheng, T. Wang, S. Yu, F. Yuan, X. Zhu and G. Jia, Sci. Total Environ., 2022, 818, 151741 CrossRef CAS PubMed.
  171. B. L. Dekker, D. J. Touw, A. N. A. van der Horst-Schrivers, M. J. Vos, T. P. Links, D. A. J. Dijck-Brouwer and A. C. M. Kobold, J. Nutr., 2021, 151(12), 3671–3677 CrossRef PubMed.
  172. G. M. Parkin, M. J. McCarthy, S. H. Thein, H. L. Piccerillo, N. Warikoo, D. A. Granger and E. A. Thomas, Bipolar Disord., 2021, 23(7), 679–688 CrossRef CAS PubMed.
  173. A. K. James, N. V. Dolgova, S. Nehzati, M. Korbas, J. J. H. Cotelesage, D. Sokaras, T. Kroll, J. L. O’Donoghue, G. E. Watson, G. J. Myers, I. J. Pickering and G. N. George, ACS Chem. Neurosci., 2022, 13(12), 1756–1768 CrossRef CAS PubMed.
  174. K. A. Luders, L. Braunschweig, A. Ziola-Frankowska, A. Stojek, D. Jakkielska, A. Wichmann, G. H. Dihazi, F. Streit, S. E. Gusewell, T. C. True, S. Luders, J. Schlie, K. Tsaknakis, H. M. Lorenz, M. Frankowski and A. K. Hell, Sci. Rep., 2022, 12(1), 10811 CrossRef CAS PubMed.
  175. K. V. Sullivan, R. E. T. Moore, M. S. Capper, K. Schilling, K. Goddard, C. Ion, D. Layton-Matthews, M. I. Leybourne, B. Coles, K. Kreissig, O. Antsygina, R. C. Coombes, F. Larner and M. Rehkamper, Metallomics, 2021, 13(6), mfab027 CrossRef PubMed.
  176. L. Ruzik and M. Jakubowska, Foods, 2022, 11(10), 1441 CrossRef CAS PubMed.
  177. H. Ahmad, B. H. Koo and R. A. Khan, Microchem. J., 2022, 175, 107179 CrossRef CAS.
  178. V. V. Rosen, O. G. Garber and Y. Chen, J. Food Compos. Anal., 2022, 111, 104600 CrossRef CAS.
  179. N. W. Sadiq and D. Beauchemin, Can. J. Chem., 2021, 99(9), 742–750 CrossRef CAS.
  180. F. Rovasi Adolfo, P. Cicero do Nascimento, L. Brudi, D. Bohrer and L. Machado de Carvalho, Food Chem., 2021, 345, 128766 CrossRef CAS PubMed.
  181. C. Ji, J. Liu, Q. Zhang, J. Li, Z. Wu, X. Wang, Y. Xie, J. Zhao, R. Shi, X. Ma, M. R. Khan, R. Busquets, X. He, Y. Zhu, S. Zhu and W. Zheng, Molecules, 2022, 27(9), 2982 CrossRef CAS PubMed.
  182. J. M. Muñoz-Redondo, D. Bertoldi, A. Tonon, L. Ziller, F. Camin and J. M. Moreno-Rojas, Food Control, 2022, 137, 108975 CrossRef.
  183. B. Quinn, P. McCarron, Y. Hong, N. Birse, D. Wu, C. T. Elliott and R. Ch, Food Chem., 2022, 386, 132738 CrossRef CAS PubMed.
  184. R. Amit, S. Jamwal, S. Kumari, S. Kelly, A. Cannavan and D. K. Singh, Curr. Res. Food Sci., 2022, 5, 545–552 CrossRef CAS PubMed.
  185. B. M. de Andrade, L. P. Margalho, D. B. Batista, I. O. Lucena, B. A. Kamimura, C. F. Balthazar, R. P. Brexó, A. K. R. Pia, R. A. S. Costa, A. G. Cruz, D. Granato, A. S. Sant’Ana, A. S. Luna and J. S. de Gois, J. Food Compos. Anal., 2022, 109, 104519 CrossRef CAS.
  186. X. Zhang, W. Jia, X. Tang, Q. Shan, Q. Chen, C. Cheng, J. Shao, Y. Ling and D. Hei, Anal. Lett., 2023, 56(4), 556–575 CrossRef CAS.
  187. M. Atikul Islam, J. Young Jeong, E. J. Kim, N. Khan, N. Jamila and K. S. Kim, Anal. Lett., 2023, 56(5), 744–757 CrossRef CAS.
  188. H. Shimizu, F. Akamatsu, A. Kamada, K. Koyama, K. Iwashita and N. Goto-Yamamoto, J. Food Compos. Anal., 2022, 107, 104377 CrossRef CAS.
  189. L. Ruggiero, M. C. Fontanella, C. Amalfitano, G. M. Beone and P. Adamo, Food Chem., 2021, 362, 130168 CrossRef CAS PubMed.
  190. F. Di Donato, F. Squeo, A. Biancolillo, L. Rossi and A. A. D’Archivio, Food Control, 2022, 131, 108451 CrossRef CAS.
  191. K. Pasvanka, M. Kostakis, M. Tarapoulouzi, P. Nisianakis, N. S. Thomaidis and C. Proestos, Separations, 2021, 8(8), 119 CrossRef CAS.
  192. D. Stefas, N. Gyftokostas, P. Kourelias, E. Nanou, V. Kokkinos, C. Bouras and S. Couris, Food Control, 2021, 130, 108318 CrossRef CAS.
  193. S. H. Gregorcic, N. Ogrinc, R. Frew, M. Necemer, L. Strojnik and T. Zuliani, Foods, 2021, 10(8), 1729 CrossRef CAS PubMed.
  194. C. Durante, L. Lancellotti, D. Manzini, M. C. Rossi, S. Sighinolfi, A. Marchetti and L. Tassi, Lwt, 2021, 147, 111571 CrossRef CAS.
  195. A. Soos, E. Bodi, S. Varallyay, S. Molnar and B. Kovacs, Talanta, 2021, 233, 122613 CrossRef CAS PubMed.
  196. N. Altunay, A. Elik, M. Farooque Lanjwani and M. Tuzen, Microchem. J., 2022, 179, 107541 CrossRef CAS.
  197. E. Yilmaz, Z. Erbas and M. Soylak, Anal. Chim. Acta, 2021, 1178, 338808 CrossRef CAS PubMed.
  198. M. Şaylan, N. E. Karlıdağ, M. Toprak, Z. Tekin, U. Balçık, E. Öztürk Er and S. Bakırdere, Anal. Lett., 2022, 55(11), 1719–1730 CrossRef.
  199. A. Elik and N. Altunay, Sustainable Chem. Pharm., 2022, 27, 100697 CrossRef CAS.
  200. A. Islam and A. Chauhan, Environ. Sci. Pollut. Res. Int., 2022, 29(45), 69068–69081 CrossRef CAS PubMed.
  201. A. C. R. Carneiro, J. G. Meneguin, P. M. dos Santos, M. Z. Corazza, M. C. Prete, A. W. Rinaldi and C. R. T. Tarley, J. Braz. Chem. Soc., 2022, 33(8), 958–968 CAS.
  202. M. Shamsipur, N. Mafakheri and N. Babajani, Food Anal. Methods, 2022, 15(5), 1203–1213 CrossRef.
  203. H. Xue, L. Jia, H. Jiang, S. Qin, Y. Yang, J. Wu and X. Jing, J. Food Compos. Anal., 2022, 110, 104569 CrossRef CAS.
  204. F. C. Pinheiro, M. A. Aguirre, J. A. Nobrega, N. Gonzalez-Gallardo, D. J. Ramon and A. Canals, Anal. Chim. Acta, 2021, 1185, 339052 CrossRef CAS PubMed.
  205. N. Manousi, A. Kabir, K. Furton, G. Zachariadis and A. Anthemidis, Separations, 2021, 8(7), 100 CrossRef CAS.
  206. S. Khodadadi, E. Konoz, A. Niazi and A. Ezabadi, Chem. Pap., 2022, 76(11), 6735–6751 CrossRef CAS.
  207. F. Tokay and S. Bagdat, Turk. J. Chem., 2022, 46(2), 459–474 CrossRef CAS.
  208. X. Dong, X. Gao, J. Song and L. Zhao, Food Chem., 2021, 360, 130023 CrossRef CAS PubMed.
  209. Z. Li, Y. Zhang, R. Wang and D. Jiang, J. Appl. Spectrosc., 2022, 89(2), 371–380 CrossRef CAS.
  210. A. Rihana-Abdallah, Z. Li and K. C. Lanigan, Anal. Lett., 2022, 55(8), 1296–1305 CrossRef CAS.
  211. A. Londonio, E. Morzan and P. Smichowski, J. Food Compos. Anal., 2022, 107, 104388 CrossRef CAS.
  212. F. C. Pinheiro, M. A. Aguirre, J. A. Nobrega and A. Canals, Anal. Methods, 2021, 13(46), 5670–5678 RSC.
  213. Z. S. Hashami, A. Taheri and M. Alikarami, Anal. Chim. Acta, 2022, 1204, 339262 CrossRef CAS PubMed.
  214. N. Altunay, B. Hazer, M. Tuzen and A. Elik, Food Chem., 2021, 359, 129923 CrossRef CAS PubMed.
  215. E. A. Erfan, W. I. Mortada, H. A. El Asmy and M. M. El-Defrawy, Egypt. J. Chem., 2021, 64(7), 3431–3437 Search PubMed.
  216. T. Dasbasi and C. Soykan, Food Chem., 2021, 365, 130518 CrossRef CAS PubMed.
  217. Q. Li, Y. Zhuo, S. You, Y. Zhang, B. Zhao and L. Xu, Microchem. J., 2022, 181, 107698 CrossRef CAS.
  218. K. Turan, R. Saygili Canlidinc and O. M. Kalfa, Turk. J. Chem., 2022, 46(2), 550–566 CrossRef CAS.
  219. J. S. Trindade, V. A. Lemos, U. M. F. Mata Cerqueira, C. G. Novaes, S. A. Araujo and M. A. Bezerra, Food Chem., 2021, 365, 130473 CrossRef CAS PubMed.
  220. S. Ulusoy, A. Demirbaş, A. Elik, N. Altunay, G. Sarp, E. Yılmaz and H. İ. Ulusoy, Chem. Pap., 2022, 76(10), 6153–6165 CrossRef CAS.
  221. S. Ozdemir, Z. Turkan, E. Kilinc, R. Bayat, M. Soylak and F. Sen, Environ. Res., 2022, 209, 112766 CrossRef CAS PubMed.
  222. S. Ozdemir, E. Kılınç, Ö. Acer and M. Soylak, Microchem. J., 2021, 171, 106758 CrossRef CAS.
  223. W. I. Mortada, K. A. Nabieh, T. E. Helmy and Y. G. Abou El-Reash, J. Food Compos. Anal., 2022, 106, 104352 CrossRef CAS.
  224. J. C. Garcia-Mesa, P. Montoro-Leal, S. Maireles-Rivas, M. M. L. Guerrero and E. V. Alonso, J. Anal. At. Spectrom., 2021, 36(5), 892–899 RSC.
  225. M. K. Bin Ateeq, N. M. Bin Durayhim, M. M. Sulayem, W. A. Al-Qahtani, N. H. Khdary, A. M. Alhassan, F. M. A. Alzahrani, K. M. M. Katubi and N. S. Alsaiari, Water, 2021, 13(13), 1829 CrossRef CAS.
  226. S. Ozdemir, E. Kilinc, O. Acer and M. Soylak, Biol. Trace Elem. Res., 2022, 200(4), 1972–1980 CrossRef CAS PubMed.
  227. Z. Yao, J. Liu, X. Mao, G. Chen, Z. Ma and B. Li, Food Chem., 2022, 379, 132116 CrossRef CAS PubMed.
  228. D. B. Zhou, Y. B. Xiao, F. Han, Y. N. Lv, L. Ding, W. Song, Y. X. Liu, P. Zheng and D. Chen, J. Chromatogr. A, 2021, 1654, 462465 CrossRef CAS PubMed.
  229. Y. Wang, A. L. Zhu, Y. Y. Fang, C. J. Fan, Y. L. Guo, Z. Q. Tan, Y. G. Yin, Y. Cai and G. B. Jiang, J. Environ. Sci., 2022, 115, 403–410 CrossRef CAS PubMed.
  230. E. Kilinc, S. Ozdemir, A. Poli, B. Niolaus, I. Romano, M. Bekmezci and F. Sen, Environ. Res., 2022, 207, 112231 CrossRef CAS PubMed.
  231. M. Burylin, E. Kopeyko, E. Bokiy, V. Konshin and D. Konshina, Chem. Pap., 2022, 76(9), 5893–5899 CrossRef CAS.
  232. F. Elahi, M. B. Arain, W. Ali Khan, H. Ul Haq, A. Khan, F. Jan, R. Castro-Munoz and G. Boczkaj, Food Chem., 2022, 393, 133384 CrossRef CAS PubMed.
  233. H. Ul Haq, R. Bibi, M. Balal Arain, F. Safi, S. Ullah, R. Castro-Munoz and G. Boczkaj, Food Chem., 2022, 379, 132085 CrossRef CAS PubMed.
  234. R. Sanmartín, V. Romero, I. Lavilla and C. Bendicho, Spectrochim. Acta, Part B, 2022, 188, 106349 CrossRef.
  235. X. Yu, C. Zhu, H. Wang and Y. Wu, Anal. Bioanal. Chem., 2022, 414(7), 2409–2418 CrossRef CAS PubMed.
  236. E. Cakmak and T. Oymak, Anal. Sci., 2022, 38(4), 703–709 CrossRef PubMed.
  237. S. Z. Chen, J. H. Liu, J. T. Yan, C. L. Wang and D. B. Lu, Anal. Lett., 2022, 55(10), 1623–1636 CrossRef CAS.
  238. M. C. L. do Nascimento, F. N. Ferreira, A. P. Benevides, D. D. da Silva, D. V. Cesar, A. S. Luna and J. S. de Gois, J. Anal. At. Spectrom., 2021, 36(5), 900–908 RSC.
  239. H. Abdolmohammad-Zadeh, Z. Ayazi and M. Veladi, J. Food Compos. Anal., 2022, 109, 104493 CrossRef CAS.
  240. A. Shishov, S. Savinov, N. Volodina, I. Gurev and A. Bulatov, Microchem. J., 2022, 179, 107456 CrossRef CAS.
  241. S. Q. Zhu, J. W. Liu, X. Gu, J. Q. Zheng and P. X. Zhu, Curr. Pharm. Anal., 2022, 18(1), 82–87 CrossRef CAS.
  242. X. Wang, Z. Qian, H. Li, H. Chen, L. Lin, M. Guo and X. Hai, J. Pharm. Biomed. Anal., 2021, 203, 114201 CrossRef CAS PubMed.
  243. C. Eum, J. Park, S. Kumar, E. Jang, Y. Lee, S. H. Nam, D. Choi and H. Chung, J. Anal. At. Spectrom., 2022, 37(4), 823–832 RSC.
  244. L. Barregard, G. Sallsten, F. Harari, E. M. Andersson, N. Forsgard, O. Hjelmgren, O. Angeras, E. Fagman, M. Persson, T. Lundh, Y. Borne, B. Fagerberg, G. Engstrom and G. Bergstrom, Environ. Health Perspect., 2021, 129(6), 67007 CrossRef CAS PubMed.
  245. C. Ma, M. Iwai-Shimada, S. F. Nakayama, T. Isobe, Y. Kobayashi, N. Tatsuta, Y. Taniguchi, M. Sekiyama, T. Michikawa, S. Yamazaki and M. Kamijima, Environ. Int., 2021, 156, 106762 CrossRef CAS PubMed.
  246. L. Liu, L. Yao, M. Dong, T. Liu, W. Lai, X. Yin, S. Zhou, L. Lv, L. Li, J. Wang, X. Jiang, B. Parveen, J. Chen and X. Sun, J. Trace Elem. Med. Biol., 2021, 68, 126823 CrossRef CAS PubMed.
  247. J. Yim, S. B. Kwon, J. S. Han, J. H. Kim, E. H. Lee and S. G. Lee, Arch. Pathol. Lab. Med., 2021, 145(7), 877–882 CrossRef CAS PubMed.
  248. J. Jablan, E. Besalu, M. Zarak, J. Dumic and E. Margui, Talanta, 2021, 233, 122553 CrossRef CAS PubMed.
  249. B. Michalke, A. Berthele and V. Venkataramani, Int. J. Mol. Sci., 2021, 22(16), 8892 CrossRef CAS PubMed.
  250. J. Dong, S. Liu, L. Wang, X. Zhou, Q. Zhou, C. Liu, J. Zhu, W. Yuan, W. Y. Xu and J. Deng, BMC Pregnancy Childbirth, 2021, 21(1), 454 CrossRef CAS PubMed.
  251. P. Ferreira, C. Pinheiro, C. M. Coelho, J. Guimaraes, G. Pereira, N. X. Moreira, A. Cortez, I. Bracchi, D. Pestana, I. B. Mota, C. Prucha, C. Martins, C. A. Ribeiro, E. Pinto, A. Almeida, C. Delerue-Matos, N. Montenegro, C. C. Dias, A. Moreira-Rosario, L. F. Azevedo, A. L. Brantsaeter, C. Ramalho, V. C. Fernandes, C. Calhau, J. C. Leite and E. Keating, Br. J. Nutr., 2021, 126(9), 1314–1322 CrossRef CAS PubMed.
  252. T. Liu, Y. Li, D. Teng, X. Shi, L. Yan, J. Yang, Y. Yao, Z. Ye, J. Ba, B. Chen, J. Du, L. He, X. Lai, X. Teng, Y. Li, H. Chi, E. Liao, C. Liu, L. Liu, G. Qin, Y. Qin, H. Quan, B. Shi, H. Sun, X. Tang, N. Tong, G. Wang, J. A. Zhang, Y. Wang, Y. Xue, L. Yang, Q. Zhang, L. Zhang, J. Zhu, M. Zhu, Z. Shan and W. Teng, Thyroid, 2021, 31(12), 1858–1867 CrossRef CAS PubMed.
  253. E. A. Miles, T. Vahlberg, P. C. Calder, N. Houttu, L. Pajunen, E. Koivuniemi, K. Mokkala and K. Laitinen, Eur. J. Nutr., 2022, 61(6), 2919–2927 CrossRef CAS PubMed.
  254. M. Stravik, K. Gustin, M. Barman, H. Skroder, A. Sandin, A. E. Wold, A. S. Sandberg, M. Kippler and M. Vahter, Front. Nutr., 2021, 8, 733602 CrossRef PubMed.
  255. C. Eum, E. Jang, H. Kim, S. H. Nam, Y. Lee, D. Choi and H. Chung, Analyst, 2022, 147(14), 3193–3200 RSC.
  256. R. Gunawan, A. Imran, I. Ahmed, Y. C. Liu, Y. W. Chu, L. B. Guo, M. S. Yang and C. Lau, Analyst, 2021, 146(16), 5186–5197 RSC.
  257. X. Fan, Q. Li, P. Lin, Z. Jin, M. Chen and Y. Ju, Ann. Clin. Biochem., 2022, 59(3), 166–170 CrossRef CAS PubMed.
  258. P. Nasr, S. Ignatova, P. Lundberg, S. Kechagias and M. Ekstedt, J. Trace Elem. Med. Biol., 2021, 67, 126772 CrossRef CAS PubMed.
  259. G. Drevin, B. Lelievre, J. Riou and M. Briet, Ann. Work Exposures Health, 2022, 66(1), 52–59 CrossRef CAS PubMed.
  260. V. Rodriguez-Saldana, J. Fobil and N. Basu, Environ. Res., 2021, 198, 110444 CrossRef CAS PubMed.
  261. X. N. Li, L. H. Jia, X. Cao, S. S. Zhang, R. Pu, X. J. Cheng and Y. Liu, J. Trace Elem. Med. Biol., 2021, 67, 126783 CrossRef CAS PubMed.
  262. S. Arshad, A. Arif and J. I. Wattoo, Dose Response, 2022, 20(2) DOI:10.1177/15593258221101744.
  263. K. Uno, N. Yoshikawa, A. Tazaki, S. Ohnuma, K. Kitami, S. Iyoshi, K. Mogi, M. Yoshihara, Y. Koya, M. Sugiyama, S. Tamauchi, Y. Ikeda, A. Yokoi, F. Kikkawa, M. Kato and H. Kajiyama, Sci. Rep., 2022, 12(1), 4513 CrossRef CAS PubMed.
  264. B. B. Chen, G. Y. Xiao, M. He and B. Hu, Anal. Chem., 2021, 93(27), 9454–9461 CrossRef CAS PubMed.
  265. Y. Cao, H. Zhang, J. Yang, Q. Man, P. Song, D. Mao, J. Lu and L. Yang, Int. J. Environ. Res. Public Health, 2022, 19(8), 4908 CrossRef CAS PubMed.
  266. L. Wang, L. Song, B. Liu, L. Zhang, M. Wu, Y. Liu, J. Bi, S. Yang, Z. Cao, W. Xia, Y. Li, Y. Tian, B. Zhang, S. Xu, A. Zhou and Y. Wang, Eur. J. Clin. Nutr., 2022, 76(5), 716–721 CrossRef CAS PubMed.
  267. E. Fireman, A. E. Mahamed, D. Rosengarten, N. N. Ophir and M. R. Kramer, Int. J. Environ. Res. Public Health, 2021, 18(14), 7237 CrossRef CAS PubMed.
  268. X. Chen, L. Huang, Q. Li, X. Zhang, X. Wang, G. Sun, N. Yi, G. Xiong, X. Yang, L. Hao, H. Yang and N. Yang, Environ. Sci. Pollut. Res. Int., 2022, 29(33), 49966–49975 CrossRef CAS PubMed.
  269. V. Kovac, B. Poljsak, M. Bergant, J. Scancar, U. Mezeg and J. Primozic, Coatings, 2022, 12(2), 190 CrossRef CAS.
  270. X. Li, C. Wang, Y. Wang, X. Zhao and N. Li, BMC Pediatr., 2021, 21(1), 483 CrossRef CAS PubMed.
  271. L. Shi, Y. Yuan, Y. Xiao, P. Long, W. Li, Y. Yu, Y. Liu, K. Liu, H. Wang, L. Zhou, H. Yang, X. Li, M. He and T. Wu, Environ. Int., 2021, 157, 106808 CrossRef CAS PubMed.
  272. S. A. Takyi, N. Basu, J. Arko-Mensah, D. Dwomoh, K. G. Houessionon and J. N. Fobil, Chemosphere, 2021, 280, 130677 CrossRef CAS PubMed.
  273. Z. Chai, G. Zhang, X. Ling, T. Dong, J. Wang, Y. Zhang, P. Zou, H. Yang, N. Zhou, Q. Chen, Y. Zheng, J. Liu, J. Cao and L. Ao, Sci. Total Environ., 2022, 828, 154395 CrossRef CAS PubMed.
  274. C. M. Cirtiu, M. Valcke, M. Gagne, M. H. Bourgault, C. Narame, S. Gadio, P. Poulin and P. Ayotte, Chemosphere, 2022, 289, 133142 CrossRef CAS PubMed.
  275. A. Kubala-Kukus, D. Banas, J. Braziewicz, I. Stabrawa and S. Gozdz, Appl. Spectrosc., 2021, 75(7), 802–817 CrossRef CAS PubMed.
  276. H. T. Phuong, N. A. Son, N. T. N. Ha, N. T. M. Sang, N. T. T. Linh, D. T. Binh, T. T. H. Loan, H. M. Dung, T. T. Anh and N. Van Dong, Spectrochim. Acta, Part B, 2021, 182, 106234 CrossRef CAS.
  277. M. Luo, X. H. Li, Q. X. He, J. Liu, H. S. Lu, Q. Liu and X. H. Yang, J. Anal. At. Spectrom., 2022, 37(4), 833–840 RSC.
  278. G. Ding, Y. Jing, Y. Han, P. Sun, S. Liang, J. Liu, X. Wang, Y. Lian, Y. Fang, Z. Jin and W. Li, Food Addit. Contam., Part B, 2021, 14(3), 177–183 CrossRef CAS PubMed.
  279. M. C. Fontanella, M. Martin, D. Tenni, G. M. Beone and M. Romani, Rice Sci., 2021, 28(4), 402–408 CrossRef.
  280. J. Li, S. Chen, H. Li, X. Liu, J. Cheng and L. Q. Ma, Environ. Res., 2021, 198, 111198 CrossRef CAS PubMed.
  281. C. C. Brombach, S. T. Lancaster, M. A. Dexter, K. C. Thompson and W. T. Corns, Anal. Methods, 2021, 13(28), 3154–3159 RSC.
  282. T. Koga, S. Hirakawa, Y. Ishibashi and T. Hori, Food Hyg. Saf. Sci., 2021, 62(3), 100–104 CrossRef CAS PubMed.
  283. L. Lancellotti, S. Sighinolfi, A. Ulrici, L. Maletti, C. Durante, A. Marchetti and L. Tassi, Curr. Res. Food Sci., 2021, 4, 807–814 CrossRef CAS PubMed.
  284. G. E. Guild and J. C. R. Stangoulis, Plant Soil, 2022, 473(1–2), 659–667 CrossRef CAS.
  285. J. M. Niu, J. Q. Lu and Y. K. Li, Laser Optoelectron. Prog., 2021, 58(17), 1730001 Search PubMed.
  286. Z. Karimi and M. Goli, Food Sci. Nutr., 2021, 9(9), 5112–5118 CrossRef CAS PubMed.
  287. S. Dubascoux, J. Richoz Payot, P. Sylvain, M. Nicolas and E. Campos Gimenez, Food Chem., 2021, 362, 130197 CrossRef CAS PubMed.
  288. C. Zhu, J. Lv, K. Liu, Q. Li, Z. Tang, R. Zhou, W. Zhang, J. Chen, K. Liu, X. Li and X. Zeng, Microchem. J., 2021, 168, 106408 CrossRef CAS.
  289. G. S. Filbido, I. M. P. Narita, A. P. de Oliveira Pinheiro, D. da Cruz e Silva, B. A. Ferreira, E. Nascimento, R. D. Villa and A. P. de Oliveira, J. Food Meas. Charact., 2021, 15(6), 5648–5656 CrossRef.
  290. A. I. Zhang, Y. Lin, J. H. Yang, L. B. He, Y. R. Deng, X. D. Hou and C. B. Zheng, Analyst, 2021, 146(12), 3852–3857 RSC.
  291. A. Cherkani-Hassani, M. Slaoui, I. Ghanname, B. Mojemmi, R. Eljaoudi, A. Belhaj, B. Kers, M. Flayou and N. Mouane, Environ. Toxicol. Pharmacol., 2021, 85, 103629 CrossRef CAS PubMed.
  292. A. Cherkani-Hassani, M. Slaoui, I. Ghanname, B. Mojemmi, A. Belhaj, B. Kers, M. Flayou and N. Mouane, Regul. Toxicol. Pharmacol., 2021, 127, 105065 CrossRef CAS PubMed.
  293. M. B. Perez, V. M. Lipinski, M. F. Fillipini, K. Chacon Madrid, M. A. Zezzi Arruda and R. G. Wuilloud, Food Addit. Contam., Part A, 2021, 38(9), 1506–1519 CrossRef CAS PubMed.
  294. S. Bodur, M. Oner, S. Erarpat and S. Bakirdere, J. Sep. Sci., 2021, 44(16), 3031–3040 CrossRef CAS PubMed.
  295. J. Cao, Y. Cheng, B. Xu, Y. Wang and F. Wang, Food Anal. Methods, 2021, 14(11), 2420–2429 CrossRef.
  296. B. C. Coldwell, N. M. Perez, M. C. Vaca, M. J. Pankhurst, P. A. Hernandez, G. V. M. Rodriguez, E. Padron, M. Asensio-Ramos, S. Ribeiro and J. F. Santos, Beverages, 2022, 8(1), 9 CrossRef CAS.
  297. Y. H. Kim, W. J. Ra, S. Cho, S. Choi, B. Soh, Y. Joo and K. W. Lee, Molecules, 2021, 26(21), 6729 CrossRef CAS PubMed.
  298. S. H. Nam, N. Y. Kim, E. S. Park and Y. Lee, Bull. Korean Chem. Soc., 2022, 43(3), 402–406 CrossRef CAS.
  299. F. H. Reboredo, W. Junior, M. F. Pessoa, F. C. Lidon, J. C. Ramalho, R. G. Leitao, M. M. Silva, N. Alvarenga and M. Guerra, Plants, 2021, 10(10), 2041 CrossRef CAS PubMed.
  300. K. von Wuthenau, T. Segelke, M.-S. Müller, H. Behlok and M. Fischer, Food Control, 2022, 134, 108689 CrossRef CAS.
  301. K. von Wuthenau, M. S. Muller, L. Cvancar, M. Oest and M. Fischer, J. Agric. Food Chem., 2022, 70(16), 5237–5244 CrossRef CAS PubMed.
  302. S. H. Bilston-John, A. Narayanan, C. Tat Lai, A. Rea, J. Joseph and D. T. Geddes, Food Chem., 2021, 363, 130179 CrossRef CAS PubMed.
  303. M. Kalpage, C. Dissanayake, S. Diyabalanage, R. Chandrajith, R. Frew and R. Fernando, Foods, 2022, 11(3), 275 CrossRef CAS PubMed.
  304. W. Huang, L. Guo, W. Kou, D. Zhang, Z. Hu, F. Chen, Y. Chu and W. Cheng, Microchem. J., 2022, 176, 107190 CrossRef CAS.
  305. B. Duarte, J. Carreiras, R. Mamede, I. A. Duarte, I. Caçador, P. Reis-Santos, R. P. Vasconcelos, C. Gameiro, R. Rosa, S. E. Tanner and V. F. Fonseca, J. Food Compos. Anal., 2022, 109, 104479 CrossRef CAS.
  306. C. Dumitrascu, Y. Fiamegos and M. B. de la Calle Guntinas, Anal. Bioanal. Chem., 2021, 413(20), 4947–4957 CrossRef CAS PubMed.
  307. H. Wang, P. Zhang, Z. Xu, W. Cheng, X. Li, Y. Yang, Y. Wu and Q. Wang, Microw. Opt. Technol. Lett., 2022 DOI:10.1002/mop.33226.
  308. W. B. Jia, X. R. Tang, X. L. Zhang, J. F. Shao, G. C. Xiong, Y. S. Ling, D. Q. Hei and Q. Shan, Spectrosc. Spectral Anal., 2021, 41(12), 3815–3821 CAS.
  309. T. Chen, X. P. Qi, Z. Y. Si, Q. W. Cheng and H. Chen, Int. J. Food Eng., 2021, 17(11), 877–884 CrossRef.
  310. A. S. Maltsev, R. A. Yusupov and S. A. Bakhteev, X-Ray Spectrom., 2022 DOI:10.1002/xrs.3283.
  311. E. Margui, I. Queralt, D. Andrey and L. Perring, Food Chem., 2022, 383, 132590 CrossRef CAS PubMed.
  312. D. Yasar, Y. Erim Kose and S. Kose, J. Food Process. Preserv., 2021, 45(12), e16021 CAS.
  313. L. Abdi, G. R. Jahed-Khaniki, E. Molaee-Aghaee, N. Shariatifar, S. Nazmara and A. Mousavi Khaneghah, Environ. Sci. Pollut. Res. Int., 2021, 28(44), 62651–62661 CrossRef CAS PubMed.
  314. H. Hernández-Mendoza, N. Lara-Almazán, A. Kuri-Cruz, E. T. Romero-Guzmán and M. J. Ríos-Lugo, Phys. Sci. Rev., 2021, 000010151520200121,  DOI:10.1515/psr-2020-0121.
  315. F. Ait lhaj, H. Elhamri, Z. Ait lhaj, A. Zarrouk, A. El Abidi, S. El Hajjaji and A. Bellaouchou, Int. J. Environ. Anal. Chem., 2021, 1–13 CrossRef.
  316. J. M. de Higuera, H. M. Santos, A. F. de Oliveira and A. R. A. Nogueira, J. Braz. Chem. Soc., 2021, 32(11), 2111–2119 CAS.
  317. J. P. Asensio, D. A. Uceda and P. J. Navarro, Spectroscopy, 2021, 36, 24–34 Search PubMed.
  318. K. Motoyama, T. Isojima, Y. Sato, A. Aihara, H. Asakura, H. Hiraike, Y. Hino, M. Mimaki, K. Nomura and H. Kodama, Pediatr. Int., 2021, 63(8), 910–917 CrossRef CAS PubMed.
  319. F. Di Donato, M. Foschi, N. Vlad, A. Biancolillo, L. Rossi and A. A. D’Archivio, Molecules, 2021, 26(22), 6875 CrossRef CAS PubMed.
  320. N. P. Kalogiouri, N. Manousi, I. Mourtzinos and G. A. Zachariadis, Anal. Lett., 2022, 55(13), 2108–2123 CrossRef CAS.
  321. M. Marques, E. Correig, E. Capdevila, E. Gargallo, N. Gonzalez, M. Nadal and J. L. Domingo, Biol. Trace Elem. Res., 2022, 200(10), 4524–4533 CrossRef CAS PubMed.
  322. M. Szukalska, T. A. Merritt, W. Lorenc, K. Sroczynska, I. Miechowicz, I. Komorowicz, J. Mazela, D. Baralkiewicz and E. Florek, Environ. Res., 2021, 197, 111090 CrossRef CAS PubMed.
  323. D. Vitali Cepo, M. Karoglan, L. Borgese, L. E. Depero, E. Margui and J. Jablan, Food Chem.: X, 2022, 13, 100209 CAS.
  324. J. Wang, S. Wang, X. Ge and M. Zhang, Eur. Food Res. Technol., 2021, 247(10), 2657–2666 CrossRef CAS.
  325. L. Xu, C. Hai, S. Yan, S. Wang, S. Du, H. Chen, J. Yang and H. Fu, J. Food Sci., 2021, 86(8), 3447–3456 CrossRef CAS PubMed.
  326. J. Proch, A. Orlowska and P. Niedzielski, Foods, 2021, 10(12), 2925 CrossRef CAS PubMed.
  327. F. Badanaro, N. Barsan, A. Tété-Benissan, V. Nedeff and M. Panainte-Lehadus, Sustainability, 2022, 14(3), 1892 CrossRef CAS.
  328. E. G. Nasr, E. N. Epova, A. de Diego, R. Souissi, M. Hammami, H. Abderrazak and F. X. D. O, Foods, 2021, 11(1), 82 CrossRef PubMed.
  329. C. M. Lim, M. Carey, P. N. Williams and A. Koidis, Curr. Res. Food Sci., 2021, 4, 45–52 CrossRef CAS PubMed.
  330. J. Griboff, M. Horacek, D. A. Wunderlin and M. V. Monferrán, Front. Sustain. Food Syst., 2021, 5, 657412 CrossRef.
  331. L. Sinkovic, M. Necemer, B. Pipan and V. Meglic, Acta Chim. Slov., 2021, 68(4), 913–920 CrossRef CAS PubMed.
  332. M. Ghidotti, S. Papoci and M. B. de la Calle Guntiñas, Spectrochim. Acta, Part B, 2022, 189, 106369 CrossRef CAS.
  333. A. Mara, S. Deidda, M. Caredda, M. Ciulu, M. Deroma, E. Farinini, I. Floris, I. Langasco, R. Leardi, M. I. Pilo, N. Spano and G. Sanna, Molecules, 2022, 27(6), 2009 CrossRef CAS PubMed.
  334. Y. Sun, Y. Zhao, J. Wu, N. Liu, X. Kang, S. Wang and D. Zhou, Food Control, 2022, 134, 108753 CrossRef.
  335. B. Duarte, R. Mamede, I. A. Duarte, I. Cacador, S. E. Tanner, M. Silva, D. Jacinto, T. Cruz and V. F. Fonseca, Molecules, 2022, 27(4), 1298 CrossRef CAS PubMed.
  336. D. Cui, Y. Liu, H. Yu, Z. Wang and X. Mao, J. Consum. Prot. Food Saf., 2021, 16(4), 323–331 CrossRef CAS.
  337. T. I. Todorov, M. M. Wolle and S. D. Conklin, Chemosphere, 2022, 294, 133651 CrossRef CAS PubMed.
  338. T. Voyslavov, E. Mladenova and R. Balkanska, Molecules, 2021, 26(23), 7219 CrossRef CAS PubMed.
  339. Y. Wang, X. Yuan, L. Liu, J. Ma, S. Fan, Y. Zhang, Q. Li and A. J. Signes-Pastor, J. Food Qual., 2021, 2021, 5536241 Search PubMed.
  340. M. Gajek, A. Pawlaczyk, K. Jozwik and M. I. Szynkowska-Jozwik, Foods, 2022, 11(11), 1616 CrossRef CAS PubMed.
  341. R. R. Kavle, A. Carne, A. E. D. A. Bekhit, B. Kebede and D. Agyei, Int. J. Food Sci. Technol., 2022, 57(10), 6270–6278 CrossRef CAS.
  342. M. L. Astolfi, E. Marconi, G. Vitiello and L. Massimi, Food Chem., 2021, 360, 130027 CrossRef CAS PubMed.
  343. E. Cala, A. Fracchia, E. Robotti, F. Gulino, F. Gullo, M. Oddone, M. Massacane, G. Cordone and M. Aceto, Molecules, 2022, 27(12), 3854 CrossRef CAS PubMed.
  344. P. Alonso Gonzalez, E. Parga-Dans, P. Arribas Blazquez, O. Perez Luzardo, M. L. Zumbado Pena, M. M. Hernandez Gonzalez, A. Rodriguez-Hernandez and C. Andujar, PLoS One, 2021, 16(11), e0258739 CrossRef CAS PubMed.
  345. M. O. Varra, L. Husakova, J. Patocka, S. Ghidini and E. Zanardi, Food Chem., 2021, 360, 129790 CrossRef CAS PubMed.
  346. M. O. Varra, L. Husakova, J. Patocka, S. Ghidini and E. Zanardi, Food Chem., 2021, 356, 129687 CrossRef CAS PubMed.
  347. J. J. Zhang, Y. M. Lin, X. H. Wei, Z. Y. Li and R. R. Li, J. AOAC Int., 2022, 105(2), 603–611 CrossRef PubMed.
  348. F. Di Donato, G. Gornati, A. Biancolillo and A. A. D’Archivio, J. Food Compos. Anal., 2021, 98, 103842 CrossRef CAS.

This journal is © The Royal Society of Chemistry 2023