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

Contents

• 1. Reviews

• 2. Metrology and reference ranges
  • 2.1 Interlaboratory studies, reference materials and measurement uncertainty
  • 2.2 Reference ranges

• 3. Sample collection and preparation
  • 3.1 Collection, storage and preliminary preparation
  • 3.2 Digestion, extraction and preconcentration

• 4. Progress with analytical techniques
  • 4.1 Mass spectrometry
  • 4.2 Atomic absorption and atomic emission spectrometry
  • 4.3 Laser induced breakdown spectroscopy
  • 4.4 Vapour generation procedures
  • 4.5 X-ray spectrometry

• 5. Nanomaterials and single cell analysis
  • 5.1 Food and beverages
  • 5.2 Biological applications
  • 5.3 Pharmaceuticals and consumer products

• 6. Speciation and imaging studies
  • 6.1 Speciation studies
  • 6.2 Imaging with MS and X-rays

• 7. Applications: clinical and biological materials
  • 7.1 Elements used for indirect determinations
  • 7.2 Multi-element applications
  • 7.3 Progress for individual elements

• 8. Applications: drugs and pharmaceuticals, traditional medicines and supplements

• 9. Applications: foods and beverages
  • 9.1 Progress for individual elements
  • 9.2 Single and multi-element applications in food and beverages

• 10. Abbreviations

• Conflicts of interest

• References

---

Abstract

This Update covers publications from the second half of 2023 to the middle of 2024, related to the analysis of clinical specimens, food and beverages and including reviews discussing specific aspects and future perspectives. The pursuit of lower detection capabilities continues to promote further research. Some works have focussed on vapour generating techniques for trapping and preconcentration prior to instrumental analysis and many more on the extraction and preconcentration of elements during the sample preparation step. The synthesis and application of nanomaterials play an important role for these developments. The interest in reagents for green chemistry was the subject of several reviews, and, increasingly, researchers are discussing the environmental impact and sustainability of new methods. In terms of analytical instrumentation, the performances of MICAP-OES were thoroughly assessed on various complex matrices, confirming the technique as highly robust to interferences and with detection capabilities similar to ICP-OES. The N₂-MICAP-MS, which is free from Ar-based polyatomic interferences, was applied to human serum and food samples, although with no significant advantage. There was continued interest in applications of scICP-TOF-MS for the evaluation of more than one characteristic of the same individual cell. The determination of isotope ratios in biological matrices was carried out by high precision MC-ICP-MS. The interest in single cell and single particle analysis and the determination of nanoparticles resulted in the publication of multiple papers specifically reviewing the development in these areas, from which the need emerged to apply multiple techniques and promote standardisation. Multielement techniques (LIBS and XRF) are increasingly applied both in the clinical and food area. Several papers reported new applications of LIBS for disease diagnosis, based on complex statistical modelling methods, whereas others have focussed on techniques to enhance its sensitivity and precision for the analysis of biological, food and water samples. Portable LIBS instrumentation offers potential for a wider application in forensic and food sciences, if calibration strategies and method validation could be improved. The indirect detection of low concentrations of molecules of clinical interest after element-tagging represents a rapidly expanding field. Applications of atomic spectrometry to food and beverages continue to grow, especially in relation to sustainable diets and food supplements. The trend of investigations of authenticity/provenance of foodstuffs using multielement profiles continues. The quality of analytical measurements should be of the utmost importance, however still too often such information is insufficient. We therefore highlight the analysis of CRMs in the papers discussed in this Update and report publications addressing interlaboratory comparisons and new CRMs.

---

1. Reviews

This latest Update follows on from last year’s¹ and is a critical review of the relevant literature published during the period from the second half of 2023 through the first half of 2024. It should be read in conjunction with the five other Atomic Spectrometry Updates published during the same period.^2–6 A number of researchers reviewed general trends or specific aspects in the areas of clinical specimens, food and beverage samples. Whereas most of them are briefly mentioned here, to avoid repetition, more specific reviews are discussed in greater detail in the context of the relevant sections.

Sample preparation is a critical step for most analytical work for elemental analysis. It is usually carried out by digestion or extraction, procedures that are often time-consuming and prone to contamination or losses of the analyte. Alternatively, samples may be reduced into small particles and dispersed into a liquid (slurry sampling) to be introduced directly into measuring systems, such as AFS, CV-AAS, FAAS, GFAAS, ICP-MS and ICP-OES. The fundamentals of the slurry technique, as well as recent developments, were reviewed by Cerqueira et al.⁷ examining 115 publications, most of them issued over the last ten years. They highlighted the criticalities in the preparation of slurries, among which the size of particles, the choice of diluent, the ratio between sample mass and liquid volume, the addition of surfactants, and the difficulty of maintaining the slurry homogeneity, all of which may limit the success of the sampling and lead to lower accuracy of the results in comparison with conventional sample preparation techniques. However, they also noted that because of the benefits of reduced sample preparation time and lower reagent consumption, new approaches were developed to improve the performances of slurry sampling, in particular combining it with sample pre-digestion or applying DSPE with the direct introduction of the adsorbent into the equipment. Slurry sampling was also used in conjunction with FIA systems, gas phase molecular spectrophotometry, TXRF and speciation analysis. The review also gave an interesting overview of the potential of slurry sampling and its future developments.

On the 50^thanniversary of ICP-OES becoming commercially available (1974–2024), Douvris et al.⁸ considered the achievements of this technique. Its capability for multielement analysis of several sample types, over a wide concentration range, has been exploited in support of clinical investigations, occupational and environmental biomonitoring, food safety and nutritional assessment. Besides an overview of the features of the technique, the authors highlighted multielement detection, expanded linear range and reduced interferences as its advantages and presented a review of some of its recent significant applications. They reported that the drawbacks, identified as poor LODs for solid samples, spectral interferences and high cost of operation, were overcome by implementation of USNs, LIBS and MIP, although there are still some challenges to be addressed in future research and development.

Laser-induced breakdown spectroscopy (LIBS) and XRF attract increasing interest for the analysis of biological specimens, food and beverage samples, as robust analytical techniques, requiring minimal sample treatment and offering the capability to map the elemental distribution. A group of researchers⁹ considered the applications of LIBS for the determination of chemical elements for biomedical applications. Through a systematic search of on-line databases over the period 2000–2023, they selected and examined 102 publications, addressing the analysis of blood and tissues (bones, hair, nails, teeth, other organs and cancer tissues) from experimental models of human diseases, including human subjects, human tissues, in vivo animal and in vitro cell line models. A useful summary of the elements detected and the various solid and soft tissues analysed in each study was provided as a figure. The authors highlighted, in particular, the application of LIBS in studies of elemental profiles associated with specific pathologies and for the classification of cancerous vs. healthy tissue. They noted, however, that the majority of studies focused on a limited number of trace elements (Ca, Cu, Fe, K, Na, Mg and Zn), i.e. those present at higher concentrations in blood and tissue samples, and reported that few researchers investigated, with variable success rate, the comparison of LIBS data with those obtained with other multielement techniques such as ICP-OES and ICP-MS. They concluded that further developments of LIBS, in terms of sensitivity, calibration interval and validation by comparison with other techniques, would be necessary to exploit its full potential in this area. Hair and nails have been proposed as useful, non-invasive matrices for the assessment of exposure to toxic elements as well as of nutritional status, as they may store such information over a longer period compared to biological fluids, although studies of essential elements may need to be further validated from a clinical perspective. Reports of recent investigations of hair and nails are mentioned elsewhere in this Update (see Sections 2.2 and 7.2.2). Bali et al.¹⁰ reviewed the major developments and advances of LIBS and XRF, over the last 10 years, for qualitative and quantitative analysis, elemental imaging and distribution of essential and non-essential elements in human hair and nails. Key findings of the examined studies were reported in tables, highlighting the role of LIBS and XRF analysis of human hair and nails in relation to the diagnosis of human diseases. Forensic settings often require rapid responses and in situ analyses of a variety of specimens, including, e.g., human remains, environmental samples, glass, paint and gunshot residues. Jurowski et al.¹¹ proposed that elemental fingerprinting obtained using field portable XRF may provide an answer to these needs. In addition, given the non-destructive nature of XRF, samples could be preserved for additional tests using other, well-established techniques, such as AAS, ICP-MS and Raman spectroscopy. The major drawback, leading to the technique being considered semiquantitative, was identified in the challenge of harmonising the analytical calibration of such instrumentation, based on specific computational algorithms. They examined 31 publications over the period between 2006 and 2022 to highlight the state-of-the-art of the application of field portable XRF in forensic science, covering, among other areas, the detection of counterfeit pharmaceuticals and the analysis of bones. The advantages and disadvantages experienced in these applications were reported in two tables, together with the analytical details available, and additional information was published as supplementary material. The authors noted that key factors underpinning the reliability of data, such as calibration strategies, LOD and LOQ, linearity coefficient and validation, were missing in most of the cited papers, possibly because of the present limited application of the technique in forensic sciences. They concluded that the relatively new, field portable XRF, offered a great potential for a wider application in forensic science, which needs to be supported by further research and validation. To illustrate this, they provided a diagram summarising what has been achieved so far and highlighting directions for future developments. Rapid responses are also necessary in the analysis of food, to ensure its compliance with stated limits for human consumption. Frydrych and Jurowski¹² reviewed the current state-of-the-art of the application of portable XRF to this task, examining 54 publications between 2005 and 2022, including technical specifications for known portable XRF spectrometers. As noted in a previous paper¹¹ with regards to forensic analysis, a wider application of the technique is hindered by the limited information on analytical calibration and performance. The authors summarised the details of applications in solid and liquid food samples in two tables and supplementary material, highlighting their positive features, e.g. provision of LOD and/or LOQ information, comparison with other techniques, external calibration or description of the calibration used, and drawbacks (mainly incomplete description of the parameters used and lack of details supporting the validity of the analytical results), and suggested analytical calibration strategies that can improve analytical performance. This thorough review of the principles, applications, benefits and drawbacks of portable XRF for food analysis may be useful to address its further developments.

The presence of chemical elements in food is the subject of continuous monitoring, to assess dietary intakes and potential risks linked to food consumption. Unsurprisingly, applications of atomic spectrometry in this area are continuously growing and several researchers have undertaken reviews of the recent literature on specific topics.

Arsenic and its species are widely investigated, because the complex incorporation of As into biomolecules, by the formation of bonds with C, leads to a large number of As-containing compounds and As toxicity depends on its chemical form (e.g. AB, AC, As^III, As^V, DMA^III, DMA^V, MMA^III, MMA^V and several others). A high consumption of vegetables is recommended as part of a healthy diet. However, geological features and/or environmental pollution may lead to the accumulation of As in some plants, calling for reliable analytical methods for As speciation in vegetables consumed as food. Sadee et al.,¹³ after considering the variety of As species that may be present in food, their relative toxicity and potential effects on human health, examined the characteristics of the analytical methods in place for this task (128 references). Procedures for the extraction of As species (As^III, As^V, DMA and MMA) from plants and vegetables, using different reagents (water, MeOH, acids, EDTA, NaH₂PO₄, NH₄H₂PO₄) were summarised in a table. The authors discussed the advantages and disadvantages of the techniques used for the separation (CE, IC, ion pair LC, HILIC, RP LC and SEC) and measurement of As species (AAS, AFS, AES, ICP-MS, XRF and chemiluminescence) and concluded that hyphenated techniques based on ICP-MS provided the most promising and realistic approach. Notwithstanding the need to prove the validity of the results of these investigations, as with other areas of research, suitable CRMs are lacking. The authors noted that in the literature examined, some CRMs, certified for the total As content, (NCRM GBW10048-GSB-26 celery, NCRM GBW10015 spinach, NCRM GBW 82301 peach leaves, BCR-279 sea lettuce (Ulva lactuca), NCRM GBW10049 green onion, NCS-ZC-85006 tomato leaves, CS-PR-2 parsnip root powder, NIST SRM 1573a tomato leaves) were also analysed for the content of As species, thus providing at least some useful indicative values for comparison. Another group¹⁴ focused on freshwater fish, the worldwide production of which amounts to about one third of that harvested from marine waters. The levels of As are lower in freshwater than in marine fish, a fact that poses challenges for the determination of As species. The authors reviewed a considerable number of articles (ca. 150) to summarise the information available as well as the details of the analytical techniques applied for the separation of As species (CE, anion-exchange HPLC, cation-exchange HPLC, RP HPLC or a combination of these) and for the identification and quantification of the As species (ESI-MS, ICP-MS). The level of total As observed for freshwater fish in different parts of the world varied from 0.010 mg kg⁻¹ to 5.97 mg kg⁻¹, except for one study (1.1–664.8 mg kg⁻¹). The As species considered in this review included AB, AC, As^III, As^V, DMA, MMA, TETRA, TMAO, arsenosugars, arsenolipids, and other molecules. However, the sum of the concentrations of the identified As species was often much lower (from 31% to 74%) than that of the total As, leaving some doubts about the efficiency of the extraction procedures as well as the presence of unidentified As compounds and their possible role. The authors recommended extraction procedures that can maintain the integrity and stability of As species. They noted that extraction efficiencies may vary with different types of food, fish or fish tissue and suggested the use of different solvents or enzymatic digestion, to improve the extraction efficiency. Unfortunately, the lack of both species-specific CRMs and pure As-species standards hinders the assessment of these procedures, that should at least be compared one against the other. The investigation of the multiple As species in living organisms remains a lively area of research. However, for the time being, there is no strong evidence of risk for human health associated with the current exposure to As species (except iAs), through vegetables other than rice and marine or freshwater fish consumption—unless in areas where the water is heavily contaminated. An exposure limit of 10 μg L⁻¹ As in drinking water has been set by WHO, the EU and the US EPA. The FAO/WHO Codex Alimentarius established a limit of 0.5 mg kg⁻¹ As for edible salt. Maximum limits for iAs are in place in the EU for rice and rice products, food for infants and young children, fruit juices, concentrated fruit juices and nectars, and salt, but not for fish/seafood (where As is present mainly as the non-toxic AB) or vegetables other than rice. Arsenolipids are a relatively new class of organic As compounds, usually found in marine organisms, including those that enter the chain of human consumption. Since the description of the structure of diacyl arsenosugar phospholipids in 1988,¹⁵ several other organoarsenical compounds have been identified, thus opening a vast area of research for their characterisation and the assessment of their occurrence and potential toxicity. Coniglio and co-workers¹⁶ reviewed the relevant literature (107 references, from 1988 to 2023) to address analytical progress and challenges for the determination of arsenolipids in marine food, noting the need for harmonisation of the relative nomenclature. The introduction of HPLC-ICP-MS, using IC, RP or SEC for separation, gave a strong input to research in this field, although hindered by the lack of analytical standards to confirm retention times and species quantification. In addition, sample pre-treatment (acid or basic hydrolysis), aimed to separate the lipophilic fractions and facilitate the chromatographic separation prior to ICP-MS analysis, was prone to induce alterations of the chemical species under examination. Another issue was represented by the use of aqueous mobile phases, to avoid plasma instability due to carbon overload, that required further hydrolysis of the lipid-soluble compounds, but the introduction of gradient elution and aqueous support flows enabled this problem to be overcome, thus allowing the investigation of arsenolipids by RP HPLC-ICP-MS. Other techniques (NMR, XAS and XRD) were applied for the characterisation of the structure of complex organo-arsenical compounds, paving the way for the synthesis, either in-house or by commercial companies, of these molecules and hence allowing for identification and quantification of these species. However, the authors noted that, to date, given the large number of arsenolipids, few standards and no CRMs are available. These circumstances prevented the quantification of arsenolipids by a single technique. However, the combination of HPLC-ESI-MS and HPLC-ICP-MS allowed for the identification of compound structures, by means of the mass spectra analysis and the simultaneous quantification of the As present in the same fraction. Focusing on the determination of three classes of arsenolipids (arsenic fatty acid, arsenic hydrocarbons and diacyl arsenosugar phospholipids) in marine foods, the authors highlighted the significant contribution of the use of both RP HPLC-ICP-MS and RP HPLC-ESI-MS to increase knowledge about the identity, amount and decomposition processes of these compounds.

Potentially toxic elements present in the aquatic environment are readily accumulated by certain living organisms, such as bivalves (clams, mussels, and oysters) that are widely used for human consumption. The assessment of metal concentrations in the soft tissues of these species is performed both to ensure food safety and for environmental monitoring purposes. A review of the relevant literature to date (170 papers, 2000–2022), carried out by Pasinszki et al.,¹⁷ confirmed a growing interest in this topic, with 55% of the examined papers published between 2021 and 2022. Although the majority of the studies performed the determination of the total content of elements in the whole body of the bivalve, some addressed the assessment of the element distribution in tissues or organs, whereas speciation studies were limited to small molecules such as MeHg. The authors provided a critical evaluation of the existing literature, presenting details of the sample preparation methods (freeze-drying, oven drying, ashing, digestion, extraction and preconcentration) and comparing the advantages and disadvantages of the different measurement techniques (AAS, AES, AFS, ICP-MS, ED-XRF, GC-MS, NAA, and electrochemical sensors). Their study covered the determination of 24 elements (As, Ag, Bi, Cd, Co, Cr, Cu, Fe, Ga, Hg, La, Lu, Mn, Mo, Ni, Pb, Sb, Sm, Sn, Th, Tl, V, U and Zn) although some are rarely measured. Another group¹⁸ addressed a similar topic. They reviewed (89 references) analytical methods for the measurements of As, Cd, Hg and Pb in fish, as part of an assessment of nutritional and toxicological aspects of fish intake. They covered the approaches used for drying fish, the first step of sample treatment, and suggested freeze-drying as the best option. Various techniques for sample digestion were applied (open or closed systems as well as MAD) whereas measurements were performed by either AFS, ETAAS, ICP-MS, ICP-OES or TDA-AAS (with a direct mercury analyser). Advantages and limitations of these approaches were discussed and compared, based on details of the analytical performance of each method, presented in a table.

The different species of Cr (Cr^VI and Cr^III) show different properties in relation to human health, with the former being linked to genotoxicity and carcinogenicity and the latter being involved in physiological processes. Although Cr^VI is readily converted to Cr^III by the organic matrix in foodstuffs, it is suggested that Cr^VI may still be present in some raw food items and therefore the accurate determination of the two Cr species in food is deemed necessary to assess dietary exposure. Such measurements are complicated by the risk of inter-conversion during the analytical steps, an issue not yet resolved. Chung¹⁹ explored recently published work (18 papers, from 2018 to 2023) to assess whether the applied procedures maintained the stability of the species under investigation, as well as whether the achieved LODs were suitable for the assessment of dietary exposure to Cr^VI. Details of matrices, sample preparation, techniques for analytical detection and analytical performances (LOD, LOQ and recoveries) were reported in a table. In most of the eleven papers reporting off-line detection, based on ETAAS, FAAS, ICP-MS and other techniques, the determination of one of the species was calculated by subtraction of the other from the total Cr content, but the author noted that this practice may lead to incorrect results in the presence of strong oxidising agents in the sample preparation step. Out of the seven papers using HPLC-ICP-MS, SSID-ICP-MS was applied to several foods (bread, breakfast cereals, dairy products and meat). After spiking the samples with ⁵⁰Cr^III and ⁵³Cr^VI, the addition of EDTA led to the selective complexation of Cr^III, followed by reduction of Cr^VI to Cr^III using 1,5-diphenylcarbazide to form a DPC-complex and analysis of the two complex species using HPLC-ICP-MS, in KED mode with He as the collision gas. The procedure allowed the determination of both Cr species during the same analytical run, avoiding species inter-conversion, and was considered the best choice for Cr speciation. The author suggested that LODs for Cr species should be at least lower than 0.03 μg kg⁻¹ and recommended the use of HR-ICP-MS or ICP-MS/MS rather than ICP-MS to achieve this.

Selenium is an important element to support human health and adequate intake through food or supplements is recommended. Therefore, Se is measured in food, beverages and supplements as part of monitoring studies of Se intake as well as for the purpose of demonstrating compliance of the product content with the level indicated on the label. de Souza and co-workers²⁰ undertook a systematic review and meta-analysis of validated analytical methods for the determination of total Se in foods and beverages. Their search strategy, described in detail, led them to consider 105 articles, published between 1989 and 2022. They noted that the most common analytical techniques used were ICP-MS (29), HPLC-ICP-MS (26), and AAS (24). Details of the matrices analysed, sample treatment and method performance (LOD, LOQ, recovery) were summarised in tables. This review provides a useful overview of the state-of-the-art for the determination of Se in food and beverages.

2. Metrology and reference ranges

Over the last 30 years, since the establishment of the Consultative Committee for Amount of Substance: Metrology in Chemistry and Biology, the word “metrology” and related terminology and concepts^21,22 have entered the analytical laboratories, supporting the implementation of international standards devoted to harmonise good practices in testing and calibration activities (ISO/IEC 17025, ISO 15189) as well as in other areas related to the work of analytical laboratories (ISO 17034, ISO/IEC 17043). As in our previous Updates, in this Section we report on relevant publications addressing ILCs, new CRMs and investigations of commercially available ones, to document and collect information about other elements or species, not yet certified. In addition, we report updates of reference ranges for chemical elements or their species, to be used in clinical practice or to assess background levels.

2.1 Interlaboratory studies, reference materials and measurement uncertainty

Two interlaboratory studies of note were carried out. Since the request for monitoring Se content in fish tissue may often face low-weight tissue samples of varying lipid content, which challenge the analytical capabilities to quantify tissue Se concentrations, an interlaboratory study²³ was conducted among four laboratories accredited to ISO/IEC 17025 on four tissue samples of various types and weights (two CRMs, dorsal muscle and ovary from pike minnow). The participants employed distinct sample preparation and analytical techniques. The quality of the data generated was compared against predefined data quality objectives for accuracy, precision, and sensitivity. Data analyses focused on evaluating how each protocol’s data quality varied as a function of sample weight and how these relationships differed for different sample types. Each protocol consisted of a different method for drying, tissue preparation, and quantitation (CV-AAS, ICP-MS, LA-ICP-MS) and had different minimum sample weight requirements (50–500 mg). Results showed that data quality tended to decrease with decreasing sample weight, particularly when the PT item weight was less than the minimum weight prescribed by the procedures in place at the participating laboratories. For all laboratories, the minimum sample weight requested in their own procedures was greater than the minimum recommended weight identified in this study. This comparison indicates that laboratories were conservative at estimating their limitations and that data quality for small weight samples can be adequate provided they meet laboratory requirements. One study²⁴ focussed on assessing the interlaboratory precision for Zn measurements in whole blood samples, within the range of 20 to 150 μmol L⁻¹, across various analytical methods, by analysing the EQA data obtained by 4283 laboratories from 2018 to 2022. Participating laboratories were grouped into peer categories based on the technique used (FAAS, GFAAS, ICP-MS, differential-pulse polarography and differential potentiometric stripping, and others). Flame AAS was the most frequently used method for Zn determination. For all samples and method groups, the robust mean and RSD% of the results were calculated. The acceptance criteria for optimal, desirable, and minimum allowable imprecision in Zn estimation (expressed as RSD%) were set as 2.50%, 5.05% and 7.55%, respectively, based on biological variation. During the period 2018–2022, when more than 4000 laboratories participated, a consistently high pass rate (>96%) was reached. The lowest interlaboratory RSDs% during the whole period were achieved with differential-pulse polarography (0.61–1.86%), while the highest values were observed for the group using ICP-MS (5.28–6.20%). The EQA data showed that ICP-MS is the fastest growing method among users, but there is a need for further enhancement in the precision of blood Zn measurement using this method. The authors noted a higher risk of error at lower concentrations of blood Zn, that did not decrease through the implementation of the EQA program.

Recent work for the preparation of new CRMs for element levels focused on beverages such as tea and coffee, to fill the need for such matrices CRMs and support the QC of these worldwide popular beverages. Rastogi et al.²⁵ reported the preparation of a CRM for major and trace elements in black tea powder (BARC-D3201), carried out in accordance with ISO 17034 and ISO Guide 35. They applied an acid MAD followed by quantitation of: Al, Ba, Ca, Cu, Fe, K, Mg, Mn, P, Sr and Zn by ICP-AES; Cd and Pb by GFAAS; As by HG-AFS and Hg by CV-AFS. Tea granules were spiked with known amounts of As, Cd, Hg and Pb since the initial concentrations of these analytes were very low. A total of 14 mass fraction values along with expanded uncertainties were assigned based on the robust estimation of ILC results received from 11 participating laboratories. The newly prepared CRM demonstrated good stability for all the assigned values during long-term and short-term stability tests. The assigned values and expanded uncertainties were between 0.458 ± 0.025 for Ca (% m/m) and 0.60 ± 0.084 mg kg⁻¹ for Hg. The development of a novel coffee bean matrix CRM (KRISS 108-10-023) for elemental analysis was described by Lee et al.²⁶ The CRM was prepared by processing green coffee beans into a dry homogeneous powder. The preparation of the candidate CRM, characterisation procedure, and uncertainty evaluation approach were described in detail. After drying of commercially available green coffee beans, the material was spiked with standard solutions of As, Cd, Hg and Pb. Finally, the material obtained (14.9 kg) was bottled into 1008 amber glass bottles. Among the ten elements tested, eight that were deemed sufficiently homogeneous were certified. Mass fractions for these eight elements (Ca, Cd, Cu, Fe, Hg, Mg, Pb and Zn), measured in the CRM using a primary method—double spiking ID-ICP-MS after MAD with HNO₃, were certified with their relative expanded uncertainty ranges of 0.66% (Cd) to 12% (Pb). Except for Ca, the uncertainties of the measured values were at least three times smaller than the interlaboratory reproducibility estimated using the Horwitz equation. An approach developed for estimating isotopic abundances and molar masses of elements with high isotopic variations for double ID-ICP-MS was discussed in detail.

Two papers reported information obtained for commercial CRMs for non-certified analytes. A study focusing on the determination of technology-critical elements in four seafood CRMs provided mass fractions for new emerging contaminants, thus allowing the application of existing CRMs for method validation in studies dealing with determination of e.g. Ga and In in seafood or other biota.²⁷ Microwave assisted closed vessel digestion using a mixture of HNO₃–HCl–H₂O₂ was applied to varying sample masses, using two different microwave systems. Within this set of CRMs (BCR-668 mussel tissue, NCS ZC73034 prawn, NIST SRM 1566a oyster tissue, NIST SRM 2976 mussel tissue) up to 19 technologically-critical elements, specifically REE and the less studied Ga, Ge, Nb, In and Ta, were determined by ICP-MS/MS. Measurement results were compared to certified values (where available) taking into account the absolute difference between the mean measured value and the certified value, as well as the combined uncertainty of the result and the certified value. If this approach was not possible, due to missing information in the certificate, the certified values were discussed based on recoveries. Combined expanded uncertainties (k = 2) were calculated for each sample replicate based on a simplified Kragten method, considering instrumental repeatability and the intermediate precision of the method. Guo et al.²⁸ reported Cu, Fe and Zn isotope data for nine biological RMs including three plant and six animal matrices. They modified a previously published one-column high precision method to purify Cu, Fe and Zn for isotope analysis. The isotope compositions of Cu, Fe and Zn were determined by MC-ICP-MS in the same laboratory. Several parameters, such as matrix components (chromium, nickel and titanium), mismatch in acid molarities, concentration between samples and bracketing standards and double-spike dosages, were evaluated to ensure the high precision and accuracy of isotope measurement. The long-term reproducibility was better than 0.04‰ for the isotopes of the elements of interest (Cu, Fe and Zn). The CRMs were divided into two groups, one from plants and marine animals with higher δ⁵⁶Fe values (−1.23‰∼0.11‰), δ⁶⁶Zn values (0.28‰∼0.87‰) and relatively homogeneous δ⁶⁵Cu values (0.39‰∼0.53‰), while the other from terrestrial animals and humans had lower δ⁵⁶Fe values (−2.40‰ to 0.06‰), δ⁶⁶Zn values (−0.65‰ to 0.24‰) and relatively homogeneous δ⁶⁵Cu values (−0.33‰ to 2.76‰).

2.2 Reference ranges

Reference values for elements with a physiological role are well established, however, it is worth continuing such investigations to highlight trends for the concentration of elements emerging as potential new environmental pollutants as well as identifying specific areas or groups experiencing nutritional deficiencies or exposures.

In the period covered by this Update, a number of researchers undertook the determination of reference values for trace elements in biological fluids of the general population of specific countries. Perrais et al.²⁹ applied ICP-MS to measure the concentration of 24 elements (Ag, Al, As, Be, Bi, Cd, Co, Cr, Cu, Hg, I, Li, Mn, Mo, Ni, Pb, Pd, Pt, Sb, Se, Sn, Tl, V and Zn) in blood plasma and 24 h urine specimens, collected from 1078 Swiss adults, aged from 18 to 90 years. After considering the influence of sex, age, BMI and smoking habits, the authors defined 50^th and 95^th percentiles for almost all the elements considered. These values were found comparable with those reported in other studies. Whereas whole blood and plasma are common matrices for the detection of elements, erythrocytes are less favoured, possibly because they require more extensive preparation. In Slovenia, another group of researchers³⁰ undertook to measure the concentrations of 24 trace elements (Ag, As, Au, Be, Cd, Co, Cr, Cs, Cu, Ga, Hg, Li, Mn, Mo, Ni, Pb, Rb, Se, Sn, Sr, Tl, U, V and Zn) in blood, plasma and erythrocyte specimens, obtained from a group of blood donors (107 women and 85 men, aged from 18 to 65 years). An aliquot of every blood, plasma, washed erythrocyte sample, calibrator or control sample was mixed with a solution of NH₄OH–Triton X-100–1-butanol–Na₂EDTA and the IS (Bi, Ge, In, Li, Lu, Rh, Sc and Tb). Analyses were carried out using an ICP-MS instrument, equipped with an ORS. The authors established reference intervals for six essential elements (Co, Cu, Mn, Mo, Se and Zn) in the three matrices as the 2.5^th to 97.5^th percentile values, according to the CLSI guidelines.³¹ For most of the non-essential trace elements, a large number of analytical results remained below the method’s LODs, therefore only an upper reference limit at the 97.5^th percentile was defined. Sex, age, seafood consumption, smoking habits and amalgam fillings were confirmed as factors influencing the concentrations of specific elements. As part of the Barcelona Health Survey 2016, another group³² enrolled a representative sample of 240 subjects, aged from 19 to 80 years, to assess the levels of 50 elements, including 26 REEs, in whole blood. Samples (100 mg) were weighed into quartz tubes for MAD with 1 mL 65% HNO₃ prior to analysis by ICP-MS. The authors considered the presence of elements, such as Au, Eu, In, Ru, Ta and Tm, used in the manufacturing of technology-based devices and released into the environment from electronic waste. These were detected in a significant proportion of the subjects (20% to 62%) and higher levels were associated with less affluent economic status. The study provides some insight and reference point about the increasing dispersion of “new” elements in the environment and in biological specimens, which may pose a risk to human health.

Although thoroughly investigated in European and North American countries, exposure to Pb continues to be a health problem in several other parts of the world. The work of Zebbiche et al.³³ addressed the evaluation of the exposure to Pb of the general population in Algeria, through an extensive national survey, involving 3674 subjects (age 3–74 years, 1622 males and 2052 females). Blood samples were collected in PET tubes, specific for trace element determination, and diluted 1 + 19 with a solution of 1% propanol–1 g L⁻¹ EDTA–0.05% (v/v) Triton X-100–2% (v/v) NH₄OH prior to analysis by ICP-MS, using Y as IS. The results showed that exposure to Pb should be of concern in Algeria. The median value for Pb in blood was 22.22 μg L⁻¹, and the 95^th percentile was 71.15 μg L⁻¹ in children and 75.43 μg L⁻¹ in adults. Based on this study, the authors proposed a reference value of 75 μg L⁻¹ for Pb exposure in Algeria and Africa, defined as the 95^th percentile of the concentrations of Pb in blood measured in a representative population sample of 2780 subjects without declared pathologies or chronic medication intake, distributed throughout the national territory. Beside age, gender and residence, specific local factors, such as dietary habits, methods for storage of water and use of the eye cosmetic kohl, influenced blood Pb levels. Another group of researchers³⁴ reported on the trend observed during a 10 year long cross-sectional survey of blood Pb levels collected from a total of 759 children (aged ∼7 years) in Montevideo (Uruguay). Blood Pb levels were measured by FAAS or GFAAS. Over the period of the study, the methods’ LODs were improved from 2.5 μg dL⁻¹ (FAAS) and 2.0 μg dL⁻¹ (GFAAS) in 2009 to 0.4 μg dL⁻¹ (GFAAS) in 2019 to meet the decreasing concentrations observed: from 4.8 ± 2.6 μg dL⁻¹ (2009) to 1.4 ± 1.4 μg dL⁻¹ (2019). The authors stressed the importance of preventive measures to further reduce the risks of exposure to Pb in childhood.

Forensic investigations may require decisions on the presence of potentially toxic elements as the causes of accidental, suicidal, and homicidal deaths. Söderberg et al.³⁵ noted that differences due to post-mortem redistribution and putrefactive changes pose a challenge for the interpretation of the results of determination of chemical elements as compared to reference values for living individuals. Therefore, they assessed the concentration of 68 elements in paired samples of femoral blood and urine of 120 death cases, classified as death by suicidal hanging, autopsied between 2020 and 2021, to serve as reference values. Samples underwent MAD with HNO₃ and HF prior to analysis by double-focusing SF-ICP-MS. Apart from Er, Eu, Ho, Lu, Rh, Sc, Sm, Tb, Th, Tm and Yb, whose concentrations were below the LODs for over 90% of the samples, the arithmetic means of the levels observed in the postmortem samples were on average 80% higher than those reported for living subjects from the same country. This set of data provides the basis to define postmortem reference values for a large number of elements of potential forensic interest. The authors stressed the importance of evaluating potential sources of contamination, such as the elemental content of preservatives, the effect of storage conditions and variability associated with other confounding factors. Another group of researchers³⁶ reported a similar investigation, although limited to measurements of As, Pb and Tl levels, in post-mortem femoral blood samples. Blood (200 μL) was incubated (room temperature, 30 min) with 500 μL of a 20% (v/v) 1-propanol, 0.5% (v/v) tergitol solution and 200 μL of 20% HNO₃. The mixture was then diluted to 5 mL with ultrapure water, centrifuged and the supernatant analysed by ICP-MS. The analysis of post-mortem femoral blood samples from 279 individuals (no cause of death mentioned) gave median concentrations for Pb (14.0 μg L⁻¹) and Tl (0.1 μg L⁻¹) in good agreement with those reported by Söderberg et al.,³⁵ whereas the median value for As was much lower (0.49 μg L⁻¹ vs. 2.25 μg L⁻¹). The authors attributed the difference to higher fish consumption among the population group examined by Söderberg et al.

Although hair represents an attractive matrix for the biomonitoring of exposure to chemical elements and/or nutritional status, its application is hindered by difficulties in establishing reliable reference ranges. This is mainly due to external contamination and other confounding factors, such as hair treatment. Christensen and LaBine³⁷ proposed a novel approach to the determination of reference ranges for 26 elements in hair, by analysing different portions along the length of hair stands (1 to 2 mm, 10 to 11 mm and 39 to 40 mm from the root bulb). Hair samples, collected from 61 adults, were washed with acetone and distilled water, then analysed by LA-ICP-MS using a double ablation technique (see Section 7.2.2 for details). For the statistical analysis, data below the LOD were converted to the respective LOD values and all data were log₁₀ transformed, because of the non-normal distribution of the original datasets. Analysis of variance confirmed significant differences between the element concentrations in different portions of the same hair stand. Reference ranges were calculated for each element in each hair layer as the median and interquartile range and, for the portion closer to the root bulb, as the 5^th to 95^th percentile range. These last values, reported for the first time, were proposed as reference ranges not affected by external contamination for 19 elements, whereas for As, Cd, Co, Mo, Ni, U and V, the levels were below the LOD of the analytical procedure. Beside this drawback, the authors also acknowledged the limited size of the study and the lack of information about potential confounding factors such as age, gender, geographical locations, ethnicity, hair colour and hair treatments.

3. Sample collection and preparation

3.1 Collection, storage and preliminary preparation

The initial step of any analytical procedure is the sample collection. Over the past years, there have been reports of micro-techniques applied to simplify the collection and/or minimise the sample volume required, especially for blood samples.

Particular interest has been paid to dried blood spots, as a simpler and less invasive technique to collect blood samples in monitoring programmes. However several problems hinder its application, such as the risk of contamination, efficiency of the extraction, variability linked to haematocrit levels and matrix effects, which are all reflected in poor agreement in PT schemes. Franco and co-workers³⁸ addressed these issues by organising a comparison among three laboratories, each using their own method to measure Pb in DBS by ICP-MS. They were asked to run parallel tests on the same 15 EQA DBS samples using either calibrators prepared by spiking Pb amounts in the extraction buffer or matrix-matched ones. The latter were obtained as dried spots of control materials and underwent the same extraction procedure as the PT items. In this exercise, the matrix-matched calibration procedure showed superior performance across the range of concentrations studied (from 5.6 μg dL⁻¹ to 38.02 μg dL⁻¹), with 100% of the results within the acceptability limits (±4 μg dL⁻¹ or 10%) and intermediate precision ranging from 5.6% to 8.1%. In addition, good agreement (r = 0.989, mean bias from −0.1 μg dL⁻¹ to 1.5 μg dL⁻¹) was achieved between the results obtained on liquid aliquots or DBS from 39 patient samples with approximate concentrations ranging from 1 μg dL⁻¹ to 80 μg dL⁻¹, analysed in these three laboratories. The authors concluded that, with matrix-matched calibration, ICP-MS can deliver reliable results for Pb from DBS within the range of concentrations covered by the study. However, they also acknowledged the limitations of DBS to provide representative samples in routine use and the need to re-evaluate these methods to address lower Pb concentrations, given the trend towards setting lower action limits.

In our Update last year,¹ we reported the application of volumetric absorptive micro-sampling (VAMS®) for the determination of Pb in blood. This device consists of an absorbent porous polymer, placed into a plastic tip and designed to absorb a fixed volume (10, 20 or 30 μL) of fluid into its pores. Besides being minimally invasive, sampling of biological fluids with this device does not require high level skills or facilities. It can be performed at home and the sample, stored in the plastic bag provided, is kept at room temperature until the subsequent analytical steps. Not surprisingly, this device appears as an interesting alternative to other sampling methods and has attracted the interest of the scientific community. This year, another group of researchers³⁹ undertook an extensive evaluation of this method of sampling for the determination of several elements in whole blood, using four PT items and the RM Seronorm™ Trace Elements Whole Blood. Thirty μL of sample, either liquid or absorbed onto the VAMS®, were placed in 2.5 mL polypropylene vials and digested in a microwave oven with 400 μL of 4.5 mol L⁻¹ HNO₃ and 100 μL 30% (w/w) H₂O₂, then the digests were diluted to 900 μL with UP water, leaving the VAMS® tip in the vial. Analysis was performed by ICP-QQQ-MS, using ⁸⁹Y, selected over other elements, as the IS. A low sample-consumption nebuliser (Meinhard High-Efficiency concentric nebuliser) was mandatory due to the limited volume of sample to be presented to the instrument. The authors observed comparable LODs and LOQs (3× and 10× SD of blank measurements, respectively) for all, but 9 elements (Al, Ba, Co, Cr, Mn, Mo, Ni, Sn and Ti), for which the VAMS® blanks showed a high variability. For the other 15 elements tested (As, Be, Cd, Cs, Cu, Fe, Mg, P, Pb, S, Sb, Se, Tl, V, and U), the VAMS® results agreed well (p > 0.05) with those obtained on the liquid aliquots and with the assigned/reference values (when available). In addition, for these elements, the concentrations of the analytes were stable over two months storage of the RM aliquots in the VAMS®. This work provides a valuable and detailed laboratory assessment of this promising micro-sampling device, highlighting both its potential and some of the challenges for its application, in particular the need for advanced equipment, skilled staff and careful control of contamination of the VAMS® devices. It paves the way to further investigations, that explores the application of VAMS® for trace element analysis in a practical context.

Defining appropriate conditions for the storage of biological samples is a current area of research, since the analysis of stored specimens may provide insights into the development of diseases later in life. Unfortunately, there isn’t a single method catering for all needs or all types of samples, considering the number and variety of parameters that may be or become of interest. Formalin fixation of tissue samples is a common practice in anatomic pathology. Given the difficulty in accessing fresh human tissues, several reports of trace element concentrations determined in FFPE specimens have appeared and were mentioned in previous Updates. They aimed to provide missing information on the background or critical levels of trace elements in human tissues as well as on their potential relation with the development of diseases. However, prolonged storage may alter the concentration of some elements within FFPE tissues. Silva et al.⁴⁰ engaged in assessing the short-term effect of formalin fixation on variations of the element content in human tissues. They applied ED-XRF, PIGE and elastic backscattering spectroscopy to six sets of FFPE human colon tissue samples. Each specimen was divided into nine cubes (approximately 1 cm³), one of them was snap-frozen, and the others placed in vials with a 10% v/v formalin buffered solution. At different times up until 48 h, one aliquot of each specimen was removed from the vial, washed with distilled water, then freeze-dried (−60 °C, 20 Pa, 48 h). The lyophilised samples were powdered and pressed into pellets for presentation to the instruments. The results of element determinations were compared with those obtained on the corresponding frozen aliquot of the same specimen. They noted a significant decrease in the concentrations of Cl and K between 1 h and 3 h of fixation. On the other hand, both Na and P levels increased within the first 15 min of fixation, possibly due to the uptake of these elements from the buffered formalin solution into the tissues.

Preliminary preparation of samples for analysis may involve steps such as washing to remove external contamination as well as procedures to improve homogeneity (e.g. crushing, pressing or milling), that may affect the content or the distribution of the elements under investigation. In this year’s Update, we report three publications of note in these areas. As part of their research on As biology, Feldmann and co-workers⁴¹ applied non-destructive techniques (XFM and XANES) to toenail samples, to investigate As speciation and the spatial distribution of As and other elements. To remove external contamination, toenail samples underwent a four-step procedure. This involved stirring the samples in an ultrasonic bath for 20 min with, in sequence, acetone, ultrapure water and 0.5% Triton X-100, then thoroughly rinsing with ultrapure water and overnight drying in a oven at 60 °C. To assess the efficacy of this preliminary preparation, the authors compared the results obtained on cleaned toenail samples with paired sets, collected from the same 13 subjects, analysed without further cleaning. Analysis by XFM identified the distribution patterns of the trace elements and distinguished between endogenous and exogenous levels. The authors concluded that the cleaning procedure allowed the removal of most of the external contamination, with no substantial effect on the trace elements, such as As, Ca, Cu, S, and Zn, that accumulate endogenously. However, they also noted residual external contamination, that may be the cause of large within-subject variabilities for elements such as Co, Fe and Mn, when destructive analytical techniques are applied, as these cannot distinguish between endogenous and exogenous levels of the elements. They suggested that the simultaneous determination of non-essential elements such as Ti and Rb might help to assess the amount of residual external contamination. Sample preparation for LIBS analysis was the subject of two papers. Lin et al.⁴² investigated the effect of different methods of soft tissue sample preparation (rapid freezing, fresh slicing, drying, and pressing) on the stability of LIBS signals. Using pork tissue, they observed reduced S/N when samples were dried or pressed, as well as improved stability of the Ca, Na, Mg, and CN bands within the sample spectra for pressed samples. The authors observed that the improvement was linked to the reduction of both the uneven distribution of elements within the pork tissue and to the fluctuation of plasma temperature across the sample. Dried and pressed samples achieved better classification performances in SVM models, as well as better ROC curves and area under the curve values. In another paper⁴³ the authors compared five methods of sample preparation for LIBS analysis, aimed to classify oral liquids intended for children according to the compliance of their Ca, Fe and Zn content with the values stated in the label of the product. Using a model built with XGBoost, they identified filter paper adsorption as the more suitable one among those studied (filter paper adsorption, filter paper adsorption with elemental Cu, addition dropwise to glass slides, addition dropwise to glass slides with elemental Cu and gel preparation). Besides a classification accuracy of 91.25%, they showed that filter paper adsorption was simpler, more efficient, and less affected by the colour of the sample than the other approaches tested.

3.2 Digestion, extraction and preconcentration

Once again, the current review period, has seen a significant increase in the numbers of publications relating to digestion, extraction and preconcentration. This is especially true for the topics of preconcentration by LLE or SPE in methods for the determination of trace elements in food. These, together with the much smaller number of papers describing the analysis of clinical materials, are summarised in Tables 1 and 2, which are discussed in more detail later in this section. Increasingly, researchers are discussing the greenness (the environmental and sustainability characteristics) of their new methods, as calculated by one or more green metric scales, and the topic is highlighted below. However, greenness is no substitute for accuracy, and so only publications in which the results of the analysis of a CRM (where available and appropriate) were reported are included in this Update. Spike recoveries are considered acceptable when a CRM is not available. A significant number of publications do not provide important details of the sample preparation procedure. Often omitted is the final volume obtained after solubilising a solid sample prior to implementation of the next stages of the method. This makes it impossible for readers to ascertain the LOD in the original sample, as almost always LODs are given in mass/volume units for the solution produced by the first step of the analysis. Many publications do not mention whether (or how) the moisture content of the sample was accounted for, an important step if results are to be compared with those on the certificate of a CRM, which are always given on a dry weight basis. Readers are also interested in the throughput of a method, but the time required for an analysis is rarely readily available; readers have to add up the times given for each of the individual steps. Researchers (and referees of submitted manuscripts) are reminded that a single-cycle of the alternating variable search method is not an acceptable optimisation strategy.

Table 1 Preconcentration by liquid- or solid-phase (micro) extraction

Analyte	Matrix	Technique	Extraction mode/reagents	Procedure/comments	LOD in μg L^−1 (unless stated otherwise)	Validation	Ref.
Ag, Cu, Pd, Pt	Pharmaceuticals (ibuprofen suspension, effervescent, and tablet)	ICP-OES	DLLME with ionic liquid (1-butyl-2-diphenylphosphino-3-methylimidazolium hexafluorophosphate)	To the sample (0.5 g) were added 20 mL of water. Mixture stirred (10 min) and filtered (0.2 μm). If adjustment of the pH and ionic strength required, then to 4 mL were added 100 μL of buffer and 500 μL of NaCl (5% m/m) solution. Next, 230 mg of the ionic liquid, dissolved in 500 μL of methanol, was injected. Following separation by centrifugation (4000 rpm 5 min), the extractant was diluted with acetic acid containing 250 μg L^−1 of Mn and Y as IS (final volume not given). Introduction to the spectrometer was by FI (25 μL valve loop). The carrier was either 1% (m/m) HNO3 or air. Enrichment factors ranging from 14 to 70 were reported, but as these were defined as the ratio of the analyte concentration in the organic phase to the initial concentration in the sample, they are meaningless for solid samples. None of the analytes were found in any of the three samples	0.2–2	ISC Science, Oviedo, Spain CRM-DW1 (drinking water) Ag and Cu spike recoveries	252
Ag, TiO2 NPs	Water (tap)	spICP-MS	DLLME into Triton X114 + 1,2-dichloroethane (49 + 1)	To 10 mL of tap water was added 1.0 mL surfactant extractant mixture and the mixture vortexed (1500 rpm 30 s). After separation by centrifugation (3500 rpm 5.0 min), the recovered lower phase was diluted to 1 mL with 1.0% (w/v) glycerol. The extracts were again vortexed before spICP-MS measurements. A real sample was analysed but the results were not given. Spike recoveries ranged from 44 to 58%. The ionic forms of the elements were not extracted	Number concentration 6.06 × 10^6 L^−1 (Ag), 2.85 × 10^6 L^−1 (TiO2); size 14.1 nm (Ag), 55.5 nm (TiO2)	Spike recoveries	170
Al	Milk (cow)	ICP-OES	LLME of complex with 8-HQ into DES (ChCl + oxalic acid)	To 50 g cow milk was added 0.2 mol L^−1 acetic acid, followed by ultrasonication (10 min 60 °C), cooling to room temperature and centrifuging (4000 rpm 10–20 min). The resulting whey milk sample was treated with 1.0% TCA and NaCl (1.0 mL of each), to precipitate whey protein, followed by vortexing (2 min) and centrifuging (10 min 3500 rpm). The clear upper phase was separated and to 25 mL (pH adjusted to 6) were added 0.2–0.5 mL (optimum value not given) of 0.1% (w/v) 8-HQ solution, 200 μL (DES) and 0.5 mL of tri-hydrofluoric acid (maybe a misprint for trifluoroacetic acid). The mixture was sonicated (50 °C 5 min) then centrifuged (3500 rpm 5 min). The upper layer containing the enriched DES phase was separated and diluted with acidic EtOH (250 μL) and analysed. The EHF was 50, and Al was found in all samples	0.042	CRM NIST SRM 1643e (water) and spike recoveries	253
Al	Beverages (milk, fruit juice)	ETAAS	CPE of complex with 2-hydroxy-5-p-tolylazobenzaldehyde in the presence of Triton X-114 in DES (histidine + glycerol)	Sample (3 mL) was diluted to 10.00 mL and proteins precipitated with 1.0 mL 20% TCA and separated by centrifugation. In the rapid synergistic method, to 10.0 mL of sample were added, ligand (0.15 mmol L^−1), and Triton X-114 (0.08%) at room temperature. The DES was then added with stirring (5 min). After centrifugation (4000 rpm 2 min), the bottom aqueous layer was removed and the remaining surfactant layer was then diluted with EtOH (volume not given). In the CPE procedure, reaction mixture heated (45 °C for 12 min). After centrifuging, the mixture was cooled (3 min) when a lower viscous cloud point layer formed. The top aqueous layer was removed and the organic phase dissolved in EtOH (volume not given), but 500 μL was injected into the spectrometer. Enhancement factors of 40 and 35 were reported but not defined. Preconcentration factors of 20, suggest that the final volume was 0.5 mL	0.1 (CPE) and 0.01 (rapid synergistic CPE)	CRM Shanghai Botong Industrial Science Co. SLRS-5 (milk) and spike recoveries	254
As	Grain (rice, wheat)	ETAAS	LLE as ion-pair with tetraoctylammonium into DES (tetraoctylammonium bromide and hexanoic acid)	Ground (1 mm mesh) sample (0.5 g) MAD with HNO3 and H2O2. After adjusting to pH 3, the final volume was made up to 25 mL. To a 12 mL subsample were added 100 μL of DES and the mixture vortexed (2 min). Phase separation was by centrifugation (6000g for 2 min), and 50 μL of the extract was diluted with 50 μL of isopropyl alcohol, and 10 μL introduced into the spectrometer. A 57-fold enrichment factor was reported, but no basis for the parameter was given. The analyte was not found in either of the two wheat or two rice samples. As the LOD in the solid was 0.5 μg kg^−1, the rice is extremely unusual, as the evidence is that all rice contains (much) higher concentrations than this. The text indicates that hexanoic acid was used, but this is not listed under “reagents” and the conclusion suggests that this should be heptanoic acid. The As concentrations were <LOD of the ICP-OES method (1 μg kg^−1)	0.01	CRM Petroanalytica Russia RM 122341-016 (rice grains), spike recoveries and comparison of results with those of an ICP-OES method	255
As	Fish (fresh, canned), water (river, tap)	HG-AAS	LLE of complex with 4-((2-hydroxyquinoline-7-yl)diazenyl)-N-(4-meth-ylisoxazol-3-yl)benzene sulfonamide (HDNMBA) with tertiary amine 4-(2-aminoethyl)-N,N-dimethylbenzylamine (AADMBA) as disperser into 1-undecanol	Water: 100 mL evaporated to 5 mL. Fish: 0.5 g MAD, As^V reduced to As^III (KI and ascorbic acid), final volume 50 mL. To 10 mL were added HDNMBA, buffer and 100 μL AADMBA (as disperser), followed by 60 μL of 1-undecanol. The floating drop was solidified by cooling in an ice bath. Finally, 80 μL was taken for analysis (the additional 20 μL was not explained). The EHF was 267. As was found only in the fish samples	0.005	CRM GSBZ GSB07-3171-2014 200 455 (fresh water), BCR-627 (fish muscle) and spike recoveries	256
As	Rice (basmati rice, polished, unpolished), water (ground, tap, river, lake)	ICP-OES	SPE on coordination polymer gel based on zirconium^IV and 2-thiobarbituric acid (ZTA)	The ground samples (1.0 g) digested (hot plate) with a mixture of HClO4, HNO3 and H2SO4. Final volume 100 mL, to 50 mL of which was added 10 mL of hydroxylamine hydrochloride to reduce As^V to As^III, and after pH adjustment (pH 5), the volume was made up to 100 mL. Then 40 mg of ZTA was added and after 90 min, the ZTA was then recovered, washed with distilled water and treated with 5 mL of 0.1 mol L^−1 NaOH for 15 min (no other details provided). As was found in all samples	0.038	Comparison of results with those of (a) a spectrophotometric method and (b) a spectrofluorimetric method and spike recoveries	257
As^III, As^V, total As	Rice	ETAAS	DLLME of DDTP complexes into DES (L-menthol and octanoic acid)	Sample mass 0.5 g hot-plate digested with 0.28 mol L^−1 HNO3 (no conversion of species) final volume 50 mL. Into 10 mL of which were injected 600 μL acetone containing 60 μL DES and 10 μL DDTP. After centrifuging (5000 rpm 5 min) the extractant (containing As^III) was collected on the surface and solidified in a freezer (few minutes). The solidified DES was separated, melted and 20 μL taken for analysis. To determine As^V, it was reduced with Na2S2O3 and KI. Parameters relating to the preconcentration were given but not defined. Both species were found in all six samples	0.1 μg kg^−1	CRM NMIJ CRM 7503a (rice flour)	81
As (inorganic species)	Olive oil	HPLC-HG-FAAS	Spray DLLME into water	The extraction solvent (water, volume not specified) was sprayed into 7.0 mL of sample, in an inverted centrifuge tube, followed by vortex mixing (45 s). Centrifugation (3.0 min at 25 000g) separated the phases and 120 μL of aqueous phase was injected directly into the HPLC-HG-FAAS system, while 60 μL was subject to reduction of As^V to As^III (by heating in presence of L-cysteine and HCl). It is not clear why HPLC was needed, as the post-column HG conditions gave a signal only from As^III	3 μg kg^−1 (As^III), 10 μg kg^−1 (t-As)	Spike recovery	258
As^III, Cd, Cr^III, Cu, Pb, V^V	Rice, milk powder	ICP-MS	SPE on poly (N-methacryloyl-L-alanine acid) grafted tartaric acid-crosslinked chitosan	Sample mass 0.2 g MAD (HNO3 and H2O2) final volume not given but could be 20 mL. This volume passed through 20 mg extractant at 1.5 mL min^−1. Elution with of 2.5 mL 0.5 mol L^−1 HNO3. Neither As nor Cd found in rice, and only Cu and Cd found in the milk powder. The researchers list As^III. as an analyte, but this cannot be correct as the digestion procedure would convert all As to As^V. Interestingly, they got the right answer for arsenic in the CRM. The LOD for As in the solid was 0.23 μg kg^−1, so this is another unusual rice sample	1–4 ng L^−1	CRM Center of National Standard Reference Material of China. GBW10043 (rice) and GBW 10024 (scallop)	259
Cd	Food (apples, spinach, salad, tomatoes, onions, oats, corn, aubergine, wheat, rice, and mushrooms), water (tap, well, bottled, hot and cold spring)	FAAS	DSPME on polyvinyl benzyl xanthate (PvbXa)	Food samples were dried, ground and 0.5 g subject to MAD with HNO3 and H2O2 with final volume of 25 mL. For extraction, to 10 mL of the digest or 200 mL of water samples were added 125 mg of PVBXa, the pH was adjusted to 4.5 and the mixture vortexed (200 rpm 8 min). The phases were separated by centrifugation (5 min at 4000 rpm) and aqueous phase removed. Finally, 1250 μL of EtOH was added and the mixture vortexed (120 s). Cd found in all 13 food samples and in tap and well waters. A PF of 160 was reported (based on the ratio of volumes for the water sample) and an EHF of 100 (based on calibration slope ratio)	0.06	CRMs INCT-TL-1 (tea leaves), NIST SRM 1547 (peach leaves), and NIST SRM 1643e (water) and spike recoveries	260
Cd	Food (green tea, rice flour, corn, tomato, spinach, mushroom, black tea, parsley, lettuce, tuna fish), water (tap, well, spring, waste)	FAAS	SPME on maleic anhydride–styrene–glycidyl methacrylate (MA-St-GMA) terpolymer	Sample mass 3 g MAD with HNO3 and H2O2 final volume 10 mL. To which 40 mg of extractant added, shaken 19 min. Centrifuged for 2 min, decanted, analyte dissolved (vortex 1 min) in 0.9 mL acidic EtOH. Injected into spectrometer. EHF 186 reported, but cannot be correct as the maximum ERF is 11. Analyte found in all samples except well water, tap water, green tea and lettuce	0.03	CRM NIST SRM 1643e (trace elements in water), 1570a (spinach leaves), 1573a (tomato leaves) and spike recoveries	261
Cd	Food (corn, rice, mushroom, spinach, leek, tomato, dried tea, walnuts, almonds, and wheat), water (spring waste, well, tap)	FAAS	DLLME as complex with tetrasodium iminodisuccinate with magnetic ionic liquid, (P6,6,6,14^+)3(GdCl6^3−)	Water samples 100 mL evaporated to 10 mL. Solid samples mass 1.0 g MAD (HNO3 and H2O2) final volume 125 mL. To 10 mL were added iminodisuccinate, MIL, acetone, repeated injection from syringe into solution. Magnetic separation, dilution to 1 mL with acidic ethanol. ENF 106 reported but makes sense only for water samples. Analyte found in six out of eight water samples, and in all food samples	0.17	CRM NIST 1570a (spinach leaves), 1573a (tomato leaves), 1643e (water)	262
Cd, Co, Ni	Water (spring, mineral, factory process)	FAAS	Magnetic DSPE on pine cones (Pinus pinea) derivatized with Fe3O4, with elution by HNO3	To 50 mL sample was added 0.100 g solid magnetic adsorbent and the mixture sonicated (2 min). Following magnetic separation, the analytes were dissolved in 2 mL 1.0 mol L^−1 methanolic HNO3. None of the analytes was found in any samples, apart from Cd in the factory process water	2.3 (Cd), 19 (Co), 12 (Ni)	CRM National Metrology Institute of Turkey UME 1204 (elements in wastewater) and spike recoveries	263
Cd, Co, Cu, Hg, Ni, Pb, Zn	Milk	ICP-OES	LLE into DES (ChCl, sulfosalicylic acid, and 8-HQ)	To 5 mL milk sample were added 1.25 g NaCl, 2 mL ACN, and 100 μL DES. After shaking and centrifuging (4000 rpm 5 min), 1.0 mL of the upper phase (1.2 mL) was injected into 5 mL 5%, w/v NaCl at pH 4. Following sonication and centrifuging (4000 rpm 5 min), the sedimented phase was injected into the spectrometer. No parameters relating to preconcentration were given, though extraction recoveries were 70–80%. Cu and Zn were found in all 10 samples	0.04–0.10 ng mL^−1	Results presented for a whole milk “CRM” but no details given. Spike recoveries	264
Cd, Co, Cu, Mn, Ni, Pb, Zn	Urine, water (sea, ground, river)	ICP-MS	SPE on N-isopropylacrylamide photocurable resin with processing by 3-D printing to fabricate a column with a {H^+}/temperature dual-responsive monolithic packing, stacked as interlacing cuboids, elution with HNO3	The idea was that the monolithic packing swelled owing to its hydrophilic/hydrophobic transition and electrostatic repulsion among the protonated groups by (a) cooling to 10 °C and (b) the use of 0.5% HNO3 resulting in smaller interstitial volumes improvements in the elution peak profiles leading to lower LODs. Visual examination of the peaks suggest that about 167 μL were need to reach the peak maxima, giving a PF of 3. All six analytes were found in all samples except for Cd in river and ground waters, and Pb in urine. The table of results of the analyses of the CRM contains some errors in the calculation of the relative measurement errors	0.2–7.2 ng L^−1	CRM NRCC CASS-4 (sea water), SLRS-5 (river water), NIST SRM1643f (fresh water), and Seronorm L-2 (trace elements in urine) and spike recoveries	265
Cd, Cr, Pb	Honey	ETAAS	DSPME with 3-aminopropyltriethoxysilane modified chrysin (CHY-APTES)	Honey (10 g) was dissolved in 100 mL water with pH adjustment to 5. To 45 mL were added 120 mg CHY-APTES dispersed in 2 mL of ethanol and the mixture vortexed (900 rpm 3 min). After separation by centrifugation (no details given), the analytes were dissolved in 0.5 mL 0.1 mol L^−1 HNO3 (ultrasonication 15 min). Preconcentration factors of 49, 9.5 and 111 were reported for Cd, Cr and Pb, respectively but no explanation given for the basis of the calculation of these numbers. Only Cr and Pb were found in one of the three samples	57 (Cd), 350 (Cr), 44 (Pb) μg kg^−1	Spike recoveries	266
Cd, Cu	Fruit juice (mango, orange, apple, cherry), water (surface, urban, well)	FAAS	DSPME of DDTC complexes onto particles formed in situ from a solution containing ibuprofen and ChCl, followed by DLLME	To 7 mL sample were added 150 μL complexant and 50 μL of the ibuprofen and ChCl solution. After vortexing and centrifuging, the particles were dissolved in 1.25 mL of MeOH. Then 275 μL of 1,1,2,2-tetrachlorethane was injected and the mixture dispersed in 5 mL water. After centrifuging, the water-immiscible phase was taken for analysis. The ERF was 32. Cu was only found in the surface water and well, and Cd was found only in the well water	0.3 (Cu), 0.87 (Co)	CRM LGC SPS WW-2 (waste water) and spike recoveries	267
Cd, Cu	Fruit juice (pineapple, peach, grape), water (fountain, dam, urban)	FAAS	SPE on MOF, NH2-UiO-66(Zr), with dissolution in DES (ChCl and ethylene glycol)	To 6 mL samples as received, 9 mg of MOF sorbent was added, after vortexing (4 min), centrifugation (6 min 6000 rpm). The supernatant was discarded and 250 μL DES was added. After further vortexing and centrifugation, two 100 μL portions of the resultant solution were injected. Cd not found in urban water, peach juice or grape juice, Cu not found in urban water or peach juice	0.19 (Cd), 0.37 (Cu)	CRM LGC SPS-WW2 (waste water)	268
Cd, Cu	Food (rice, cocoa powder, garlic, scallion, parsley, spinach, radish, potato, carrot, lettuce, bulgur wheat, green pepper, tomato, leek, walnut), soft drinks (iced tea, cherry juice, apple juice), water (tap, mineral, drinking)	FAAS	DSPME on a magnetic polystyrene-b-poly dimethyl siloxane block copolymer (pores filled with magnetic iron oxide)	Solid samples (1 or 2 g) were hot plate digested with HNO3 and H2O2 and made up to 50 mL. The pH was adjusted to 6.0 and 250 mg of the magnetic PSt-PDMS material added and mixed (orbital shaker and vortexing 10 min). Following magnetic separation, the analytes were extracted with 1 mL of 0.5 mol L^−1 HCl. Both analytes were found in all food samples. Cu was found in all soft drink and water samples, but Cd was found only in the tap water	1.7 (Cd), 0.8 (Cu)	NIST SRM 1570a (spinach leaves) and CRM BCR-032 (phosphate rock) and spike recoveries	269
Cd, Cu, Pb	Water (drinking, ditch, river)	FAAS	LLME of complexes with APDC into DES (menthol and heptanoic acid)	The procedure, carried out automatically in a syringe containing a floating magnetic stirrer connected to a multi-position valve, started with the sequential aspiration of 300 μL 0.1% m/v APDC solution, 4000 μL sample, 200 μL DES, and 500 μL air. Following stirring (30 s) phase separation occurred in 60 s. Finally, the piston was moved upwards delivering the extract to the nebulizer of the spectrometer. The EHFs were 31, 48, and 55 for Cd, Cu, and Pb, respectively, Cd was found in ditch water, Cu was found in all three samples, but Pb was not found in any	0.54 (Cd), 0.78 (Cu), 4.5 (Pb)	CRM BCR 278-R (trace elements in mussel tissue), NIST SRM 1643e (trace elements in water)	270
Cd, Cu, Zn	Honey	FAAS	LLME (salt induced) of complex with 8-HQ into DES (ChCl and 4-chlorophenol) with ACN and NaCl	Honey (30 g) dissolved in 100 mL water. To 27 mL were added 7 g NaCl, 200 μL complexing agent, and 6.5 mL of ACN containing 375 μL DES. The homogeneous solution was passed through a column of 7 g of solid NaCl, when droplets of the mixture of the DES and ACN formed at the interface of the solid and solution, and collected as a separated layer. Only 8 mL of solution was loaded and 1.2 mL of organic phase was collected and injected into 5 mL H2O. After centrifugation, three aliquots (100 μL) of the settled phase (306 ± 5 μL) were injected into the spectrometer. The ERF was around 80. Five samples were analyzed: Cd was found only in one, Cu in three and Zn in two	0.35–0.48 μg kg^−1	Spike recoveries	271
Cd, Fe, Mn, Pb, Zn	Coffee (Turkish, Dibek, espresso, filter, granular, gold, cappuccino)	FAAS	DLLME of lysine complexes in DES (phenyl acetic acid and dimethylglycine)	Sample mass 2.5 g MAD (HNO3 and H2O2) final volume not given but could be 10 mL, the volume taken for extraction, to which were added 0.9 mL 10^−4 mol L^−1 serine, followed by 1.25 mL DES, 0.3 mL CH2Cl2 and vortexing (75 s). The DES phase was collected on the aqueous solution after waiting for 1 min and made up to 3 mL with EtOH. The ERFs ranged from 105 to 175, numbers that are hard to reconcile with the volumes involved. Fe, Mn and Pb were found in all samples, Cd was found in espresso, granular and gold, and Pb was found in espresso, granular and cappuccino	0.015 (Cd), 0.01 (Fe), 0.27 (Mn), 0.18 (Pb), 0.02 (Zn) μg kg^−1	CRM SRM-1548a (typical diet), SRM 2385 (slurried spinach), and GBW07605 (tea)	272
Cd, Fe, Pb, Zn	Baby foods (mixed vegetables, mixed fruit, vegetable soup with bone broth added, organic custard)	FAAS	LLE of complexes with L-proline with a with switchable-hydrophilicity solvent N,N-dimethylcyclohexylamine (DMCHA)	Sample mass 0.1 g MAD (HNO3 and H2O2) final volume not definite but could be 15 mL. To 10 mL (pH adjusted to 5.5) 375 mmol L^−1 of L-proline, 750 μL of DMCHA solution (as the extracting phase) and 0.9 mL of 2 mol L^−1 acetic acid, were added. Then 0.45 mL of 10 mol L^−1 NaOH solution was added and the solution vortexed (100 s). Following centrifugation (4000 rpm 2 min), the aqueous phase was removed and the remaining phase injected into the spectrometer. Enhancement factors (not defined) of between 106 and 133 were reported. Fe and Zn were found in all 13 samples, Cd was found in five and Pb was found in six	0.6 (Cd), 0.009 (Fe), 0.2 (Pb) 0.03 (Zn) ng L^−1	CRM INCT-MPH-2-(mixed Polish herbs), GBW10011 (wheat), and GBW10010 (rice)	273
Cd, Hg and Pb	Serum, urine	ICP-MS	SPE on monolithic column of poly(sulfhydryl)-modified covalent organic framework (1,3,5-tris(4-aminophenyl) benzene-dimethylterephthalaldehyde-2,5-bis(2-propynyloxy)-terephthalaldehyde) with high internal phase emulsion	Sample volume 1 mL MAD (HNO3 and H2O2) final volume 2 mL, 0.5 mL of which was pumped onto a 2 cm capillary monolithic column, with elution in 50 μL of thiourea (4% m/v). Column cleaned with 50 μL dithiothreitol followed by 100 μL H2O. Analytes found in both samples, but results for only one element presented for each of the three CRMs	6.1 (Cd), 44 (Hg), 28 (Pb)	CRM of GBW(E)091031, GBW(E)091030, GBW09104 (freeze-dried human urine)	274
Cd, Pb	Oil (sunflower, colza, fish), butter	FAAS	DSPE of DDTC complexes on powdered walnut green husk followed by DLLME into tetrachloromethane	Sample (1 g) diluted to 5 mL with ethyl acetate. Extracted with 0.2 g of husk (vortex, centrifuge) and elute into 1.5 mL methanol (vortex, centrifuge). 250 μL of tetrachloromethane added and mixture injected into water. After phase separation (centrifuge), 100 μL injected into microsample system. ERFs, based on concentration ratios in the ethylacetate to tetrachloromethane phases, were 21 and 23 for Cd and Pb, respectively. Both analytes found in all samples, except Pb in butter	0.12 (Cd), 0.32 (Pb)	CRM Analytichem EnviroMAT HU-1 (used oil) and spike recoveries	275
Cd, Pb	Food (fish), water (tap, river)	ICP-OES	SPE on a phosphorylated aluminum oxide disk with elution by HCl	Sample mass 1.0 g MAD (HNO3 and H2O2) final volume 10 mL. This may have been further diluted to 100 mL, which was probably the volume of water samples taken. After pH adjustment, the SPE disk (300 mg) was loaded at 6.0 mL min^−1 and the sorbed metal ions were eluted with 3 mL, 0.5 mol L^−1 HCl. Both analytes were found in one fish sample and the river water sample, but only Pb was found in the tap water	0.02	CRM NIES 10c (rice flour) NIST SRM 1572b (citrus leaves) and spike recoveries	276
Cd, Pb	Hair (human)	ICP-OES	CPE (in-syringe) of complexes with 8-HQ into CTAB-rich phase	Sample mass 0.1 g hot-plate digested (HNO3 and H2O2) final volume 50 mL (adjusted to pH 4.5). 2.0 mL was drawn into a syringe followed by 2.5 mL of 0.002 mol L^−1 8-HQ solution and 3.0 mL of 0.002 mol L^−1 CTAB solution. Following manual shaking (2 min), 2 mL of 0.2 mol L^−1 KI was introduced into the syringe to separate the phases (shaking 1.0 min, standing 5 min), Finally, the analyte-rich phase was dissolved in 1.0 mL of EtOH. Analytes were found in 10 real samples	3 (Cd), 62 (Pb)	CRM BCR 397 (human hair)	277
Co	Food (cabbage, cucumber, spinach), water (tap, mineral, river, sea)	FAAS	Magnetic DSPME of 1-nitroso-2-naphthol complex on toner powder	Sample (1.0 g dried and ground) were hot-plate digested (HNO3 and H2O2) and made up to 50.0 mL. The pH was adjusted to 2.0 and 1.0 mL of 1-nitroso-2-naphthol solution (0.01 mol L^−1) added followed by 100 mg of toner powder. After sonication (2 min), the dispersed toner powder was collected magnetically (and the aqueous phase discarded), and Co extracted with 1.0 mL of 1.0 mol L^−11 HNO3 in ethanol (sonication for 1 min). Finally the sorbent was separated magnetically. Co was found only in the spinach	0.7	LGC Standards (UK) CRM SPS-SW2 (surface water), and Sigma Aldrich QC-1187 and spike recoveries	278
Co	Food (broccoli, spinach), water (sea, river, mineral, tap)	FAAS	DSPME of complex with 1-nitroso-2-naphthol on benzoic acid	1 g of dried ground sample digested with HNO3 and H2O2 (multi-step, hot plate) final volume 50 mL. To which (after pH adjustment to pH 6) were added 1-nitroso-2-naphthol, and 50.0 mg of benzoic acid after shaking at 60 °C, the benzoic acid dissolved. On cooling in an ice bath for 2 min, the solution became cloudy as the benzoic acid precipitated. The cloudy mixture was centrifuged at (4000 rpm for 7 min), the aqueous phase discarded and the solid dissolved in 1.0 mL of ethanol and injected into the spectrometer. The preconcentration factor was 50. Co was detected in both foods and in river and sea water	0.55	CRM LGC Standards SPS-SW2 (surface water), Sigma Aldrich QC-1187 (unknown)	279
Co	Food (tea, rice), water (tap, river, geothermal)	CS-FAAS	SPE on activated C from grape stalk	Food samples (0.5 g) were hot-plate digested with HNO3 and H2O2 final volume not given. 750 mL passed through 400 mg activated C column at 4 mL min^−1 (3.1 h) and eluted with 5 mL of 2 mol L^−1 HCl. The ENF was 145. Co was found in all samples, except the black tea	0.27	CRM NIST SRM 1640a, (trace elements in water) and spike recoveries	280
Co, Ni	Fruit juice (peach, orange), pea, lentil, rice, milk (raw, powder), water (waste, sea, tap)	FAAS with SQT	Solidified floating drop DLLME of PAN complexes into 1-dodecanol) with DES (tetrabutylammonium bromide : acetic acid) as disperser and ACN as demulsifier	Fruit juices and raw milk (10 mL) were digested with conc. HNO3 and H2O2, filtered and diluted to 10 mL. Solids were dried and ground and 1 g digested with conc. HNO3 and H2O2 and diluted to 10 mL. After adjusting to pH 5 and adding PAN, the solution was injected into the extraction mixture of 75 μL 1-dodecanol (extraction solvent) and 250 μL of DES (disperser) and the mixture sonicated (1 min). Then, 250 μL of ACN was added to break up the emulsion and the mixture transferred to an ice bath. After 2 min, the upper solidified layer was separated, diluted 100 μL with ethanol. Enhancement factors (based on calibration slope ratios) were 95 and 94 for Co and Ni, respectively. Both analytes were found in all 10 samples, except for the tap water and Co in milk	0.2 (Co), 0.4 (Ni)	CRM NIST SRM 1643b of (trace elements in water), SRM 1568a (rice flour), and FAPAS T07370 QC material (brown rice) and spike recoveries	281
Cr	Medicinal plant infusions (calendula, chamomile, carqueja, green tea, leather hat, melissa, Passiflora), water (not specified)	FAAS	DSPME of complex with PAN on cetyltrimethylammonium perchlorate particles	To 10.0 mL of sample were added 500 μL PAN, 500 μL buffer (pH 9.5), and 900 μL of the surfactant CTAB. After shaking 600 μL of NaClO4 solution was added and the volume made up to 12.0 mL with water. The resulting cloudy solution was vortexed agitation then centrifuged (7000 rpm 20 min). After separation, the Cr was dissolved in 1.0 mL of HCl. No parameters relating to preconcentration were given. No analytes were detected in the infusions or four water samples	1.7	CRM APS-1071 (drinking water) and spike recoveries	282
Cr	Food (apple, apricot, banana, grape, hazelnut, walnut, tomato, plum, mulberry, pear, fig, rosehip, broccoli, garlic, mushroom, onion, potato, spinach, tea, egg, fish, honey, meat, salami, and sausage)	ETAAS	DSPME copolymer composed of polystyrene and polylinoleic acid with the sodium salt of iminodiacetate	Sample (1 g) MAD (HNO3 and H2O2) final volume 10 mL. To which was added 40 mg of extractant. After mixing and centrifuging, the Cr was eluted in 1.5 mL HCl. The EHF was 73. Cr was found in all 26 samples	0.003	CRM NIST SRM 1515 (apple leaves), INCTTL-1 (tea leaves)	283
Cr^III and Cr^VI	Beverages (orange juice, soda, cola, apple juice, sprite), water (well) and food RMs (tea, milk powder, scallop, water, wheat flour)	ETAAS	LLE dual droplet immersion. HTTA-chloroform for Cr^III and APDC-chloroform for Cr^VI	Beverages vortexed (2.0 min), and then diluted five-fold 0.1000 g of CRMs MAD digested but no details given. Final volume could be 20 mL. Analytes were extracted from 20 mL into 10 μL drops (stirring 25 min 35 °C) diluted to 50 μL and 10 μL introduced to spectrometer. ENFs were 354 and 326 for Cr^III and Cr^VI, respectively. Cr^III was detected in all samples and RMs except the water RM. Cr^VI was detected in the well water, and the water, milk and scallop RMs	3.0 (Cr^III), 4.1 (Cr^VI) ng L^−1	CRM Institute for Reference Materials, State Environmental Protection Administration (Beijing), GSBZ50027-94 (water), IGGE GBW10052 (tea leaves), GBW 10017 (milk powder), GBW10024 (scallop) and GBW10011 (wheat flour)	284
Cr^III	Tea (green infusion), water (tap)	ETAAS	SPE on ion-imprinted polymer, Cr^III-1,10-phenanthroline complex plus (a) styrene, or (b) (styrene and 4-vinylpyridine	Tea (one bag) extracted with 200 mL of hot water (80 °C 5 min). Tap water passed though cation-exchange membrane. Up to 20 mL passed through column containing 0.1 g polymer at 1 mL min^−1, rinsed, elution with 1 mL 0.1 mol L^−1 EDTA. PF was 20. No analyte found in either sample	0.02	CRM NIST SRM 1643e (water)	285
Cr, Pb	Vegetable oil (sunflower, olive)	FAAS	Extraction by emulsion breaking followed by DLLME of DTZ complexes into CHCl3	To 7.0 g sample was added 0.5 mL of surfactant (1.5% and 3.5% Triton-X 100 for Cr and Pb, respectively) and 2.0 mL of nitric acid (10^−2 and 10^−3 mol L^−1 for Pb and Cr, respectively). The mixture was shaken (15 min) heated to 40 °C (2 min) and then shaken for 3 min. The phases were separated by centrifugation (4000 rpm 2 min). To 1.8 mL were added 250 μL of 200 mg L^−1 DTZ, and the pH adjusted to 8 and 9 for Cr and Pb, respectively and the volume made up to 2 mL. Then, 0.4 mL of a mixture of dispersing solvent (acetonitrile) and extracting solvent (50.0 μL of CHCl3) was injected and the mixture shaken. The phases were separated by centrifugation (4000 rpm for 15 min), and 40.0 μL of the extracting solvent (total volume 45 μL) was introduced to the spectrometer via a microsampling system. ERF (based on the ratio of concentration in the oil to concentration in the DTZ) of 70 and 54 for Cr and Pb, respectively were reported. Neither analyte was found in either sample. The concentrations in the samples spiked with 100 μg kg^−1 were <LOD of the wet digestion method	0.5 (Cr), 1.5 (Pb) μg kg^−1	Spike recoveries and comparison of results with those of a wet digestion method (2 g and 10 mL final volume)	76
Cu	Tap water	FAAS	SPE on ion-imprinted sorbent, Cu N-methoxymethyl melamine plus activated carbon	Sample (25 mL) passed through 0.100 g sorbent and eluted with 5 mL 4 mol L^−1 HNO3. Flow rates not given in mL min^−1. Analyte not found in sample	0.038 (based on a sample volume of 750 mL)	ERML-CA021e (soft drinking water), NIST SRM 1643e (water) and spike recoveries	286
Cu	Milk	FAAS	SPE on restricted access ion-imprinted polymer (doubly coated with hydrophilic monomer and bovine serum albumin)	Sample (20 mL) passed through 330 mg of extractant in a mini-column at 5.5 mL min^−1 and the retained Cu eluted with 1.0 mol L^−1 HCl solution at same flow rate and delivered directly the spectrometer. No preconcentration factors were given. The concentrations found in three samples were not significantly different from those found by the European Committee for Standardization's approved method, EN 14082 (dry ash and FAAS)	30	No CRM but comparison of results with those of another method	287
Cu, Fe, Pb	Oil (soybean, sunflower, rapeseed, sesame, and olive)	MIP-OES	DLLME into DES (ChCl and glycerol)	Samples were diluted 1 + 9 with petroleum ether and to 8.6 g were added 100 μL of the DES followed by mixing (vortex 1 min) centrifuging (3000 rpm 3 min). The supernatant was removed and approximately 80 μL of the extract was collected and introduced directly into the MIP-OES instrument. No parameters relating to preconcentration were given. Cu was found in all samples, except sesame oil; Fe was found in all samples, but Pb was not found in any	0.7 (Cu), 2 (Fe), 3 (Pb) μg kg^−1	Spike recoveries	288
Cu, Mn, Zn	Oils (olive, soybean, colza, canola, corn, rice bran, camellia, peanut, sunflower, sesame, rattan pepper, capsicol)	ICP-OES	DLLME into a supramolecular DES containing DL-lactic acid and 2-hydroxypropyl beta-cyclodextrin	To 10 mL of oil were added 800 μL of supramolecular DES (extractant) and 50 μL of 5% HNO3 solution (dispersant). After vortexing (5 min) and heating (75 °C 20 min) in a thermostatic water bath, the phases were separated by centrifugation (8000 rpm 5 min) and 0.2 mL of the lower phase removed and subjected to MAD (HNO3 and HClO4) with final volume 10 mL. The ENF were 24, 35 and 21, for Cu, Mn and Zn, respectively. Cu and Zn were found in all samples except rice bran and capsicol; Mn was found in all except capsicol	0.89–1.30	Spike recoveries	289
Cu, Ni	Food (black tea, rice flour, wheat, green pepper, spinach, apple, eggplant, pomegranate, parsley, mint, tomato, mushroom, potato), water (bottled, waste, well, river, spring)	FAAS	Magnetic DSPME of methyl violet complexes into DES ferrofluid composed of lauric acid and menthol and toner powder@aliquat 336 magnetic particles	Solid sample mass 10 g MAD (HNO3 and H2O2) final volume 10 mL. Waters 100 mL evaporated to 10 mL. For both sample types, 10 mL was taken and after pH adjustment (6.0), methyl violet was added followed by 0.8 mL ferrofluid-based DES. After shaking, the ferrofluid was collected on a magnetic stir bar and the analytes desorbed in 300 μL THF and the resulting solution (1.0 mL) was analyzed. The ENFs were 113 for Cu and 135 for Ni, but these seem high given the volumes involved. Cu was found in all food samples, and Ni was found in all except rice flour, wheat and pomegranate. Neither was found in the bottled water or mineral water	0.03 (Cu), 0.15 (Ni)	CRM GBW10015 (spinach), GBW10016 (tea), GBW10019 (apple) and spike recoveries	290
Cu, Pb	Dairy products (milk, doogh, and cheese)	FAAS	DLLME of DDTC complexes into DES (choline chloride and p-chlorophenol)	Milk was diluted 1 + 1 with water. Cheese (3 mm layer 2.0 g) was extracted with 10.0 mL hot water (70 °C), vortexed (5 min) and centrifuged (7000 rpm 6 min). Doogh was used directly. Proteins were precipitated from 5 mL by 75 mg of TCA, followed by vortexing and centrifugation. The pH was adjusted (value not given), the solution heated to 70 °C and 125 μL of sodium DDTC solution (0.1 mol L^−1) and 350 μL of the DES injected. After 3 min, the solution was cooled in ice bath at 0 °C. About 210 μL of DES was separated by centrifuging. No details of how this solution was delivered to the spectrometer were given. It is possible that extraction was aided by salt addition. Cu was found in all five samples, but Pb was detected in only one sample (a cheese). An enrichment factor was defined, but not given	0.04 (Cu) and 0.18 (Pb)	NIST SRM 1549 (milk powder) and spike recoveries	291
Cu, Pb, Sn	Corn, soil	ICP-OES	CPE of complexes with 2-(4-sulphonylamidebenzo)hydrazide-1-dithiocarbamate and Triton X-114	Sample mass 1.0 g (corn) or 0.25 g (soil) MAD (HNO3 and HCl) final volume 100 mL. To an unspecified volume were added 1 mL of 0.01 mol L^−1 ligand and 0.1 mL of 5% (v/v) Triton X-114 and the volume made up to 50 mL. After pH adjustment (8), the solution was heated (70 °C 20 min), then centrifuged (4000 rpm 20 min). To the surfactant-rich phase was added 4.5 mL of 2 mol L^−1 HNO3. No parameters relating to preconcentration were given. Cu was found in all three corn samples, Pb in one, and Sn in none	0.0002 (Cu), 0.002 (Pb), 0.0004 (Sn)	CRM GSBZ GBW10011 (wheat flour) and GBW10012 (corn flour)	292
Cu, Zn	Food (fish, chicken, apple, tomato, spinach)	FAAS	DLLME of complexes with an organoselenium Schiff base, (2-((4-((2-hydroxy-4-nitrobenzylidene)amino)phenyl)selanyl)-N-phenylacetamide) (HOASe), with solidification of floating organic drop (1-undecanol)	Samples, dried, ground, sample mass 0.5 g and MAD (HNO3 and H2O2) with final volume 25.0 mL, to 10 mL of which were added buffer (pH 6), 150 μL 0.01 mol L^−1 ethanolic solution of complexant and 75 μL 1-undecanol. The mixture was ultrasonicated (2 min), centrifuged (3 min, 3000 rpm), and cooled in an ice bath (5 min). The solidified 1-undecanol phase was removed, melted and made up to 500 μL with EtOH. The ENF was 20. Both analytes were found in all samples	1.9 (Cu), 2.4 (Zn)	CRM NIST SRM 1570a (spinach leaves), SRM 1573a (tomato leaves) and spike recoveries	293
Fe^II, Fe^III	Rice wine	ETAAS	Dual-drop microextraction based upon the selective reaction of 1-(2-pyridylazo)-2-naphthol (PAN) with Fe^II and N-benzoyl-N-phenylhydroxylamine (BPHA) with Fe^III	Sample (5.0 mL) de-alcoholised (80 °C ultrasonic bath with a reduced pressure evaporator) diluted to 30 mL with H2O. Two 15 μL CHCl3 drops containing PAN and BPHA were immersed in 20 mL solution, stirred at 40 °C for 25 min. The drops were diluted (CHCl3) to 100 μL, 10 μL taken for analysis. Preconcentration factor of 300 reported. Both analytes found in all four samples, but only Fe^III in the CRM because of oxidative sample digestion	0.067 (Fe^II), 0.054 (Fe^III)	CRM GSBZ GBW 10010 (rice) and spike recoveries	294
Mn	Food (green pepper, onion, spinach, eggplant, parsley, tomato, oat, bean, green tea, pine nuts, wheat, almond, walnut, hazelnut, chickpeas)	FAAS	LLME of complex with 2,6-pyridinedicarboxylic into a switchable solvent based on octanoic acid	Samples (5.0 g homogenized) MAD (HNO3 and H2O2) final volume 10 mL, to which were added complexant, 1.2 mL fatty acid switchable solvent, and 1.25 mL of 0.75 mol L^−1 NaOH (to deprotonate the switchable solvent). Following separation by centrifugation, the switchable solvent phase containing Mn was made up to 1 mL with acidic MeOH and injected into the spectrometer. Reported values for preconcentration factor (150) and enhancement factor (118) do not make sense. Analyte found in all samples except eggplant, green tea and walnut	0.3	CRM INCT MPH-2 (mixed Polish herbs), NIST SRM1568b (rice flour), SRM 1548a (typical diet)	295
Pb	Rice	ETAAS	Gas–liquid extraction of the PAN complex into toluene vapour	Dried sample (1 g) was dry ashed (450 °C), then wet ashed (HNO3), then dry ashed again, then dissolved (6 mol L^−1 HCl) and evaporated. The residue was dissolved in 0.1 mol L^−1 HNO3, but the volume was not given. The extraction procedure is difficult to understand, but it appears to involve bubbling argon saturated with toluene vapour through a vertical column (20 mL) of sample solution and collection of liquid toluene (volume not given) from the top of the column. No figure of merit relating to preconcentration was given. Analyte was found in all six samples	0.11	CRM NCS ZC73027 (rice)	296
Pb	Food (wheat flour, milk powder), water (tap, waste)	FAAS	SPE on EDTA-modified activated C derived from apricot pits	1.0 g hotplate (HNO3 and H2O2). Final volume not specified but maybe 20 mL. This volume passed though (no details given) 40 mg at 46 °C (water bath) at 10 mL min^−1, eluted with 5 mL HCl at 5 mL min^−1. Enhancement factor (not defined) of 130 reported but makes no sense. Analyte only found in wastewater	0.25	CRM ERM-CA011 (water) and spike recoveries	297
Pb	Food (lettuce, dill), water (tap, waste)	FAAS	DSPME on thiosemicarbazide-modified, sulfonamide-based poly(styrene) adsorbent	UBM (BARGE) saliva extraction (4.5 mL) of 0.30 g of dried, ground sample, filtered, diluted to 90 mL. To 30 mL was added 100 mg of extractant, vortexed, centrifuged. Pb dissolved (vortex) in 2 mL of 2 mL L^−1 HCl and separated (centrifuge). The PF was 15. Pb was not detected in the saliva extracts or the tap water	5.1	CRM TMDA-70.2 (lake water) and BCR-482 (lichen) and spike recoveries	298
Pb	Fruit juice (orange, pear, multi-fruit and herbal)	ETAAS	SPE on hybrid material composed of Rhodococcus erythropolis AW3 bacteria and Brassica napus hairy roots	The samples were diluted (1 + 1) with 0.5% (v/v) HNO3 and filtered (0.22 mm). Then 0.5 mL (to which NaNO3 was added) was pumped at 1 mL min^−1 through a 15 mm-long microcolumn containing 0.1 g of the extractant with elution by 80 μL of 0.05 mol L^−1 HNO3. An enrichment factor of 62.5 was reported (basis not defined, but is the volume ratio). Analyte found in all eight samples	5 ng L^−1	NIST SRM 1643e (trace elements in water) and spike recoveries	299
Pb	Food (ground vegetables, butter, wheat flour), water (tap)	ETAAS	SPE on ion-imprinted polymer (details given) with elution by HNO3 followed by LLE (of complex with ethyl (2Z)-3-{(5-chloro-7-methyl[1,3]thiazolo[5,4-d]pyrimidine-2-yl)amino}-2-cyano 3-(methylsulfanyl) prop-2-enoate) (TPAS) into 1-undecanol with solidification of floating drop	Samples MAD but no details given. To 500 mL of solution were added 100 mg ion-imprinted polymer particles with elution in 1.0 mL of HNO3. After pH adjustment (7.5), 100 μL of TPAS in 1-undecanol and 150 μL SDS were added and the volume diluted to 10.0 mL. After shaking and centrifuging, the floating organic phase was separated by cooling in an ice bath and 20 μL was mixed with NH4H2PO3 (chemical modifier) and injected into the graphite furnace. An ENF of 670 was reported and analyte was found in all samples	1 ng L^−1	CRM Geological Survey of Japan JR-1 (igneous rock) and spike recovery	300
Pb	Food (chicken, tomato, rice, black tea powder)	ETAAS	SPME and CPE. The complex with PAR was adsorbed onto montmorillonite and transferred into micellar phase of Triton X-114, elution with HNO3	Sample mass 0.2 g MAD (HNO3 and H2O2) final volume 50 mL to 45.0 mL were added 50.0 μL of PAR (pH adjustment to 7.0), Triton X-114 (0.2% v/v) and 2.0 mg of montmorillonite. and the mixture heated (water bath 15 min at 31 °C). After centrifugation (4000 rpm 5 min), the surfactant rich phase was vortexed with 1.0 mL of HNO3 (0.5 mol L^−1) to desorb the Pb ions, finally the supernatant was analysed by GFAAS. The ENF was 96. Analyte was found in all samples	0.006	CRM NIST SRM 1643 f (trace elements in water), SRM 3255 (green tea extract)	301
Pb	Food (mushroom, green chili pepper, butter, scallions, potato, leek, garlic, tomato, lettuce, cucumber, courgette, bulgur wheat, rice, carrot, walnut), beverages (iced tea, Coca Cola energy drink), fruit juices (pomegranate, apple, orange, pineapple, tangerine, cherry), water (drinking, tap, mineral)	FAAS	SPE on magnetic poly linoleic acid–polystyrene–polydimethylsiloxane hydrophobic graft copolymer	Solid samples were dried and 1 g hot plate digested with (HNO3 and H2O2), the final volume was 50 mL. Liquid samples were filtered. To 50 mL (adjusted to pH 6) were added 400 mg of the extractant and the mixture shaken for 15 min. After magnetic separation, the Pb was dissolved in 0.3 mL 0.5 mol L^−1 HCl (orbital shaker 5 min). The ENH was 167. Pb was found in all 27 samples except butter, lettuce courgette, rice, carrot, apple juice cherry juice, energy drink and drinking water	0.5	CRM LGC-6010 (hard drinking water), NCS ZC73032 (celery) and CS-M-3 (microelements in mushroom powder) and spike recoveries	302
Pb	Food (canned green lettuce, mustard green, corn, and cucumber; fresh chili pepper, tomato paste, garlic, and onion), water (tap, drinking)	ETAAS	LLME electrically driven (30 V) through 1-octanol in a supported liquid membrane hollow fibre into DES (ChCl + phenol)	Sample mass 0.5 g MAD (HNO3 and H2O2) final volume 50 mL. The hollow fibre was placed in 30 mL sample solution (adjusted to pH 9) and one graphite electrode (2B pencil lead) was inserted into fibre (as the anode) with another inserted into the donor phase (as the cathode). The voltage was applied (25 min) and the acceptor phase (40 μL) was extracted and analysed. The ENF was 111. The analyte was found in all samples	0.011	CRM NRCC SLR-6 (river water), NIST SRM 1515 (apple leaves) and spike recoveries	303
Pb	Food (spinach), water (tap, river, dam, spring, battery)	CS-FAAS	SPME of complex with 2-(5-bromo-2-pyridylazo)-5-(diethylamino)phenol on Sepabeads SP70 resin impregnated with IL (1-butyl-3-methylimidazolium hexafluorophosphate) elution with HCl	Sample (0.5 g) hot-plate digested (HNO3 and H2O2) final volume 5 mL. To this was added 150 μL of 0.1% (w/v) ligand solution, the pH was adjusted to 7, and 5 mg of suspend beads added, the mixture ultrasonicated (10 min), vortexed (3 min) and the phases separated by centrifugation (5 min). The Pb was eluted with 0.4 mL of 2 mol L^−1 HCl solution (vortexed, centrifuged). Analyte only found in battery water. The preconcentration factor of 125 must apply to the water samples (presumably 50 mL were taken)	0.21	Environment Canada TMDA-64.2 and TMDA-53.3 (fortified water) and NIST SRM1570a (spinach leaves)	304
Pb, Sn	Water (bottled, tap, river, sea, rain)	CS-ETAAS	Magnetic DSPME of complexes with DDTP on ferrite particles coated with MIL (trihexyl (tetrahexyl) phosphonium tetrachloroferrate), elution with ACN	To a 10 mL sample aliquot were added 600 μL concentrated HCl, 600 μL 30% m/v NaCl and 500 μL of 0.1 mol L^−1 DDTP, after standing (5 min) 50 μL of the MIL mixture with ACN was added. Following stirring, 50 μL of the 10 mg mL^−1 ferrite suspension were added. After magnetic separation, the acceptor phase was vortexed (5 min) with 50 μL of ACN and the magnet reapplied. Finally, 10 μL and 10 μL of modifier were transferred to the atomizer. The EHFs were 195 and 165 for Pb and Sn, respectively. Pb was found in all samples except spring water and two of the five bottled waters; Sn was found in all samples except spring water and two of the five bottled waters	25 (Pb) and 72 (Sn) ng L^−1	NIST SRM 1640a (natural water), Environment Canada TM-23.4 and TM-25.4 (Lake Ontario water)), Spectrapure Standards SPS-SW1 and SPS-SW2 (surface water), Train-04 (deep water)	305
^226Ra	Drinking water	ICP-MS	Five different SPE materials evaluated	For five procedures, already described in the literature, figures of merit were tabulated, these included LOD, time, sample volume, PFs, flow rates, costs and volume of waste generated. The researchers noted that labour costs were the most significant, but that these could be decreased by automation of the separation step	2–7 fg L^−1 (0.1–0.3 mBq L^−1)	CEAEQ, Québec, QC, (tap water)	306
Sb^III	Beverages in PET bottles (water, coke, orange juice, apple juice, vinegar, lemonade, cherry juice, ice tea, energy drink)	HG-AAS	DSPME on polyoleic acid–polystyrene block/graft copolymer with elution by ACN	Samples (5 mL) hot plate digested with HNO3 and H2O2 with final volume 50 mL, to 5 mL of which (adjusted to pH 5) was added 105 mg of extractant and the mixture sonicated (16 min) and the phases separated by centrifugation (4000 rpm 2 min). The Sb^III was dissolved in 1 mL of ACN, with vortexing (60 s) and presumably centrifugation. The EHF was 82. The analyte was not found in two of the four waters, orange juice, lemonade or iced tea	1 ng L^−1	CRM NIST SRM 1643e (trace elements in water)	307
Se	Tea (black leaves)	HG-AAS	CPE (air-assisted at room temperature) of the complex with 1-(2-hydroxy-5-p-tolylazo-phenyl)-ethanone (HPAPEO) and Triton X-100	Samples were dried, ground and sieved (100-mesh) and 0.2 g subjected to MAD with 2.0 mL conc. HNO3, 2.0 mL H2O2 (30% v/v), and 2.0 mL water. The final volume was 25.0 mL, to 10 mL of which (adjusted to pH 9) were added 1 mL of 1.0 × 10^−4 mol L^−1 HPAPEO and 0.5 mL of 0.05% (v/v) Triton X-100. Air was blown through the mixture (80 s) and the phases separated by centrifugation (8 min 4000 rpm). After cooling (ice bath 5 min), the top aqueous layer was removed and the surfactant-rich mixed with 1 mL of ethanol. An EN of 150 (based on calibration slope ratios) was reported, as was a preconcentration factor of 100, though it is difficult to ascertain the basis for this. The researchers claimed that Se was present in “trace amounts” in the samples, but the table of results showed no values in the “found” column	0.02	CRM GBW08513 (tea leaves) and spike recoveries	308
Se	Food (rice, egg white), water (not specified)	Slurry HG-ICP-MS	SPE on MOF UiO-66-NH2 (100 nm particles)	Sample (0.2 g) acid MAD then heated with HCl (to reduce Se^VI to Se^IV) final volume 50 mL, 10 mg UiO-66-NH2 was added, filtered (suction) and particles dispersed by ultrasound in NaBH4 + KOH to yield 2.5 mL of slurry. The ENH was 15.7. Analyte found in all samples except one water sample	0.001	CRM NRCC TORT-3 (lobster hepatopancreas), DOLT-5 (dogfish liver), NCS GBW(E)100697 (selenium-rich rice)	309
V	Food (potato, egg, spinach, tomato, chicken, cucumber mushroom, milk (cow, sheep), wine (white, red), coffee), water (canal, tap, waste)	ETAAS	SPE on imino diacetate functionalized polystyrene elution with HNO3	Food samples MAD no details of sample mass or final volume. SPE performed by aspirating and expelling solution with syringe. Elution by same process. No details of volumes given. Enhancement factor of 100 reported but basis not given	30 ng L^−1	CRM NRCC SLRS 4 (riverine water), NIST SRM1515 (apple leaves) and spike recoveries	310
V	Food (mint, lettuce, cabbage, spinach, tomato, apple, and eggplant), water (waste, river, hot spring, bottled, well)	FAAS	Magnetic DLLME of complex with bis(acetylpivalylmethane) ethylenediimine into DES {ChCl/p-cresol}{FeCl4}	Sample mass 1.0 g MAD (HNO3 and H2O2) final volume 50 mL. Water samples filtered and 100 mL evaporated to 10 mL. For both samples, to 10 mL (pH adjusted to 6) were added 1.2 mL of 100 mmol L^−1 ligand and 800 μL of the DES and the mixture vortexed (80 s). Phases were separated by a neodymium magnet (1.17 T), the upper aqueous phase decanted, and the remaining phase diluted to 1 mL by acidic ethanol. The EHF was 120. Analyte was found in all food samples and in the waste, hot spring and well waters	0.3	CRM NIST SRM 1570a (spinach leaves) and 1643f (trace elements in water)	311

---

Table 2 Preconcentration using nanomaterial-based methods

Analyte	Matrix	Technique	Extraction mode/reagents	Procedure/comments	LOD in μg L^−1 (unless stated otherwise)	Validation	Ref.
As	Food (egg, fish, meat, mushroom, rice, salami, sausage, tea, tomato, pepper, flour, cabbage, carrot, parsley, mint)	HG-AAS	Ultrasound assisted-dispersive SPME was utilised with composite adsorbent poly(3-hydroxy butyrate)-b-poly(dimethyl amino ethyl methacrylate) amphiphilic block copolymer containing gadolinium oxide NPs (PHB-PDMAEMA-Gd2O3-NPs)	Solid samples MAD was utilised prior to extraction. 10 mg of copolymer was used as composite adsorbent for extraction of tAs from liquid samples at pH = 6.8. The mixture was placed in ultrasonic bath, then centrifuged. For desorption of the absorbed As, an elution step was performed using 2 mL of 1 mol L^−1 HNO3. ERF was 128	0.02	CRMs: SRM 8418 (wheat gluten), IPE-10 (clover honey-stalk)	312
Be	Rice wine	ETAAS	Nanofiber membrane-based syringe SPE with TiO2@SiO2	Nanofiber membrane was cut into a disk shape (2 cm of diameter) and put into a reusable syringe filter. The filter was connected to the tip of the syringe and conditioned with 5 mL of 1 mol L^−1 HNO3 and ultrapure water. Then, 30 mL of sample (pH 6) was pulled and pushed for six cycles. Be retained was eluted with 0.4 mL of 1 mol L^−1 HNO3 for 4 cycles. ERF achieved 150	0.00052	CRMs: GBW 10016 (tea leaves), GBW 10010 (rice), spike recoveries	313
Bi	Water (tap), food (corn, chicken liver, beef liver, lentils, chickpea), cosmetics	HR-CS-FAAS	Micro SPE with MgAl2O4@MoSe2 NC was used to separate–preconcentrate Bi	To 30 mL solution (pH 5) 10 mg of MgAl2O4@MoSe2 was added, shaken and centrifuged. 3 mL of 0.1 mol L^−1 HNO3 was added for elution, vortexed and centrifuged. EHF = 8.69	0.012	CRMs: NCS ZC73028 (rice), NCS ZC73036 (green tea), spike recoveries	314
Cd	Wine, seawater	Microplasma OES (slurry sampling)	DSPME based on GO	Wine samples were wet digested prior to analysis. 30 mL of sample solution was adjusted to pH 6, the GO dispersion was introduced and the mixture shaken. After centrifugation the resulting GO precipitate was rinsed with water and Cd was eluted with diluted HNO3 while shaking dispersion and analysed by OES. The tolerance of Cd to K was improved 25 times	0.003	Comparison with a reference method (ICP-MS, standard additions)	315
Cd	Water (dam, waste, river, well, sea), food (tea, nut, rice, cacao, chocolate, leek, cinnamon, parsley)	FAAS	DSPME without vortexing using Ni-Mn-Co tetragonal spinel ternary oxide NC as sorbent	The adsorbent (100 mg) was mixed with 20 mL of sample, pH was adjusted to 6. Adsorption took place without vortexing. The mixture was centrifuged. Three mL of 2 mol L^−1 HCl were added to elute the analyte, then the solution was centrifuged. PF = 23.3	0.49	CRMs: TMDA-54.6 (lake water), SPS-WW1 121 Batch (wastewater), SRM 1573a (tomato leaves), TORT-3 (lobster hepatopancreas) and spike recoveries	316
Cd	Food (chili powder, black pepper, tea, chocolate), wastewater, tobacco	HR-CS-FAAS	Magnetic DSPME using MOF Fe3O4-SiO2-MIL-53 (Fe) NC	Solid samples underwent MAD with HNO3 and H2O2 prior to extraction. 20 mL of sample/digest (pH = 9) were mixed with 20 mg of NC and subjected to a vortex. A magnet was used to separate the nanomaterial. With the aid of 4 mL of 1 mol L^−1 HNO3 and 1 min. vortex the analyte was released. PF was 5	1.3	CRMs: TMDA-64.3 (fortified water), SRM 1573a (tomato leaves), SRM 1570a (spinach leaves)	317
Cd, Pb	Water (tap), food (pear, apple, tangerine, mint leaves, chickpea, corn, flour, starch, baking powder)	FAAS	Micro SPE using functionalized nanodiamonds@CuAl2O4@HKUST-1 NC	To the sample volume of 25 mL (pH 7) 10 mg of the adsorbent material were added, vigorously shaken and centrifuged. 1 mL of 1 mol L^−1 HNO3 was added, and the mixture vortexed, then centrifuged, and the resulting liquid layer was separated for analysis. EHF was 29.5 for Cd and 21.7 for Pb	0.031 μg kg^−1 (Cd), 0.052 μg kg^−1 (Pb)	CRMs: ZC 73032 (celery), ZC 73033 (scallion), SPS-WW2 (wastewater), spike recoveries	318
Cd, Pb	High salt food (salt, soy sauce, pickled cucumber, aginomoto, salted fish, salted casing, ham)	ICP-MS	SPE with sulphur-functionalized magnetic MOFs (Fe3O4@UiO-66-SH)	0.5 g of sample was digested with HNO3 using MAD and evaporated near dryness, then diluted to 50 mL (pH 6). 20 mg Fe3O4@UiO66-SH were added into 50 mL of prepared sample solutions and sonicated. MOF was separated using a magnet. Elution with 5 mL 5% HNO3 (v/v) was carried out during vortexing. More reliable results for high salt samples were achieved due to reduction of matrix effect	0.17 μg kg^−1 (Pb), 0.21 μg kg^−1 (Cd)	CRM CFAPA-QC1728-4, P41839 (soy sauce) and spike recoveries	319
Cd, Cr, Ni, Pb	Food (green lentil, corn flakes, corn starch), tobacco, wastewater	FAAS	SPME with magnetic Luffa@MOF-199	Solid CRMs and food samples (250 mg) were digested with HNO3 and H2O2 on hotplate prior to extraction. 40 mL of digest/sample and 1 mL of 2-nitroso-2-naphthol solution (0.1% w/v) were mixed at pH = 7 to obtain the complexes of metals. 20 mg of adsorbent was added, the volume completed to 50 mL and the mixture vortexed. The solid layer was collected using a magnet. 1 mL of 1 mol L^−1 HCl was added and vortexed. Using a magnet, the solid adsorbent was collected, and the upper layer was analysed by FAAS. Achieved EHF was 50. The main disadvantage of the procedure is loss of quantitative adsorption properties after ten times usage	0.35 (Ni), 0.38 (Pb), 0.55 (Cr), 0.12 (Cd)	CRMs: TMDA-64.3 (fortified water), GBW07424 (soil), SRM 1573a (tomato leaves)	320
Co	Water (tap, waste), food (black tea, green lentil, corn starch), tobacco	HR-CA-AAS	Micro SPE with magnetic date palm fiber-WSe2 NC	Solid samples (0.25 g) were digested using MAD. 10 μg of 1-nitroso-2-naphthol and 5 mg of adsorbent were added to 30 mL of sample/digest (pH = 7) and vortexed. After centrifugation, the liquid layer was separated using an external magnet. 4 mL of 3 mol L^−1 HNO3 in EtOH was added, vortexed and centrifuged. EHF value was 7.3	1.73	CRMs: NZ ZC73032 (celery), NC ZC73033 (scallion), TMDA-64.3 (fortified water), spike recoveries	321
Co, Ni	Food (juice, parsley, tomato, cabbage, potato, spinach, hazelnut, tea), tobacco, water	FAAS	Ultrasound assisted DSPME on oxidized multiwalled carbon nanotubes (ox-MWCNTs) as adsorbent and 3-(2-hydroxy-5-acetylphen-1-ylazo)-1,2,4-triazole (HAPAT) as complexing agent	Food samples (1 g) digested by MAD were diluted to 100 mL. To 100 mL of studied solutions containing Co^II and Ni^II, 2 mL of HAPAT (10^−3 mol L^−1) and 0.3 mL of ox-MWCNTs suspension (5 mg mL^−1) were added to 100 mL of the digests/water samples, pH was adjusted to 7 and the mixture laced into an ultrasonic bath. The adsorbed Co^II and Ni^II ions were eluted with 2 mL of HNO3 solution (1 mol L^−1). An ERF up to 200 was mentioned	0.3 (Co), 0.6 (Ni)	CRMs: SRM 1570a (spinach leaves) and TMDA-52.3 (fortified water) and spike recoveries	322
Cr^VI	Water (tap, waste, sparkling, pond)	ED-XRF and TXRF	Dispersive micro SPE using GO modified tetraethylenepentamine (GO-TEPA)	The well-dispersed homogeneous suspension of GO-TEPA (500 μg mL^−1) was pipetted into a suitable sample volume to acquire desired adsorbent dose, pH was adjusted to 3.5 and stirred. The adsorbent was separated from the solution using nitrocellulose membranes. The implementation of EDXRF or TXRF eliminates the elution of adsorbed analyte from the GO-TEPA. Due to high PFs (865 for EDXRF and 100 for TXRF) extremely low LODs were obtained	0.053 (ED-XRF), 0.0035 (TXRF)	CRMs: QC1088 (water), QC3016 (seawater)	323
Cr^VI	Water (tap, mineral)	ICP-MS	Preconcentration by IL-ferrofluid (IL-FF) where the surface of Fe3O4 magnetic NPs was coated with silica and functionalized with L-cysteine (SCMNPs-Cys)	IL-FF was injected into 20 mL sample solution, cloudy solution was sonicated for 4 min. SCMNPs-Cys were collected using an external magnet. 1 mL of 2 mol L^−1 HNO3 was added to the collected SCMNPs-Cys, followed by sonication at 50 °C to desorb analytes. The SCMNPs-Cys were again separated using an external magnet. The eluates were analysed by FI-ICP-MS. Achieved ERF was 20	0.003	Spiked water samples recoveries	324
Cr^VI, Cu^II	Edible vegetable oils	ICP-MS	A magnetic effervescence-enhanced emulsification microextraction based on the employment of effervescent tablets composed of dicationic ionic liquids, MFe2O4 NPs, and acidic and alkaline sources	The reagent tablets were added to the oil–acid mixed solution of 0.2 g of oil in 10 mL of diluted HCl. After the tablet dissolution (3∼5 min), the Cr^VI- and Cu^II-enriched extractant was adsorbed onto the Fe based MNPs. The MNPs were collected with an outer magnet and the elements eluted by 3× washings with 10% HNO3	0.086 μg kg^−1 (Cr^VI), 0.076 μg kg^−1 (Cu^II)	Spike recoveries	325
Cr, Cu, Pb	Tea infusion	LIBS	Extraction on Ag NPs modified resin	10 mL of sample were added to weighed amounts of the Ag NPs modified resin and the mixture shaken. Next, 0.01 g of resin was isolated from the containers and left to air-dry at room temperature before it was fixed to the slides for LIBS. The enhancement of the resin absorption performance via Ag NPs were 3 to 7 times greater than those observed with the resin-LIBS	0.22 (Cr), 0.33 (Cu), 1.25 (Pb)	Reference method (ICP-MS), spike recoveries	140
Cu	Food (black tea, and rose hip powder), olive leaves, water (tap, ground, well, fish farming)	HR-CS-FAAS	SPE on BaTiO3 nanomaterial from liquid samples	Food (0.25 g) subjected to MAD with HNO3 + H2O2, but final volume not given. To (volume not given), 2 mL of pH 7 buffer were added followed by 20 mg of the BaTiO3 nanomaterial. After mixing and centrifugation, the retained Cu was eluted with 0.5 mL of 1 mol L^−1 HNO3. An ENH of 18 was calculated	1.6	CRM: TMDA-64.3 (fortified lake water) and spike recoveries	326
Cu	Tap water, human hair, cigarette, black tea	Microsampling FAAS	DSPME with Fe-Ni@ACC NC	30 mL of a sample at pH = 9 with 5 mg of Fe-Ni@ACC was used. The adsorbent was stable at least 20 times without any loss of its adsorption properties. The optimum eluent was determined as 0.75 mL of 2 mol L^−1 HNO3. The 40-fold ERF was calculated	0.69	CRMs: TMDA-53.3 (water), TMDA-64.2 (water) and spike recoveries	327
Cu, Pb	Water (tap, waste, drinking, mineral), fruit juice (apple, cherry, grape)	FAAS	SPE with a magnetic polymeric ionic liquid NC-coated hollow fiber membrane (HFM) developed in house	For the extraction of Pb^II, 7.5 mL of a sample at pH = 10 was placed in a 15 mL centrifuge tube. Afterward, 0.75 cm of HFM was added to the solution and shaken. The HFM was collected using tweezers and eluted with 1 mL of a 0.5mol L^−1 HNO3, the resulting solution was then shaken and introduced to FAAS using a home-made microinjection system. For Cu^II, 15 mL of sample, pH 6, 1.5 cm of HFM, and 1 mol L^−1 of desorption solution were applied. PFs were 7.4 for Pb and 14.6 for Cu	20.4 (Cu), 16.5 (Pb)	CRM: SPS-WW2 Batch 116 (wastewater)	328
Cu, Pb	Water (lake, waste, sea), food (radish, spinach, lettuce, celery)	FAAS	DSPME on magnetic mesoporous carbon (Fe3O4@C, MMC), synthesized in house	Food samples were prepared by wet digestion with HNO3. 20 mL of sample/digest were mixed with100 mg of MMC and pH adjusted to 6 (no shaking). An external magnet was used to separate MMC after centrifuge was made. The adsorbed ions were eluted with 3 mL of 2 mol L^−1 HCl with vortexing, centrifuging and magnetic separation. PFs were 83 for Cu and 167 for Pb	0.87 (Cu), 2.8 (Pb)	CRMs: TMDA-53.3 (fortified lake water), SRM 1573a (tomato leaves), spike recoveries	329
Cu, Pb	Water (dam, river, sea, waste), food (sumac, tea, green lentils, chocolate)	FAAS	DSPME with nanoflower Al2O3@carbon spheres composite	Food samples were prepared by wet digestion with HNO3 and H2O2. Soil sample was digested by aqua regia. 100 mg of reagent were dispersed in 20 mL of sample/digest and pH was adjusted to 7. The mixture was centrifuged without mixing. Elements were eluted using 2 mL of HCl (2 mol L^−1) without shaking, then the resulting mixture was centrifuged. PF was 125 for Cu and 75 for Pb	0.69 (Cu), 2.8 (Pb)	CRMs: TMDA-53.3 (fortified lake water), TMDA-54.6 (lake water), SRM 1573a (tomato leaves), SRM 8704 (buffalo river sediment), spike recoveries	330
Cu, Hg, Ni, Pb, Zn	Juice, water (tap, waste)	ICP-OES	Dispersive micro-SPE with deep eutectic solvent functionalized cobalt ferrite NPs	5 mL of sample solution was taken and transferred into an 8 mL glass test tube. After adjusting its pH at 5, 0.25 g NaCl (5%, w/v) was dissolved in the solution. Then 25 mg of the sorbent was added into the solution and vortexed. The sorbent particles were collected using external magnet. The adsorbed ions onto the sorbent surface were desorbed by 100 μL of ammonia solution (5%, v/v). Fruit juice samples and tap water were directly subjected to the extraction. Wastewater samples were diluted with buffer	0.544 (Cu), 1.33 (Hg), 1.12 (Ni), 0.622 (Pb), 0.962 (Zn)	CRM: SPS-WW2 Batch 108 (water)	331
Hg	Food (mung bean sprout)	CVG-PD-OES	SPE with Au NPs that extract Hg^2+ by forming gold amalgam	A total of 44 mL of sample solution and 6 mL of Au NPs dispersion were reacted for 40 min in a centrifugal tube, then the precipitate was obtained via centrifugation. 2 mL of desorption solution was used to generate Hg^0 CV with KBH4, which was transported to OES	0.16	CRM: GSB07-3173-2014 and spike recoveries	332
Hg	Water, fruit juice	CV-AFS	Miniaturized DSPME using novel hybrid bionanomaterial formed by the bacteria Bradyrhizobium japonicum and GO nanomaterial	To facilitate extraction, a complex was formed with Hg and ammonium pyrrolidinedithiocarbamate. The optimal Hg extraction conditions for 25 mL of sample were 3 mg of BJ@GO, NaNO3 at 1% (w/v), vortex agitation and centrifugation. For back extraction, 500 μL of 14 mol L^−1 HNO3, vortex agitation and centrifugation were used and followed by Hg detection	0.08	CRM: SRM 1641e (water)	333
Hg (iHg, tHg)	Water (well, waste) food (tuna, lettuce, spinach, tomato, pepper, cucumber, cowpea), human blood	CV-AAS	Cloud point assisted ionic micro SPE implementing a novel nitrogen-doped porous graphene nanostructure (NDPG)	A mixture of 20 mg of NDPG, 200 mg of ionic liquid, and 500 μL acetone, as dispersant solution, was mixed and rapidly injected into 10 mL of sample at pH = 7.5 and shaken. The Hg species were extracted by a pyrrolic/pyridinic NDPG. After centrifugation iHg was eluted by adding HNO3 (0.5 mL, 0.2 mol L^−1). Total Hg was determined using the same procedure but after MAD of samples. ERF = 9.8 for human blood, ERF = 20.2 for water and food	0.0026 (blood), 0.0012 (water, food)	CRMs: SRM 1570a (spinach), SRM 1573a (tomato), SRM 1946 (tuna), ERM CE464 (tuna), SRM 955c (blood), spike recoveries	334
Hg, Pb, V	Urine, sea water	HPLC-ICP-MS	Magnetic SPE with a new material (GO MNPs (M@GO-TS))	50 mg of M@GO-TS in a preconcentration system was loaded with sample at pH = 3.5 for 10 min. Elution was done with 1 mL of 7 mmol L^−1 thiourea + 40 mmol L^−1 H3PO4. Achieved ERF was 27	0.005 (Pb^II), 0.02 (trimethyl Pb), 0.002 (Hg^II), 0.010 (MetHg), 0.004 (V^V)	CRM: TMDA-64.3 (spiked water), spiked samples for species recoveries	335
Mn	Blood, serum, urine	ETAAS	Ultrasound-assisted-ionic liquid-DSPME implementing immobilization of 2-(aminomethyl) thiazole on the multi-walled carbon nanotubes	To 2.0 mL of sample 20 mg of nanotubes as adsorbent, 0.2 mL of acetone and 0.1 g of ionic liquid were injected at pH of 5.5 and shaken. The suspension of Mn on nanotubes was collected by centrifuge. The ions were eluted by HNO3 (0.1 mL, 0.4 mol L^−1) and eluent phase separated from sorbent by centrifuging. EHF reached 10.2	0.015	CRMs: SRM 955c (blood), SRM 2668 (urine), SRM 2670a (urine), spike recoveries	336
Mn, Pb	Water (dam, sea, waste), food (tea, cinnamon)	FAAS	DSPME with NiCo2O4@ZnCo2O4 nanomaterial, synthesized in house	50 mg of nanomaterial were added to20 mL of sample, pH = 4 (no shaking). The mixture was centrifuged. Then 3 mL of 2 mol L^−1 HCl was added to the sorbent, vortexed and centrifuged. PF was 16.7 for Mn and 33.3 for Pb	1.7 (Mn), 4.0 (Pb)	CRMs: BCR-482 (lichen), TMDA-70.2 (lake water), SRM 8704 (buffalo river sediment)	337
Ni	Food (tea, chocolate powder), water	FAAS	Magnetic DSPME on ion-imprinted polymer (IIP) based on n-allylthiourea in the presence of PAN as ligand	Chocolate powder underwent MAD prior to extraction. 45 mL of a sample (pH 10) was transferred into a 50 mL tube containing 50 mg of the IIP-PAN and stirred. The polymer was separated from the solution by using a magnet and adsorbed Ni^II ions were eluted using 0.5 mL of 2.0 mol L^−1 HCl solution containing 1% (v/v) thiourea, under vortex stirring. PF of 76.7 was achieved	0.26	CRMs: SRM 1573a (tomato leaves) and BCR 191 (brown bread)	338
Ni	Water (tap), food (tea, coffee beans, carrot, tuna, herring), tobacco, soil	FAAS	DSPME with nanodiamonds@CuAl2O4@TiO2 NC	30 mL of a sample (pH = 6) + 10 mg of adsorbent were shaken on a vortex. The mixture was centrifuged. 3 mL of 0.5 mol L^−1 HNO3 was used for elution (on vortex), and adsorbent separated by centrifuging the mixture	0.29	CRMs: TMDA-64.3 (fortified water), SRM 1573a (tomato leaves), spike recoveries	339
Ni, Pb, Zn	Mineral water, food (coffee, tea, sugar, salt, rice, corn, cucumber, biscuit, lettuce, parsley, onion, walnut, spinach, watermelon, melon)	ICP-OES	SPE with Bacillus subtilis loaded MWCNT as biosorbent	Solid samples were digested using MAD (HNO3, HCl, H2O2). 50 mL of sample was applied to SPE at pH = 5 and passed by peristaltic pump. Before metal elution, the column was cleaned with 10 mL of distilled water. HCl (5 mL of 1 mol L^−1) was used to elute the retained metal ions on the column. 80-fold PF was achieved. The column could be used even after 25 reuses	0.029 (Ni), 0024 (Pb), 0.019 (Zn)	CRMs: CWW-TM-D (wastewater), NCS ZC73014 (tea leaves), NWTM-15 (fortified water), EU-L-2 (wastewater), NCS ZC73350 (poplar leaves)	340
Pb	Food (fish)	FAAS	Micro-SPE with specific absorption of Pb^II on three-dimensional imprinting polymer based on GO-mesoporous silica nanobeads	350 mg of sample underwent MAD with 65% HNO3, H2O of 30% H2O2. An appropriate amount of polymer was added to 50 mL of digest and the solution was ultrasonicated. The adsorbent was filtered and Pb^II was determined after elution by MeOH and acetic acid. Obtained ERF was 89.5	0.003	CRMs: DORM-3 (fish protein), SRM 1947 (lake Michigan fish tissue) and spike recoveries	341
Pb	Food (salt, green tea, sage tea, kefir, oat, olive leaves, canned tuna fish, garlic), tobacco, multivitamin tablets	HR-CS-FAAS	SPE with a novel NC (magnetic MWCNTs@MgAl2O4@TiO2) as adsorbent	Solid CRM (0.15 g) and food samples (1 g) were digested with HNO3 and H2O2 on a hotplate prior to extraction. 5 mg of adsorbent were added to 50 mL of digest/sample (pH = 5) in a centrifuge tube and after that the solution was: mixed on a vortex, sonicated, and centrifuged. The remaining solid phase was washed with 0.5 mL of 3 mol L^−1 HNO3 to elute Pb. Finally, 100 μL of eluent presented to FAAS using a home-made microinjection system. Published EHF = 100	0.42	CRMs: TMDA-64.3 (fortified lake water), NCS DC 73349 (bush branches and leaves) and spike recoveries	342
Pb	Food (black tea, dried coriander, baby food, cocoa powder, starch, carrot, dill), tobacco, tap water	HR-CS-FAAS	SPE utilising a novel magnetic luffa@TiO2 NC	Food samples and CRMs were digested using MAD with HNO3 + H2O2. To 50 mL of a sample (pH = 6) 5 mg of adsorbent was added. The mixture was shaken via a vortex and put into a centrifuge. The tube content was decanted using a magnet to separate the solid phase. After decantation 1 mL of HNO3 (3 mol L^−1) was added for elution. Vortex and centrifugation steps were repeated, then using a magnet the solution was decanted and analysed. The EHF was 50	0.04	CRMs: SRM 1577b (bovine liver), TMDA-53.3 and TMDA-64.3 (fortified waters)	343
Pb	Food (instant noodles), water (waste, tap)	FAAS	DSPME with a synthesized nanodiamond@Bi2MoO6 NC	Food samples (0.25 g) were digested using MAD prior to extraction. 2.5 mg of adsorbent + 2 mL of buffer (pH = 8) were added to 5 mL of sample/digest in a centrifuge tube, then mixed on a vortex. After the separation of solution from solid, 3 mL of 0.5 mol L^−1 HNO3 were added for elution of Pb^II from the adsorbent (vortex 0.5 min). After centrifugation, the upper layer was used for Pb analysis. PF was 16.7	1.75	CRMs: RM8418 (wheat gluten), BCR 505 (estuarine water) and spike recoveries	344
Pb	Fruit juices, water	FAAS	SPME with nanodiamonds@NiCoFe layer double hydroxide	Different flavour fruit juices were digested (10 mL) by MAD with conc. HNO3 and H2O2 and transferred to a final volume 30 mL. SPME procedure: 5 mg of adsorbent, vortexed, centrifuged. For elution, 2 mL of 0.5 mol L^−1 HNO3 ware added as an eluent (vortex, centrifuge). Obtained PF 25	0.62	CRMs: SRM 1515 (apple leaves), BCR 505 (estuarine water) and spike recoveries	240
Pb	Bovine liver, kidney, muscle, lung	ICP-MS	SPE with magnetic NPs synthesized using lemon peel (Fe3O4-LP)	Samples (0.025–0.1 g) were digested by MAD. 25 mL of digest (pH = 5) were spiked with 50 mg Fe3O4-LP and shaken. A strong magnet was used to separate the adsorbent on the wall of the tube. The supernatant was discarded. Then, 2.5 mL of 3% HNO3 was added to the test tube to elute the adsorbed Pb, while vortexing. PF = 10	0.039	CRMs: BCR 185 (bovine liver), SRM 2976a (mussel tissue)	345
Pb	Water (tap, mineral)	FAAS	Dispersive SPE using magnetic GO aerogel (M-GOA), metal-loaded magnetic sorbents were directly inserted into FAAS	9 mg of M-GOA were added into a 50 mL tube containing 28 mL of water sample (pH = 6.5) and vortexed, then placed over a neodymium magnet for 10 s. A magnetic metallic probe was inserted into the tube, particles were attracted to the probe tip. After 10 s, the probe was removed from the solution, and the particles, in a “drop” form, were magnetically fixed in the probe tip. The probe was then inserted directly in the base of the flame	1.3	CRM: SRM 1643e (trace elements in water)	346
Pb	Water (river, sea, dam, waste), food (black pepper, lentils, tea)	FAAS	DSPME with Zn-doped CeO2 nanorods prepared in-house	Solid samples were digested with HNO3 and H2O2. 10 or 20 mL of water (depending on type) containing 100 mg of Zn-doped CeO2 (pH = 3) was dispersed by vortexing. The Zn-doped CeO2 was separated from solution with centrifugation. 1 mL of 2 mol L^−1 HCl was added to tube containing only sorbent (no vortexing) and centrifuged. PF was 60	2	CRMs TMDA-54.6 (lake water), SPS-WW1 Batch 114 (wastewater), SRM 1573a (tomato leaves)	347
Pb	Water (spring, lake, fish farming, sea, dam, ground, waterfall, river), food (oils, tea)	HR-CS-AAS	Dispersive SPE with Zn-doped MoO3 nanorods	To 20 mL of sample 10 mg of nanorods were added, pH was adjusted to 5. The mixture was hand stirred. Nanorods were separated from the sample solution using a centrifuge. The adsorbed ions on the nanorods were removed with 2 mL of 0.5 mol L^−1 HNO3 and vortexing. EHF = 10.9	3.1	CRMs: TMDA-64.3 (lake water), INCT-OBTL-5 (tobacco leaves)	348
Se	Water, beverages, seafood, milk, vegetables	TXRF	DSPME with a new thiosemicarbazide-incorporated graphene (G@TSC)	MAD in closed Teflon vessels was used for solid samples and determination of tSe. To determine Se^IV G@TSC suspension was added to 100 mL of the sample, the solution was sonicated. The pH was adjusted to 1.5, the solution stirred for at least 60 min, then passed through a nitrocellulose membrane placed in the Eppendorf and further ultrasonicated. 10 μL of the suspension of G@TSC is deposited on a quartz reflector and dried (60 °C) till the droplet entirely evaporates. The proposed method eliminates the necessity of using acids for the elution, organic solvents, and/or gas for the measurement	1.7	CRMs: SRM 1640a (natural water), ERM-CA713 (waste water), QC3163 (sea water), BCR-414 (plankton), TORT-2 (lobster), ERM-BB422 (fish), M-3 HerTis, M-5 CodTis, ERM-BD151 (skimmed milk), NCS ZC73013 (spinach), NCS ZC73029 (rice), NCS ZC73032 (celery), NCS ZC73033 (scallion) and spike recoveries	102
Se	Food (rice, egg white), water (not specified)	HG-ICP-MS (slurry sampling)	SPE of complex with on UiO-66-NH2 MOF	Sample mass 0.2 g prepared by MAD to final volume 50 mL, to which 10 mg of MOF was added. After 5 min the suspension was filtered, the Se-enriched NPs mixed with NaBH4 and KOH to yield 2.5 mL of slurry that was dispersed by ultrasound. The ERF was 20	0.001	CRMs: DOLT-5 (dog shark liver), TORT-3 (lobster hepatopancreas), GBW(E)100697 (selenium-enriched rice)	309
V^V	Water, food (rice, parsley, spinach, carrot, banana)	ETAAS	SPE with ion-imprinted polymer synthesised on magnetic multiwalled carbon nanotubes and using 8-HQ as the ligand	Solid samples (0.1 g) are digested prior to extraction with HNO3 and H2O2. 12 mg of the magnetic sorbent were added to 200 mL of the sample (pH ∼5), and the mixture stirred. By holding the sorbent with the magnet, supernatant was decanted. V was desorbed with 0.5 mL of 0.5 mol L^−1 thiourea in 0.2 mol L^−1 HCl. tV was determined by the oxidation of V^IV to V^V before extraction	0.0029	Spiked water samples recoveries	349
Various (Cd, Cr, Co, Cu, Fe, Mn, Ni, Pb, Zn)	Food (fish, pork liver and kidney)	ICP-OES	DSPME using GO with neocuproine or batocuproine as complexant	A 1 mL of the GO suspension was transferred to 50 mL of the analysed sample solution containing 1 mL of neocuproine solution or 0.2 mL of batocuproine solution and metal ions at pH = 8. The sample solutions were stirred. After the solutions were passed through cellulose filters. The adsorbed metal ions were eluted with 5 mL of 0.5 mol L^−1 HNO3. The PF for GO/neocuproine and GO/batocuproine in the range 10–100 and 40–200 was obtained for the analytes	0.035–0.84 (GO/neocuproine), 0.047–0.54 (GO/batocuproine)	CRMs: M-3 HerTis, M-4 CormTis, and M-5 CodTis	350
Zn	Nuts	FAAS (with slotted quart tube)	SPE with citric acid-coated magnetic NPs (CAMNP)	30 mL of a sample at pH = 7 was mixed with 50 mg of CAMNP and stirred for 30 s. Using a magnet, the precipitated CAMNP was collected as the solid at the bottom of the tube and isolated from the supernatant. The isolated CAMNP were mixed with 1 mL of 65% HNO3 and vortexed. Achieved EHF was 13.25	12.3	CRM: SRM 1515 (apple leaves)	351

---

While it is relatively straightforward to categorise a procedure as a “digestion” (the matrix components are degraded, possibly to simple oxides of carbon, hydrogen and nitrogen) or an “extraction” (the analytes are transferred to an acceptor liquid phase whereas the matrix components are not), reports featuring preconcentration as the only novel aspect are harder to identify and so some papers that describe an overall procedure that includes preconcentration are covered in the sub-sections dealing primarily with digestion or extraction.

3.2.1 Reviews. Aspects of the sample preparation for the elemental analysis of foods have been the subject of two reviews: one devoted to MAD procedures and the other to alternatives to mineralisation. In the first (90 references), Sheehan and Furey⁴⁴ presented a tutorial overview of MAD that is slightly out-of-date. They did not appear to be aware of the performance characteristics of the current generation of commercially available systems; there was no mention of the single-vessel or autoclave mode of operation. The review is not recommended, even though the green chemistry aspects (a topic of increasing interest) were mentioned. On the other hand, Assis et al.⁴⁵ presented a critical evaluation (147 references) of the possibilities for direct solid sampling, slurry sampling, solid–liquid extraction, alkaline treatment, enzymatic digestion, emulsified systems-based procedures, and liquid sample dilution. For many of the procedures, a useful summary table was included. Although preconcentration was barely mentioned, the suitability of many of the procedures for speciation analysis was pointed out.

Two reviews have focused on solidified floating organic drop microextraction procedures. Kumar et al.⁴⁶ considered both metals and pharmaceuticals in their review. Of the 145 references cited, at least 46 were concerned with the determination of PTEs, but the summary table did not list the matrix, so it is not possible to ascertain how many of these might describe methods for the analysis of foods or clinical materials and how many might be concerned with environmental samples. There was heavy emphasis on preconcentration; however, speciation was only mentioned once in the text, even though the word “speciation” appeared in the titles of six of the articles cited (four of which were concerned with As). The table included values of LODs and “PF”s, but the abbreviation (presumably preconcentration factor) was not defined, so may simply be based on the volume ratio, and not necessarily be indicative of the sensitivity enhancement. There was barely any discussion of the green characteristics of such procedures. In contrast, Hussein et al.⁴⁷ placed a significant emphasis on greenness in their review (101 references) of procedures for the extraction of metals. The contents of some 69 articles that described the determination of metals by atomic spectrometry techniques were summarised in a table, but again the matrix was not listed, so the extent of application in the analysis of foods and clinical materials is unknown. The table included LODs and ERFs. The reviewers evaluated the procedures according to (a) the GAPI and (b) the AGREE scale. According to GAPI, an in-syringe, vortex-assisted procedure was the greenest, whereas according to the AGREE methodology, procedures with ion-pair solvents were the greenest. The application of magnetic SPE in the determination of PTEs in food was reviewed (65 references) by Zhao et al.⁴⁸ Surprisingly, “preconcentration” was not mentioned, but this was because the reviewers’ preferred terms were “enrichment” and “pre-enrichment”. Numerical values of (undefined) enrichment factors were sprinkled throughout the text, but not included in the summary table, which featured the LODs for the determination of the elements reported in 23 publications. A considerable portion of the review was devoted to the preparation of magnetic NPs and their subsequent modification to render them capable of the specific adsorption of the target element. Not surprisingly, the relative ease of phase separation by the application of a strong external magnetic field was stressed. The reviewers concluded that future developments should include consideration of automation as well as the synthesis of materials capable of capturing more than one element.

Several reviews of sample preparation for the determination of elements in biological materials have appeared. Kowa et al.⁴⁹ concentrated on flow-based methods with ICP-MS, the majority of which would be classified as FI. Their review (162 references) was sub-divided according to sample matrix: urine; blood, serum and plasma; CSF; cells and tissues; and others, such as aqueous humour and brain micro-dialysate. The reviewers considered the vast majority of the examples presented in their review concerned urine analysis, and they pointed out that FI procedures have been designed for preconcentration and for matrix separation as well as just exploiting the basic FI features of micro-sampling and on-line dilution. The contents of the relevant summary table indicated that of the 17 articles selected for inclusion, 13 were descriptions of the results of column SPE. The 22 papers selected for inclusion in the summary table for blood, serum or plasma, were mostly descriptions of procedures in which reproducible micro-sampling was the most important feature. The contribution of FI features to the greenness of the overall method was also pointed out. Oviedo et al.⁵⁰ focused on liquid- and solid-phase preconcentration procedures for the determination of trace elements by ICP-MS. They described their paper (75 references) as an “exhaustive overview” of the literature from 2000 to the present. Several topics were highlighted in summary tables: (1) main characteristics and properties of widely used nanomaterials; (2) main advantages/limitations of various preconcentration techniques; (3) total trace element preconcentration; and (4) methods involving elemental speciation. The reviewers concluded by noting that LLME has not been extensively used, and as such is a future research direction that would involve novel, greener solvents. However, in the case of ILs or DESs, the difficulty of introducing high-viscosity solvents into the spectrometer would need to be overcome. Shahtaheri et al.⁵¹ reviewed (107 references) the application of porous materials for SPE of iHg from biological and environmental samples. They identified five categories of porous materials: metal oxide NPs; carbon nanostructures; carbon active silica gel and Amberlite; imprinted polymers; and metal–organic frameworks. The performance characteristics of methods reported in 59 publications were summarised in a table, from which it can be seen that the vast majority of the methods involved quantification by an atomic spectrometry technique. The reviewers concluded with a list of seven recommendations, two of which mentioned the need to pay attention to the green characteristics of the methodology.

Several research groups have, in fact, focused on the green characteristics of sample preparation. Ferreira et al.⁵² determined As, Cd, Pb, and V in plant material by ICP-MS. They compared three sample preparation procedures: MAE with a DES, UAE with a DES, and MAD with HNO₃ and H₂O₂. They prepared three different DESs, though it is not clear why. The sample material was a RM (Brazilian Agricultural Research Corporation FO-01/2012 forage grass). No details were given, so it is not clear (a) to what extent getting the right answer was part of the evaluation, and (b) whether the material would be homogeneous at masses of 90 mg, the amount taken for the DES procedures. Five chemical metrics for the evaluation of greenness of the procedures were considered: NEMI, GAPI, Analytical Eco-Scale, AGREE and WAC. These metrics were described in some detail, from which it appears that only the WAC metric takes accuracy into account; even so, the researchers selected results that were deemed “satisfactory recoveries”: for MAD, the data for As and Pb; for MAE the data for As with the DES consisting of xylitol, citric acid and water; and for UAE, the data for Pb with the DES consisting of malic acid, citric acid, and water. It appeared that no satisfactory results for Cd were obtained, and V was not mentioned at all—so may have been listed by mistake in the abstract. They found that (a) unsurprisingly, the MAD with HNO₃ and H₂O₂ was identified as the least green by all the metrics, (b) the NEMI metric found no differences between the MAE and UAE procedures, (c) the GAPI metric did distinguish between the procedures (and identified the disadvantageous aspects of each step of the methods), (d) the Analytical Eco-Scale and AGREE identified the MAE procedure as the greenest, whereas (e) the WAC metric indicated the UAE procedure as the greenest. Zhang et al.⁵³ presented an extensive review (373 references) of micro-extraction procedures with sustainable, green solvents for mass spectrometric analysis. They noted that over the past 20 years, consideration of the principles of green analytical chemistry in the choice of solvent, which requires a balance between availability, price, recyclability, grade, synthesis, toxicity, biodegradability, performance, stability, flammability, storage, and renewability, has been responsible for the emergence of a number of green solvents including ILs, DESs, amphiphilic solvents, switchable solvents, sub-/supercritical fluids, and bio-solvents. The first part of the review (some 210 references) is a tutorial concerning the various types of solvents. The second part, devoted to applications, concerned mostly molecular analytes (and the corresponding molecular MS) with a relatively small number of ICP-MS examples. The relevant summary table only contains 12 entries, most of which are reports of methods for analysing environmental water samples. The review contains a number of colourful eye-catching figures that analytical chemistry educators might find useful. The authors of several other studies, discussed elsewhere in this Section, have evaluated their methods against one or more green metric scales.^47,54–59

3.2.2 Matrix isolation. A MC-ICP-MS method⁶⁰ for the determination of Am, Pu, and U, in the hair and nails (toe and finger) of occupationally or medically exposed individuals, involved the separation of the three analytes by a sequential SPE following MAD with HNO₃ and H₂O₂. No information about washing the samples to remove surface contamination was provided. The digest, after adjusting the oxidation state of Pu to Pu^IV, was passed through three resin SPE columns containing TEVA® (trialkyl methylammonium nitrate) to retain Pu; UTEVA® (dipentyl pentylphosphonate) to retain U and N,N,N′,N′-tetra-n-octyldiglycoamide, to retain Am; connected in series. After rinsing (30 mL of 3 mol L⁻¹ HNO₃), the elements were separately eluted: Pu, with 15 mL of 0.02 mol L⁻¹ HCl–0.005 mol L⁻¹ HF–0.0001 mol L⁻¹ TiCl₃; U, with 6 mL of 0.02 mol L⁻¹ HNO₃–0.005 mol L⁻¹ HF, and Am with 15 mL of 0.02 mol L⁻¹ HCl–0.005 mol L⁻¹ HF. The overall dilution factors for the 31 samples ranged from 8 to 104. The method was validated by the analysis of a CRM (NCS DC 73347 hair), and the LODs for ²⁴¹Am, ²³⁹Pu and ²³⁸U were 15 pg kg⁻¹, 20 pg kg⁻¹ (both with dilution factor 28) and 36 ng kg⁻¹ (dilution factor of 54), respectively. The researchers also measured the ²³⁹Pu : ²⁴⁰Pu, ²³⁵U : ²³⁸U, and ²³⁶U : ²³⁸U isotope ratios and concluded that measurement of these elements in hair and nails supported the use of keratinous materials for the long-term monitoring of actinide exposure.

Guo et al.²⁸ determined the isotope composition of Cu, Fe and Zn in nine biological reference materials by MC-ICP-MS following separation of the analytes by SPE. The CRMs analysed included three plant materials (skimmed soybean, whole soybean and pumpkin) and six animal materials (human hair, human serum, prawn, pig liver, pig kidney and yolk). Samples (100–600 mg) were first mineralised in a multi-step procedure that consisted of (1) wet ashing with HNO₃ and H₂O₂, (2) dissolution of dried residue in HF, HNO₃ and HCl aided by pressurised microwave heating, and (3) dissolution of the dried residue in 1 mL 6 mol L⁻¹ HCl. This solution was then loaded onto a column containing 2 mL of an anion-exchange resin (AG MP-1M, 100–200 mesh). Matrix elements were removed using 6 mL of 6 mol L⁻¹ HCl. Then Cu, Fe and Zn were eluted in sequence, using 26 mL of 6 mol L⁻¹ HCl, 6 mL of 0.5 mol L⁻¹ HCl and 13 mL of 0.5 mol L⁻¹ HNO₃, respectively. For some samples, a further clean-up with a second column was implemented (consisting of AG MP-IM for Cu and Zn; and AG1X8 for Fe). For eight of these CRMs, this was the first time δ⁵⁶Fe, δ⁶⁵Cu and δ⁶⁶Zn values have been reported; values for the pig kidney material were already available.

The rapid determination of radioactive pollutants released after a nuclear accident or explosion is crucial. To this end, Kavasi and Sahoo⁶¹ have developed a procedure for the determination of ⁹⁰Sr in milk by TIMS. The procedure involved dry ashing, SPE to separate interfering matrix components (such as zirconium, whose major isotope is ⁹⁰Zr, and ⁹⁰Y, which may be present in equilibrium with its parent isotope ⁹⁰Sr, a product of nuclear fission) and concentration by evaporation. They started with 1 mL of milk, evaporated the water (20 min) and after ramping the temperature in 100 °C increments, ashed the residue at 700 °C for 10 min. After cooling, the ash was dissolved in 2 mL of 8 mol L⁻¹ HNO₃ (total time 90 min). The 2 mL were loaded first onto a SPE column (42 mm × 5 mm) of N,N,N′,N′-tetra-n-octyldiglycolamide (100–150 μm particle size) that retained zirconium and yttrium and then onto a similar-sized column of 4,4′(5′)-di-tert-butylcyclohexano 18-crown-6 (crown ether) in 1-octanol that retained the Sr together with barium, calcium, magnesium and potassium. After rinsing, Sr was eluted with 1.1 mL of 0.05 mol L⁻¹ HNO₃ and the eluent passed through a third column containing an uncoated, inert polymeric support to remove traces of organic compounds. Finally, the solution was evaporated to a few μL (200 °C, hot plate) and transferred to the filament of the TIMS instrument. The time needed for the column clean-up and evaporation was about 40 min. The minimum detectable activity concentration of ⁹⁰Sr, which is affected by the stable Sr concentration, was approximately 500 mBq kg⁻¹ (100 ag per g). They were able to analyse 198 samples in 30 h and for the first time reported ⁸⁷Sr : ⁸⁶Sr isotope ratios in Japanese milk samples.

3.2.3 Digestion. A major problem in the determination of Si by atomic spectrometry is the loss of Si by volatilization that occurs when HF is needed for sample preparation. Arslan and Lowers⁶² showed that significant losses occurred when evaporating solutions of the water-soluble Si species hexafluorosilicic acid (H₂SiF₆), and sodium metasilicate (Na₂SiO₃) that also contained HF. For the analysis of biological materials, they developed a two-stage procedure in which a closed-vessel hot plate (140 °C) digestion was performed with concentrated acids, i.e. 4 mL HNO₃, with either 1 mL HCl or 1 mL HCl–1 mL HClO₄, followed by evaporation to incipient dryness at 120 °C. This was followed by a second closed-vessel digestion with 0.5 mL of conc. HNO₃ and 0.5 mL of conc. HF at 130 °C to dissolve silicates. Digestates were diluted to 10 mL and the solution, containing about 5% HNO₃ and 5% HF, was analysed by ICP-MS with an instrument fitted with an HF-inert sample introduction system. The LOD was 110 μg L⁻¹ and the method was validated by the analysis of CRMs (NIST SRM 1547-peach leaves, SRM 1566b-oyster tissue, SRM 1572-citrus leaves, SRM 1573a-tomato leave, SRM 1575-pine needles, SRM 1577b-bovine liver; NRCC DOLT-3 and DOLT-4-dogfish liver, TORT-1-lobster hepatopancreas). Analysis of lung tissues from individuals occupationally exposed to silica dust indicated the presence of some water-soluble silicates, but most Si was present in the particulate fraction.

A number of digestion procedures that do not involve microwave irradiation have been described. For the determination of Ni in peach juice by FAAS, Ayyildiz et al.⁶³ developed a procedure in which samples were digested by a UV-assisted Fenton reaction. To 0.2 mL of sample were added 7.5 mg of hybrid magnetic NPs (Fe₃O₄ and stearic acid), 200 μL of conc. H₂O₂ and 40 μL of conc. HNO₃. The mixture was subject to UV irradiation (4 W) in a lab-made reactor system for 90 min. After magnetic separation the solution was analysed without further treatment. The LOD was 0.14 mg L⁻¹ and the procedure was applied to some real samples, but these did not appear to contain a measurable concentration of Ni. The only results presented were for recoveries of four spikes measured by calibration against standards prepared by spiking one of the samples. These ranged from 92% to 106%. It is not clear what would have happened if the peach juice had been filtered and introduced directly into the spectrometer.

Vieira et al.⁶⁴ reported results obtained following digestions in closed vessels with a convective heating system. The commercially available device was described in detail, as was the optimisation that resulted in the following conditions: 250 mg sample, 2.0 mL of HNO₃, 1.5 mL of H₂O₂, and block temperature set to 240 °C, producing liquid-phase temperatures up to 190 °C. The final volume was 25 mL. The procedure was part of the ICP-OES determination of B, Ca, Cu, Fe, K, Mg, Mn, P, S, and Zn in 16 test items from an interlaboratory trial, consisting of leaves and grains from a variety of plants, with a wide range of macro- and micro-nutrient concentrations: four samples of maize (Zea mays) grains and one sample each of soy (Glycine max), sorghum (Sorghum bicolor), eucalyptus (Eucalyptus sp), Figueira (Ficus sp), acerola (Malpighia emarginata), persimmon (Diospyros kaki), macadamia (Macadamia sp), teak (Tectona grandis), grass (Poaceae sp), palm tree (Arecaceae sp), pau rei (Pterygota brasiliensis) and caja mirin (Spondias mombin) leaves. The method was also applied to the analysis of three CRMs (NIST SRM 1547-peach leaves, SRM 1515-apple leaves, SRM 1570a-spinach leaves), all of which were also analysed using two other procedures: a single reaction chamber MAD with HNO₃ and H₂O₂, and a hot-block open-vessel digestion with HNO₃ followed by HClO₄. In terms of “digestion efficiency”, measured as RC, the convective heated system achieved a significantly poorer performance (RC 700 mg L⁻¹) than the MAD procedure (visual inspection of the relevant figure suggests a value of around 350 mg L⁻¹) and the HClO₄ procedure (about 10 mg L⁻¹), although the authors considered values below 2000 mg L⁻¹ to be acceptable. The LOQs ranged from 0.1 mg kg⁻¹ (Mn) to 60 mg kg⁻¹ (Ca). Accurate results were obtained for all elements in the three CRMs, except for Fe in SRM 1515 and Zn in SRM 1570a (the results for these elements by the other two procedures were also significantly lower than the certificate values). The statistical evaluation of the agreement between the results for the analyses of 16 real samples was not really satisfactory. The researchers applied an unpaired t-test concluding that there was “no significant difference among digestion methods at a 95% confidence level”, and displayed the data as log–log plots, whose correlation coefficients and slopes were calculated (possibly incorrectly, as different values for the same plot are reported), but no conclusions were drawn.

For the determination of Ba, Ca, Cd, Cr, Cu, Fe, Hg, K, Mg, Mn, Pb and Zn in instant soups by MIP-OES, Luckow et al.⁶⁵ digested samples in glass tubes in a heating block. To 200 mg of sample were added 1.0 mL of conc. HNO₃ and 1.0 mL of 30% (v/v) H₂O₂ and the tubes were heated at 150 °C for 3 h. The heating block was mounted in a box that allowed cool air to be blown over the upper parts of the tubes during the digestion. The final volume was 20 mL. The procedure was applied to CRMs (NIST SRM 1568-rice flour, SRM 1846-infant formula) and to eight real samples. The LODs ranged from 0.001 mg kg⁻¹ (Cr) to 0.3 mg kg⁻¹ (Cd, Fe). No detailed statistical analysis of the results for the CRMs was presented, but visual inspection of the results shows that the relative measurement errors ranged from −20% to +19%, several of which are probably significant. All the analytes were found in all the samples except for Cu in three samples. Zhang et al.⁶⁶ investigated a dry ashing procedure in the preparation of food samples for the determination of Se by ICP-MS. Heating in a stream of O₂ was achieved by focused IR irradiation in a lab-made system that automated all the steps: pre-oxidation, dry ashing, sample transfer, dissolution of the ash, constant volume control and homogenisation of the solution (via ultrasound). The system could also generate O₃ continuously. The researchers indicated that since O₃ is toxic, operations were conducted with a gas mask worn in a well-ventilated environment. Several metal nitrate salts as ashing aids were investigated, and palladium nitrate was selected as the most effective, based on the prevention of loss by volatilization. To a 250 mg sample, 500 μg of Pd were added (dropwise as a solution, but concentration not given). After an initial oxidation with O₃ at 150 °C (to convert organoselenium to SeO₂), the sample was ashed at 500 °C in a stream of O₂ for 30 min and the ash dissolved in 25 mL 10% HNO₃. The LOD was 0.021 μg g⁻¹ and the procedure was validated by the analysis of seven CRMs (NCS GBW07602-bush leaves, GBW10020-citrus leaves, GBW10022-garlic, GBW10023-laver, GBW10025-spirulina, GBW10048-celery, GBW10052-green tea) with relative errors ranging from −2.4% to 5.7%. The procedure was applied to three real samples (scallion powder, Astragalus propinquus powder and Se-enriched tea) in all of which Se was found.

Pereira Jr et al.⁵⁷ determined the concentrations of As, Ca, Cd, Cr, Cu, Fe, K, Mg, Mn, Na, P, Pb, Sr and Zn in medicinal herbs by ICP-OES following digestion in a commercially available closed digestor block. The researchers employed a fractional factorial design to optimise the experimental parameters, which included concentrations of HNO₃ and H₂O₂, temperature and time, but not sample mass or final volume. The optimisation featured the analysis of a “certified” RM (Agro C1003a tomato leaves from CENA - Universidade de São Paulo, Piracicaba, SP, Brazil) but the figure of merit was not understandable and it was not possible to source the CRM certificate. The researchers also measured the RC by ICP-OES, using citric acid as standard, and residual acidity (by titration with NaOH, though the wrong indicator was used), but neither of these parameters were considered in the optimisation. The optimised procedure consisting of adding 1.38 mL of conc. HNO₃ and 1.0 mL of 30% (m/m) H₂O₂ to 1.0 g of sample in a PTFE flask followed by dilution to 5.0 mL with H₂O. The flasks were closed and heated at 180 °C for 120 min. The final volume was 15 mL. Under these conditions, the RC was 4.9 mg L⁻¹ (the lowest value obtained for all 12 optimisation experiments), the residual acidity was 2.73 mol L⁻¹ (68% of the original value of 4.0 mol L⁻¹) and the LODs ranged from 0.06 mg kg⁻¹ (Cd) to 1.9 mg kg⁻¹ (P). The method was applied to the analysis of two additional CRMs (Agro C1005a-sugar cane and NCS DC 73351-tea). Values not significantly different from the certified ones were obtained for almost all analytes in all three CRMs, though, in several instances, this was because of the relatively large ± term associated with the certified values. The method was applied to the analysis of 10 medicinal herbs, in all of which all of the analytes were found, except for As (<LOD of 0.5 mg kg⁻¹ in all samples) and Cu (<LOD of 2.2 mg kg⁻¹ in three samples). Finally, the green characteristics were evaluated according to the AGREE scale and were found to be better than those of a published MAE procedure, involving a DES, for the same kind of samples.

Several research groups have investigated modifications to the widely used digestion with conc. HNO₃ and H₂O₂ in a commercially available microwave oven device in which a number of vessels are mounted in a carousel and subjected to the same microwave irradiation. Shabanova et al.⁶⁷ investigated digestions of four plant-based CRMs in a MultiVIEW tunnel-type system (SPC SCIENCE, Canada) in which reaction vessels (75 mL) with PTFE inserts were installed in a single, 12-position rack located in a Teflon-lined stainless steel tunnel. Each position was exposed to individual focused microwave radiation from a magnetron (power 350 W). Temperature was measured by 12 independent IR sensors whose output was used to control the magnetron output. After extensive optimisation, the best digestion conditions were identified as three-stages of heating, a sample mass of 0.5 g and the separate and sequential addition of 4 mL HNO₃, 1.5 mL H₂O₂, 1 mL HCl and 0.05 mL HF. After dilution of the digests to a final volume of 15 mL with water, the determination of Al, B, Ba, Ca, Cu, Fe, K, Mg, Mn, Na, Ni, P, Rb, Si, Sr, Ti, V and Zn was achieved using ICP-OES. The researchers monitored residual acidity (by acid–base titration) but not RC. The introduction to the paper contains a comprehensive survey of MAD. The effects of UV irradiation on the MAD of starch and skimmed milk with HNO₃, together with effects of the addition of other reagents including O₂, H₂O₂ and HCl, were investigated by do Nascimento et al.⁶⁸ The UV lamp was a Cd EDL powered by the microwave radiation inside the oven cavity, an accessory available from the supplier of the oven. The output varied (1–10 W) as the microwave power in the oven cavity varies. They determined Ca, Cu, Fe, K, Mg, Mn, Mo, Na and Zn by ICP-OES, and Cd and Pb using ICP-MS. They investigated a number of parameters, but did not discuss sample mass, which could have been 100 mg. Nor did they discuss the final volume, which could have been the 15 mL needed to make best use of the UV radiation. They concluded that the best conditions were (approximately 15 mL) 0.1 mol L⁻¹ HNO₃, 2 mL of 30% H₂O₂, and 20 bar of O₂, with 30 min of microwave heating and UV irradiation, with simultaneous cooling by air delivered at 190 m³ h⁻¹. To get accurate results for Fe, they added 100 μL of 0.08 mol L⁻¹ HCl to the digestion. They monitored the RC content by measuring the C by ICP-OES with tartaric acid standards, but expressed the results as a percentage of the original C rather than as a concentration in, say, mg L⁻¹, which would have been more informative when comparing the performance with that of other procedures reported in the literature. The LODs ranged from 0.004 μg g⁻¹ (Cd) to 17 μg g⁻¹ (Na). The procedure was applied to the analysis of CRMs (BCR-151 skim milk powder and NIST SRM 8414 bovine muscle) with results that were not significantly different from the certified values (apart from Fe, for which HCl was needed), though BCR-151 is only certified for the values of Cd, Cu, Fe and Pb. The procedure was applied to two real samples, skim milk powder and potato starch: Ca, K, Mg, Mn, Na and Zn were detected in the skim milk powder, and K, Mg, Mn and Na were detected in the starch, though the value given for Mn (0.066 μg g⁻¹) was below the LOD (0.12 μg g⁻¹). The RC content was 0.3%. The researchers pointed out that it was possible to obtain acceptable results for MAD-UV digestions with much milder reagents, possibly even just water. Cauduro et al.⁶⁹ compared the results obtained by microwave-induced combustion (MIC) for the ICP-MS determination of As, Cd, Hg, and Pb in honey with those obtained by two MADs (with HNO₃ or HNO₃ and H₂O₂), one of which involved the single-reaction chamber protocol. The procedure involved mixing 0.8 g of honey with 0.4 g of microcrystalline cellulose and 100 μL of 6 mol L⁻¹ NH₄NO₃, then wrapping in PE film squares. These were then placed on a filter paper disc wetted with 50 μL of 6 mol L⁻¹ NH₄NO₃ on the quartz holder, which was inserted in the vessel already containing 6 mL of the absorbing solution (0.1 mol L⁻¹ HNO₃). A three-step heating program was applied (total time 8 min) followed by 20 min cooling and the final volume made up to 25 mL. The LODs were 1.1 ng g⁻¹, 1.7 ng g⁻¹, 2.9 ng g⁻¹ and 4.6 ng g⁻¹ for As, Cd, Hg and Pb, respectively. Validation was by spike recovery at three different concentrations indicting no analyte loss. There were no significant differences between the result of the three procedures for the analysis of nine real honey samples, but Hg could not be detected in any of them, Cd in only two, Pb in three, and As in six. The researchers pointed out that the advantages of the MIC procedure were reduced time (only 28 min including the cooling step), high sample mass, negligible residual acidity and C content (<25 mg L⁻¹). Apine et al.⁷⁰ claimed to have developed an improved method for the determination of Cu, Fe and Mn in non-centrifugal sugars (the unrefined product of sugarcane juice evaporation) by ICP-OES in which MAD was used to prepare the samples. A lengthy optimisation of the MAD based on response surface methodology and central composite design was described, but the details of the procedure selected were not obvious. A partial listing of parameters is given in the abstract. Unnecessary space was devoted to the routine optimisation of the ICP operating conditions. The LODs were 0.006 mg kg⁻¹ (Cu), 0.005 mg kg⁻¹ (Fe) and 0.005 mg kg⁻¹ (Mn), respectively. The method was applied to 10 real samples in all of which all three analytes were found. Spike recoveries (from 0.5 mg kg⁻¹ to 10 mg kg⁻¹) to an 11^th sample ranged from 94% to 102%. Givelet et al.⁷¹ examined the role of MAD in an investigation of what they called “a relatively large variation” in the results for Al reported by the laboratories participating in a PT on trace elements in cocoa powder. The original study was carried out in 2020 by the European Union Reference Laboratory for metals and nitrogenous compounds in feed and food, and the 28 participating laboratories were, presumably, considered competent. They assumed that the variation was due to the digestion conditions and not due to extraneous factors, such as sample contamination, and they collected information about the digestion procedures used from 17 of the participating laboratories, but as only limited conclusion could be draw from these data, they investigated the effects of sample mass, reagents, temperature and time. The digests were analysed by ICP-MS. In addition to the cocoa powder, they also investigated the solubilisation of Al from three CRMs (NIST SRM 1566b-oyster tissue, SRM 8436-durum wheat flour and SRM 1572-citrus leaves). They implemented a single cycle of the alternating variable search protocol and concluded that food samples were digested satisfactorily with a mix of 3 mL H₂O–2 mL HNO₃ for 25 min at a temperature of at least 240 °C. The maximum sample mass investigated was 500 mg. Interestingly, they did not find it necessary to add H₂O₂. In contrast, Senger et al.⁷² found that to digest five active pharmaceutical ingredients used in hypertension treatment for the determination of Ag, As, Ba, C, Cd, Cr, Cu, Li, Mo, Ni, Pb, Sb, Se and V by ICP-OES, only H₂O₂ was needed. In comparing the H₂O₂ digestion to one employing both H₂O₂ and HNO₃, they found that for a 500 mg sample mass, 6 mL of 50% (w/w) H₂O₂ and a digestion temperature of 250 °C, the RC was <2400 mg L⁻¹ for a final volume of 15 mL. When comparing the RC contents of the digests of compounds containing aromatic rings, there was no difference; however, not surprisingly, (a) the residual acidity of the H₂O₂ digest was lower than that of the HNO₃ digest and (b) the H₂O₂ procedure scored higher on the Analytical Eco-Scale (a previously described greenness metric). The LODs ranged from 0.02 μg g⁻¹ (Ba, Co, Li) to 0.3 (Se) μg g⁻¹, but even so for Ag, As, Ba, Cd, Cu, Li, Pb, Sb and Se the concentrations in all samples were <LODs. Cobalt was found in one sample, Cr in three, Mo in two, Ni in two, and V in one. Given the wide availability of commercial rotor-based and autoclave-style MAD systems, it seems unlikely that there would be much interest in the construction of one’s own flow-based MAD instrumentation. Nonetheless, Hallwirth et al.⁷³ have constructed and evaluated such a system. The samples were injected as slurries (12 mL, 1% m/v), sandwiched between two 2.5 mL zones of 6 mol L⁻¹ HNO₃, into a single-line FI system and carried at 5 mL min⁻¹ through 22 mL of PFA tubing(id 2 mm) carefully coiled (according to the results of extensive simulation calculations) in a pressurised (40 bar) waveguide. The solution emerging from the system was collected, but the volume was not specified—maybe 50 mL. They applied the procedure to three CRMs (BCR-185R bovine liver, IAEA-A-13 animal blood and NIST SRM 1547 peach leaves), which were digested with a 5 mL mixture of 6 mol L⁻¹ HNO₃–3 mol L⁻¹ HCl. After 5 min in the digestion system, the RC concentration was <50 mg L⁻¹. They measured Al, As, B, Ba, Ca, Cu, Fe, K, Mg, Mn, Na, Rb, Sr and Zn by ICP-OES and ICP-MS, and they also determined the elements following a conventional closed-vessel, batch MAD. For the flow through procedure, relative measurement errors ranged from −10% to +10%, though there were problems with Fe in NIST SRM 1547 (−14%) and Cd in BCR-185R (+21%). No statistical evaluation of the results was presented. The residual acid concentration was about 1.6 mol L⁻¹ for all CRMs and the throughput was 12 h⁻¹.

3.2.4 Extraction. Several research groups have reported the determination of metals in edible oils by methods that do not involve an initial acid digestion. Sorouraddin et al.⁷⁴ separated Cd and Pb as the 8-HQ complexes from 0.25 g oil samples diluted with 3 mL of CHCl₃ to which 6 mg of solid cobalamin had been added. After mixing on a vortex for 4 min, the solid was collected, centrifuged for 5 min, dissolved in 220 μL of 5% (v/v) HNO₃ and the analytes determined by FAAS. The LODs were 0.48 μg kg⁻¹ and 0.34 μg kg⁻¹ and the method was applied to the analysis of a CRM (Enviro MAT HU-1 used hydrocarbon oil). However, the certificate values quoted in the paper are not correct and the method had a significant negative bias. The recovery of spikes (10 and 50 μg kg⁻¹), added as nitrate salts in acetone, ranged from 86% to 101%. When applied to the analysis of omega 3, olive, almond, hazelnut, sunflower-1, sunflower-2 and fish oils, both analytes were found in fish oil, almond oil, olive oil and sunflower-2 oil. Pb was not detected in omega 3, hazelnut or sunflower-1 oil, and Cd was not found in omega 3 oil. A procedure based on disruption of a three-component solution was devised by Trevelin et al.⁷⁵ in which 2.45 mL octanol and 0.05 mL of 3.5 mol L⁻¹ HNO₃ were mixed with 2.5 mL of the vegetable oil forming a homogeneous solution. On the addition of 1 mL of H₂O and shaking, two phases formed, which were separated by centrifugation, and Cu and Ni in the lower aqueous phase determined by GFAAS. The procedure was applied to several oils, virgin and extra virgin olive, canola, coconut, sunflower, corn and soybean, and evaluated, using a paired t-test, by comparison with the results of the analysis of the same materials by an ICP-MS method in which oils and organometallic standards were diluted in xylene then directly injected into the spectrometer. The LODs for the GFAAS method in the oil samples were 0.05 ng g⁻¹ for both metals, which were detected in all 11 samples. Recoveries of spikes, added as organometallic compounds, to four extra virgin olive oils ranged from 91% to 106%. Several research groups reported methods based on emulsion breaking. For the determination of Cr and Pb in edible vegetable oils (sunflower and olive) by FAAS, Saleem et al.⁷⁶ extracted the metals into the aqueous phase produced by breaking an emulsion formed by shaking 7.0 g of pristine oil sample with 0.5 mL of surfactant (1.5% and 3.5% Triton X-100 for Cr and Pb, respectively) and 2.0 mL of HNO₃ (10⁻³ and 10⁻² mol L⁻¹ for Cr and Pb, respectively). The phases were separated by centrifuging (4000 rpm, 2 min) and the bottom aqueous layer taken through a DLLME preconcentration procedure (details in Table 1). The LODs of 0.5 μg kg⁻¹ and 1.5 μg kg⁻¹ for Cr and Pb, respectively, were not low enough to determine either element in any of the samples. Costa et al.⁷⁷ devised an emulsion-breaking method for the determination of Al, Ca, Cr, Cu, K, Mg, Mn and Zn in vegetable oils, in which the emulsion was formed by mixing 5 mL of hot (50 °C) oil with 1.7 mL of a DES (ChCl–oxalic acid–water) and 1.3 mL of 5% (v/v) Triton X-100. After vortexing for 36 s and standing for 5 min to allow the extraction to proceed, the emulsion was broken by microwave heating in a domestic oven (350 W, 10 s), and the aqueous phase, suitably diluted, was analysed by ICP-OES. The LODs ranged from 0.001 mg kg⁻¹ to 0.01 mg kg⁻¹ and the method was applied to the analysis of sesame, soybean, Brazil nut and copaiba oils and cupuacu butter, in all of which all analytes were found. No CRM was analysed, but the results were compared with those of a method involving MAD with HNO₃ and H₂O₂. Spike recoveries for three elements from three samples at three concentrations ranged from 91% to 107%. Spikes were added as a oil standard (from SCP Science, Quebec, Canada), but no further information was provided. The researchers pointed out that, in addition to a favourable score on the AGREEprep metric for sample preparation scale, the procedure was fast, with the capability of processing 78 samples per h. Liu et al.⁷⁸ determined As, Cd, Cr, Hg, Ni and Pb in palm oil by MIP-OES following breaking of the emulsion formed with Triton X-100 and HCl by heating in an ultrasonic bath (90 °C, 40 min). The LODs ranged from 0.025 μg L⁻¹ to 0.29 μg L⁻¹ and the method was validated by spike recoveries and comparison of the results with those of a method involving MAD and ICP-MS/MS. The original paper is in Chinese and no further information is available. Adolfo et al.⁷⁹ applied the procedure to the ETAAS determination of Co and Ni in chocolate bars. The emulsion was created in a 10 mL plastic syringe by shaking 0.08 g of melted chocolate with 4 mL of a warm solution containing 4% (m/v) Triton X-100 and 12% (v/v) HNO_3, added through the syringe tip by micropipette. To break the emulsion, it was passed through a 0.22 μm nylon membrane that retained both solid chocolate particles and oil. The researchers showed that the method, with a total extraction time of only 1 min, was as effective as those in which heating (90 °C, 50 min) or centrifugation (4000 rpm, 25 min) were used to break the emulsion. Results were compared with those obtained by MAD and ETAAS, indicating recoveries from 92% to 105%. The LODs in the solid were 25 μg kg⁻¹ and 49 μg kg⁻¹ (the procedure involved a dilution of approximately 50). Both analytes were found in all 11 samples (both milk and dark chocolate).

A number of procedures featuring DESs as the extraction solvent have been reported. For the determination of Se in cereal and biofortified materials by ICP-OES, Sihlahla et al.⁵⁵ extracted the samples with a DES of ChCl and phenol (1 + 3 molar composition). To 100 mg of sample were added 4 mL of the DES, the mixture shaken (vortex, 3 min), heated at 125 °C for 25 min, cooled, acidified with 4 mL of 3 mol L⁻¹ HNO₃ and the resulting solution made up to 25 mL. For the comparison ICP-MS method, samples were extracted/digested with aqua regia and H₂O₂. In addition, the method accuracy was established by the analysis of CRMs (IRMM 804 rice flour, NIST SRM 1567b wheat flour). The LOD was 0.0011 μg g⁻¹ and the procedure was applied to the analysis of corn flakes, wheat biscuits, instant porridge, meat, bone meal, salmon meal whey meal, shrimp meal, pure whole mussel meal, tuna meal, dried plasma, dried haemoglobin, mixed fish meal and crushed seashells; Se was found in all samples. The researchers also evaluated the greenness of the method according to three metrics: NEMI, the Analytical Eco-Scale, and AGREE. They were critical of the NEMI, indicating that it was the least informative tool for the evaluation of greenness, because it does not provide a detailed visualization of the analytical process appraisal and is restricted to qualitative criteria. They considered that the AGREE metric comprehensively included integrated information on the entire method and extensively assessed all individual steps of the method. Needless to say, their new method scored sufficiently highly (0.76 out of a maximum of 1.00) on this scale to qualify as green. In an extensive study of the effectiveness of DESs and solutions of their components (ChCl and various carboxylic acids) Ninayan et al.⁸⁰ showed that the most effective extraction of Cd, Co, Cr, Cu, Fe, Pb and Zn from solid-phase food products was with a DES containing ChCl, lactic acid and water. To 100 mg of sample was added 1 g of extractant and the mixture sonicated at 40 °C for 20 min. The phases were separated by centrifugation (5 min), then 500 μL of the DES phase was mixed with 2 mL of 0.1 mol L⁻¹ HNO₃. The ICP-OES results were compared with those obtained following MAD with HNO₃ and H₂O₂. The LODs ranged from 0.1 mg kg⁻¹ to 0.25 mg kg⁻¹ and the method was applied to two bovine liver CRMs (BCR-186 and NIST SRM 1577c). The concentrations of Co, Cr and Pb could not be measured in the BCR-186 material. The only real sample was a spiked chicken breast that did not appear to contain measurable concentrations of any of the analytes. In developing a method for the determination of total As and inorganic As species in rice, Fattahi et al.⁸¹ evaluated the performance of 16 DESs and selected L-menthol with octanoic acid in the molar ratio of 1 : 1.5. Further details of the method are given in Table 1. Botella et al.⁵⁹ also evaluated the performance of several DESs in developing a method for the determination of Se in nuts by ICP-MS. However the DES extraction was only the third step in a procedure that started by defatting 25 g of ground sample with three 30 mL portions of CHCl₃. The next step was extraction of 0.5 g of the dried defatted sample with 10 mL of 0.2 mol L⁻¹ H₂SO₄ and then, to a 5 mL portion were added 42 μL of 6 mmol L⁻¹ APDC, to form a complex with Se^IV, and 50 μL of a DES made from thymol and decanoic acid. After vortexing and centrifugation, the DES was solidified in an ice-bath and removed with a spatula. This phase could not be introduced into the spectrometer because it was too viscous and too hydrophobic, and so the Se^IV was determined from the difference of the Se in the acid extract (assumed to be Se^VI and Se^VI) and the Se remaining after the DES extraction. Organo-Se was estimated by subtracting the acid-extract Se (assumed to be inorganic Se) from the total Se determined following a hot plate digestion with HNO₃ and H₂O₂. The LODs were 0.31 μg L⁻¹ for Se^IV and 0.24 μg L⁻¹ for Se^VI. The method accuracy was determined by the analysis of a CRM (NIST SRM 1643e trace elements in water). The procedure was applied to seven different nut samples, in all of which both Se^IV and Se^VI were found, except for Se^IV in peanut and Se^VI in hazel nut. The values in the table of results were given in the rather unhelpful units of μg L⁻¹. The researchers compared their method to seven previously reported methods for Se speciation analysis in food samples by ICP-MS using the AGREE software. Although they found that their method was greener than the others, they did not include the defatting step, in which 90 mL of CHCl₃ was used and then discarded. Nor did they include the acid/peroxide digestion step in the determination of total Se. For the determination of Pb in biological materials by ETAAS, Germer et al.⁵⁸ developed a procedure in which 40 mg of sample was extracted with 400 μL of a DES made from ChCl and oxalic acid in a 1 : 1 molar ratio. After adding 10 to 15 glass microbeads (1 mm diameter) to improve homogenisation, the mixture was heated for 30 min at 95 °C, centrifuged for 10 min (3500 rpm) and the supernatant separated, diluted, and acidified with 1% (v/v) HNO₃ (no details of volumes were given). The LOD was 0.04 μg g⁻¹ and the procedure was applied to the analysis of four CRMs (BCR-186 pig kidney, ERM® CE278 mussel tissue, NIST SRM 8414 bovine muscle and SRM 1566b oyster tissue), whose suppliers were not given but are probably BCR, JRC and NIST, respectively. The results for the bovine muscle are problematic, the “agreement %” is calculated incorrectly (it should be 136% not 110%) but, as the certificate value is given as “0.308 ± 0.240” μg g⁻¹, (such a large uncertainty seems unlikely) there is, in fact, no significant difference with the measured value of 0.42 ± 0.07 μg g⁻¹. It is also of interest to note that the certified Pb content of the bovine muscle, pig kidney and oyster tissue were the same. No real samples were analysed. The greenness of the proposed method was evaluated applying the green star and ecoscale metrics. For the former metric a value of 95 (out of 100) was calculated, which the researchers characterised as “excellent.” Although the sample material was a forage grass RM, a comparison of two DES procedures is of interest.⁵² The procedures involved either microwave assistance or ultrasound assistance and the three different DESs were evaluated, whose extraction capabilities varied. The paper is discussed in more detail in Section 3.2.1 above.

Patel et al.⁸² compared the results obtained for the determination of 11 elements in bread, noodles and chips following treatment with hydrolytic enzymes with those obtained by “conventional acid digestion”. To 100 mg of ground sample were added some buffer solution, followed by 2 mL of enzyme (cellulase, pectinase, xylanase, or amylase or mixtures thereof) solution, then the mixture was shaken at 37 °C for 24 h. After settling for an hour, 1 mL was diluted 100-fold with 0.1% HNO₃ and filtered (0.45 μm) before analysis by ICP-MS. For acid digestion, 50 mg of ground sample was transferred to a 10 mL vial to which 5 mL of a 3 + 1 mixture of 70% HNO₃–30% H₂O₂ were added, followed by heating on a hot plate at 150 °C until a clear solution was obtained. This solution was also diluted 100-fold. Not surprisingly, because of the overall dilution, several elements (Cd, Hg, Mo and Ni) were not detected in any of the food samples by either method. In addition, As and Cu, were not detected in bread, Cr and Mn were not detected in noodles, and Zn was not detected in bread and chips. For those elements that were detected, notably Pb and Sn, the enzyme treatment results were higher than those obtained following the acid digestion. The researchers indicated that this was due to “superior release from the food matrix” an explanation that seems highly unlikely given that clear solutions were produced by the acid digestion. The possible role of contamination was not discussed. In the introduction to their paper, the researchers raised the possibility of analyte loss, but did not follow-up with any experiments. Amazingly, there were no results for the analyses of CRMs or of spike recovery experiments.

Several researchers have devised UAE methods. For the determination of Ca, Fe, Mg and Mn in cane syrup by FAAS, Alves et al.⁸³ mixed 0.5 g of sample with 5 mL H₂O (for the determination of Fe and Mn) or with 5 mL of dilute HNO₃ (for Ca and Mg). The mixtures were sonicated (37 KHz, 100 W, 25 °C) for 10 min and diluted to 10.0 mL (Fe and Mn) or 50.0 mL (Ca and Mg). For accuracy evaluation, samples were also subjected to MAD with HNO₃ and H₂O₂ (sample mass 1.0 g, final volume 25 mL), a procedure that took 50 min. All four elements were found in five samples, and there were no significant differences between the results of the two methods, nor were there significant differences between the slopes of the external calibration plots and the slopes of the standard addition plots (for the same elements). The LODs, interestingly, were lower for the UAE procedure, ranging from 0.04 μg g⁻¹ to 0.3 μg g⁻¹, but the researchers did not comment. The UAE method got scores of 0.66 out of 1.00 (Fe and Mn) and 0.39 (Ca and Mg) on the AGREEprep scale. In contrast, the MAD method scored 0.21. For the determination of Sn in canned tomatoes by HG-ICP-OES, Welna et al.⁸⁴ investigated UAE with various reagents, including HCl, HNO₃, CH₃COOH, aqua regia or TMAH. The optimised procedure, selected on the basis of the best LOD, involved UAE of 2 g of drained pulped sample with 2.0 mL of freshly prepared aqua regia for 15 min at room temperature. No details of power or frequency were given, and although time and temperature could be controlled, their possible effects did not appear to have been considered. The resulting suspensions were made up to 25.0 g with water, centrifuged (12 000 rpm, 10 min) and filtered. For comparison, samples were subjected to MAD with HNO₃ and H₂O₂ (sample mass 2.0 g, final solution 25 g), in a procedure that took 45 min for the digestion. The UAE procedure was applied to seven samples, but comparison of the results with those of the MAD method was presented for only one sample, and spike recoveries (at two concentrations) were performed on only one sample. All samples contained measurable concentrations of Sn, with the tomatoes in unlacquered cans having concentrations an order of magnitude higher than those in the lacquered cans. The LOD was 0.74 ng g⁻¹. In the analysis of rice for 11 elements (Al, Ba, Ca, Cu, Fe, K, Mg, Mn, Na, Sr and Zn), Pohl et al.⁸⁵ sonicated 0.5 g with 4 mL of conc. HNO₃ at 60 °C for 16 min. The resulting solution was made up to 20 g with H₂O and filtered (0.45 μm) before analysis by ICP-OES. For validation, results were compared with those obtained by a method involving a multi-stage, open-vessel wet digestion procedure with sequential attack by HNO₃ and H₂O₂. They also analysed a CRM (NIST SRM 1568b rice flour), which they also spiked with several elements (this apparently unnecessary consumption of an expensive resource was not explained). They presented results for the analysis of the CRM by a closed-vessel MAD procedure, but no details were provided (there was no mention of a MAD system in the list of instrumentation). The results table for the analysis of the CRM was missing the certified value for Na (6.74 mg kg⁻¹), possibly because the value found was 10.8 mg kg⁻¹, and contained a number of problems with the statistical analysis; however, the results for most elements were accurate. The LODs ranged from 4.0 ng g⁻¹ for Mn to 2.7 μg g⁻¹ for K and all analytes were found in all 13 samples. To determine a similar number of elements (Ca, Fe, K, Mg, Na, P, and Zn by ICP-OES, and As, Cd, and Pb by ICP-MS) in chocolate, Santana et al.⁸⁶ found that UAE with dilute HNO₃ was suitable. They sonicated 200 mg of sample at 50 °C for 10 min with 10 mL of 2.0 mol L⁻¹ HNO₃. After centrifugation (3800 rpm, 10 min), 5 mL were removed and diluted to 25 mL. Results were compared for the analysis of one sample (with 70% cocoa) with those obtained by two published MAD methods, for which no details were provided and, again, no relevant instrumentation was listed. The only significant differences were for Mg and Mn. Spike recoveries for one sample, evaluated according to AOAC criteria, were found to be acceptable for As, Cd, Mg, Mn, Na and Zn, but for Ca, K, F, P, and Pb there was a “slight discrepancy.” The LODs ranged from 0.009 mg kg⁻¹ (As, Pb) to 4.6 mg kg⁻¹ (P). The procedure was applied to the analysis of 17 samples of craft chocolate from two different suppliers, but the individual results were not given. In the one sample analysed by the three methods, all analytes were detected except for As and Pb. The researchers concluded that the method was a fast and simple alternative to traditional acid digestion by microwaves, and required a minor amount of reagents resulting in decreased costs and residues. They did not go so far as to invoke any of the green scales. An even simpler method was developed by da Silva et al.⁸⁷ for the determination of Cr, Cu, Fe, Mn and Zn in dietary supplements by MIP-OES that involved sonicating (40 kHz, 110 W) 1.0 g of sample with 25 mL of 1.0 mol L⁻¹ HNO₃ for 5 min. For accuracy determination, results were compared with those obtained after complete mineralization by MAD with conc. HNO₃, and for the UAE method the slopes of the external calibration plots were compared with those of the standard additions plots. No spike recovery experiments or analyses of SRMs were performed. The LODs ranged from 0.01 μg g⁻¹ (Mn) to 0.18 μg g⁻¹ (Zn). Not surprisingly, all analytes were found in all samples though the concentration of Cr was not in accordance with the label value. There were no significant differences between either the results of the two methods or the slope of the calibration plots. For the determination of Fe and Mn in beef, Silva et al.⁵⁶ extracted 0.35 g of dried ground sample with 15.00 g of a mixture of 0.7 mol L⁻¹ HNO₃–0.6 mol L⁻¹ HCl for 5 min. The ultrasonic bath operated at 47 kHz, but no information about power or temperature was given. After centrifuging (5 min), the solution was analysed by MIP-OES. The method accuracy was evaluated by the analysis of a CRM (IRMM ERM®-BB184 bovine muscle) for which relative measurement errors of −17% and +7% were obtained for Fe (75 mg kg⁻¹) and Mn (0.28 mg kg⁻¹), respectively. The different value for Fe is most likely significant, but did not elicit any comment. The LODs were 2.6 mg kg⁻¹ and 0.06 mg kg⁻¹ for Fe and Mn, respectively. No results for the analysis of the real samples (six pooled samples obtained from 12 animals) were presented, but the authors discussed in some detail the evaluation of their method by various green analytical chemistry metrics, including the WAC approach, which was described as an interesting proposal with a holistic vision.

For the determination of total I in abalone by ICP-MS, Doh and Lee⁸⁸ evaluated two alkaline extraction methods. In one, an aliquot of 0.2 g was mixed with 5 mL of H₂O and 1.0 mL of 25% TMAH, then heated in an oven at 90 °C for 3 h. After dilution, centrifugation (4000 rpm, 15 min) and filtering, the supernatant was diluted to 3 mL. In the other, 0.2 g were mixed with 5 mL of H₂O and 5.0 mL of 25% TMAH then heated in a sealed vessel in a microwave oven at 200 °C for 5 min. After centrifugation (4000 rpm, 15 min) and filtering, the supernatant was diluted to 3 mL. The LODs were 0.11 ng g⁻¹ for both procedures, which were applied to the analysis of a CRM (ERM®-BB422 fish muscle). The concentrations found were not significantly different from each other or from the certificate value. The researchers also presented results for an in vivo bioavailability study. Photi et al.⁸⁹ reported on the outcome of a collaborative study designed to harmonise the performance of laboratories in Thailand when determining I in direct-iodized sauce by ICP-MS. During a hands-on training workshop, participants working with NIST SRM 1869 (infant formula), evaluated both alkaline and acid extractions and concluded that alkaline extraction with TMAH was superior. The procedure involved heating 0.5 g of sample with a 4–5% TMAH solution (volume not given) at 90 °C for 3 h. The LOD was 0.06 mg L⁻¹. The authors reported that most of the participating laboratories (7 or 8) achieved a satisfactory performance (|z| < 2.0) for all six test samples from two rounds of interlaboratory comparison.

To determine As in homeopathic medicines by HG-ICP-OES, Welna et al.⁹⁰ investigated sample preparation procedures as alternatives to MAD with HNO₃ and H₂O₂. Fortunately, they were able to make comparisons between the methods, as they found As in four of the 13 samples examined (eight pellets Arsenicum iodatum and Arsenicum album in four different potencies, one solid sample containing several active compounds per one tablet, including 30 mg Arsenicum album with the potency of D6 (dilution of the original by a factor of 10⁶), and four liquid samples—alcohol-based drops, containing Arsenicum iodatum in potencies of D6 and D8, as well as Arsenicum album in potencies of D8 and D10). Based on these results, together with spike recoveries and comparison of the slope of the external calibration with that of the standard additions calibration, they concluded that the simpler preparation procedures were satisfactory. These included dissolution or dilution with water (where appropriate) or UAE (15 min at room temperature) of 0.5 g portions of powdered tablets with 2.0 mL of freshly prepared aqua regia, followed by centrifugation, dilution and filtration. The LOD was 0.088 ng g⁻¹ in the prepared solution, and the concentration in all but four of the samples was below this value. In the other samples, the concentrations ranged from 12 to 234 ng g⁻¹. The researchers noted that for the analysis of drops, it was crucial to measure at 193.7 nm as the line at 188.9 nm suffered spectral interference from the alcohol remaining in the prepared solution. For the determination of Ca, Cu, Fe, K, Mg, Mn, P and Zn in swine feed by MIP-OES, Jofre et al.⁵⁴ compared two sample preparation procedures. In the UAE method, 500 mg of sample was sonicated (60 °C, 30 min) with 2 mol L⁻¹ HNO₃ and 27% H₂O₂ (volume not given). In the IR irradiation method, performed with a prototype device containing two commercial IR lamps (250 W) operated at a fixed maximum operating temperature of 190 °C, an aliquot of 500 mg was extracted for 23 min with 2 mL of 2 mol L⁻¹ HNO₃ and 23% H₂O₂. Prior to extraction, samples were dried at 60 °C for 24 h and ground under liquid N₂. The methods were also applied to the analysis of a PT material, MRC 20 (ground corn) produced by EMBRAPA Pecuária Sudeste (São Carlos, SP, Brazil). The LODs ranged from 0.06 mg kg⁻¹ (Cu) to 22 mg kg⁻¹ (P), and there was no significant differences between the results of the two methods for the ground corn, except for Zn, for which the UAE value was significantly higher than that obtained by the IR-assisted extraction. All analytes were found in all six samples. The researchers devoted significant effort to the evaluation of the greenness of the overall methods, using no fewer than five different tools. Perhaps not surprisingly, the only metric that distinguished between the two procedures was the AGREEprep scale (the UAE method scored slightly lower), this being the only metric that focusses specifically on the sample preparation stage. To determine Hg in multivitamin dietary supplements by CV-ICP-MS, Viana et al.⁹¹ extracted the Hg with a solution of L-cysteine. To 50 mg of sample (ground to <100 μm) was added 6.0 mL of 1% (m/v) L-cysteine solution and after manual stirring, the mixture was allowed to stand at room temperature for 20 min. The final volume was made up to 10 mL with H₂O, and 200 μL were injected into a water carrier in a FI system that was merged with HCl and then with BH. The accuracy was checked by analysis of a CRM (NIST SRM 2710a Montana soil). In addition, results were compared with those obtained after sample decomposition by MAD (150 mg sample plus 3.0 mL of conc. HNO₃, 1.0 mL of conc. HCl and 1.0 mL of H₂O, final volume 15 mL). The L-cysteine procedure (LOD 2 ng g⁻¹) was applied to five dietary supplements, in which Hg was found in four. The MAD procedure only detected Hg in two of the samples, but there was no significant difference between the results of the two procedures for these two samples or, possibly, between result of the analyses of the CRM the two procedures and the certificate value. The value for the L-cysteine procedure appears to be significantly higher than the certificate value as the mean of the experimental value did not fall within the expanded uncertainty interval around the certificate value, and the mean of the certificate value did not fall within the 95% CI about the experimental mean, although the two intervals overlap slightly. The researchers noted that it was not necessary to centrifuge or filter because the solid material settled down in 3 min.

3.2.5 Speciation. Clearly, for elemental speciation analysis, sample preparation conditions must be selected so that the target species are extracted from the matrix without transformation. For the simultaneous speciation analysis of Hg and Se in fish by HPLC-ICP-MS, Ribeiro et al.⁹² developed a microwave-assisted enzymatic hydrolysis procedure. A key component of both the extraction and subsequent chromatographic separation was BME, which is a mixture of ammonium acetate, MeOH, L-cysteine, and 2-mercaptoethanol, but the exact composition was not given. Much shorter time was required when microwaves were used to heat the sample, and so the optimum conditions involved adding 3.0 mL of a solution containing 6.67 g L⁻¹ protease XIV in ammonium phosphate buffer (pH 7) containing 500 μL of BME and heating to 37 °C for 60 min at 50 W power. After hydrolysis, the pH was adjusted by addition of 570 μL of HNO₃ and of 500 μL of BME (to obtain a final BME concentration of 0.2%) and the solution diluted to 50 mL. Four Se species, Se^IV, Se^VI, SeMet and SeCys, and two Hg species, iHg and MeHg were then separated in 15 min on a mixed anion/cation exchange column with a mobile phase of 5% (v/v) MeOH, 45 mmol L⁻¹ HNO₃, 0.015% BME, and 1.5 mmol L⁻¹ sodium 3-mercapto-1-propanesulfonate, whose pH was approximately 1.5. As the CRM are not certified for the element species, accuracy was assessed by the “accuracy profile approach,” which involved spiking at eight different concentrations for MeHg and six different concentrations for the Se species. All of the recoveries were within the tolerance limits. The LODs ranged from 0.011 mg kg⁻¹ (MeHg) to 0.26 mg kg⁻¹ (Se^IV) on a wet weight basis and the method was applied to the analysis of CRMs (JRC ERM®-CE101-trout muscle, JRC ERM®-BB422-fish muscle, NMIJ 7402-a-codfish tissue and NRCC DOLT-5-dogfish liver tissue). Neither Se^VI nor SeCys were found in any samples, whereas all other species were found in all samples, except Se^VI in the trout. The researchers pointed out that the small volumes needed allowed the processing of up to 64 samples in a single run. Enzymatic digestion was also featured in a method for the determination of I species in kelp by HPLC-ICP-MS.⁹³ The authors investigated the enzymatic hydrolysis with trypsin, pancreatin, Protamex, and neutral protease and found that only trypsin digests were satisfactory in terms of the release of the highest amounts of 3-iodo-L-tyrosine and 3,5-diiodo-L-tyrosine. They also showed that these species were not released in the absence of the enzyme. Kelp samples were first washed, dried and ground. Then to 0.2 g of sample were added to 0.04 g of trypsin and 20 mL of NaH₂PO₄/NaOH (pH 8) buffer solution and the mixture sonicated at 50 °C for 10 h (no further details were given). The mixture was centrifuged and filtered (0.45 μm) repeatedly and diluted to 50 mL. The LODs ranged from 1 μg L⁻¹ to 2.6 μg L⁻¹. Results for the analysis of and spike recovery from one sample were presented. For the determination of up to 11 As species in vegetables by HPLC-ICP-MS, Zhao et al.⁹⁴ investigated a number of possible extraction procedures. These included two ultrasound-assisted, two microwave-assisted and one oven-assisted extraction, together with a range of possible extractants including H₂O, 1% HCl, 1% H₃PO₄, 1% HNO₃, 1% H₂SO₄, 1% CH₃COOH, 50% (v/v) MeOH–H₂O and 50% (v/v) ACN–H₂O. In the method selected, to 2 g of homogenised vegetable sample were added 8 mL 1% (v/v) HCl, and the mixture heated at 90 °C for 3 h. After cooling, the mixture was centrifuged (8000 rpm, 5 min) and the supernatant filtered (0.22 μm). The LODs ranged from 0.032 μg kg⁻¹ to 0.12 μg kg⁻¹. The method was applied to two rice CRMs (NCS GBW(E)100361 and GBW(E)100348a) and to the analysis of 142 vegetable samples, including spinach (25), tomato (24), leek (25), pepper (24), celery (24) and canola (20). Only four As species were detected (AB, DMA, As^III and As^V). The organic As was detected only in spinach, leek, pepper and celery, whereas iAs was detected in all vegetable samples. The researchers removed a background peak arising from As^V in the mobile phase (As^V accumulated during the early part of the gradient elution and was then released as a “ghost” peak later in the gradient) by an on-line SPE procedure. Hu et al.⁹⁵ determined iSb species in tea by a procedure in which the analytes were preconcentrated by SPE before determination by HPLC-ICP-MS. They evaluated five SPE columns and chose the HLB (a universal polymeric RP sorbent) column. Further details of the method are given in Table 1.

Several research groups have devised sample preparation protocols for speciation analysis of NPs. Sajnóg et al.⁹⁶ investigated 10 procedures for the extraction of Se NPs from yeast prior to determination by spICP-MS. They classified the procedures as (a) enzymatic digestion, (b) mechanical cell lysis or (c) chemical extraction. The enzymatic procedure gave the highest recoveries in terms of mass of Se NPs for three of the four samples studied, although broader size distributions were obtained, which was attributed to partial agglomeration. In this procedure 10 mg yeast were sonicated for 1 h and centrifuged, then the pellet was resuspended in 4% disease in 30 mmol L⁻¹ TRIS buffer (pH 7.5), incubated for 17 h at 25 °C and centrifuged, then the pellet was resuspended with 4 mg mL⁻¹ protease in TRIS buffer, incubated for a further 17 h at 37 °C then, after centrifuging, the pellet was sonicated in 4% SDS for 1 h, centrifuged and diluted 50 000-fold with water. In the mechanical procedures designed to disrupt cell walls by vigorous shaking in the presence of beads, the effects of various combinations of glass and metal beads were investigated. Yeast, suspended in water, was mixed with the beads and shaken (10 Hz oscillation) for 5 min followed by centrifuging for 10 min. For the chemical procedures, yeast was mixed with solutions of either SDS, NaOH or TMAH at several concentrations. A mechanical procedure, involving two cycles of sonication and glass bead milling was faster than the enzyme procedure and provided more realistic size distributions but recoveries were lower. These two procedures were considered complementary, whereas the chemical procedures were unsatisfactory because of partial dissolution of the Se NPs. In contrast, for the determination of TiO₂ NPs in food sauces by spICP-MS, Klein et al.⁹⁷ concluded that the best procedure of the several they investigated (including ultrasonication and enzymatic digestion) involved TMAH. The food additive E171 is known to contain a nanoparticulate fraction (<100 nm) whose proportion varies between 5% and 50% and over which concern has been expressed about possible adverse health outcomes. The optimised alkaline extraction conditions were sonication (30 min) of a dispersion of 0.2 g of sample in 10 mL 10% (v/v) TMAH followed by vortexing and dilution prior to analysis. The method was validated by the analysis of a Ti-free mayonnaise spiked with the E171 additive or the JRC NM-100 raw material (anatase). Agreement in terms of particle sizes was obtained between those of the spiked mayonnaise and the TiO₂ suspensions used for spiking, even though it was found that the presence of fatty compounds improved particle dispersion. When the researchers applied the method to E171 labelled food sauces (white sauce, mayonnaise, aioli, cocktail sauce, Caesar sauce and white chocolate), they found concentrations of TiO₂ particles between 0.6 g kg⁻¹ and 6 g kg⁻¹ with median diameters between 170 nm and 320 nm. Between 2% and 12% of the TiO₂ was nanoparticulate. No Ti was detected in food sauces not labelled as containing E171. In the determination of Ag NPs, the presence of dissolved Ag ions can lead to an elevated baseline that obscures signals from smaller Ag NPs in spICP-MS. Zhang et al.⁹⁸ separated the dissolved Ag from the NPs by selective SPE. They synthesized a Ag ion-imprinted magnetic adsorbent, Fe₃O₄@SiO₂@Ag-IIP, and applied it in the analysis of an antibacterial gel containing Ag NPs obtained from a local pharmacy. They extracted 0.1 g of the gel, which contained various matrix components such as gelatine, alginate, hyaluronic acid, sodium carboxy-methylcellulose, pectin, and chitosan, with 50 mL of H₂O (sonication 10 min). After filtering, 10 mg of extractant was added and the mixture ultrasonicated for 30 min. The supernatant was collected after magnetic separation. The LODs were 12.6 nm (particle size) and 6.3 × 10⁵ L⁻¹ (particles number concentration). The spICP-MS particle size and number accuracy was determined with NIST SRM 8012 (gold NP 30 nm). The real sample contained 12 ± 3 μg g⁻¹ of 36 nm sized particles at a particle concentration of 196 ± 13 × 10⁸ g⁻¹. The size results were in agreement with those of TEM. A spike with Ag NPs of 15 nm nominal size was only 85% recovered, but this was accounted for by about 15% of the spike particles being below the size LOD.

3.2.6 Solid sample analysis. Several research groups have studied sample preparation resulting in a solid for analysis. For the determination of Cr, Ni and Pb in mineral and well waters by LIBS, Santini et al.⁹⁹ extracted the analytes onto carbon-based adsorbent films. The work reported is a continuation of work previously published, which is perhaps why it is almost impossible to follow what they did. The thin film micro-extraction supports were prepared by drop-casting graphene and graphite dispersions onto chemically etched glass substrates, and the extraction time was 15 min. No information about how the analytes were extracted was given, though it appeared that, after extraction, the supports were dried on a hot plate. A table of optimised parameters indicated that for two adsorbents (graphite and graphene) the best results were obtained with (a) 20 μL, pH 9, and agitation, and (b) 60 μL, pH 9 and agitation, respectively. They concluded that graphene was the superior material and obtained LODs of 1.0 mg L⁻¹ (Cr), 0.8 mg L⁻¹ (Ni) and 0.9 mg L⁻¹ (Pb). They also showed that applying a top coat of Ag NPs following analyte collection on the graphene improved the LODs for mineral water, but for the well water no satisfactory results could be obtained. For the XRF analysis of ovarian tissues for Ca, Cl, Fe, K, P and S, Gianoncelli et al.¹⁰⁰ evaluated three sample preparation procedures. In addition to FFPE, they prepared samples by two cryo-fixing procedures: (a) freezing, cryo-slicing and freeze-drying and (b) freezing, cryo-slicing with measurement under cryogenic conditions. The measurements were made at the ID21 beamline of the European Synchrotron Radiation Facility (Grenoble, France). They encountered problems with both procedures, concluding that cryo-fixation with slow freezing of soft tissue was very delicate causing the hydrated tissue to lose integrity and almost collapse, so that light elements were redistributed and thus not representative of morphological features. They suggested two possible modifications: (a) high-pressure freezing and (b) the use of cryoprotectants. For the determination of Cu, Fe, Se and Zn in murine liver, Lossow et al.¹⁰¹ showed that results obtained by TXRF following ultrasound-assisted lysis in radioimmunoprecipitation assay buffer were not significantly different from those obtained by ICP-MS following MAD with HNO₃ and H₂O₂. For TXRF determination, 10 μL of the lysates (to which 1 mg L⁻¹ Y was added as IS) were applied to siliconized sample carriers and dried at 40 °C for 5–10 min followed by measurement for 1000 s using a bench-top total TRXF spectrometer. Serum was also analysed, though in this case for the TXRF analysis Ga was the IS. The ICP-MS analysis of the liver, lysates, pellets (after centrifugation) showed that between 90% and 100% of the elements were extracted; the slight loss of some elements could be corrected by normalizing the TXRF values to protein concentration in the lysate instead of to the mass of tissue taken. They concluded that the buffer-based procedure allowed simple and fast preparation of small amounts of tissue for trace element determinations. To decrease the LOD of a TXRF method for the determination of Se, Musielak et al.¹⁰² devised a DSPE preconcentration procedure in which the Se was retained on graphene impregnated with thiosemicarbazide (G@TSC). To 100 mL of sample was added 0.5 mL of a 2 mg mL⁻¹ G@TSC suspension and the mixture sonicated (10–15 min) and stirred (60 min). After filtration and resuspension with sonication to add Y as the IS, 10 μL was transferred to a quartz reflector and dried at 60 °C. Solid samples were first subject to MAD with HNO₃ and H₂O₂ (100 mg, final volume 100 mL). The procedure was applied to 14 environmental and food CRMs, as well as locally sourced tap water, mineral water, beer, orange juice, apple juices and four different types of wine. In addition to the preconcentration procedure, they also prepared samples as (a) powdered solids suspended in the solution with IS and (b) mineralized solutions with IS, though no details of these procedures were given. The LOD (of the G@TSC method) was 1.7 pg mL⁻¹ and Se was quantified in all CRMs and the samples, except for the tap and mineral waters and the apple juice. However the other two procedures were unable to detect Se in five of the food CRMs (JRC ERM®-BD151 skimmed milk, NCS ZC73013 spinach, NCS ZC73029 rice, NCS ZC73032 celery and NCS ZC73033 scallion), in which the concentration ranged from 0.53 mg kg⁻¹ to 0.19 mg kg⁻¹. As the G@TSC procedure was selective for Se^IV, the researchers showed that it could be used, in combination with the analysis of a second sample in which the Se^VI had been reduced to Se^IV, for the speciation analysis of the liquid samples. As it happened, none of them contained any Se^VI, but recoveries of spikes at 10 ng mL⁻¹ ranged from 91% to 117%. A somewhat similar comparison of sample preparation procedures was made by Alov et al.¹⁰³ for the determination of a number of elements in three different dietary supplements by TXRF. Preliminary investigations showed that (a) the mineral contents were not water soluble and (b) it was not possible to prepare stable slurries either in H₂O or HNO₃, even with the addition of Triton X-100. So one procedure involved preparing slurries in ethylene glycol (10–15 mg sample plus 1 mL of ethylene glycol and 1 μL of IS solution, ultrasonication for 10 min). The other method was MAD with HNO₃ and H₂O₂ (200 mg, final volume 50 mL including IS). For the TXRF analysis, 2 μL was dried on a quartz reflector and data acquired for 300 s. The acid digest was also analysed for, Al, Co, Cr, Cu, Mo, Ni, Se, Si and Sn by ICP-OES; a further 50-fold dilution was needed for Ca, K, Fe, Mg, P and Zn. The LOQs ranged from 1 μg L⁻¹ to 100 μg L⁻¹ (0.01–1 mg per tablet). No statistical evaluation of the results was presented, which would have been difficult for comparisons involving the label amounts of the various elements in each tablet as no ± terms were available. Visual inspection of the tables of results suggests that for many of the ICP-OES analyses, the 95% CI about the mean would not include the label value. This number would be smaller for the TXRF results, as the ± terms are larger. Visual inspection also suggests that on the basis of a paired t-test, there is no significant difference between the TXRF results and the ICP-OES results. The MAD TXRF method could not detect Cr, Co, I, Mg, Mo, Ni, Se, Sn, P, and V of which Cr, P and Se could be detected by the slurry TXRF method. The ICP-OES method could not detect Cl, I, Se, Sn and V. The researchers noted that the slurry TXRF procedure required 30–40 min for the total analysis, whereas the MAD procedures required 3–4 h.

3.2.7 Preconcentration. Several research groups have devised coprecipitation methods. For the determination of Se isotope ratios in a number of different sample types, including wheat flour and enriched yeast, Karasinski et al.¹⁰⁴ precipitated the Se with iron hydroxide from 20 mL of acidified sample to which 3.6 g of NH₄Cl and 150 mg of Fe(NO₃)₃·9H₂O were added, followed by heating, neutralization with NaOH pellets, and then adjusting “to pH of 2.40 with 0.2 M solution of sodium hydroxide”, which is presumably a typo for pH 12.4. After further heating for 2 h (temperature not given), the mixture was centrifuged (5 min), the precipitate was dissolved in 5 mL 12 mol L⁻¹ HCl and diluted to 10 mL, then Se was determined by HG-MC-ICP-MS. The samples (JRC ERM®-BC210a selenized wheat flour and NRCC SELM-1 Se-enriched yeast) were treated with MAD with HNO₃ and H₂O₂ (0.13 g, final volume 20 mL). The researchers concluded that with the combination of coprecipitation and HG it was possible to measure the relative isotopic fractionation (δ values) by the simplest possible method (SSB) with no need for additional corrections. They also pointed out that compared with the commonly used ion-exchange separation procedure, their method was much more robust as well as simpler and faster. They also acknowledged the potential of the method for preconcentration. For the determination of Sr isotope ratios in red wine by TIMS, Li et al.¹⁰⁵ devised a coprecipitation method in which the majority of the Sr, together with most of the aluminium, calcium and magnesium, was precipitated together on the addition of HF. Wine (3 mL) was mineralised in a multi-step wet ashing stage, the residue from which was dissolved in 0.5 mL of 0.4 mol L⁻¹ HCl at 100 °C. Then, 1 mL of 22 mol L⁻¹ HF was added and the mixture heated in a closed vessel at 120 °C for 1.5 h. Finally the precipitate was separated by centrifuging and dissolved in 0.3 mL of 7 mol L⁻¹ HNO₃. The procedure separated Sr from the associated interference from ⁸⁷Rb. The results agreed with those obtained by a method in which Sr was separated by a selective cation-exchange resin, and the researchers concluded that the new procedure decreased costs (no expensive resin was needed) and time (by about 50%) without compromising accuracy, but did require handling conc. HF. For the determination of ²³⁸Pu, ²³⁹Pu and ²⁴⁰Pu concentrations in urine samples from potentially exposed workers by ICP-MS and alpha spectrometry, Macsik et al.¹⁰⁶ devised a three-stage procedure. In the first step, 10 mL of 8 mol L⁻¹ HNO₃ and a ²⁴²Pu spike were added to 200 mL of sample and the mixture allowed to stand overnight to pre-digest the samples and allow Pu isotopic equilibration. The second step was preconcentration by coprecipitation with calcium phosphate from alkaline solution. After adding the reagents, the solutions were stirred (10 min) and left to settle (2 h). After separation, the precipitate was dissolved in 1.5 mL conc. HNO₃, then diluted to 8 mol L⁻¹ and heated to just below boiling (125–130 °C) for 1 h to remove the excess NH₄OH and convert Pu isotopes to Pu^IV. In the final step, the samples were loaded onto a 1.8 mL anion-exchange column (AG MP-1 M, 50–100 mesh), which was rinsed with 30 mL of 8 mol L⁻¹ HNO₃, to remove americium and uranium, then with 20 mL of 9 mol L⁻¹ HCl to remove thorium. The Pu was then eluted with 15 mL conc. HBr and 5 mL of 0.1 mol L⁻¹ HCl. The ICP-MS LODs were 0.015 pg (²³⁹Pu) and 0.002 pg (²⁴⁰Pu) per 200 mL. As chelation therapy can have a number of adverse effects, the researchers concluded that their rapid method would be important in deciding not to chelate if it can be quickly established that the likely committed effective dose was less than 200 mSv.

To quantify tumour cells circulating in the blood (via the Zn measured by ICP-MS), Nian et al.¹⁰⁷ fabricated a microfluidic chip device to preconcentrate the cancer cells. The human breast cancer (MCF-7) cells were selectively labelled with epithelial cell adhesion molecule (EpCAM) aptamer-modified magnetic beads (as EpCAM is overexpressed in most cancer cells, EpCAM-based immunomagnetic separation is the most commonly used strategy for circulating tumor cell enrichment) and then introduced into a microfluidic chip featuring three regions (sorting, purification, and release zone). In the release zone, the beads with the MCF-7 cells were fixed by strong magnetic and hydrodynamic forces, and a continuously flowing nuclease solution cleaved the aptamer on the trapped MCF-7 cells, causing gentle release of MCF-7 cells for subsequent ICP-MS analysis (or further cultivation). Based on the Zn content, the average cell recovery from spiked blood samples was 84%. The method was applied to the analysis of real blood samples from healthy people and breast cancer patients, and circulating tumour cells were successfully detected in all 16 patient samples.

The large number of publications featuring various forms of LLE and SPE are summarised in Tables 1 and 2. Procedures in which nanoparticulate materials were used for the enrichment are reported in Table 2. Despite the exclusion of reports that did not feature the analysis of a CRM (where appropriate), the entries in the two tables total 107, compared with the 79 papers described in a table in our previous year’s Update.¹ In an effort to make the summaries more manageable, not all the details of the procedures are given. To allow readers to calculate an LOD in a solid sample, the sample mass and final volume after the first dissolution step are given, if they were disclosed in the paper. Another useful figure of merit is the extent of preconcentration. There is no uniformity in the terminology used by researchers who give values for “preconcentration factor”, “enhancement factor” or “enrichment factor” often without defining the basis of the calculation. As far as possible, the following convention has been adopted for the tables: the preconcentration factor, PF, is based on volume ratio, the enhancement factor, EHF, is based on the ratio of the calibration slopes, and the enrichment factor, ERF, is based on concentration ratios. In addition to the shortcomings of published articles mentioned above in the introduction to Section 3.2, many papers make it difficult for the reader to ascertain (a) what the values of the relevant parameters of the final, optimised method were, (b) how spikes were added to solid samples, and (b) what calibration procedure(s) was/were used.

4. Progress with analytical techniques

4.1 Mass spectrometry

There was continued interest this year in applications of scICP-TOF-MS, the only technique with which true simultaneous detection of multiple m/z ratios is possible, to allow evaluation of more than one characteristic of the same individual cell. Two proof of concept studies applying scICP-TOF-MS to human cells were published over the review period. The first aimed to determine the relative concentrations of three target proteins, hepcidin, metallothionein-2 and ferroportin, via specific antibodies labelled with Ir, Pt and Au nanoclusters respectively, in individual cells from a human retinal pigment epithelial cell line (ARPE-19).¹⁰⁸ In addition to detection of these label elements by ICP-TOF-MS, Ru from labelling of the cell membranes with an Ru-containing inorganic dye, was also monitored, enabling both cell counting, and calculation of the relative cell volume of each individual cell. The lack of sensitivity of the ICP-TOF-MS to simultaneously measure low m/z endogenous cell elements and the high m/z labels, hindered use of endogenous cell element markers for counting cells. Introducing cells at a rate of 1 × 10⁻⁵ cells per mL via a microFAST Single Cell System yielded a reasonable transport efficiency for the ARPE-19 cells of 51 ± 4%, however, this may be an underestimation as only events where both Ru and a label from one of the metal nanocluster probes were considered. The approach allowed relative concentrations to be expressed as protein mass per relative cell volume and the LODs reported were 3.8 ± 0.4 ag per cell (hepcidin), 9 ± 1 ag per cell (metallothionein-2) and 4.4 ± 0.6 fg per cell (ferroportin). In the second report,¹⁰⁹ the multielement detection capabilities of scICP-TOF-MS allowed for the direct assessment of Pd-doped nanoplastic particles association with individual human cells (THP-1 and A54). In contrast to the first study, paired endogenous cell elements, ³¹P and ⁶⁴Zn, were utilised to identify single cells in a label-free manner whilst ¹⁰⁶Pd, ¹⁰⁷Pd and ¹⁰⁸Pd isotopes were used to detect the nanoplastics. Although nebulisation efficiency was determined to be 83 ± 3% using 100 nm Au NPs, the transport efficiency of the cells was markedly lower, from 14% to 33% for THP-1 cells and from 2% to 6% for A54 cells, owing to their larger size. Heterogeneous association between cells and nanoplastic particles, with increased association at higher exposure concentrations, was observed; findings that were confirmed by TEM imaging.

Single cell phenotype analysis in cancer research using scICP-MS was the subject of two publications from the same group this year and these both investigated more economical alternatives to scICP-TOF-MS for evaluating multiple characteristics of single cells. Zhang et al.¹¹⁰ reported the development of a dual detection analytical strategy consisting of a bespoke single cell focusing microfluidic chip with detection by LIF and ICP-MS. The set-up was used to assess expression levels of a target protein, PTK7, which may be associated with tumour cells, by labelling of the protein with biofunctionalised fluorescent nanoprobes containing Au NPs. Furthermore, scICP-MS was used for detection of ¹⁹⁵Pt to assess the uptake of the drug oxaliplatin, by single tumour cells. A high transport efficiency of 90.6 ± 7.6% was achieved through the use of a HEN (Enya Mist) with a low cell flow rate, which increased throughput speed (500 cells per min) with respect to previous studies. The method was applied to high and low PTK7-expression cancer cell lines and the dual detection strategy was found to achieve superior accuracy in cellular phenotyping than either detection method alone. Real life applicability of the method was demonstrated by detection of circulating tumour cells in blood from ten breast cancer patients. No labelled cells were found in five healthy volunteers. The focus of the second piece of work¹¹¹ was dual isotope detection of highly specific aptamer functionalised Au and Ag NP probes, bound to target proteins on cancer cells, using a similar scICP-MS set-up to that in the first study. To prove dual isotope capability, simultaneous detection of two protein surface markers, EpCAM and MUC1, via Au and Ag respectively, was used to discriminate between three different cell types. Dual isotope detection was achieved using conventional ICP-MS by setting appropriate dwell (100 μs) and settling (500 μs) times such that both ¹⁹⁷Au and ¹⁰⁷Ag could be detected within the ion plume time of a cell (700 μs). To optimise cell typing accuracy, data were excluded for cells with an ion plume of greater than 700 μs, which reduced the heterogeneity of the biomarker expression information among individual cells. The ICP-MS method was found to be superior in discriminating the difference in biomarker expression levels among the various cell types compared with flow cytometry. Breast cancer stem cells spiked into serum and whole blood matrices were used to demonstrate the set-up.

Other applications of scICP-MS of note included development of a highly sensitive method both to monitor the Legionella pneumophila cell concentration in drinking water, using Mg as an endogenous cell marker, and to study the Cu mass distribution in the cells following disinfection with Cu.¹¹² The scICP-MS method was found to be much faster and more reproducible than conventional cell counting using a haemocytometer (RSD%: 4.2% vs. 10.2%). Utilising a short dwell time of 50 μs, an LOD for Cu measurement of 38 ag per cell was achieved. Meanwhile, in other work, an application was described for assessing Hg accumulation in single cells from the human neuroblastoma line, SH-SY5Y, following exposure to Hg and Se.¹¹³ A useful finding in this study was that cell fixation with 4% (v/v) paraformaldehyde greatly improved transport efficiency (93 ± 6% vs. 10 ± 3% with no fixation), possibly due to the high fragility of human cell lines. However, as conventional ICP-MS was used for detection, and the acquisition time of the transient signals originating from single cells are generally too short to scan signals at different m/z ratios, monitoring of ²⁰²Hg for assessing Hg accumulation and ⁵⁵Mn and ⁵⁹Co for identifying cells had to be performed on separate runs (and thus on different cell samples), which presented a significant limitation to the work.

A review, covering 49 papers, by Loeschner et al.¹¹⁴ provided a useful summary of the analytical approaches and performance characteristics of spICP-MS applications in food. The authors had collated information including the type of instrumentation and sample introduction system used, dwell time, sample uptake rate, NP calibration standard and methods for transport efficiency assessment and data analysis. It was interesting to note that whilst the majority of the studies employed ICP-QMS, approximately a quarter used ICP-QQQ-MS, primarily for Ti NP analysis, but only two employed SF-ICP-MS and one, ICP-TOF-MS. The most commonly used nebuliser for this type of analysis was Micromist, followed by Meinhard and PFA concentric nebulisers, and the most frequently used spray chambers were cyclonic and Scott. Dwell times were collated with 100 μs being the most popular, followed by 3 ms and 50 μs, although choice of dwell time will depend on the type of ICP-MS and sample introduction system used. Quantifying small diameter (<100 nm) TiO₂ NPs in food and food additives, with the aim of establishing whether E171, a TiO₂ food additive, which has been recently banned in the EU, should be classified as a nanomaterial, was the focus of two further papers this year. Both utilised the interference removal capabilities of ICP-QQQ-MS, operating in “mass-shift” mode with O₂ and H₂ gases, to avoid isobaric interferences on ⁴⁸Ti⁺ from ⁴⁸Ca⁺ and on ⁴⁸Ti¹⁶O⁺ from ⁴⁸Ca¹⁶O⁺, thereby improving size LODs. The first analytical set-up featured a highly efficient APEX sample introduction system, which nebulised samples into a heated cyclonic spray chamber as well as a two-stage desolvation strategy to reduce oxides and increase solvent removal.¹¹⁵ The method was able to reliably detect TiO₂ NPs with diameters from 12 to 800 nm in various food samples, following extraction with a surfactant solution. A comparison was performed with HR-ICP-MS and SEM using two E171 food additives to demonstrate the method before it was applied to food samples purchased from a local Spanish market. Ojeda et al.¹¹⁶ also reported use of a similar analytical set-up for evaluating TiO₂ NPs in food, which was optimised to achieve a comparable size LOD of 15 nm. Factors contributing to the low size LOD here were use of the 3σ criterion approach to distinguish particle events from background, purging of the O₂ gas line for 2 h prior to analysis and use of axial acceleration to increase the sensitivity of the ion products. Moreover, a dual analytical strategy was employed to confirm the accuracy of the spICP-MS analysis; AF4 coupled online to both multi-angle light scattering and ICP-QQQ-MS. Calibrators consisting of SiO₂ NPs, with nominal sizes of 20 and 180 nm, were used for TiO₂ characterisation by AF4-ICP-MS, which were detected as ²⁸Si using 2.0 mL min⁻¹ H₂ cell gas to eliminate polyatomic interferences. Application of the methods to the E171 raw material (JRC NM-100 anatase) and food samples showed the presence of an independent TiO₂ NP fraction with diameter <35 nm that had a significant number-based concentration; a phenomenon that has previously been observed with TEM.

High precision MC-ICP-MS methods for determining isotope ratios in biological matrices featured in two recent publications. Guo et al.²⁸ developed a one-column method to simultaneously purify samples for Cu, Fe and Zn isotope ratio measurement, enabling data on the same sample aliquot to be obtained. With regards to the MC-ICP-MS method, Fe and Zn isotopes were measured in HR mode, and Cu, in low-resolution mode. For Cu and Fe, instrumental mass bias of the isotope ratios was corrected by bracketing with a RM, whereas for Zn, the double-spike technique, using ⁷⁰Zn and ⁶⁷Zn, was utilised. A comprehensive assessment of the robustness of the method demonstrated that the concentrations of samples and standards needed to agree to ±10% so as not to affect δ⁵⁶Fe accuracy and furthermore, identified thresholds for interfering matrix elements (Cd, Cr, Ni and Ti) that when exceeded indicated that a second purification step was required. The long-term intermediate precision was better than 0.04‰ for the three metal isotope ratios. The optimised method was applied to nine biological RMs, one of which (ERM®-BB186, pig kidney) has established Cu, Fe and Zn isotope ratio values. Meanwhile, Hobin et al.¹¹⁷ focused on a highly precise MC-ICP-MS method comprising a one-step column purification step to determine K isotope ratios in acid-digested, microvolume CSF samples. Both K isotopes, ³⁹K and ⁴¹K, were measured in extra HR mode with Faraday cups connected to 10¹³ Ω resistors, which resulted in a two-fold increase in precision compared with conventional 10¹¹ Ω resistors in samples with a low final K concentration of 25 ng mL⁻¹. Use of hot rather than cold plasma conditions in the MC-ICP-MS provided higher robustness at the low K concentration of 25 ng mL⁻¹ in terms of concentration mismatch between samples and standards, and reduced the effect of matrix elements (Ca, Fe, Mg and Na) on the K isotope measurements. Recovery for K in CSF microsamples pre- and post-column separation was 99.6 ± 0.7%. Excellent agreement was observed with results obtained using the conventional 10¹¹ Ω resistor method, and a sample volume of 100 μL, rather than 5 μL, for the QC material Randox L2 CSF. Results were also in good agreement with previously published isotope ratio values for this material. Repeatability and intermediate precision for the δ⁴¹K value was 0.11 and 0.10‰ respectively. The method was applied to K isotopic analysis of ten murine CSF samples.

Radionuclide ⁹⁹Tc is challenging to measure accurately by ICP-MS because it lacks a commercially available standard. Isobaric dilution analysis, an ID-like approach involving an isobaric ⁹⁹Ru spike was used for quantification of ⁹⁹Tc in urine and radiopharmaceuticals in two papers published by the same group. Horstmann et al.¹¹⁸ aimed to quantify a stable species of ⁹⁹Tc, [⁹⁹Tc]TcO₄⁻, in urine by ICP-MS, combining AEC, to preconcentrate the ⁹⁹Tc-containing species, and aerosol desolvation nebulisation (achieved with an APEX-2 High Sensitivity Desolvating System equipped with a MicroFlow™ PFA ST nebuliser) to increase transport efficiency and sensitivity with respect to conventional sample introduction. To accurately ascertain the ICP-MS sensitivity for ⁹⁹Tc, as required for the IBDA approach, an in-house prepared standard, manufactured from decayed medical ^99mTc-generator eluates and previously quantified using TXRF, was used. With the developed method, [⁹⁹Tc]TcO₄⁻ was well separated from interferences (MoO₄²⁻ and RuCl₄⁻) in less than 3 min. Very good agreement with TXRF was obtained for measurement of ⁹⁹Tc in the in-house quantified standard. The general uncertainty of the whole method was less than 4% and the LOD and LOQ were 0.67 ± 0.04 ng kg⁻¹ and 2.2 ± 0.1 ng kg⁻¹, respectively. The method was applied to untreated neat urine from a patient who had undergone scintigraphy with a ⁹⁹Tc-tracer. In the second paper,¹¹⁹ the group utilised the online post-column IBDA calibration approach in an RP HPLC-ICP-MS method, this time to separate and quantify Tc species in radiopharmaceuticals, in combination with a HPLC-ESI-HR-MS method for absolute compound identification. The ability to characterise novel experimental tracers in nuclear medicine without the need for “cold” reference compounds was a distinct advantage of the approach. In contrast to the work described in the first paper, to correct for element-specific differences in sensitivity, the response of Tc was interpolated using sensitivities of its neighbouring elements with 4d electron shells. The LOD was 0.5 nmol L⁻¹ and the LOQ 1.7 nmol L⁻¹. The method was applied to separate Tc species in four different commercial tracers and two experimental tracers.

Yang et al.¹²⁰ developed a novel ICP-QQQ-MS method for radionuclide analysis, suitable for rapid screening of ⁹⁰Sr in urine. Samples were digested in HNO₃ and purified using stacked 2 mL DGA-N (N,N,N′,N′-tetra-n-octyl DGA) and Sr resin cartridges before injection into the ICP-QQQ-MS via a high efficiency sample introduction system (Apex-Q), equipped with a membrane desolvation unit. Use of CO₂ as the reaction cell gas, with “on-mass” detection of ⁹⁰Sr, was found to achieve more effective interference removal compared with O₂ gas. Naturally occurring ⁸⁸Sr in urine was used as a yield tracer and this approach was demonstrated by spiking ⁹⁰Sr into urine and showing that the recoveries of stable ⁸⁸Sr and spiked ⁹⁰Sr were comparable. Under optimised conditions, the abundance sensitivity, a measure of peak tailing, was determined to be 6.27 × 10⁻¹⁰, which was better than the requirement of 10⁻⁸ for the intended application. The LOD was 0.978 pg L⁻¹ in 10 mL urine or 2.74 × 10⁻² pg L⁻¹ in 400 mL urine. Meanwhile, Peng et al.¹²¹ reported a single method to determine transuranium radionucleotides (²⁴¹Am, ²⁴⁴Cm, ²³⁷Np, ²³⁹Pu, ²⁴⁰Pu, ²⁴¹Pu) in urine with separation using a single DGA resin column. To achieve simultaneous extraction of the radionuclides, it was necessary to digest the urine at 150 °C for at least 5 min in 8 mol L⁻¹ HNO₃. The column elute was directly injected into the ICP-QQQ-MS without need for evaporation. Use of the cell gases, O₂ (0.30 mL min⁻¹) and He (10.0 mL min⁻¹), and detection of the elements of interest “off-mass”, as oxides, achieved effective removal of the peak tailing ²³⁸U interference, which is difficult to accomplish using chromatographic separation. Elements, ²⁴¹Am and ²⁴¹Pu, were successfully separated by ICP-QQQ-MS based on their different reaction efficiencies with O₂. The ²⁴²Pu isotope served as a tracer for Np and Pu yield monitoring, whereas ²⁴³Am was used for Am and Cm. The chemical behaviours of these pairs were consistent during chromatographic separation. The LODs achieved were very low, ranging from 0.03 fg (²⁴⁴Cm) to 1.18 fg (²³⁷Pu) in 20 mL urine with the LODs for the Pu isotopes being nearly one to two orders of magnitude lower than those required for emergency screening. The method accuracy was established using urine spiked with standard solutions, giving mean relative deviations from the target values of 1.11%, 0.33% and 0.35% for ²³⁷Np, ²³⁹Pu and ²⁴¹Am, respectively.

Interference from Ar-based polyatomic molecules is a challenge in the ICP-MS determination of certain elements. Two papers explored the use of N₂-sustained microwave inductively coupled atmospheric pressure plasma mass spectrometry (N₂-MICAP-MS), which has an Ar-free plasma, to overcome this. In the first paper, use of the N₂-based plasma allowed Ca, Fe and Se to be determined in human serum using the most abundant isotopes, ⁴⁰Ca, ⁵⁶Fe and ⁸⁰Se, which are not usually feasible to measure owing to the Ar-based polyatomic interferences, ⁴⁰Ar⁺, ⁴⁰Ar¹⁶O⁺ and ⁸⁰Ar²⁺ respectively.¹²² This methodology was combined with IDA (monitoring isotope ratios ⁴⁰Ca : ⁴⁴Ca, ⁵⁷Fe : ⁵⁶Fe and ⁸²Se : ⁸⁰Se) to avoid the influence of matrix effects, particularly on Se. However, in the presence of high matrix Na concentrations, intensity suppression was observed, and this was noted to cause isotopic fractionation for ⁴⁰Ca and ⁴⁴Ca. Accurate quantification of the elements was confirmed using nine serum CRMs (NIST SRM 909C, JRC BCR-304, BCR-637, BCR-638 and BCR-639, Seronorm™ L1 and L2, and ClinChek L1 and L2), for which the results obtained were metrologically comparable to the certified values. Others reported coupling of AEC and N₂-MICAP-MS for As speciation in rice, following a mild acid hydrolysis extraction.¹²³ In the case of As, use of the N₂-based plasma avoided interference of polyatomic ion, ⁴⁰Ar³⁵Cl⁺, on ⁷⁵As⁺. However, the LODs achieved for the As species were 0.26 μg kg⁻¹, 0.22 μg kg⁻¹, 0.26 μg kg⁻¹, 0.13 μg kg⁻¹ and for As^III, As^V, DMA and MMA, respectively, which were approximately an order of magnitude higher than those obtained with conventional Ar-plasma ICP-MS, likely reflecting the higher m/z-non-specific background observed in N₂-MICAP-MS. The method was validated using SRM NIST 1568b rice flour, for which recoveries of all species were between 85% and 110%, before application to eight varieties of rice. Neither of these studies demonstrated a significant analytical advantage of the N₂-MICAP-MS approach over Ar-plasma ICP-MS but the methods developed were fit-for-purpose and reaped the benefit of using low cost and accessible N₂ gas.

There were two papers of interest that employed online isotope dilution analysis, an approach that has the advantage of largely eliminating the influence of sample matrix on measurement accuracy. In work by Tukhmetova et al.,¹²⁴ a non-species specific ID strategy was employed to quantify albumin in serum through detection of S via CE-SF-ICP-MS. Addition of the inorganic ³⁴S spike post-column meant that the molecular form of the spike does not have to be the same as the target compound. Separation of proteins in the sample by way of CE afforded the advantages of requiring a very small injection volume and necessitating no sample preparation. Use of a special capillary-coating procedure to reduce adsorption of proteins onto the capillary wall, improved separation efficiency and robustness. The method measured SI-traceable S mass concentrations through use of a metrologically traceable ³⁴S spike. A pure albumin RM (NIST SRM 927 bovine serum albumin), spiked into albumin-depleted serum at a physiological concentration to replicate the native matrix, was used to validate the method. The relative expanded uncertainty of the method for quantification of albumin was rather high at 6.7% (k = 2), with the greatest contribution to the uncertainty being the isotope ratio measurement when calculating ratios from transient signals from the online CE. Although the estimated LOD for albumin, at 2.6 mg L⁻¹ S, was two orders of magnitude higher with respect to other published methods, the absolute LOD for S of 96.2 pg S was comparable when the very low injection volume (5 μL) was taken into account. In other work, online ID was employed in conjunction with HPLC-ICP-MS for the simultaneous monitoring of Cd and iAs in polished rice.¹²⁵ Following a single heat assisted extraction procedure and addition of the enriched ¹¹¹Cd ID spike, Cd and As species were measured by HPLC-ICP-MS in the same analytical run. In the proposed method, Cd concentration was calculated using the reverse ID method whereas As^V, DMA and MMA species were determined by a one-point (As^V) calibration method using the spiked ¹¹¹Cd as an IS; an approach that produced comparable results for iAs to those obtained with conventional calibration methods in white rice flour CRMs (NMIJ 7501-a, 7502-a and 7503-b). All measurement uncertainties were less than 3% and similar to those obtained by conventional ID-ICP-MS. The method was also adapted for the rapid screening of Cd and iAs content of rice by monitoring the intensity ratios of natural ¹¹¹Cd and iAs to the additional ¹¹¹Cd and then using pre-determined thresholds to indicate that accurate ID (for Cd) and/or full species determination (for iAs) was required for food safety assessment.

A method discussed last year in this review for the direct analysis of whole blood by way of LA-ICP-MS combined with an aerosol local extraction cryogenic ablation cell was applied to CSF.¹²⁶ The concentrations of twelve elements (Al, Ba, Ca, Co, Cr, Cu, Fe, Mg, Mn, Ni, Pb and Zn) were determined directly from 5 μL of CSF, with no sample preparation required. It was necessary to use matrix-matched calibrators (in 0.4% bovine serum albumin) to optimise the accuracy of the method. Solidifying the sample in a cryogenic ablation cell prior to ablation led to a marked improvement in precision compared with ablation at room temperature, resulting in RSDs between 2.2% (Pb) and 6.5% (Mn). A laminar flow aerosol local extraction strategy using a specially designed ablation cell improved transmission efficiency of the aerosols and increased signal intensity four-fold with respect to the standard commercial ablation cell. The LODs ranged from 0.17 μg L⁻¹ (Mn) to 8.67 μg L⁻¹ (Mg). To assess the accuracy of the method, standard addition recovery experiments in synthetic CSF were performed, achieving recoveries of 93.3% to 114%, and analysis of Seronorm™ serum L1 and L2 demonstrated generally good agreement with certified values. The method was applied to 24 CSF samples from patients with various neurological diseases. Although the method was suited to very small volumes of biological fluids and facilitated high-throughput analysis, it is questionable whether it would offer any real advantage over analysis by simpler and cheaper conventional solution nebulisation ICP-MS, following “dilute and shoot” sample preparation, as is routinely employed in clinical laboratory practice.

4.2 Atomic absorption and atomic emission spectrometry

Atomic absorption spectrometry is a reliable analytical technique, well-established in routine practice, with known pros and cons. Research highlighting novel aspects is nowadays frequently carried out using HR-CS-AAS instruments and GF atomisation. In this Update, we report about two works of note. Aramendía et al.¹²⁷ presented a new perspective for the direct B determination in solid biological samples at the mg kg⁻¹ level by means of GF molecular absorption spectrometry. They detected B as the diatomic BF molecule, deploying a gas phase reaction with an Ar–CH₃F mixture as fluorinating agent, with which the GF was filled at the end of the pyrolysis step of the temperature programme. A mixture of permanent W, citric acid and Ca(NO₃)₂ as chemical modifiers was used. With this approach, memory effects, very common in GFAAS, were avoided and a straightforward calibration with aqueous standard solutions was possible. The reliability of the method was assessed analysing two solid CRMs (NIST SRM 1570a-spinach leaves and SRM 1573a-tomato leaves) with recoveries ranging from 85% to 95% and precision (RSD%) in the range of 15%. For NIST SRM 1570a, the time-absorbance profile correction method for PO interferences was needed, while for NIST SRM 1573a, no correction was necessary. Achieved LOD was 0.6 mg kg⁻¹, which is quite similar to previously published data for B determination by AAS, but required milder furnace conditions with the additional advantage of also providing isotopic information (¹⁰B, ¹¹B), due to monitoring bands at λ = 200.993 nm for ¹¹BF while that at λ = 201.160 nm, was used for ¹⁰BF. The other paper¹²⁸ implemented solid sampling HR-CS-GFAAS for the determination of Tl in water from an area where a Tl deposit was discovered, after sorption on chromatographic filter paper, achieving a PF of 55 and an LOD of 0.018 μg L⁻¹. The following parameters were optimised: temperature programme, pH, extraction time, sample volume and effect of potential interferences. In the optimised procedure, 9 mL of the sample and the weighed chromatographic paper disk (Whatman P81, 2.5 mm diameter) were added to a 50 mL polypropylene tube. After adjusting the pH to 7, 0.2 mL of 0.01 mol L⁻¹ EDTA were added. The sample was then diluted to 10 mL and shaken for 3 min. Subsequently, the paper disk containing the retained analyte was removed from the solution and the determination was performed using the HR-CS-GFAAS directly on the paper. Calibration curves were obtained by analysing standard Tl solutions that underwent the same preconcentration procedure. Recovery tests and analysis of a water CRM (NW-TMDA-52.4) were performed to evaluate trueness, and the results ranged from 91% to 110%. The procedure was used to determine Tl levels in river and tap water samples and the results were comparable with those obtained using ICP-MS. Thallium concentrations in the analysed samples ranged from <0.059 to 0.80 μg L⁻¹.

Single cell analysis and NPs determination are still an important research topic showing the technical capabilities of atomic spectrometry. Apart from the wide application of ICP-MS for these tasks, some progress was recently made using ICP-OES. A new conical ICP torch, with better analytical performance than conventional ones, and a newly designed heated chamber coupled with an HEN were applied by Guo et al.¹²⁹ to enable SiO₂ single-particle and Ca determination in single cells. For the detection of single SiO₂ particles, the number of particle events obtained with the new sample introduction system was found to be up to 9-fold higher than that with the conventional system, without sacrificing the signal intensity. Subsequently, Ca was successfully detected in human and mice breast cancer cells as well as mice osteocytes. The cell detection efficiency turned out to be around 2–3%, which is much higher than the previously reported values. Finally, as a new application, the effect of Yoda1, a recently identified activator of Piezo1 Ca channel, on osteocytes was investigated, allowing to observe a 36% increase of the Ca content in Yoda1-treated cells compared to the control sample.

Sensitive and high-throughput analysis of trace elements in biological samples with limited volume is of great significance especially for clinical studies. The group led by Liu¹³⁰ developed a simple, sensitive, high-throughput and portable droplet cathode GD-OES method for the rapid analysis of micro-volume serum samples. The digestion of 10 μL samples was carried out in PFA jars with 0.5 mL conc. HNO₃ and 0.1 mL H₂O₂ at 100 °C for 4 h. The digests were heated at 80 °C to nearly dry, then dissolved and diluted with 0.1 mL of HNO₃ and subjected to analysis. The cathode component consisted of 20 nylon columns, which served as holders for the droplet samples to be tested. The anode component was a tungsten rod. A micropipette was used for the introduction of 10 μL samples. To initiate and sustain the discharge a high voltage, direct current power supply operating in a constant current mode, was employed. The resulting spectral signal of GD was transmitted through a quartz window and entered into a collimating lens, completing the spectral acquisition by the hand-held CCD spectrometer. The signal acquisition of each droplet sample could be completed within 300 ms, enabling the detection of approximately 240 samples per h. Since the samples being analysed by this instrumentation were single droplets rather than flowing samples, it was necessary to perform a comprehensive optimisation of experimental parameters. The process included the type of electrolyte, pH of the solution, sample volume and discharge current. Under optimised conditions, the LODs for Li, Ca and K were 0.002 mg L⁻¹, 0.078 mg L⁻¹ and 0.005 mg L⁻¹, respectively. The trueness of the proposed methodology was verified through the analysis of two CRMs (Seronorm™ serum L-1 and L-2) with recoveries ranging from 98% to 103%. Results from the analysis of 20 serum samples were compared with values obtained by ICP-OES. The obtained RSD% (n = 5) was approximately 5% and the linearity of calibration curves for all three analytes was confirmed by R² > 0.998.

Among the new generation of MIPs it is worth to mention the microwave-sustained inductively coupled atmospheric-pressure plasma for which previous works have shown similar detection capabilities to those afforded by ICP-OES. Serrano et al.¹³¹ applied this instrumentation with axial view to complex matrix sample analysis. The goal of the work was to evaluate MICAP-OES performance (analytical figures of merit, matrix effects, etc.) for elemental analysis of samples of different nature. The method was tested on seven CRMs, representing a variety of matrices (CRM-DW1 Drinking water; BCR-146 sewage sludge – industrial origin; BCR-185 bovine liver; BCR-278R mussel tissue; NIST SRM 1549 non-fat milk powder; ERM® EC681k polyethylene – high level; BCR-483 sewage sludge amended soil). The drinking water CRM was analysed directly. Solid samples (BCR-146, BCR-185, BCR-278R and NIST SRM 1549) were analysed after ultrawave MAD, with 4 mL conc. HNO₃ added to 0.1 g sample, except for ERM® EC681k, where the further addition of 1 mL of conc. H₂SO₄ was necessary. Digests were brought to a final weight of 15 g with ultrapure water. To assess bio-availability, BCR-483 (sewage sludge amended soil) underwent four different single-step extractions with 0.05 mol L⁻¹ EDTA, 0.43 mol L⁻¹ CH₃COOH, 0.01 mol L⁻¹ CaCl₂ and 0.1 mol L⁻¹ NaNO_3, as recommended in the CRM report, and the extracts of each step were centrifuged. Both the dilute digests and the extracts were filtered (0.45 μm), then stored at 4 °C until analysis. Both spectral and non-spectral interferences were investigated for 19 elements (Ag, Al, As, B, Ca, Cd, Co, Cr, Cu, Fe, Ga, In, Mg, Mn, Ni, Pb, Sr, Tl, Zn) in the presence of inorganic acids (H₂SO₄, HCl), organic substances (glycerol, CH₃COOH) and saline solutions and compared to a 5% HNO₃ solution. The MICAP technique was found to be highly robust with regard to influences of inorganic acids and organic matrices. Even though this system was still prone to matrix effects caused by easily ionisable elements, changes in both atomic and ionic emission were significantly lower than those traditionally reported for MIPs. Analyte recoveries ranged between 82.6% and 113%. The LODs in digested samples ranged from 1 mg kg⁻¹, for Ca and Mg, to 90 mg kg⁻¹, for Ag. Data also showed that there was not a universal IS to correct matrix effects and improve long-term performance, thus requiring two ISs to correct matrix effects for atomic (i.e. Rh or Pd) and ionic (i.e. Sc and Y) emission lines.

4.3 Laser induced breakdown spectroscopy

Review articles are discussed in Section 1 of this Update, however, two review papers have specifically covered the use of LIBS for biomedical research applications. Manrique et al.¹³² covered the instrumental developments, describing the improvements and differences between laser systems, such as the wavelength, pulse duration (ns vs. fs), energy, sample focussing, grating type and detector type. This was followed by sample preparation considerations for soft tissues, calibration procedures and statistical analysis of the spectra. Finally, key application papers from the last 20 years were highlighted for tissue differentiation, imaging/mapping, cancer and disease diagnosis, and biomarker detection with elemental NP tags. Hou et al.¹³³ focussed on recent advances with LIBS for sample preparation, signal enhancement, spectral processing and data modelling. The authors discussed progress and trends since 2017 across multiple areas of biomedicine, namely biological tissues (blood, bones, hair, nails, teeth and various cancer tissues), medicinal plants (leaves, roots, stems) and microorganisms (bacteria, fungi and viruses). Both reviews provided useful resources for progress with bioanalytical applications of LIBS.

The use of LIBS for disease diagnosis is a regular feature in this section and within this review period, a number of works have shown new applications and use of higher complexity statistical modelling methods. Blanchette et al.¹³⁴ implemented LIBS for detection of bacteria in blood. Whole blood samples, spiked with known amounts of representative human pathogens, namely Enterobacter cloacae, Escherichia coli, Pseudomonas aeruginosa and Staphylococcus aureus, to simulate a bacterial infection, were compared to control samples. The blood was deposited directly onto nitrocellulose filters and analysed by LIBS. The intensities of 15 emission lines in the spectra and the ratio of those intensities (n = 92) were used as the input data for the PLS-DA model to differentiate between the controls and blood spiked with bacteria. Additionally, PCA was applied as a preprocessing method for the whole spectra from 200 nm to 590 nm, followed by ANN analysis to identify the specific pathogen. For both models, 100% sensitivity and specificity were observed for detection and identification of the pathogen. When tested using PCA-ANN with only 18 samples to build the model and the remaining as the test samples, the average sensitivity, specificity and classification accuracy were 85.5%, 95.0% and 92.5% respectively. This demonstrated the potential of LIBS at rapid detection and identification of bacteria in whole blood with very little sample preparation. The authors noted that speed of analysis compared to classical methods could save days which can be critical in cases such as sepsis. Lin et al.¹³⁵ reported on the combination of LIBS and Raman spectroscopy to diagnose different stages of lung cancer tumours. Samples of cancerous tissue and adjacent normal tissue were collected from 44 patients, covering cancer stages 1 to 4. A platform was developed which could obtain both LIBS and Raman spectra simultaneously. Using the emission lines of Ca I, Ca II, Cu II, Fe I, Mg I and Mg II and molecular peaks, such as phenylalanine (1033 cm⁻¹), tyrosine (1174 cm⁻¹), tryptophan (1207 cm⁻¹), amide III (1267 cm⁻¹) and proteins (1126 cm⁻¹ and 1447 cm⁻¹), ratios of the signal changes between the cancer stages were calculated and entered into a Bayesian fusion model to create the combined dataset. Then a 1D CNN process was applied for classification. The final model achieved accuracy, sensitivity and specificity of 99.17%, 99.17% and 99.88% respectively, demonstrating a potentially powerful tool for rapid diagnosis and stage identification of lung tumours. In a second study,¹³⁶ the same group implemented LIBS analysis for the detection of benign and malignant lung tumours. Spectra were collected from tumours along with normal lung tissue and underwent preprocessing with background subtraction, drift correction, standardisation and feature extraction. This highlighted Ca II 393.4 nm, Cu II 518.3 nm, Mg I 285.2 nm, Mg II 279.6 nm and Na I 589.6 nm as the key lines showing differences between the three sample types. Then a combination of a RF and a 1D residual network model was created to distinguish between the tissue types. The approach achieved classification accuracy, precision, sensitivity, and specificity of 91.1%, 91.6%, 91.3% and 91.3%, respectively, demonstrating the capability to differentiate between benign and malignant lung tumours and normal tissue.

Several papers have focussed on techniques to improve the sensitivity of LIBS for the analysis of biological samples. Khumaeni et al.¹³⁷ described the use of an alternative laser source for the analysis of blood serum. The Nd:YAG laser is the most common source, however, in this work, a transversely excited atmospheric CO₂ laser at a wavelength of 10.64 μm and a pulse duration of 200 ns was used to overcome sensitivity issues with Nd:YAG lasers with soft or liquid samples. Furthermore, a copper mesh target was used for deposition of the serum sample which enhanced the plasma formation, leading to a 5- to 6-fold increase in signal intensities. Qualitative elemental identification was achieved by comparison to the NIST atomic spectral database with C, Ca, Fe, H, K, Mg, Na, O and P, for neutral, atomic and ionic lines. Quantitative analysis was performed by the ‘calibration free’ method, which relies on the plasma achieving local thermal equilibration to calculate the temperature (Boltzmann distribution) and electron density (Stark broadening), in combination with further fundamental data from the NIST atomic spectral database. It was possible to quantify C, Ca, Fe, K, Mg, Na and P, with this approach. The LIBS values were compared against XRF spectrometry, achieving good agreement for Ca, Fe, K, Mg, Na and P. The authors noted that it is not possible to measure elements such as C, H and O by XRF spectrometry, leading to an additional advantage of the metal mesh enhanced CO₂ LIBS system. Zhang et al.¹³⁸ implemented NPs to improve sensitivity and precision of LIBS analysis. The ‘nanoparticle-enhanced’ approach has been widely used by Surface Enhanced Raman Scattering researchers but not significantly applied to LIBS. Here, 5 μL of serum was spotted onto the filter paper impregnated with Ag NPs which was compared to the commonly used aluminium and silicon substrates, also coated in Ag NPs. The analysis focussed on Ca, K and Mg and showed that the Ag NP filter paper increased the intensities, with improvement factors of 1.76, 3.10 and 1.85 for Ca, K and Mg, respectively. When compared to the aluminium and silicon substrates, differences were found between the elements, but in general Ag NP filter paper had the same or increased intensities compared to silicon, whereas aluminium had higher intensities compared to Ag NP filter paper. However, the RSD% was significantly larger with aluminium, for example, the K (I) emission at 766.49 nm was 26.89% on the aluminium substrate compared to 3.48% for the filter paper. On further investigation, this was likely due to the ‘coffee ring’ effect on both the aluminium and silicon substrates, whereas the filter paper was effective in forming a uniform distribution. The study showed the dual benefit of using NP loaded filter paper for electrolyte serum analysis (enhanced intensities and improved measurement precision) with small sample volumes and minimal preparation.

Two papers have reported methods to enhance the sensitivity and precision for the analysis of food samples. Park et al.¹³⁹ developed a modified silicon substrate for the analysis of edible salts to reduce the heterogeneity from simple drop evaporations. Here, a grid was laser etched on to a silicon wafer and was surrounded by hydrophobic tape. Commercial edible salts were dissolved in water and a 15 μL drop was placed on the surface of the bare wafer and modified wafer. Once dried, it was clear the salt on the bare wafer was very inhomogeneously dispersed, with the outline of the drop forming a crust containing significant large salt crystals, whereas the sample was more evenly distributed across the grid of the modified wafer. However, it was observed that crystals tended to form around the grid trenches, so some degree of inhomogeneity was still present. To further improve the precision of the measurements, the authors proposed a data sampling scheme where the spectra were separated into three by n + 3, i.e. spectra numbers 1, 4, 7, followed by 2, 5, 8, and finally 3, 6, 9, etc. This significantly reduced the intensity RSD% from 5–25% to below 1%. This enabled detection and quantification of Ca, K, Mg and S after signals normalisation with the Na signal, achieving LODs of 14 mg kg⁻¹, 0.64 mg kg⁻¹, 1.7 mg kg⁻¹ and 530 mg kg⁻¹, respectively. The results for five edible salts from Australia, Bolivia, France and South Korea were presented showing significant differences in the elemental levels between them. Wen et al.¹⁴⁰ assessed various elemental NPs (Ag, Au, Pd and Pt) on modified resin as a preconcentration method to enhance the detection of Cr, Cu and Pb in tea infusions. The NPs were produced on resin carrier particles which were exposed to the tea infusions to capture the analytes of interest. The resin was filtered, dried and analysed by LIBS. It was found that Ag NPs produced the greatest enrichment, 3- to 7-fold improved compared to the resin alone. Impressive LODs for LIBS were obtained; 0.22 μg L⁻¹ for Cr^III, 0.33 μg L⁻¹ for Cu^II and 1.25 μg L⁻¹ for Pb^II. Spike recoveries ranged from re 83.3% to 114.5%, demonstrating its performance as a direct and sensitive technique for measurement of Cr, Cu and Pb.

The application of LIBS detection for the elemental analysis of water has featured in a number of works in this review period. Wang et al.¹⁴¹ described the addition of a Tesla coil discharge device to the sample stage to improve the sensitivity of Cr and Pb for the analysis of water samples. An aluminium substrate was used to place 120 μL droplets of the calibration standard or sample, before drying. The Tesla coil device created a discharge arc which is positioned to fire at the same location as the laser target. The purpose of the coil was to pre-ionise the sample surface immediately prior to the laser pulse. This increased the sample temperature and improved the coupling efficiency between the sample and laser, therefore leading to an increased ablation mass and ultimately improving the spectral intensity. Once the setup was optimised, the sensitivity improvement was approximately 2-fold for Cr and Pb and achieved LODs of 12.2 ng mL⁻¹ for Cr and 5.5 ng mL⁻¹ for Pb. Furthermore, the RSD% also improved from 7.02–13.22% without the coil to 4.70–6.40% with the device. The study demonstrated the benefit of using the Tesla coil discharge as a means to improve the sensitivity of LIBS for heavy metal determination in waters which typically requires LODs in the low ppb region. Yang and co-workers¹⁴² used a different approach to improve the sensitivity of P in water. Here, the sample was exposed to the laser as an aerosol, produced from a concentric nebuliser with argon and a cyclonic spray chamber to remove larger droplets. The fine droplets entered a heated quartz tube aligned with the laser. Following optimisation of the aerosol temperature and laser energy density, the optimal conditions were 100 °C and 18.70 J cm⁻² respectively. Compared to 25 °C and 26.00 J cm⁻², the LOD for the atomic line of P (I) at 213.618 nm decreased from 5.22 ppm to 1.81 ppm and achieved a 3-fold improvement in the S/N. The accuracy was assessed by the ‘leave one out’ cross validation and calculation of the RMSECV for the predicted calibration standard concentration, which was 2.91 ppm. Additionally, water samples spiked at three concentration levels had recoveries between 99.08% and 104.59%.

Two publications applied LIBS for the analysis of food products. Ribeiro et al.¹⁴³ implemented LIBS for the discrimination between genetically modified maize grain and non-genetically modified. Samples were collected from an experimental farm, with grains from two conventional, non-genetically modified maize varieties and four varieties of genetically modified ones. Spectra (n = 30) were collected over two wavelength ranges, namely 175–330 nm and 275–770 nm, with spectral lines identified for Al, Ca, CN, Fe, K, Mg and Na at similar intensities in all samples. Therefore, the data was pretreated by PCA to reduce the dimensions and noise before implementing machine learning techniques such as k-nearest neighbour, SVM and pairwise combinations of the ‘leave one out’ method, with the latter achieving 100% accuracy. The study demonstrated the potential of LIBS with chemometric tools for discrimination of transgenic maize without direct quantification. Fayyaz et al.¹⁴⁴ implemented calibration free LIBS analysis of the medicinal plant Saussurea simpsoniana to determine its elemental levels. In this work, a large wavelength range was measured due to the combination of six miniature spectrometers to capture the emissions. Leaves, roots and seeds were analysed and 13 major and minor elements were detected, namely Al, Ba, C, Ca, Fe, H, K, Li, Mg, Na, Si, Sr and Ti. By ensuring local thermodynamical equilibrium was achieved, quantification was performed using the fundamental parameters determined by the plasma excitation temperature (Boltzmann distribution) and electron number density (Stark broadening). To validate the results, the same samples were analysed by EDX, finding good agreement between the two techniques.

4.4 Vapour generation procedures

Typically, this section of the Update covers both AFS and VG applications. However, within this review period, no papers with significant developments for AFS in clinical and biological materials or food and beverages were identified.

A number of works have focussed on the use of trapping and preconcentration prior to analysis by AAS as a means to significantly enhance the detection capability. In two papers,^145,146 a tungsten coil coated in either gold or palladium was implemented for the ultrasensitive analysis of As and Se, respectively, in drinking water. In both cases, the hydrides were generated, trapped and preconcentrated on the coil before release at elevated temperatures. For As,¹⁴⁵ the enrichment factor was 15.3 compared to non-trap HG-AAS and achieved an LOD of 4.8 ng L⁻¹, which was comparable to HG-ICP-MS. The method was validated with CRMs (NIST SRM 1640a natural water, EnviroMAT drinking water high, CRM-023 sandy loam 7 and NRCC DOLT5 dogfish liver) and spiked tap water samples. For Se,¹⁴⁶ a 38-fold enhancement was found compared to non-trap HG-AAS, with an LOD of 2.6 ng L⁻¹. The method was also validated with CRMs and spiked tap water samples and t-tests showed no significant differences between the results and expected values. Both studies demonstrated the potential of coated coils for a significant improvement in sensitivity for the determination of hydride forming elements by AAS. Schlotthauer et al.¹⁴⁷ described improvements to the HG process for the determination of Pb in water samples by AAS, as typically the formation of PbH₄ requires additional reagents to promote the formation of Pb^IV and suffers from interferences from common elements such as Cu. In this work, Na₂B₄O₇ was used as a buffer in the NaBH₄–NaOH solution to provide a wider working pH range and the addition of K₃Fe(CN)₆ further enhanced the formation of PbH₄. Furthermore, KSCN was added to the sample solution as a masking agent for interfering elements. Finally, FI was used to minimise the impact of the matrix on the hydride formation. The fully optimised method achieved very respectable figures of merit, with an LOD of 0.3 μg L⁻¹ and an LOQ of 0.9 μg L⁻¹ which is more than an order of magnitude below the maximum guidance limit for Pb in drinking waters (10 μg L⁻¹). This approach was then applied to 50 water samples, finding levels between <0.3 μg L⁻¹ and 88.8 μg L⁻¹ Pb. Manousi and Anthemidis¹⁴⁸ reported the development of a dual preconcentration and vapour separation device using polyurethane foam in a piston driven syringe, therefore creating in situ CVG. The process complexed Hg with APDC online before entering the syringe system. The polyurethane foam was immobilised on the piston which trapped the Hg-APDC complex in the pores of the foam. The online addition of the reductant enabled vapour formation within the syringe, essentially providing a membrane-less gas–liquid separator. Based on 90 s preconcentration time, the LOD was 0.02 μg L⁻¹ achieving 3.8% RSD at the 0.5 μg L⁻¹ level and was linear to 4.0 μg L⁻¹ Hg. The accuracy of the method was determined using CRMs (BCR-278R mussel tissue, Seronorm™ urine L1, IAEA-433 marine sediment), with recoveries of 92.9%, 94.4% and 103.6%, respectively, as well as spiked samples of blood, hair and urine achieving recoveries of 93.6%, 98.0% and 96.4%, respectively. Overall, the approach demonstrated a novel and rapid method to significantly improve the detection capability by CV-AAS.

The combination of VG processes with plasma-based techniques has shown development across this review period. Dong et al.¹⁴⁹ described a PVG method with ICP-MS for the sensitive detection of Sb in water. Here, V^IV was added as a modifier with organic acids to enhance the formation of the volatile Sb species by UV irradiation. The optimal conditions were 10% (v/v) HCOOH, 10% (v/v) CH₃COOH and 80 mg L⁻¹ V^IV with 100 s UV exposure time, which achieved an LOD of 4.7 ng L⁻¹. It was also found that the method was insensitive to Sb species, as both Sb^III and Sb^V reacted, forming (CH₃)₃Sb as the volatile species. The repeatability (RSD%) for 0.5 μg L⁻¹ Sb^III and Sb^V standard solutions (n = 7) was 1.9% and 2.3%, respectively. The method provided a 19-fold improvement in sensitivity compared to conventional nebulisation ICP-MS. Accuracy was assessed with water samples spiked with 0.2 μg L⁻¹ Sb, obtaining recoveries ranging between 100% and 110%. The researchers applied the approach to various drinking water samples. Lan et al.¹⁵⁰ reported the development of a system to support the coupling of ETV and ICP-MS for the analysis of As, Cd, Pb and Se in food samples via slurry sampling. This comprised of the addition of a new gas line in the GF to create a turbulent flow which prevented deposition on the transfer joints and minimised carry over effects. Additionally, a PFA cyclonic spray chamber was added as a ‘signal delay device’ which effectively slowed the introduction of the analyte vapour into the ICP-MS, enabling the number of signal points to increase from 8 to 35. This significantly improved the precision in real food samples from between 14.8% and 16.2%, without the device, to between 1.2% and 8.9% with it in place. The system was fully optimised, achieving LOQs of 1.0 ng g⁻¹ As, 1.0 ng g⁻¹ Cd, 1.9 ng g⁻¹ Pb and 1.7 ng g⁻¹ Se, and R² > 0.999 across the range of 1 ng g⁻¹ to 4000 ng g⁻¹ using standard addition calibration. Five CRMs (GBW10010a rice, GBW10045a rice, GBW10013 soybean, GBW10016a tea and GBW10052a tea) were also analysed, obtaining recoveries between 86% and 118% of the certified values and repeatability (RSD%) ranging from 1.2% and 8.9% (n = 6). The analysis time was <3 min and the slurry preparation time was <5 min, demonstrating a fast, accurate and precise method without the need for complete sample digestion. Bitencourt et al.¹⁵¹ published an analytical method for the simultaneous measurement of As and Hg using CVG-MIP-OES for dietary supplements. The instrumental setup included a ‘Multimode Sample Introduction System’ (MSIS) which enabled elemental determination using conventional nebulisation and concurrent generation of volatile species with CVG. However, in this work, only CVG was applied to produce AsH₃ and Hg⁰. The reagent concentrations (NaBH₄ and HCl), MSIS configuration and flow rates were optimised and the use of a pre-reduction step for the conversion of As^V to As^III was investigated to ensure accurate analysis of As due to the poor reaction with As^V. It was found that 0.26 mol L⁻¹ thiourea and 1.2 mol L⁻¹ HCl for 10 min was effective, with potential residual acid from sample preparation having no impact on the AsH₃ formation (10–80% (v/v) HNO₃ was tested). The final method was applied to 13 dietary supplement samples which were prepared by MAD (between 0.5 g and 1.6 g sample, digested using 8 mL HNO₃ and 2 mL HCl). The method LOQs were in the range 5–15 ng g⁻¹ for As and 29–93 ng g⁻¹ for Hg, which is impressive for MIP-OES. The sensitivity improvement was three-fold and 12-fold for As and Hg, respectively, compared to conventional nebulisation MIP-OES. The evaluation of the accuracy was determined using spiked samples at three concentration levels, giving recoveries between 93% and 108% for As and between 94% and 106% for Hg, and analysing CRMs (NIST SRM 1572-citrus leaves and SRM 1575a-pine needles). Additionally, the same sample solutions were analysed by ICP-MS, finding no significant difference (t-test, confidence level of 95%). Overall, the approach demonstrated the accurate and sensitive analysis of As and Hg using MIP-OES which is more resource friendly with lower cost of operation than ICP-MS.

4.5 X-ray spectrometry

The ASU Update on advances in XRF spectrometry² complements the applications of XRF to clinical and biological materials, foods and beverages covered within this Update. Imaging applications of X-ray spectrometry are described separately in Section 6.2.

The use of TXRF spectrometry for cosmetics, food and pharmaceutical analysis was the focus of a review by Marguí et al.¹⁵² who compared TXRF spectrometry to other common techniques such as acid digestion followed by analysis by AAS, ICP-OES and ICP-MS. Particularly for cosmetic and pharmaceutical samples, it can be difficult to obtain fully digested samples with MAD due to the presence of excipients, waxes, pigments, silica, TiO₂, etc., whereas direct solid analysis is possible with TXRF spectrometry. Examples of published works were presented and the remaining challenges discussed, such as quantification of low Z elements and volatile elements, spectral overlaps and lack of standardised methods. It was also noted that TXRF spectrometry represents an opportunity to apply green and sustainable laboratory practices, due to its low energy usage, rapid analysis and potential to reduce the volume of chemicals used. Visentin et al.¹⁵³ employed ED-XRF spectrometry to analyse I in cow’s milk. Given the nutritional importance of I and difficulties with sample preparation, the approach offered a direct and fast measurement technique. The workers used the Lα spectral peak at 3.94 keV and acquired data over 10 min for milk samples (n = 56) across a large concentration range (from 100 μg L⁻¹ to 2350 μg L⁻¹). The samples were also analysed by ICP-MS after dilution in 0.6% NH₄OH and heating to 90 °C for 1 h, then a subset (n = 20) was used to generate the calibration line for the ED-XRF data. The accuracy of the model was determined by linear regression analysis, achieving a cross validation coefficient (R²) of 0.60, demonstrating suitable accuracy only for the purposes of screening or detection of very high or very low concentrations, e.g. discrimination between natural and fortified I levels. Weinberger et al.¹⁵⁴ systematically tested the application of ED-XRF spectrometry to quantify Ca, Cu, Fe, K, Mn, Rb, Sr and Zn in coffee. Two methods of sample preparation were assessed, namely loose powders in a Teflon cup and powder pressed into pellets with wax as a binder. It was found that the pellets with 10% wax addition provided the best compromise between sensitivity and precision, as higher wax effectively diluted the sample but pellets reduced the impact of particle size. The concentration of the coffee samples was calculated using fundamental parameters and the accuracy was assessed by comparison to ICP-OES data. Good equivalence was found for Ca, K, Rb and Sr but the slope and intercept were significantly different from 1 and 0 for Cu, Fe, Mn and Zn, most likely due to the lower levels for these elements (generally <40 mg kg⁻¹ in most samples). Therefore, the ICP-OES data was used as the calibration concentrations instead of the fundamental parameters for identification of product types (pure coffee, instant and with additives) with chemometric processes. It was found that using PLS-DA with the raw spectra provided a model with 100% accuracy using the ‘leave one out’ approach, therefore, quantification was not strictly required. The straightforward sample preparation and speed of analysis was demonstrated as a potential tool for coffee product identification.

There were two papers published by the same research groups describing the use of monochromatic excitation XRF spectrometry for heavy metal analysis in biological samples for forensic purposes. The relatively new development of monochromatic excitation (ME) uses a doubly curved crystal to convert the polychromatic X-ray source to monochromatic, leading to significant improvements in peak intensity and background reduction. In the first study,¹⁵⁵ multiple tissue types (namely brain, heart, intestines, kidney, liver, lungs, pancreas, spleen, stomach and testes), blood and urine from rat models, dosed with Tl to mimic acute poisoning (at 5, 15 and 30 mg kg⁻¹ bw), were directly measured by ME-XRF spectrometry with very minimal sample preparation. The values obtained by calibration with fundamental parameters were compared with ICP-MS data after MAD, to assess the accuracy. Paired t-tests showed no significant difference between the methods for all sample types at different Tl dosing levels. The Tl levels measured by ICP-MS in the control group ranged between 0.001 mg kg⁻¹ and 0.1 mg kg⁻¹ but could not be detected by ME-XRF, which does limit the use of the approach. However the researchers did demonstrate its use in a forensic investigation of suspected acute Tl poisoning. Blood and urine from three human subjects were analysed by both ME-XRF spectrometry and ICP-MS, finding relative differences between −2.19% and 13.22% at concentrations ranging approximately from 0.11 μg mL⁻¹ to 0.55 μg mL⁻¹ Tl. In the second study,¹⁵⁶ the method was extended to As, Hg and Pb with a systematic investigation of the ME-XRF spectrometry for matrix effects, moisture content, sample mass and tissue type (comparisons were made between pork liver, lamb liver, pork heart, pork kidney and pork lung). The analytical figures of merit were determined with the optimised method, demonstrating linearity from 0.2 μg g⁻¹ to 200 μg g⁻¹, recoveries of spiked samples in the range between 91% and 113% and both intra- and inter-day precision between 0.12% and 18.11%. The LOQs for As, Hg, Pb and Tl were 0.12 μg g⁻¹, 0.12 μg g⁻¹, 0.20 μg g⁻¹ and 0.18 μg g⁻¹, respectively. Finally, the optimised method was applied to two forensic cases, suspected of heavy metal poisoning. The same samples were also analysed by ICP-MS, showing no significant differences between the two techniques. Whilst the potential of ME-XRF spectrometry as a forensic tool was clearly demonstrated, the application would be limited to poisoning cases by the LOQ.

In a paper by Pandolfi et al.,¹⁵⁷ the impact of formalin-fixed paraffin-embedding of biological tissue samples for elemental analysis by ED-XRF spectrometry was investigated. The researchers compared this standard histological method for sample preparation to pellets produced from adjacent tissue sections (n = 18) after lyophilisation and homogenisation. Five biologically relevant elements were selected for evaluation, namely Ca, Cu, Fe, S and Zn, and the intensities obtained from the FFPE sections and pellets were plotted to generate a LS regression line. One sample was used to test the correlation, finding that the maximum deviation ranged between 2% and 9% across the elements, demonstrating sufficient correction from the matrix effect due to the FFPE. In order to achieve quantification, multiple CRMs (BCR-414 plankton, ERM® BB186 pig kidney, ERM® BB422 fish muscle, ERM® BD151 skimmed milk, ERM® CA713 wastewater, MODAS-3 HerTis herring tissue, MODAS-5 CodTis cod tissue, NCS ZC73013 spinach, NCS ZC73029 rice, NCS ZC73032 celery, NCS ZC73033 scallion, NIST SRM 1640a natural water, NRCC TORT-2 lobster, Sigma Aldrich QC3163 seawater) were analysed as pellets to generate the calibration equations which were applied to both the FFPE and pellet samples, showing no statistically significant difference between the sample types at the 95% CI. The work demonstrated the potential to gain further data from pathohistological tissue sections and historical tissue bank collections.

5. Nanomaterials and single cell analysis

This section covers developments for the detection of NPs in food and beverages, as well as in biological samples and pharmaceutical and consumer products. The use of NPs as a reagent for element preconcentration is covered in Table 2. Review articles are discussed within Section 1 of this Update, however, multiple papers specifically reviewing the analysis of nanomaterials and single cells were published, therefore, they are covered in this Section. An IUPAC Technical Report¹⁵⁸ provided an authoritative evaluation of the current state of the art for the detection of ENMs in complex samples. A full assessment of the analytical process was provided, using recently published articles as examples for sample preparation in solid and liquid matrices and the various analytical techniques available. Key points were ensuring that the preparation and separation methods did not impact the NP composition, structure or concentration, the difference between pristine NPs and real-world NPs, and the types of measurements required (i.e. diameter, shape, number concentration, composition, surface chemistry). The examples covered ENMs in foods, beverages, nutraceuticals, cosmetics, consumer products, biological samples (serum, plasma, urine, organs and plants) and environmental systems (water, air, soil and sediments). The discussion included comments on the need for multiple techniques due to the complexity of ENMs in real-world samples and for standardisation and RMs. Loeschner et al.¹¹⁴ focused specifically on the use of spICP-MS for the analysis of inorganic NPs in beverages, food products and additives, such as E171 TiO₂, E174 Ag and E551 SiO₂. A total of 49 papers were selected from their search criteria which were summarised and evaluated for sample collection and preparation, measurement conditions, calibrants, validation and data analysis methods. The authors noted that all studies assumed a spherical shape when calculating NP size, some papers provided evidence confirming this, but several did not include comment on the shape or presented SEM/TEM images with no discussion, despite clear deviations from spherical such as rods, ellipses and polygonal. This is a limitation with spICP-MS, as additional techniques are required to confirm the shape. The authors discussed the various calculation approaches and the software available, which was lacking in the IUPAC report. Furthermore, limited availability of inorganic NP RMs and QC control materials for size and number concentration was highlighted, also noting the lack of matrix materials. However, the benefits of spICP-MS for NP detection in foods and beverages were emphasised. Fernandez-Trujillo and co-workers¹⁵⁹ reviewed ICP-MS methods for the investigation of NP biological fate in toxicology studies. The paper included total elemental analysis, spICP-MS, scICP-MS, LA-ICP-MS and coupling with separation techniques such as AF4, HPLC and CE. These were applied to in vitro and in vivo toxicological assays to determine NP toxicity and NP transformations. There was a lengthy discussion on future perspectives and upcoming developments to further understand NP behaviour for toxicological research, also concluding that multiple techniques are required to obtain full understanding. Single cell analysis was reviewed by Davison et al.¹⁶⁰ covering spICP-MS, LIBS and LA-ICP-MS. The advantages and disadvantages of each technique, alongside sample preparation and cell fixation were evaluated. For example, LIBS can detect bulk elemental levels but lacks sensitivity for trace elements, though new developments with nano LIBS has shown promise. With LA-ICP-MS, a high degree of spatial resolution can be obtained at sub-ppm levels, but the ICP-MS type can limit the number of elements that can be analysed in time resolved mode. Similarly for scICP-MS, detector dwell time is critical to ensure single events are accurately detected, limiting the number of isotopes measured, but the technique can be used to count the number of cells as well as gaining elemental information for individual cells. The use of ICP-TOF-MS was also highlighted which, as a quasi-simultaneous technique, can therefore detect many isotopes in a single cell event or laser pulse. Calibration and validation were discussed, again noting the need for standardisation and accurate quantification. Zhou et al.¹⁶¹ produced a systematic review of the use of various metal NPs as tags for the identification of pathogens (bacterial and viral) with ICP-MS detection. The methods were split into four main categories of tagging mechanisms, namely nucleic acid probe modified NPs, protein probe modified NPs, quantum dots and hybrid NPs. The mechanisms, advantages and disadvantages were discussed, as well as the potential for multiplexing multiple pathogens in one assay. The paper provided a useful summary of this emerging field.

5.1 Food and beverages

The food additive TiO₂ or E171 can be used as a whitener and opacifier, but a proportion of the powder can be found in the nano range. Although the EU has banned E171, it is still permitted in other countries (e.g. USA, Great Britain) and there are concerns over NP migration from food packaging. Three papers have reported methods for the determination of TiO₂ NPs in food products. Bastardo-Fernández et al.¹⁶² described the comparison of two high efficiency sample introduction systems with spICP-MS to improve the size LOD for TiO₂ NPs in food simulants. Here, two models of the APEX desolvation systems were used (APEX Q and APEX Omega) which both contained a heated cyclonic spray chamber and Peltier cooled condenser, with the Omega also having a desolvation membrane to further improve sensitivity. These were compared with conventional nebulisation with a Scott double pass spray chamber, with all analyses performed by ICP-MS/MS with O₂ reaction gas and mass shift mode (⁴⁸Ti¹⁶O⁺). A TiO₂ powder referred to as NM-10200a from the EC JRC was used for determination of method accuracy. The samples were prepared in ultrapure water and food simulant solutions, namely 5%, 10%, 20% and 50% (v/v) EtOH and 3% (w/v) CH₃COOH. The results showed the APEX Omega system provided consistent data for mass concentration across the range of simulants, whereas the APEX Q had an approximate 5-fold reduction in mass concentration compared to ultrapure water. The conventional sample introduction approach could only analyse up to 10% EtOH and was positively biased by 250% due to signal enhancement from the simulated food matrix. Generally, the particle number concentration was not affected as this did not rely on instrumental sensitivity. In terms of particle size, the APEX Omega was in good agreement for all solutions (∼185 nm) with the expected diameter of 190.6 nm for the NM-10200a, whereas the APEX Q underestimated the diameter (∼100 nm) and the conventional setup overestimated it (∼275 nm). Therefore, the APEX Omega system was used to analyse a commercial grade E171 TiO₂ additive, obtaining a mean diameter of 318 nm which compared well to the SEM value of 300 nm. In a second paper,¹¹⁵ the method was extended to real food matrices in collaboration with a Danish research group. An eclectic mix of samples containing the E171 additive were acquired prior to the ban, namely chewing gums, chocolate bars, coated candies, cream buns and cakes, cured sheep’s cheese, nougats, powdered drinks and sauces. After homogenisation in a ball mill, the NPs were extracted with 0.1% (w/v) SDS and ultrasonicated for 3 min then diluted with ultrapure water. The NM-10200a TiO₂ powder and two pure E171 additives, characterised by SEM and DLS, were used as QCs. Total Ti levels were also determined by MAD with ICP-MS analysis, with concentrations ranging from 0.15 mg kg⁻¹ (cheese) to 3723 mg kg⁻¹ (nougat). However, this did not correlate with the number concentration, which ranged from 10¹⁰ kg⁻¹ for stuffed white chocolate to 10¹⁴ kg⁻¹ for coated chocolates, or the % NP fraction (i.e. % below 100 nm), which ranged from 4% for cheddar sauce to 87% for chewing gum. Additionally, a recovery factor was calculated based on the mass concentration from spICP-MS divided by the total Ti to estimate the NP recovery, which was between 1.3% for chewing gum and 58.8% for the coated chocolates. The authors suggested that the extraction protocol may not be applicable for all the matrices investigated. The paper provided an interesting insight into the TiO₂ levels in real samples as well as the challenges with such complex matrices. Espada-Bernabe et al.¹⁶³ also focussed on TiO₂ as E171 but specifically in coloured chocolate candies. Here, the researchers investigated the presence of TiO₂ in the nano form and the biological fate when ingested. The candy coating contained the E171 so this was removed by shaking in ultrapure water until dissolved. The extract solution was analysed by spICP-MS and TEM after sonication for 10 min and the pH adjusted to 6 to avoid agglomeration, then diluted in water prior to measurement. The total Ti content was also determined by ICP-MS where the extract was subjected to MAD. The ICP-MS and spICP-MS were operated in no-gas mode as the authors established Ca was not present in the samples. For the assessment of bio-accessibility, a sequential extraction procedure was applied to mimic gastrointestinal digestion using α-amylase for saliva, pepsin acidified to pH 1.8 with HCl for gastric solution and pancreatin for intestinal solution. This was applied to the aqueous coating extract and the whole candy. The extracts from each step were analysed by the same techniques as previously described. The average total Ti concentration in the different coating colours was 1219 ± 83 μg g⁻¹ for blue, green, orange, red and yellow but the brown coating was 29 ± 2 μg g⁻¹ whereas the white was 2365 ± 20 μg g⁻¹. The mean particle size determined by spICP-MS across all colours was similar, ranging between 138 nm and 151 nm, with <30% of the particles below 100 nm, which was under the EU limit of <50%. No significant difference for the mean particle size was identified between spICP-MS and TEM by ANOVA at 0.05 p-values. The results of the gastrointestinal digestion found that the particle size and the % of TiO₂ which was <100 nm remained consistent through the different stages for the aqueous extract and the whole candy. However, the % of bioavailable Ti was different between the two sample types, as the extract contained between 37% and 79% as bio-accessible Ti whereas, in the whole candy, it ranged from 6.0% to 14%. The authors extended the study by exposing Caco-2 cells (resembling intestinal epithelial cells) to the intestinal extract for 4 h and 24 h. Cell viability was 88 ± 9% at 4 h but decreased to 23 ± 4% after 24 h. The study provided an interesting insight into the biological fate of Ti in foods containing E171.

Bruvold et al.¹⁶⁴ described the validation of an enzymatic extraction method for NPs from mussel tissue, as a proxy for seafood, to investigate potential food safety risks and environmental impact. Homogenised mussel tissue was spiked with Au NPs (30 or 60 nm) as a matrix to optimise the method and determine the figures of merit. The extraction was performed using 200 g L⁻¹ protease. The size LOD was 18 nm, mass concentration LOD was 1.7 ng g⁻¹ and particle number concentration LOD was 4.2 × 10⁵ particles per g in the mussel tissue. The recovery of the spiked samples for particle mass concentration was between 76% and 77% and for particle number concentration ranged from 94% to 101%. Additionally, the researchers analysed a sample from the ACEnano PT scheme, achieving recoveries of 94% for size and 112% for particle number concentration. The method was then extended to determine endogenous Cr, Cu and Ti NPs in mussels from 3 disparate coastal locations, showing differences between the sites in terms of particle mass and number concentrations. Chalifoux et al.¹⁶⁵ compared two hydrolysis procedures (alkaline and enzymatic) for the extraction of Ag NPs from meat products. The researchers homogenised ground beef (∼19% fat content, ∼19% protein) with a suspension of 40 nm Ag NPs and a control meat, prepared with ultrapure water. The homogenates were split into two fractions, as one part was frozen and the other was lyophilised. For alkaline digestion, 2.5% TMAH was used and for the enzymatic procedure, 1.5 mg L⁻¹ protease was mixed with 1.5 mg L⁻¹ lipase and 5 mmol L⁻¹ HEPES (two types of protease were compared, namely porcine pancreatin and proteinase K). Ultrasonication was applied for both processes for 15 min and samples were centrifuged before analysis by spICP-MS. The TMAH achieved recoveries over 90% on a mass basis compared to only ∼60% for either pancreatin or proteinase. The authors suggested it was likely due to the higher digestion efficiency of TMAH in the high adipose content of the sample. However, in all cases, there was no evidence of agglomeration or dissolution as particle size measurements were all in agreement with the original NP suspension and frozen or lyophilisation treatment had no effect. The method LOD for number particle concentration was 4.8 × 10⁶ particles per g with a size LOD of 10.9 nm for the enzymatic extraction and LODs for the alkaline hydrolysis were 5.7 × 10⁷ particles per g and 10.5 nm.

Yeasts enriched with Se are common dietary supplements and there is interest in the potential of Se NPs as a novel nutritional supplement. Therefore, Sajnóg et al.⁹⁶ tested 10 extraction methods for the determination of Se NPs in Se enriched yeasts. A combination of enzymatic (n = 1, disease, protease, TRIS, SDS), mechanical (n = 6, ultrapure water with either sand, glass beads or metal beads) and chemical procedures (n = 3, NaOH, TMAH and SDS) were compared. Analysis was performed by spICP-MS in H₂ gas mode and monitoring ⁸⁰Se to determine the NP size and mass. The chemical procedures were not suitable as partial dissolution of the NPs was observed. The various mechanical methods generally were in good agreement with each other in terms of particle size distribution, which was a smooth Gaussian-like distribution. The 500 μm glass beads resulted in the highest number of particles extracted whereas the metal beads had the lowest particle number and mass, probably because only 2 or 4 beads were used, limiting the number of collisions possible. However, the enzymatic procedure extracted the largest number and mass of particles overall but had a broader size distribution profile with significant tailing at the larger diameter, indicating agglomeration. The authors concluded that the enzymatic process produced the highest recovery of Se NPs but at the expense of a wider size distribution whereas the glass beads method could be performed in a significantly shorter time frame (∼40 h vs. 1 h).

It is rare to feature alcoholic beverages in this section of the Update, however, two studies have employed ICP-MS for the detection of various elemental NPs in spirits and liquors. Edible gold is a permissible food additive within the EU (E175) though there is no acceptable daily intake established due to lack of toxicological assessments, whilst the US FDA has not approved edible gold. The potential formation of Au NPs in liquor containing gold flakes was investigated by Li et al.¹⁶⁶ with A4F-ICP-MS to understand if NPs could be present in liquid conditions. A commercial liquor (1 L) containing 43.5% alcohol and ∼13 mg Au flakes plus a pristine E175 product were acquired. A blank matrix was prepared from EtOH blended with ultrapure water at 43.5% and E175, which was compared to the commercial liquor. The analysis method was optimised with the use of Au NP standards at various sizes (5, 20, 40, 60, 80, 100 and 200 nm) in the 43.5% EtOH matrix, achieving sufficient separation for a peak deconvolution model to be developed to overcome the overlap in the fractograms of the different particle sizes. The commercial sample and synthetic sample were then compared, with and without prior application of sonication. It was found that NPs only formed with sonication treatment, with a mean particle size of 123.7 nm (range between 8.3 to 398 nm). The work demonstrated there was no immediate risk from NP exposure, but the developed method was capable of providing data relevant for toxicological studies. de Vega et al.¹⁶⁷ described the development of a data processing method to support untargeted NP analysis by spICP-TOF-MS through application to whisky, vodka, gin, and liqueurs. One of the challenges with ICP-TOF-MS analysis is the large amount of data that can be generated extremely quickly due to the quasi-simultaneous detection of the large mass range at the μs level. Therefore, the paper presented an algorithm to model background levels and discern real signal spikes for multiple isotopes with differing decision limits. The first step decided whether to use Gaussian or Poisson statistics based on the count rate observed, so that if the counts were >5, then Gaussian was implemented, otherwise, compound Poisson distribution with a log-normal approximation was applied. As the file size can easily reach several GB per measurement, the algorithm limited the evaluation to the first 500 000 data points per isotope from 45 amu to 210 amu with at least 25 data points above the decision limit per million. The output provided data meeting these criteria. Proof of concept was demonstrated with the untargeted analysis of various samples of whisky, vodka, gin, and liqueurs. The algorithm detected NP events for Ag, Au, Cu, Fe, Mn, Sn and Ti, which enabled the researchers to apply quantitative methods to characterise the NPs identified. The paper demonstrated the need for mathematical approaches to assist with large data sets. Furthermore, the authors freely provided the Python script for the algorithm.

5.2 Biological applications

Hendriks et al.¹⁰⁹ implemented scICP-TOF-MS for the detection of nanoplastics in cells. Previously published studies for microplastics have used C to detect single events but the poor ionisation of C limits the achievable size LOD, so nanoplastics cannot be detected. Furthermore, when considering cell experiments, plastic particles and single cells cannot be distinguished by C. Therefore, using metal doped materials combined with the use of ICP-TOF-MS circumvented these issues. Here, two human cell lines (A549 alveolar epithelial cells and THP-1 monocytes) were exposed to Pd-doped nanoplastics to assess their uptake and distribution. With the quasi-simultaneous detection of endogenous elements from the cell (Cu, Fe, Mg, P and Zn) and Pd-doped nanoplastics, it was possible to differentiate between cells, nanoplastics and nanoplastic incorporated cells. Three concentration levels of nanoplastics were used (0.5, 5 and 50 μg L⁻¹) which was reflected proportionally in the number concentrations of the cells with Pd signals. The majority of cells did not take up the nanoplastics after 24 h exposure (67% of A549 and 72% of THP-1 cells for 50 μg L⁻¹). The work demonstrated the benefits of scICP-TOF-MS for toxicological investigation of nanoplastics. Sanchez-Cachero et al.¹⁶⁸ performed a systematic evaluation of data processing methods for calculation of particle size and concentration by spICP-MS. Four algorithms were tested, namely 5-σ criterion, K-means, npQuant (proprietary software from Thermo Electron) and deconvolution. To generate test data, Pt NPs at 30 nm and 50 nm were used at different particle number concentrations, with and without the presence of ionic Pt. This enabled an assessment of the procedures to handle noise, linear range and the discrimination levels. It was found that the deconvolution method provided the largest working range (from 1.06 × 10⁴ particles mL⁻¹ to 3.02 × 10⁵ particles per mL) and lowest LODs for size and concentration (16.70 nm and 0.059 μg L⁻¹ respectively). Additionally, it was also able to discriminate between a mixture of 30 nm and 50 nm NPs at a range of particle number concentrations between 1.06 × 10⁴ and 1.06 × 10⁵ mL⁻¹ for each NP. Several transport efficiency calculation methods were also tested but no significant difference was identified (p < 0.05). The level of ionic Pt caused overestimation of the size for the 30 nm NP when above 500 ng L⁻¹ and above 1500 ng L⁻¹ for the 50 nm. However, the particle concentration accuracy was more sensitive to ionic Pt as poor recoveries were found when the ionic Pt was above 100 ng L⁻¹ for 30 nm and 750 ng L⁻¹ for 50 nm. The optimised analysis conditions were then tested by spiking Pt NPs in real matrices, namely hard drinking water and Dulbecco’s modified eagle medium with 10% foetal bovine serum with antibiotics, showing no significant impact from the matrix for size or number concentration.

5.3 Pharmaceuticals and consumer products

Garcia-Mesa et al.¹⁶⁹ described the use of HR-CS-GFAAS for size determination of ZnO NPs and ionic Zn concentration in pharmaceutical and cosmetic samples. This was possible due to differences in the atomisation profile of ionic species compared to NPs. Samples were simply dispersed in 0.55% Triton X-100, sonicated for 3 min and the slurry was deposited on the graphite platform. Calibration was achieved with Zn²⁺ standards and NP materials from 18 nm to 500 nm. For comparison, spICP-MS (dispersion in 1% Triton X-100, 70 min sonication) and TEM were used as confirmatory techniques. The matrices tested were eyeshadow, creams and lotions. The concentration of Zn as Zn²⁺ and ZnO NP was calculated using the upslope of the atomisation peak and the sum was compared to the total Zn obtained from MAD and HR-CS-GFAAS analysis. For the eyeshadow and creams, ZnO NPs were the main species present whereas Zn²⁺ was the larger component in the lotions, but the total Zn levels were the lowest. A paired t-test showed no significant difference between the summed species and total Zn. When analysed by spICP-MS, it was not possible to calculate the ionic content due to the high background level from the dissolved fraction. The NP size was determined by HR-CS-GFAAS, spICP-MS and TEM with statistical testing by ANOVA showing no significant difference. Overall, the study demonstrated the potential of HR-CS-GFAAS as a NP characterisation technique with straightforward sample preparation and the ability to differentiate between NPs and ionic species even at high ionic background levels. Justo-Vega and co-workers¹⁷⁰ published a sample preparation method for the extraction of Ag and TiO₂ NPs from tap water. Surfactant assisted DLLME was implemented for the extraction and preconcentration of the NPs. Following optimisation, 1 mL of a 49 + 1 mixture of 2.5% (w/v) Triton X-114 and 1,2-dichloroethane was added to 10 mL of tap water and vortexed for 30 s. The organic phase was recovered by centrifugation and dissolved in 1 mL of 1% (w/v) glycerol. The procedure led to an ERF of 10. The analysis was performed by spICP-MS. It was shown that the ionic fraction was effectively reduced during the preparation which is an advantage to improve the detection of smaller NPs. Spiked samples were prepared in two ways using fixed number concentrations of 40 nm and 60 nm Ag NPs and 100 nm TiO₂ NPs. First, the DLLME extracts were spiked to test the measurement recovery and secondly, the tap water was spiked to determine the full method recovery. The spiked extracts had particle number recoveries of between 83% and 114% demonstrating the accuracy of the spICP-MS measurements but for the full method, the recoveries were between 44% and 58%. Although the particle size was maintained, further experiments to improve the analytical recovery were unsuccessful. The authors concluded that the surfactant assisted DLLME method was valuable to preconcentrate samples needed for environmental studies of water, and that the ionic background can be reduced without impacting the particle size. However, a NP spiked blank or matrix sample should be performed to determine the quantitative recovery for particle number concentration.

6. Speciation and imaging studies

6.1 Speciation studies

The ASU Update, by Clough et al.⁶ provides a comprehensive review of recent advances in speciation and related applications. In the current Section, we specifically address advances in speciation techniques related to the analysis of clinical and biological materials, foods and beverages. Speciation in these areas, when based mainly on extraction procedures, is considered in Section 3.2 and Tables 1 and 2 of this Update. Besides work addressing As, Hg and Se speciation, over the period covered by this Update there were also reports of speciation of other elements, notably, Cu, I, Sb and Zn.

In the clinical area, a few papers reported advances in the speciation of As in biological fluids or tissues. Urine is a matrix of choice to investigate As exposure and studies on the presence and levels of As metabolites are important both to provide insights into As metabolism and to assess potential toxic effects. Zhou et al.¹⁷¹ proposed a method for the simultaneous determination by HPLC-ICP-MS of six As species (AB, AC, As^III, As^V, DMA^V and MMA^III) and, in addition, of four I species (I⁻, IO₃⁻, 3-iodo-tyrosine and 3,5-diiodo-tyrosine) in human urine. Separation was achieved on an anion exchange column, using a gradient elution with 0.5 mmol L⁻¹ NH₄CO₃, followed by 50 mmol L⁻¹ NH₄CO₃ 100 mmol L⁻¹ NH₄NO₃ and 4% MeOH. The method achieved LOQs ranging from 0.045 μg L⁻¹ to 2.26 μg L⁻¹ for As species and from 0.060 μg L⁻¹ to 0.180 μg L⁻¹ for I species. The average recoveries at spiked levels of 10.0, 20.0 and 50.0 μg L⁻¹ of each analyte were between 87.4% and 107.4% (As species) and varied from 88.6% to 113, 1% for I compounds. The RSD%, evaluated in the same set of experiments, ranged from 0.5% to 17.2% for As compounds and between 0.4% and 10.7% for I species. These figures agreed with the ratio of the sum of the species to the total elemental content, measured by ICP-MS, and varied from 77.4% to 121.2% for As and from 70.7% to 114.7% for I. Out of concern for the instability of organic As metabolites, such as DMA^III and MMA^III, another group of researchers¹⁷² considered the whole analytical process, from sampling to report, and identified the critical step being sample collection and transport to the laboratory. During these phases, As metabolites may undergo oxidation and therefore no longer represent the As species actually present in urine. To preserve the integrity of these analytes, they devised a urine self-sampling kit, that allowed for the removal of air (and oxygen) in contact with the sample, immediately after collection, by pumping argon into the urine sample with a syringe. On arrival at the laboratory, the sampling vial was loaded onto an automated unit for processing under anaerobic conditions. This involved dilution with 0.02 mol L⁻¹ CH₃COOH, filtration and introduction into an HPLC-ICP-MS instrument. The determination of seven As species (AC, As^III, As^V, DMA^III, DMA^V, MMA^III and MMA^V) was performed within 18 min using an ODS column (150 mm × 4.6 mm, 3 μm) at 50 °C and isocratic elution with 4.7 mmol L⁻¹ TBAH, 2 mmol L⁻¹ malonic acid and 4% MeOH (pH 5.85). The LODs for the seven As species ranged from 0.03 μg L⁻¹ (As^III and MMA^V) to 0.10 μg L⁻¹ (DMA^III), respectively, and the RSD% was <10%. Analysis of a CRM (GBW 09115 freeze-dried human urine) provided results in good agreement with the certified values (As^III, DMA^V and MMA^V). The stability of the As species in urine collected using the new device was assessed by analysing spiked samples stored for 6 h, 24 h and 48 h at 4 °C or −20 °C. The experiment yielded recoveries between 95% and 104% for AC, As^III and As^V, but lower ones for other species, although the stability of MMA^III and DMA^III was improved by storage in the dark and with anaerobic treatment. The group of Feldmann⁴¹ explored the effect of external contamination and cleaning procedures on trace element status (of which we reported in Section 3.1) as well as on As speciation in toenail clippings, a matrix increasingly used in epidemiological studies. Both synchrotron XFM and laterally resolved XANES were applied to investigate the in situ speciation of As in toenails. In washed toenails, only one As species, corresponding to As bound to sulfhydryl groups, was detected by XANES, whereas varying levels of As^V were also observed in non-washed samples. Microplastics (plastic particles with o.d. ≤ 5 mm) dispersed in the environment are a source of concern for public health and also as possible carriers of other toxins. To investigate the potential transfer into biological fluids of As species loaded onto microplastics ingested with food and beverages, Zhang et al.¹⁷³ linked a novel 3D-printed dynamic stomach model to a CE-ICP-MS system. The dynamic stomach model, controlled by specific software, simulated the various steps of human digestion of food. The extract resulting from this process of artificial digestion passed through a spiral channel, to remove any large particles, and underwent on-line filtering to obtain a sample suitable for analysis by CE-ICP-MS. Sample introduction was achieved using a PEEK cross-connector. One end of the silica capillary was inserted directly into the centrifuge tube containing the gastric extract and the other passed through the PEEK cross connector into the ICP-MS nebuliser. A connection with an air compressor allowed for the injection of air into the centrifuge tube to drive the sample into the capillary. The efficiency of the dynamic stomach model in comparison with a static one was validated using cooked white rice, whereas liquid food samples were used in the subsequent study of the release of As from microplastics loaded with different As species. Four As species (As^III, As^V, DMA and MMA) were measured with LODs of 0.5 μg L⁻¹, 0.9 μg L⁻¹, 0.6 μg L⁻¹ and 0.8 μg L⁻¹, respectively, and RSD% <8%. Recoveries of spiked amounts ranged from 92% to 109% and the results compared well (p > 0.13) with those obtained using a manual pre-treatment method. The authors found that the release rate of As from microplastics was higher for As^III and As^V than for DMA and MMA. In addition, co-digestion experiments revealed that the concentration of As species in the digested fluids was influenced by both the stage of digestion and the type of food (animal vs. plant origin) present.

The possible beneficial effects to human health of proper Se intake continue to attract scientific interest. A group of researchers¹⁷⁴ collaborated to achieve a systematic assessment of the effect of supplementation with different Se compounds on plasma biomarkers of Se status in cancer patients. The study involved a small group of 24 patients, divided into three groups, each receiving 400 μg per day of either Na₂SeO₃, SeMet or SeMeSeCys for 28 days. Plasma samples, collected before and after the supplementation period, were stored at −80 °C until analysis. The authors devised an extensive procedure for the comprehensive assessment of Se species, applying ICP-MS/MS and HPLC-ICP-MS/MS, with a CRC using O₂ and H₂ gases, and a mass shift of m/z +16 to reduce and overcome interferences on the selected Se isotopes (⁷⁷Se, ⁷⁸Se and ⁸⁰Se). The analytical protocol involved four aliquots of each sample, that underwent different analytical strategies. The total Se concentration was determined after 1 + 19 dilution of 0.30 g plasma with 0.5% (v/v) HNO₃. Another aliquot of 0.05 g was analysed without further treatment by HPLC-ICP-MS/MS. The chromatographic separation was achieved combining two affinity columns (blue Sepharose and heparin) for the retention of SeAlb and SelP, respectively, followed by gradient elution with CH₃COONH₄ buffers at different molar strengths at a flow rate of 0.5 mL min⁻¹ and quantification of Se mass fraction in each chromatographic peak. Selenoalbumin and SelP were eluted in separate fractions, whereas other Se species (GPx and low M_r species) remained in the void peak. High and low M_r fractions (cut-off: 10 kDa) were separated by ultrafiltration of another 0.30 g aliquot. After reconstitution of both fractions to the original volume with 50 mmol L⁻¹ TRIS–HCl, at pH 7.4, the high M_r fraction was digested with a 2 + 1 (v/v) mixture of HNO₃ and H₂O₂, then diluted 1 + 19 with 15% (v/v) HNO₃ and presented to ICP-MS/MS to determine the total Se content in the plasma protein fraction (GPx, SeAlb and SelP). The low M_r fraction was analysed without further treatment by ion-pairing RP HPLC-ICP-MS, to determine the concentrations of SeMet, SeMeSeCys, Se-sugar-1 and other low M_r Se species. To assess the non-specific incorporation of SeMet into SelP after SeMet administration, the SelP fraction, isolated from 0.20 g of plasma submitted to double affinity HPLC, was preconcentrated by ultrafiltration (30 kDa, 5000 rpm, 30 min) then enzymatically digested. The content of SeMet in the digest was determined by ion pairing RP HPLC-ICP-MS. This complex protocol was supported by assessment of the quality of the analytical results, based on analysis of CRMs (BCR-637 human serum, for total Se, and NIST SRM 1950 metabolites in human plasma, for which indicative values are provided for GPx, Se, SeAlb and SelP), as well as spiking experiments with selenosugar-1, SeMet and SeMeSeCys. From the results of this study, the authors concluded that Se supplementation, under any of the proposed species, had little influence on the low M_r pool, but for a higher level of selenosugar-1, whereas the Se concentration in the higher M_r pool, consisting of GPx, SeAlb and SelP, was significantly increased, depending on the Se chemical form, dose, duration of treatment, baseline Se status and health conditions.

The determination of As species in food often involves a preparation step, but it has been reported that the extraction process either with HNO₃ or TMAH may alter the species originally present. Matsumoto et al.¹⁷⁵ studied the thioarsenical species generated from MMA during acid extraction of rice by combining LC-TOF-MS and HPLC-ICP-MS. Rice and corn starch, spiked with MMA, were incubated with 0.15 mol L⁻¹ HNO₃ at 100 °C for 2 h. These extracts, alongside those of unspiked samples, were analysed using LC-TOF-MS with a C₁₈ column (2.0 mm × 150 mm, 5 μm) and 0.3% CH₃COOH (pH 3.0) as the mobile phase, at a flow rate of 0.2 mL min⁻¹. The new peak appearing at 3.1 min in the extract of the spiked samples was confirmed as an As species, using HPLC-ICP-MS under the same conditions used for LC-TOF-MS. The LC-TOF-MS mass spectrum was consistent with the structure CH₅AsO₂S for this product. To confirm the hypothesis, CH₅AsO₂S was synthesized by bubbling H₂S into a MMA solution and then compared with the product generated during the acid extraction of rice or corn starch. The comparison was achieved using LC-TOF-MS as well as HPLC-ICP-MS with He as the collision gas, using a C₁₈ column (4.6 mm × 250 mm, 5 μm) and a mobile phase consisting of 10 mmol L⁻¹ sodium 1-butanesulfonate, 4 mmol L⁻¹ malonic acid, 4 mmol L⁻¹ TMAH and 0.05% MeOH (pH 3.0) at a flow rate of 0.75 mL min⁻¹. The study confirmed the conversion of MMA into thioarsenical species, the major one being MMMTA, by reaction with sulfur present in the sample. Experiments with DMA also suggested the conversion into another compound, although at a low generation rate, that was not further investigated. Because new As species can be artefacts generated during the extraction process, the authors recommended verification of the origin of any unknown As compounds found during the development of an analytical method.

Arsenic speciation in food maintained a prominent position with developments related to matrices such as seaweed and insects. The use of seaweeds, as a more sustainable food and feed source and a beneficial food supplement, has been increasing in recent years. Although they are known to contain As mostly as arsenosugars, the toxic iAs species may also be present. A group of researchers¹⁷⁶ considered the impact of high concentrations of arsenosugars on the determination of iAs species in seaweeds, performed using anion exchange HPLC-ICP-MS. They proposed a simplified extraction procedure, allowing for the measurement of the sum of As^III and As^V, and investigated both the extraction parameters and the HPLC-ICP-MS conditions using a custom fractional factorial design of the experiments and several CRMs. Brown (Phaeophyta), red (Rhodophyta) and green (Chlorophyta) seaweed samples were stored at 2 °C after collection, then cleaned, freeze-dried and made into a powder. The authors treated an aliquot (100 mg) of the samples by ultrawave MAD for 40 min at 90 °C with 10 mL 1% (v/v) HNO₃ and 3% (v/v) H₂O₂. After a two step centrifugation (4000 rpm, 15 000 rpm), 40 μL of the extracts were injected onto an anion exchange column (250 × 4.1 mm, 10 μm). Arsenic species were eluted within 7 min with a mobile phase of 60 mmol L⁻¹ (NH₄)₂CO₃ containing 3% MeOH, pH 9.0, at a flow rate of 1 mL min⁻¹. Detection was achieved using ICP-MS with an octopole CRC, fed with He, and Ge, continuously added post-column during the analysis, as the IS. The extraction procedure led to the oxidation of As^III to As^V, without conversion of organic As species to iAs. The sum of iAs species could be measured without interferences from arsenosugars, that were unretained. Analysis of a CRM (NMIJ Hijiki 7405-b) yielded a recovery of 99 ± 9% of the certified value for As^V (24.4 ± 0.7 mg kg⁻¹). Recoveries of spiked amounts (1, 5, and 10 μg L⁻¹) to a seaweed known for high levels of arsenosugars were 112%, 92%, and 101%, respectively, and precision (RSD%) ranged from 7% to 15%. The LOD and LOQ calculated as 3.3-fold and 10-fold the SD achieved by measuring blanks spiked with 0.1 μg L⁻¹ As^V, multiplied by an average dilution factor, were 0.006 mg kg⁻¹ and 0.018 mg kg⁻¹, respectively. Although the reported application was limited to seaweeds, the authors noted the minimal sample preparation and high throughput of the method, that may be applied also to other food and feed. Arsenosugars were the main subject of another paper,¹⁷⁷ where the authors addressed the lack of specific calibration standards for the reliable identification and quantification of arsenosugars. They investigated the conditions for the application of centrifugal partition chromatography to the isolation and purification of arsenosugars from algae extracts, monitoring the As species in the respective fractions by IC-ICP-MS. Out of an extensive investigation, a biphasic solvent system consisting of 1-butanol, EtOH, saturated (NH₄)₂SO₄ and water at a volume ratio of 0.2 : 1 : 1 : 1 provided adequate separation of the analytes and allowed for the isolation of the three main arsenosugars present in edible algae (As-phosphate sugar, As-sulfate sugar and As-sulfonate sugar) with a purity of 98.7%, 96.1% and 90.4%, respectively. Edible insects are consumed by several population groups worldwide. More recently, ingredients derived from selected insect species have been allowed for food preparations in the EU and other countries. Accumulation of As in insects depends on the species, which may also influence the distribution into different As forms. Matsumoto and Matsumoto¹⁷⁸ validated an analytical procedure based on HPLC-ICP-MS and determined the concentration of As species in nine types of dried, whole, edible insects (Asian forest scorpions – two individuals, diving beetle, giant water bug, grasshopper, June beetle, mole cricket, male rhino beetle, female rhino beetle, sago worms, and silkworm pupae). Samples were homogenised using a food processor. An aliquot of 0.1 g was mixed with 2 mL 0.3 mol L⁻¹ HNO₃ and heated on a dry block bath at 100 °C for 2 h. After addition of 3 mL water and centrifugation (2100g, 10 min), the supernatant was removed and the precipitate washed twice with 5 mL water and subsequent centrifugation (2100g, 10 min), collecting two more supernatant fractions. The resulting liquid was adjusted to pH 3 then diluted to a final volume of 20 mL with water and filtered. The chromatographic separation of AB, As^III, As^V, DMA and MMA was carried out using a C₁₈ column (4.6 mm × 250 mm, 5 μm) and a mobile phase consisting of 1.458 g L⁻¹ 25% TMAH, 1.602 g L⁻¹ sodium 1-butane sulfonate, 0.416 g L⁻¹ malonic acid and 0.5% (v/v) MeOH, at pH 3.0 and a flow rate of 0.75 mL min⁻¹. The ICP-MS was operated in KED mode, using He as the collision gas. Total As concentrations were determined using ICP-MS, with Te as IS, on samples (between 0.1 g and 0.25 g) digested with 5 mL conc. HNO₃ in a ultrawave microwave oven and diluted to a final volume of 50 mL containing 1 mL CH₃COOH to compensate for the effect of high carbon content in the samples. The LODs (3 × SD) and LOQs (10 × SD) for each As species were evaluated from the analysis of samples of grasshoppers spiked at 0.1 mg kg⁻¹ and ranged from 0.007 mg kg⁻¹ (DMA) to 0.012 mg kg⁻¹ (As^V) and from 0.021 mg kg⁻¹ (DMA) to 0.038 mg kg⁻¹ (As^V), respectively. Overall, precision (RSD%) was <4.5%. Recovery tests for all As species were carried out at two concentration levels (0.1 mg kg⁻¹ and 1 mg kg⁻¹), spiked to grasshopper samples before treatment and gave a mean value of 100 ± 2% for all species but MMA (90.7% and 74.7%). This was attributed to loss of MMA during the extraction, due to conversion to thioarsenical compounds in the presence of sulphur, as also indicated by the appearance of additional chromatographic peaks. Therefore MMA was excluded from further analysis and the results for As^III and As^V were combined and expressed as iAs, also due to possible interconversion between these species during the extraction process. The extraction efficiency of the sample treatment procedure was >85%, as assessed by analysis of three CRMs certified for the total As content (NRCC DORM-4 fish protein, NMIJ CRM 7405-a Hijiki seaweed, NMIJ CRM 7533-a brown rice flour) using both extraction and MAD sample treatments. Two of the same CRMs also provided certified values for As species (AB, iAs or DMA), which were in good agreement with the values measured by the proposed procedure (n = 3, recoveries between 96% and 100%). In the 11 insects analysed iAs was always measurable, whereas AB was detected in only two of them and DMA in another two. The level of As were variable among the insects tested, that was attributed to the type of feed consumed. The lowest As levels (<0.1 mg kg⁻¹) were found in grasshoppers and mole crickets, now approved as novel foods by the EU.

Antimony is a potentially toxic element present in the environment. In both the EU and the US limits have been set for the maximum level of total Sb in drinking water (at 10 and 6 μg L⁻¹, respectively). However, the two Sb species, Sb^III and Sb^V, show different properties, behaviours and toxicity, calling for further research for their speciation. Whereas this is usually achieved by hyphenation of HPLC with AF or ICP-MS, a group of researchers¹⁷⁹ proposed a different approach (frontal chromatography) for the separation of Sb^III and Sb^V in mineral water samples. This was achieved on a short home-made column (2.5 mm × 60 mm), packed with a strong cationic exchange resin (Dowex® 50WX8 hydrogen form 200–400 mesh), connected to the ICP-MS nebuliser. The ICP-MS instrument was operated in KED mode, using He as the collision gas. Samples, acidified with ultrapure HNO₃ at a final concentration of 0.5 mol L⁻¹ and spiked with Ge as the IS, were fed to the column using a peristaltic pump. With 0.5 mol L⁻¹ ultrapure HNO₃ as the eluent, at a flow rate of 1.7 mL min⁻¹, iSb species were completely separated in less than 3 min, with a total analysis time, including the washout step, of 6 min. Limits of detection (3 × SD) were estimated for both species using standard solutions of Sb^III and Sb^V at 25 ng kg⁻¹. The concentrations determined for Sb^III and Sb^V were (n = 10, mean ± 3 SD) 24.3 ± 0.9 ng kg⁻¹ and 25.7 ± 0.4 ng kg⁻¹, respectively. The values of 0.9 ng kg⁻¹ (Sb^III) and 0.4 ng kg⁻¹ (Sb^V) were lower than those previously reported for other procedures based on HPLC-ICP-MS (ranging from 0.009 μg L⁻¹ to 0.4 μg L⁻¹ (Sb^III) and from 0.014 μg L⁻¹ to 0.2 μg L⁻¹ (Sb^V), respectively). Recovery was evaluated using four drinking mineral waters, with different characteristics (oligo-mineral, chlorine-rich, bicarbonate-rich and sulphate-rich, respectively) spiked with 0.1 μg kg⁻¹ Sb^III or 1 μg kg⁻¹ Sb^V, since no CRM for the content of Sb species was available. The observed values were very close to 100% for both species in all the samples and, in addition, the sum of the two Sb species agreed well with the total Sb content determined by ICP-MS. The authors noted the possibility of systematic errors when the ratios between the concentrations of the two species were >100, a situation rarely encountered in practice. The method, although currently applied only on a limited number of samples, may offer a suitable alternative to HPLC-ICP-MS procedures and favourable LODs.

The speciation of Hg in food was addressed in two papers. Acknowledging the potential risks associated with the presence of MeHg in certain foods, Anil et al.¹⁸⁰ undertook a comprehensive evaluation of MAE procedures for the determination of MeHg in fish, rice and soil, using HPLC-ICP-MS. The authors compared the efficacy of five previously published MAE procedures for MeHg, based on: (a) 25% (w/w) KOH in MeOH; (b) 5 mol L⁻¹ HCl and 0.25 mol L⁻¹ NaCl; (c) 10 g L⁻¹ EDTA, 0.2% (v/v) 2-mercaptoethanol and 2% (v/v) MeOH; (d) 25% (w/v) tetramethyl ammonium chloride; (e) 0.25% L-cysteine and 0.25% 2-mercaptoethanol. Procedures (a) and (b) were deemed superior, as they achieved extraction efficiencies >75% across all the three matrices. The authors then proposed as optimal MAE conditions: 0.5 g of homogenised samples of rice, fish and soil mixed with 6 mL of 25% (w/w) KOH in MeOH, irradiated to a temperature of 70 °C for 8 min. Chromatographic separation was achieved on a C₁₈ column (4.6 × 150 mm, 5 μm), using isocratic elution with a mobile phase of 0.06 mol L⁻¹ CH₃COONH₄, 0.1% 2-mercaptoethanol and 5% MeOH, pH 6.8, at a flow rate of 1 mL min⁻¹. An Ultra High Matrix Introduction system was used to increase matrix tolerance up to 25% total dissolved solids. The ICP-MS operated in time resolved analysis mode. The LOD for MeHg was estimated as 12 μg kg⁻¹ across the three matrices and the LOQ between 37 μg kg⁻¹ (rice) and 41 μg kg⁻¹ (soil). Intermediate precision (RSD%) was <7%. Recovery was evaluated on sets of three samples for each matrix, spiked with MeHg at concentrations of 40 μg kg⁻¹, 200 μg kg⁻¹ and 400 μg kg⁻¹, corresponding to 1, 5, and 10-fold the LOQ, respectively. The observed values, ranging from 93.9% to 98.1% (rice), from 81.5% to 89.3% (fish) and from 75.3% to 77.2% (soil), were deemed acceptable, given the benefit of the application across different matrices, although better values were noted for the analysis of soil samples with procedure (b). However, these conclusions are based on a limited number of samples, that may hardly represent the range of variability even within the same matrix type and are not supported by the analysis of CRMs. Another group of researchers¹⁸¹ exploited the advantages of developments in analytical instrumentation, namely UHPLC-ICP-MS, to improve an existing method for Hg speciation based on HPLC-ICP-MS. They used a shorter C₁₈ column (2.1 mm × 50 mm; 1.8 μm) and an injection volume of 5.0 μL. Elution was achieved with a solution of 0.05% 2-mercaptoethanol, 0.40% L-cysteine, 0.06 mol L⁻¹ CH₃COONH₄ and 5% MeOH, at a flow rate of 500 μL min⁻¹. Under these conditions, the analysis time was reduced from 15 min to 1.5 min. The method was tested on lyophilised samples (muscle and liver) of Bonito fish (Katsuwonus pelamis), 100 mg of which were extracted with 10 mL of 0.10% HCl, 0.05% L-cysteine and 0.10% 2-mercaptoethanol, by sonication for 30 min in an ultrasonic bath, followed by centrifugation (3500 rpm, 5 min). The UHPLC separation of Hg species was achieved with retention times of 0.4 min (Hg²⁺) and 0.6 min (MeHg⁺), respectively. The sum of the species concentrations yielded values not significantly different from the concentration of total Hg, determined in the same samples by ICP-MS after acid digestion, with a recovery ranging from 92% to 105%. The authors highlighted the advantages of the shorter analysis time, leading to a 90% reduction in mobile phase consumption, with no worsening of performances but also noted that these initial results need to be supported by a more extensive validation in the future. An additional paper¹⁸² addressed the more complex question of the protective effect of Se to help reduce Hg toxicity and whether the ratios between Hg and Se species may have a role in the assessment of the safety of fish for human consumption. The authors undertook speciation studies of fish and fish products, based on both HPLC-ICP-MS and HPLC-ESI-MS/MS. Samples included fish fillets of tuna (Thunnus thynnus), swordfish (Xiphias gladius), farmed salmon (Salmo salar) and wild salmon (Oncorhynchus gorbuscha) as well as fish roe from lumpfish (Cyclopterus lumpus), trout (Oncorhynchus mykiss) and wild salmon (Oncorhynchus gorbuscha). In addition, crab sticks, surimi-derived elvers, cod noodles and salmon noodles were also selected as examples of highly processed food. A TDA-AAS (with a direct mercury analyser) and an ICP-MS instrument were applied to the measurement of the concentrations of total Hg and total Se, respectively, using validated procedures with LOQs of 0.033 μg kg⁻¹ (Hg) and 0.006 mg kg⁻¹ (Se). The accuracy of the results were confirmed by the analysis of CRMs (Hg: BCR-710 oyster tissue, Hg and Se: ERM®-CE 278 k mussel tissue). For Se speciation, enzymatic extraction was carried out at 37 °C for 24 h, using 0.5 g of sample, 5 mL of 30 mmol⁻¹ TRIS, at pH 7.5, and 20 mg of Protease type XIV. Se species (Se^IV, Se^VI, SeCys₂, SeMeSeCys, SeMet) in the extracts were separated and detected using HPLC-ICP-MS with an anion exchange column (250 × 4.1 mm, 10 μm). For Hg species (Hg²⁺ and MeHg⁺), after incubation with 1% (w/v) L-cysteine·HCl·H₂O at 60 °C for 2 h, the extracts were centrifuged, diluted as appropriate and analysed using RP HPLC-ICP-MS with a C₁₈ (150 × 3 mm, 5 μm) column. Both recovery experiments of spikes (30 μg L⁻¹) for the Hg species and analysis of a CRM (BCR-710 oyster tissue), certified for MeHg, confirmed the validity of the procedure. Attempts to measure EtHg⁺ did not succeed, a finding that was attributed to instability of EtHg⁺ during the extraction procedure and confirmed by recovery experiments. The LOQs (10 × SD) were 0.02 mg kg⁻¹ (Hg²⁺) and 0.002 mg kg⁻¹ (MeHg⁺). The authors found that MeHg⁺, SeMeSeCys and SeMet were the main species present in all analysed samples, although with large variations of both the total and species specific contents. From these data, the Se : Hg molar ratio and the selenium health benefit value were calculated, where the last one is a measure of the amount of Se still available after interaction with Hg. In this set of samples, the molar ratios ranged from 5.44 to 4280, indicating that Se molar concentration always exceeded that of Hg. Therefore the selenium health benefit values were also positive in all cases.

Iodine plays an essential role in human health. Foods of marine origin, are particularly rich in I₂ and are considered a good dietary source of this element. To better understand the chemistry of I₂ in these types of foods, and the potential advantages or disadvantages for human health, Sloth and co-workers¹⁸³ engaged in the development and optimisation of an analytical procedure for I speciation in seaweeds, based on HPLC-ICP-MS. Iodine was extracted from samples of brown (Saccharina latissima, Fucus serratus, Chorda filum, Laminaria digitata), red (Furcellaria lumbricalis, Porphyra spp, Dumontia contorta, Chondrus crispus, Palmaria palmata, Delesseria sanguinea) and green (Ulva spp., Ulva lactuca, Ulva intestinalis) seaweeds with a two-step procedure. A mixture of 0.3 g of dry sample and 40 mg of pancreatin in 5 mL water was incubated for 15 h in a shaking water bath (37 °C, 200 rpm) then, after the addition of 100 mL of 25% TMAH, heated in a oven at 90 °C for 3 h, cooled and diluted to 50 mL with ultrapure water. Chromatographic separation of the I species (I⁻), monoiodotyrosine (MIT) and diiodotyrosine (DIT) was obtained using IC-ICP-MS, at 30 °C, with a 5 μL sample, an anion hydroxide column (2 mm × 250 mm) and isocratic elution with 400 mmol L⁻¹ NH₄HCO₃ at a flow rate of 0.25 mL min⁻¹. The LODs (3 × SD) and LOQs (10 × SD), based on the analysis of blanks (n = 10) and expressed as ng L⁻¹ of I, were: 0.03 and 0.10 (I⁻); 0.04 and 0.12 (DIT); 0.01 and 0.03 (MIT). Recoveries of spiked amounts of the three analytes ranged between 84% and 116%, but the comparison of the sum of species with total I concentrations gave variable recoveries (brown seaweeds: between 66% and 100%; green seaweeds: between 41% and 92%; red seaweeds: between 12% and 45%), which was attributed to differences in the binding of I within the matrix. The method was applied to 15 seaweed samples. Both the total I content and the chromatographic profiles showed large differences between seaweed types. Iodide was the predominant species in all of them and DIT and MIT were also found in most samples. The speciation profile also showed six other I-containing species, not identified at present. This work provides a valuable assessment of the extraction procedures for I speciation and highlights the complexity of I chemistry in seaweeds, thus promoting further research in this area. Another group of researchers¹⁸⁴ looked at the possible release of a toxic I compound (monoiodoacetic acid) into drinking water, as the result of reactions of I⁻ with organic matter during water disinfection processes. The authors proposed the speciation of I forms in drinking water using CE-ICP-MS with a spray-efficient nebuliser interface. They investigated the CE parameters, considering the effects of chemicals, concentration level and flow rate on the separation of I⁻, IO₃⁻ and monoiodoacetic acid, and identified 1.5% HNO₃ at a flow rate of 0.5 mL min⁻¹ as the best choice as sheath fluid. The measurement of the three I species was achieved in about 8 min. Under these conditions, the LODs as μg L⁻¹ were: 0.16 (I⁻), 0.40 (IO₃⁻) and 1.22 (monoiodoacetic acid). The authors noted that, although higher than those reported for HPLC methods, these LODs compared well with those expected for CE-ICP-MS procedures. Recovery of spiked amounts (10 μg L⁻¹) of the analytes to three real sample of water (drinking, pond and seawater) yielded recoveries ranging from 85.5% to 96.8%. However, only seawater contained detectable levels of I species.

Given the growing market for vegetarian and vegan products, of which plant-based drinks are claimed to be the most popular, there is interest in understanding the total content of elements and of their species that may influence their bioavailability. Sowik et al.¹⁸⁵ exploited the capabilities of complementary techniques (ICP-MS/MS and ESI-MS/MS) hyphenated with chromatographic separation methods (SEC and HILIC) for the speciation of Cu and Zn in rice- and millet-based drinks. Size exclusion chromatographic separations were achieved by isocratic elution, using a 10 mmol L⁻¹ CH₃COONH₄ at pH 7.4 as the mobile phase and ICP-MS/MS detection. A mixture of thyroglobulin (670 kDa), γ-globulin (158 kDa), ovalbumin (44 kDa), myoglobin (17 kDa) and vitamin B12 (1.35 kDa) were used for size calibration of the column. Copper was distributed in four fractions, two of high and two of low M_r (<44 kDa), in rice-based drinks, but only in one high and one low M_r fractions in millet-based drinks. In comparison, Zn was eluted in only one fraction of low M_r in both types of drink. The low M_r fractions were further examined by HILIC (2 × 150 mm, 3 μm), using gradient elution with 10 mmol L⁻¹ HCOONH₄ pH 5 and ACN, connected to either ICP-MS/MS or ESI-MS/MS for species identification. This rather elaborated strategy allowed for the identification of the ligands forming complexes with Cu and Zn in the vegetable-based drinks, which consisted of amino acids and organic acids.

Musielak and co-workers¹⁰² described the use of XRF spectrometry for the speciation of Se in water and beverage samples. Highly selective extraction of Se^IV was achieved using graphene particles impregnated with thiosemicarbazide which was mixed directly with the liquid samples, namely mineral water, beer, fruit juice and wine. The suspension underwent sonication, pH adjustment and filtration to separate the graphene particles which were then mixed with a solution of 1 μg mL⁻¹ Y as the IS. The final suspension (10 μL) was deposited on the quartz reflector and dried before analysis by TXRF spectrometry. The sorbent was shown to be specific for Se^IV, as spiking experiments demonstrated Se^VI was not extracted from the suspension. Validation was performed with spiked samples of beer, apple juice and wine, achieving recoveries between 97.4% and 110% and RSDs (n = 7) of 3.8%, 5.7% and 6.2%, respectively. An impressive method LOD of 1.7 pg mL⁻¹ was calculated which was significantly lower than others reported in the literature. Furthermore, three water CRMs (NIST SRM 1640a natural water, Sigma Aldrich QC3163 seawater and ERM® CA713 wastewater) were analysed using the procedure and good recoveries were found (99.0–104%) by assuming that the total Se was Se^IV. Solid CRMs were also assessed as suspensions but the particulates interfered with the results so these were first prepared by MAD prior to adding the thiosemicarbazide–graphene substrate. Again, assuming the total represents Se^IV, the recoveries ranged between 85.1% to 100% for various CRMs of animal and fish tissue, milk powder, rice and vegetables across three orders of magnitude (n = 11). To assess the potential for Se speciation, two approaches were tested using samples spiked with Se^IV and Se^VI. The first was simply the difference between the total Se (determined by direct TXRF spectrometry) and Se^IV which worked well, achieving recoveries between 95% and 104.4% for Se^IV and from 92% to 117% for Se^VI. This was verified by the direct TXRF analysis of the residual solution after filtration (i.e. the Se^VI fraction), demonstrating the approach was suitable for high level samples where direct TXRF can be used, but for ultratrace Se in liquid samples, the second approach was used. After filtration, the remaining filtrate was boiled with 4 mol L⁻¹ HCl for 30 min to reduce Se^VI to Se^IV, then fresh graphene sorbent was applied to isolate this fraction. The recoveries of Se^IV and Se^VI were between 96% and 108% and between 93% and 110%, respectively. The study clearly demonstrated the potential of TXRF spectrometry as a non-chromatographic speciation technique with sufficient detection capability relevant for waters and beverages.

6.2 Imaging with MS and X-rays

Review articles are discussed within Section 1 of this Update, but it is worth highlighting relevant reviews describing the use of spectrometric imaging techniques for research applications in the clinical and food science areas. Mervič et al.¹⁸⁶ provided a comprehensive overview of the development of accurate quantitative bio-imaging by LA-ICP-MS. Calibration strategies and internal standardization approaches were evaluated for multiple tissue types (e.g. tissue sections, bones, teeth, hair) whilst considering their advantages and disadvantages. The remaining challenges, such as calibrants for NP quantification in tissues, were also examined. Gorman et al.¹⁸⁷ covered the practical issues of multimodal MS imaging for biological tissue samples and cells via LA-ICP-MS, DESI, MALDI and SIMS. This covered sample preparation techniques, instrumental methods and key examples for trace metals (e.g. Cu, Fe, Zn), metal-based drugs and biomolecules. It was concluded that the intrinsic link between metals, important bioprocesses and proteins requires the combination of multiple technologies for full understanding of these interactions. In a review by Graziotto and co-workers,¹⁸⁸ the combination of X-ray and optical fluorescence microscopy was discussed for the imaging of biological samples at the cellular level. Key elements such as Cu, Mn and Zn were examined as biologically relevant examples. The combination of the two techniques was regarded as key, but this required careful consideration of the sample preparation process for the fixing substrate to ensure dual compatibility e.g. silicon nitride, polycarbonate films, formvar coated TEM grids, silicon carbide and polypropylene. Examples covering native metals (Cu, Mn and Zn) in health and disease, metal-responsive fluorescent probes, NP-tagging, luminescent metal complexes and heavy metal tagged fluorophores were presented. As with the previous articles, implementing multiple techniques provided complementary information leading to enhanced understanding of complex bioprocesses. The topic of imaging of food products was reviewed by two publications. Shen and co-workers¹⁸⁹ evaluated LA-ICP-MS, DESI, MALDI and SIMS to determine the composition, contamination, authenticity and quality control of foods. The authors noted the challenges across the multiple platforms due to different sample preparation techniques and limited availability of multianalyte RMs covering elements, nutritional components, metabolites and residues. Zhou et al.¹⁹⁰ focussed on laser-based imaging techniques, namely LA-ICP-MS, LIBS and Raman spectroscopy. Similarly, the authors reviewed the analysis of metabolites, nutritional elements and toxic metals in food crops to determine compositional changes, adulteration and product quality. Interestingly, they also concluded that multiple methods are required for decisive data and that a lack of standardization is still a key issue to overcome.

Fundamental characteristics of the ablation plume from soft tissues was investigated by Van Helden et al.¹⁹¹ using LA-ICP-MS. The researchers demonstrated that the laser energy density can influence whether the elements of interest were transported in the gas or particulate phase, resulting in different transport efficiencies of the ablation plume. The paper systematically explored this with two laser systems at 193 nm and 213 nm wavelengths in gelatine standards spiked with 25 elements across the mass range (²³Na to ²³⁸U). Using single pulse response profiles, bimodal peaks were found for certain elements, such as As, C, Cd, I, Se, Te and Zn. By installing a filter to remove particulates and a cryotrap to remove gaseous components, the phase could be determined for the elements at various laser parameters. It was established that the effect was element specific by comparing the signal intensity ratio in the gas phase against the total signal; for example, it was 0% for Na, 43% for Zn and up to 95% for C with a 193 nm system. The two phases have different transport efficiencies leading to the bimodal peaks. The laser wavelength also impacted the phase distribution as higher gas phase formation was found with the 213 nm system compared to the 193 nm, even those elements in the particulate phase with the 193 nm ablation converted to gaseous form. The impact of this effect could lead to poor spatially resolved signals (i.e. blurring of the bioimage) and inaccurate quantification without closely matrix matched calibrants. Mello et al.¹⁹² reported a simple approach to boost the sensitivity of LA-ICP-MS analysis by the addition of N₂ to the ICP carrier gas post-ablation. By comparing the S/N at various flow rates of N₂, the sensitivity changes were determined for 38 elements covering the full mass range (²⁷Al to ²³⁸U) in gelatine calibration standards at five concentration levels. For the majority of elements, S/N improvements ranged from 1.2-fold to 7.8-fold increases. Additionally, it was found that between Al and Te masses, higher N₂ flow rates (12–20 mL min⁻¹) were required for maximum enhancement compared to the lanthanides and higher mass elements which were best at 6 mL min⁻¹. Furthermore, certain elements showed either no improvement or worse S/N due to the presence of N-based polyatomic interference such as Sc (¹⁴N₂¹⁶O¹H⁺), V (³⁶Ar¹⁵N⁺, ³⁶Ar¹⁴N¹H⁺) and Mn (⁴⁰Ar¹⁴N¹H⁺, ⁴⁰Ar¹⁵N⁺). The optimum flow rate for the majority of elements was then tested by imaging endogenous elements in mouse brain and antibody-conjugated elements in muscle tissue, comparing with and without the N₂ addition. Overall, significant improvements in the image resolution were found, demonstrating the power of using an additional gas post-ablation.

The use of LA-ICP-MS analysis for the investigation of various diseases was the focus of a number of papers within this Update period. Tisza et al.¹⁹³ determined the quantitative distribution of Pt in tissue by LA-ICP-MS and total Pt in serum by solution ICP-MS from patients diagnosed with pleural mesothelioma, an aggressive cancer with poor survival outcomes, and treated with Pt-based chemotherapy. Lung tissue samples were collected from 25 patients during surgery, which were also assessed by traditional haematoxylin and eosin staining and immunofluorescence labelling for collagen. The results revealed Pt levels were higher in non-malignant, collagen-rich fibrotic regions compared to tumorous areas, suggesting the cisplatin cannot adequately penetrate the tumour tissue which may link with limited effectiveness of these drugs for pleural mesothelioma. Whilst serum Pt and the average Pt level in the full tissue sections were positively correlated, the authors suggested further investigation was warranted to understand the reason for poor drug penetration, so as to improve survival rates. In a paper by de Vega et al.,¹⁹⁴ the impact of bacterial infection on the concentrations of trace elements and transport proteins was investigated. Lung tissue was harvested from mice infected with Streptococcus pneumoniae fed with either Zn enriched or depleted diets, alongside a control group. The tissues were quantitatively analysed by LA-ICP-MS for Cu, Fe and Zn as key immune response elements using doped gelatine calibrants. Additionally, the samples were treated by immunohistochemistry methods to visualise the metal transport proteins ZIP8 and ZIP14, without affecting the elemental detection. The study demonstrated the potential of multiplex methods for disease research. Niehaus et al.¹⁹⁵ combined LA-ICP-TOF-MS, μXRF spectroscopy, solution-based ICP-MS/MS and MALDI-MS to analyse human brain tissue from two multiple sclerosis patients. The lesions (as identified by MRI) and unaffected white matter were initially analysed by quantitative solution ICP-MS/MS after acid digestion, finding significant differences in the levels of Ag, Bi, Gd, K, Li, Mg, Mn, P, Rb and V at the 95% CI between the tissue types. The use of MALDI-MS enabled selective detection of lipids in the lesions which are important for the control of inflammation and signalling pathways, which is known to change in multiple sclerosis patients. Subsequently, the tissues were analysed using a benchtop μXRF spectrometer as a screening approach at a resolution of 29 μm to identify lesion zones. Only P and S had sufficient signal intensity to generate bioimages, however this was sufficient for P which correlated with the phospholipids. Finally, LA-ICP-TOF-MS at a resolution of 35 μm was implemented, with quantification via doped gelatine standards. Images for Cu, Fe and Zn were presented with clear differentiation between lesions and the surrounding tissue. Furthermore, the use of ICP-TOF-MS enabled quasi-simultaneous measurement of the mass range, therefore, it was also observed that Cr, K, Na, Sn and Te were present at higher levels in the lesions compared to the surrounding tissue, whereas Hg, Mo, Se and U were at lower concentrations. Overall, the combination of molecular and elemental MS analysis provided complementary and powerful data to further understanding of the disease mechanisms.

The use of LIBS for biological imaging has shown development within this Update period, with a number of publications demonstrating its potential. Imaging of Ca and Mg element distributions by LIBS in various subtypes of malignant melanoma tissue from 17 patients were investigated by Koprivova and co-workers.¹⁹⁶ Very high relative intensities of Mg spectral lines were observed in tumour tissue with respect to non-cancerous tissue and the distribution maps coincided with the marked tumour areas on conventional histology examination slides from adjacent sections. Lower signals for Ca were generally observed in the tumour zones compared with healthy tissue, but the discrimination was less distinct. The results also indicated differing elemental concentrations in the various subtypes of the malignant melanoma tissues although this would require confirmation in a larger sample set. Gardette et al.¹⁹⁷ described the quantification of Ti in lung tissue from rats exposed to Ti NPs using a novel data processing strategy. The approach was developed using rat models exposed to air containing TiO₂ NPs at three concentration levels in comparison to control animals. Lung tissue was prepared for LIBS analysis by FFPE. The researchers used the endogenous signal from P to demarcate relevant tissue structures, while the Ti signal indicated the exogenous exposure following application of cut-off thresholds for the LOQ. Quantitation was achieved by calculating the ratio of the total Ti intensities divided by the area of interest as identified by the P signals. This was plotted against the total Ti concentration obtained by traditional acid digestion and solution ICP-MS analysis from an adjacent lung lobe. The optimum resolution obtained was 50 μm, achieving an R² of 0.999. The approach was then successfully applied to three human lung biobank specimens, demonstrating that standard histological preparation methods could also be analysed by LIBS. The results demonstrated the potential of the technique for toxicological and respiratory exposure studies. The paper by Ferreira et al.¹⁹⁸ described the development of a multivariate spectral data analysis method using the “convex envelope” approach to detect unusual elements in lung biopsy tissues from patients with occupational exposure-related diseases. The authors termed this data reduction technique the “Interesting Features Finder”, which ultimately was capable of identifying unique attributes from a few spectra out of the tens of thousands generated from LIBS analysis from multiple samples. The authors suggested that the tool has the potential to highlight potentially useful data that could be masked by the bulk. Furthermore, the Matlab and Python codes were available in the supplementary information.

As noted in our previous Update,¹ imaging applications utilising XRF spectrometry for food products are not a regular feature, however, two papers have covered this topic. The development of tomographic mapping method was reported by Graefenstein et al.¹⁹⁹ who employed XRF spectroscopy to generate 3D data through multiple 2D cross sections. The researchers analysed the leaves of chives (Allium schoenoprasum) to demonstrate the application. Improvements to the corrections for self-absorption, Compton and Rayleigh scattering and peak fitting were presented to highlight the ability to quantify light elements such as Ca, K, Mn and Zn. The authors suggested this could be extended to biological matrices such as tissue, organoids and bones. Webb and co-workers²⁰⁰ published data from the analysis of human brain tissue biobank samples from Alzheimer’s patients without cognitive impairment by SR-μXRF spectroscopy. Following initial low resolution scans, key areas were identified for high resolution imaging. In two samples, Hg particles correlated with Se were detected. Furthermore, it was observed that the particles were co-localised with Zn structures and potentially associated with S in some particles. It was postulated that S and Se may play a protective role against Hg through binding mechanisms, warranting further investigation.

7. Applications: clinical and biological materials

7.1 Elements used for indirect determinations

As one of the most sensitive and versatile atomic spectrometry techniques, ICP-MS, appears to be the primary choice for the development of analytical methods for the indirect detection of molecules present at low concentrations in biological fluids. In the period covered by this Update, several such applications were reported and two reviews were published to cover recent developments of this rapidly expanding field. To achieve the indirect determination of biomolecules using ICP-MS, they are usually labelled with specific element tags through direct antibody–antigen interaction or DNA hybridization, followed by the separation between bound and unbound elemental tag. An alternative to direct labelling is to use specific reactions to free the element of interest for detection by means of ICP-MS. Hu et al.²⁰¹ in a review covering 114 papers over the past 20 years, described the principles of three of these approaches, such as enzymatic cleavage of metal-labelled substrates, release of immobilised metal ions from DNA backbone and nucleic acid amplification-assisted aggregation, and discussed their advantages and disadvantages. The authors noted that, because the element of interest was generated in situ by a specific reaction, chromatographic separation was not necessary and, in addition, the use of metal-labelled primers in nucleic acid amplification significantly increased the amounts of metal ions available for detection, resulting in a reduction in the background and interference from the matrix. These features and the elimination of synthesis of labelled biomolecules, increased the speed and ease of the assays. The main drawbacks were identified as interferences generated from biological matrices when endogenous elements (P, S) were used as tags and non-specific adsorption of metals on DNA structures, affecting the S/N and the LOD, all of which limited the reliability of these approaches for clinical diagnostics. The authors concluded that further insights into the biomolecular interactions may help the optimisation of the substrate design and release the full potential of the label-free approach. The state of the art of quantitative element-tagged bioassays with ICP-MS detection in clinical practice was described by Du et al.,²⁰² who examined 187 publications over the period from 2001 to 2022. They noted that the complexity of clinical samples or in vivo studies posed severe challenges to the application of this technique, but binding interactions in vivo, such as antibody-antigen affinity binding, nucleic acid base pairing, enzyme-substrate interactions, and aptamer-affinity binding, offered better specificity and stability for the incorporation of element tags into biomolecules. The authors considered the type of element tags (endogenous or exogenous, the latter involving lanthanides or NPs) and described in detail the specific biomolecular interaction (antibody–antigen affinity binding, nucleic acid base pairing, enzyme–substrate interactions, and aptamer-affinity binding) exploited for these bioassays. Recent applications to the quantification of biomarkers (e.g. cancer biomarkers, hormones, immunoglobulins, serum proteins, human and viral nucleic acids) were listed in a table, summarising the biomarker detected, the tag, the type of biological fluid/tissue or cell type analysed, the LOD achieved and the available details concerning method accuracy. Time-resolved ICP-MS was deemed the best suited technique for these applications, allowing for the observation of changes overtime within the sample, to better understand the kinetics and dynamics of biomolecular interactions. The authors also discussed the properties and suitability of different instrument types (LA-ICP-MS, ICP-QMS, scICP-MS, SF-ICP-MS, ICP-TOF-MS) for investigating the morphology and constituents of single cells, studying cellular uptake and cell-to-cell variability, tissue imaging and the simultaneous determination of hundreds of biomolecules, using multiple elemental tags, using mass cytometry, a specific application of ICP-TOF-MS. The authors noted that several challenges still lay ahead to fully implement quantitative bioassays by ICP-MS in clinical practice and reckoned that this area of research will be further exploited and rapidly developing.

Two research groups developed new assays as proof of concept for future application in clinical practice, after a more extensive validation. One group²⁰³ considered L-thyroxine as an alternative molecular tag, instead of NPs or lanthanide chelates generally used in aptamer assays with ICP-MS detection. The molecule, commercially available with a purity >98%, carried four I atoms, for ICP-MS detection, and reactive groups allowing its covalent conjugation to a DNA binder. The new tag was tested initially for the determination of a small molecule (L-tyrosinamide) using a competitive approach, resulting in lower I release in the presence of higher concentrations of L-tyrosinamide. The method was further developed for the detection of a protein (alpha-thrombin) in urine, using a sandwich mode, that gave an increasing I signal in the presence of increasing concentrations of alpha-thrombin. For both procedures, the extracts obtained from the incubation with the aptamer, obtained in 50% HNO₃, were diluted 1 + 49 with pure water prior to analysis using ICP-QMS with Rh as IS and a 5 min rinsing step with 1% HNO₃ between samples to avoid memory effects. The LODs, calculated as 3 × SD of samples collected in the absence of the molecule to be tested, were estimated to be 60 nmol L⁻¹ for L-tyrosinamide and 40 nmol L⁻¹ for alpha-thrombin. Finally, tests carried out on human urine spiked with alpha-thrombin (at 62.5 and 500 nmol L⁻¹) did not show evidence of matrix interferences. Rodriguez-Penedo et al.²⁰⁴ synthesized monodisperse Pd nanoclusters (diameter 2.49 ± 0.02 nm), each containing 550 Pd atoms on average. They noted that, besides serving as a tag in bioassays with ICP-MS detection, these Pd nanoclusters could trigger electrochemical reactions, allowing detection by linear sweep voltammetry (LSV), a technique that may be easily adapted for point-of-care analysis. The authors developed a competitive immunoassay based on Pd nanoclusters for the determination of glial fibrillary acidic protein (GFAP) in human serum, a biomarker for ischemic stroke vs. haemorrhagic stroke, and compared the performance of the two techniques as detectors. The target protein was determined by ICP-MS measuring the Pd signal and by LSV from the current intensity generated at a fixed potential by the catalytic activity of the Pd nanoclusters on a specific electrochemical reaction (hydrogen evolution reaction). The inverse relationship between the concentration of the GFAP standards and (a) the concentration of Pd in the bioassay extracts or (b) the corresponding current intensity was obtained as 4-parameter-logistic (4-PL) curves, with R² values of 0.993 and 0.991, respectively. The LODs, determined from these curves as the inhibitory concentration IC₁₀, were both found to be 0.03 pmol L⁻¹ GFAP, but when calculated with the more robust error profile method, that takes into account the 95% CI of the 4-PL fit, the LODs were 0.03 pmol L⁻¹ GFAP for ICP-MS and 0.11 pmol L⁻¹ GFAP for LSV, owing to the better precision of ICP-MS measurements. In comparison, the LOD for the determination of GFAP using ELISA was 0.31 pmol L⁻¹. The method was tested on serum samples from 12 subjects (four controls, five patients with ischemic stroke and three with haemorrhagic stroke) and gave results in agreement with a conventional ELISA assay, except for the control subjects, whose serum GFAP levels were below the ELISA LOD and could only be measured using ICP-MS.

The interest in more sensitive methods for the determination of biochemical parameters led a few groups to propose alternative methods, based on ICP-MS, to those already in clinical practice. He et al.²⁰⁵ described a label-free approach for the determination of alkaline phosphatase activity (ALP), based on previous applications of MnO₂ nanosheets (MnO₂ NS), a two-dimensional nanomaterial, in colorimetric and fluorescence assays. The principle of the proposed method is to exploit MnO₂ NS as the element target source, releasing stoichiometric amounts of Mn²⁺ ions via a reduction reaction. The reducing agent was ascorbic acid, produced by the hydrolysis of L-ascorbic acid 2-phosphate, in amounts proportional to ALP activity. The concentration of Mn²⁺ ions was determined using ICP-MS, after separation of the remaining MnO₂ NS by filtration. This simple procedure allowed for the determination of ALP activity with an LOD of 0.007 U L⁻¹, without the need for labelled compounds. The method, although tested on only three samples of human serum, gave results in agreement with those obtained by a commercial kit and recoveries of 40 mU mL⁻¹ ALP spiked into the same samples was between 91% and 102%. Another group²⁰⁶ designed a dual-element Ir–Eu tag, which would allow simultaneous fluorescence imaging and ICP-MS quantification. They tested this approach for the visualization and quantification of signal regulatory protein alpha (SIRPα), which is overexpressed on immunocytes in blood samples and reflects the immune response from diseased people. The proposed method achieved the quantification of this protein with lower LODs (3 × SD in the absence of targets) than that achieved in the same study using fluorometry (65.2 pmol L⁻¹ at 612 nm), i.e. 0.5 pmol L⁻¹ (¹⁵³Eu) and 1.1 pmol L⁻¹ (¹⁹³Ir). Based on these figures, the LODs for the counting of the host cells were estimated as 220 cells (¹⁵³Eu) and 830 cells (¹⁹³Ir), respectively. Sixty-five blood samples from patients suffering from liver cancer, cholecystitis, gastritis, pneumonia or lithiasis were analysed, but conclusive remarks about the reliability of the data or their clinical significance were not reported. A procedure for the simultaneous detection of Escherichia coli (E. coli O157:H7) and Salmonella in human blood was described by another group of researchers.²⁰⁷ They combined immunomagnetic separation, specific antibodies labelled with Ag NPs or Au NPs, respectively, and ICP-MS detection. The results of spiked experiments carried out at three concentration levels (5 × 10², 20 × 10² and 100 × 10² CFU mL⁻¹) on three human blood samples gave recoveries ranging from 85% to 115% and were consistent with those obtained by the blood culture method. Unfortunately, LODs of 220 CFU mL⁻¹ for E. coli O157:H7 and 164 CFU mL⁻¹ for Salmonella were too high to allow the detection of these bacteria in patient blood samples (estimated between 1 and 10 CFU mL⁻¹), calling for substantial improvement of the proposed method.

Given the importance of early diagnosis of cancer and the assessment of therapeutic treatments, there is continuous interest in techniques which can reliably quantify easily accessible tumour biomarkers, including cells and other structures present in biological fluids at very low concentrations. Wang et al.²⁰⁸ devised a complex strategy to improve the detection of cancer cells released into the bloodstream. First, they applied an enrichment procedure, based on stepwise centrifugation, to isolate circulating tumour cells from blood samples of patients suffering from lung cancer. Then they developed a bioassay with ICP-MS detection for the measurement of mucin 1, a potential tumour marker, and the quantification of lung cancer A549 cells, which have multiple expression sites for mucin 1 on their surface. To do this, they exploited the findings of their previous studies with a catalytic hairpin assembly, an enzyme-free amplification method assisted by nucleic acid, including a C–Ag⁺–C moiety, followed by selective recognition reactions and competitive reduction of ascorbic acid by Cu²⁺ and Hg²⁺ ions. The selective binding between the aptamer probe and the target released DNA, that triggered the catalytic hairpin assembly amplification, freeing Ag⁺ that, in turn, reacted with added CuS NPs, releasing free Cu²⁺. The following step involved the addition of Hg²⁺ and ascorbic acid and the competition between Cu²⁺ and Hg²⁺ for its reduction. The final concentrations of Hg²⁺ in the extracts were measured using ICP-MS and reflected the levels of the targets. The LODs (3 × SD of the ICP-MS measurements of extracts obtained in absence of target) were 0.3 ag mL⁻¹ for mucin 1 and 0.25 cells per mL for A549 cells. The method, applied to the analysis of blood samples from 58 patients, identified all negative cases (11) and the majority (43 out of 47) of those who tested positive for lung cancer, based on clinical CT images and pathological findings. These results support the potential of the method for a wider application in clinical settings, subjected to more extensive analytical and clinical validation. Another group²⁰⁹ focused on exosomes, i.e. membrane-bound extracellular vesicles, which may carry proteins derived from parental cancer cells, thus offering potential for early cancer diagnosis. They proposed a bioassay based on the proximity ligation of cholesterol and an epithelial cell adhesion molecule aptamer on the exosome membrane, followed by the assembly of the exosome with a multicomponent nucleic acid enzyme. This, in turn, triggered the cleavage of DNA substrates containing RNA bases and led to the release of the Tb labels from the substrate DNA. The measurement of Tb concentrations using ICP-MS allowed the quantification of exosomes in the range of 5 ×10⁵ to 2 × 10⁷ particles per μL, with an R² of 0.9986. The LOD, estimated as 3 × SD of the ICP-MS measurements of extracts obtained in absence of the target, was 2.52 ×10⁵ particles per μL and the RSD% was 6.2% (at 5 × 10⁶ particles per μL, n = 7). The analysis of human plasma samples from ten oral squamous cell carcinoma patients and ten healthy volunteers showed significant differences (p = 0.0059) between the groups. The authors concluded that the developed approach showed good potential for application in early cancer diagnosis, but noted that several difficulties, such as low throughput, low degree of automation and scarce availability of critical reagents, limited the impact of ICP-MS based bioassays on clinical practice. Sun and co-workers²¹⁰ devised a multiplex, high-throughput platform for immunoassays based on REE labelling and simultaneous ICP-MS detection. They implemented a semi-automatic process, including sample introduction from a 96-microwell plate into the ICP-MS, using a modified autosampler allowing for high throughput, reduced sample uptake and rinse time. The analytical performance of the system was tested using, as a model, immunoassay kits for five lung cancer biomarkers (CEA, CYFRA21-1, NSE, SCC and ProGRP) labelled with Lu, Eu, Tm, Ho and Tb, respectively. The lower LODs (defined as 2 × SD of the blank, n = 20) were 0.26 ng mL⁻¹ for CEA, 0.10 ng mL⁻¹ for CYFRA21-1, 0.42 ng mL⁻¹ for NSE, 0.10 ng mL⁻¹ for SCC and 1.37 pg mL⁻¹ for ProGRP, respectively. Recoveries of CEA, CYFRA21-1, NSE, SCC and ProGRP concentrations were 108.4%, 91.0%, 107.2%, 113.3% and 95.8%, respectively. Repeatability RSD% (n = 10), determined at a low and a high concentration levels on clinical samples, were: 5.4% and 5.8% for CEA; 8.2% and 8.6% for CYFRA21-1; 7.0% and 4.5% for NSE; 8.7% and 9.6% for SCC; 4.8% and 5.9% for ProGRP, respectively. The entire analytical process was completed in 40 min, but the analytical step could reach a throughput of 100 samples per h. Using REE labelling, up to 12 biomarkers could be simultaneously detected in the same sample, thus notably increasing the efficiency of biomarker determination. Paired tests between standard immunoassay kits and ICP-MS measurements based on REE-labelling, carried out on 70 human serum samples for each parameter, gave r values of 0.9900, 0.9879, 0.9853, 0.9898 and 0.9705 for CEA, CYFRA21, NSE, SCC and ProGRP, respectively. This work demonstrates an innovative approach for the practical application of ICP-MS based tests in a clinical setting, paving the way for its wider use.

A thorough investigation of an ICP-MS based bioassay in routine clinical practice was reported by Tang et al.,²¹¹ building on the previous work of their group. The study addressed the issue of improving the quantification of lactoferrin (LTF) and serum amyloid A (SAA), as important factors in the diagnosis of infection. The proposed method consisted of an immunoassay system, allowing for their simultaneous determination by ICP-MS, using Eu and Ho, respectively, as labels. A sandwich-type immunoreaction, with hydrophilic streptavidin magnetic microspheres as carriers, led to the formation of a immunomagnetic beads-antigen-labelled antibody complex. After thoroughly washing, Eu and Ho were removed from the sandwich complex using a 1% HNO₃ solution, containing Rh as the IS, and their concentrations were measured using ICP-MS. The method validation was carried out according to the CLSI guidelines.^212–214 The limits of blank (LoB, defined as the highest apparent analyte concentration expected to be found when replicates of a blank sample containing no analyte are tested), calculated as the mean value of 60 measurements of the blank plus 1.645 × SD, were 0.85 ng mL⁻¹ for LTF and 1.09 ng mL⁻¹ for SAA, respectively. The LOQs (defined as the minimum concentration at which the RSD% was <20% and the deviation was <8.33%) were 2 ng mL⁻¹ for LTF and 3 ng mL⁻¹ for SAA, respectively. Recoveries for both parameters ranged from 95.01% to 106.26%, and RSD%, estimated at three concentration levels, was <8% (intra-batch) and <11% (inter-batch). The effect of possible interferences was found to be less than ±10%. The application of the method to 131 plasma samples confirmed its ability to discriminate infected subjects vs. non-infected ones, based on positive and negative predictive values. The simultaneous detection of both markers in a single reaction also offered a reduction in cost, time and effort.

Strategies for the detection of nucleic materials associated with virus infections were proposed in two papers. In the first one²¹⁵ the simultaneous detection of miR296 and miR16 in human serum was reported. These microRNAs are associated with enterovirus 71 which causes hand, foot, and mouth disease. The procedure involved a hybridization chain reaction (HCR) with the biomolecules involved and CdSe quantum dots and Ag NPs for the specific labelling of the two target molecules, followed by magnetic separation of the HCR complex, acidification with HNO₃ and measurement using ICP-MS. Although LODs were indicated as 0.33 nmol L⁻¹ for miR296 and 0.38 nmol L⁻¹ for miR16, respectively, the authors noted that, in the absence of the target, the HCR complex would not be released, resulting in a low background signal value. An assessment of recovery from serum samples spiked with 10, 20 and 30 nmol L⁻¹ of the analytes yielded values between 88.5% and 92.0%. Intra-day and inter-day precision (RSD%), evaluated at the same concentration levels, were <10%. Another paper²¹⁶ described the synthesis of Tm NPs and the assessment of their performance in bioassays using the detection of hepatitis B virus-DNA as a model. The analytical process involved mixing streptavidin magnetic beads with the sample, followed by addition of the Tm NPs. After magnetic separation of impurities, the sandwich-complex was dissolved with aqua regia and Tm was determined in the resulting solution using ICP-MS. The LOD (3 × SD in absence of the target, n = 12) was 0.473 pmol L⁻¹ and recoveries ranged between 95.9% and 111.8%. The authors concluded that optimising the synthesis of Tm NPs enhanced the sensitivity of the detection, eliminating the need for an amplification step and simplifying the analytical process. The same approach could be extended to other lanthanide NPs.

The sensitive detection of food allergens is necessary to support people’s health and food safety regulations. Torregrosa et al.²¹⁷ investigated the potential of aptamer assays with ICP-MS detection, for the determination of the protein β-conglutin, a recognised allergen from lupin, in flour samples. The study examined two competitive aptamer assay schemes: thiolated aptamers chemisorbed onto Au NPs and biotinylated aptamers linked with streptavidin–Au NPs. They also compared two ICP-MS configurations—ICP-MS/MS vs. spICP-MS, to verify whether, as suggested for non-competitive bioassays, LOD and sample throughput can be improved using spICP-MS. The thiolated aptamer assay achieved LODs (3 × SD of blank, n = 15) of 200 pmol L⁻¹ and 150 pmol L⁻¹, respectively, with ICP-MS/MS and spICP-MS detection, whereas the corresponding values, for the biotinylated aptamer, were 2 pmol L⁻¹ and 10 pmol L⁻¹. The better LODs of the biotinylated aptamer assay were attributed to the higher number of NPs per aptamer. Using spICP-MS did not provide the expected improvement or gave a worse LOD than that achieved with ICP-MS/MS. The authors suggested that, in competitive assays, it may be more difficult to distinguish signal pulses for low analyte standards from those from the blank and that the high number of events detected may lead to significant uncertainty of the measuring pulse frequency, thus worsening the LOD. The paper includes a comparison of the analytical parameters studied for this method (LOD, half maximal effective concentration and linear range) with those reported for other aptamer assays for the determination of β-conglutin based on other detection approaches (e.g. UV-vis, lateral flow assay, fluorescence resonance energy transfer, real time PCR and real time recombinase polymerase amplification). The biotinylated aptamer assay with ICP-MS/MS detection was further studied using different flour samples. Recovery tests, carried out on chickpea and soy flour samples spiked with 100 mg kg⁻¹ and 600 mg kg⁻¹ of lupin gave values ranging from 85% to 115%. Intra-day and inter-day precision (RSD%) were <10%.

7.2 Multi-element applications

7.2.1 Specimens analysed to investigate metallic implants and biomaterials. One paper of interest in the field of metallic implants, involved the assessment of total Ti concentration and TiO₂ particle fraction in periprosthetic tissue from 19 patients with a failed total hip arthroplasty.²¹⁸ Following MAD of the tissues with a mixture of 4 mL ultrapure water–1.5 mL HNO₃–0.5 mL H₂O₂–0.05 mL HF, to completely digest any TiO₂ particles, total Ti was determined by detection of ⁴⁷Ti using ICP-MS with a CRC and He gas, using ⁷²Ge as IS. There was a wide variation in the total Ti concentrations measured in the different tissue samples, with ranges between 14.6 ± 3.9 μg g⁻¹ and 71 665.0 ± 19 062.9 μg g⁻¹ reported. An estimate of the particle Ti fraction was obtained through determination of the soluble Ti in each sample. This was achieved by MAD of the tissues with a non-HF-containing acid mixture, which was not able to solubilise TiO₂ particles, before centrifugation of the digests and measurement of the supernatant Ti concentration. Of the 19 patients, six had a significant Ti particle fraction (ranging from 38% to 94% of the total Ti concentration after HF digest). The TiO₂ NPs in these samples were then further characterised by spICP-QQQ-MS and monitoring of ⁴⁸Ti¹⁶O⁺ using 10% O₂ gas in the reaction cell. The median diameters of the released NPs in the different tissue samples were between 39 nm and 187 nm and the particle concentrations ranged from 7.2 × 10⁷ to 2.3 × 10¹¹ particles per g. The median particle size in the sample with the highest TiO₂ particle concentration was determined using SEM-EDS, to be 230 nm compared with 187 nm by spICP-QQQ-MS. The authors suggested that the slight difference in the median NP size between the two techniques may have resulted from either the limited resolution of the SEM instrument or the spherical shape assumption used with spICP-QQQ-MS measurement. Meanwhile, Raman spectroscopy demonstrated that the isolated particle fraction consisted of TiO₂ NPs with an anatase structural phase.

7.2.2 Biological fluids and tissues. Hair attracts attention as a non-invasive matrix for biomonitoring studies of environmental exposure or nutritional status and for its capability to show changes in these over time. The main drawbacks to its wider use stem from the difficulty in separating information about exogenous and endogenous levels of the elements, as well as the lack of reference data from non-exposed hair portions, e.g. those below the scalp. Two studies explored the capabilities of LA-ICP-MS for the analysis of single hair strands. Fernandes et al.²¹⁹ investigated the application of the technique to determine the levels of 13 elements (Ag, Ba, Co, Cr, Cu, Fe, Hg, Li, Mg, Mn, Ni, Sr and Zn) in single hair strands. Out of concern for the potential exposure of dentists to toxic metals, in particular Hg, they examined hair strands from a female student, who had been exposed to dentistry environments for two years at the time of sample collection. Hair strands, cut at 2 cm from the root, were washed with acetone, then rinsed with distilled water and left to dry. Imaging of hair samples by means of SEM-BSE and SEM-EDS confirmed the uneven distribution of the elements within the layered hair structure (from rim to core: cuticle, cortex and medulla), with higher levels detected in the outer layer. Therefore, the authors considered whether the practice of partial ablation along hair strands, which helps to avoid contamination from mounting materials (e.g. double-sided tape or adhesive agents) provided representative results. To address this matter, hair strands were embedded in epoxy resin and analysis were performed both along the hair strands and their cross sections, using full and partial ablation. Due to its homogeneous distribution, sulphur was used as IS. Calibration was based on compressed pellets of the CRM NCS DC73347 (human hair powder), that carried certified values for most elements, except Ag, Li and Mg. The results obtained by LA-ICP-MS were compared with those determined by solution nebulisation ICP-MS, after digestion of hair strands (ca. 150 mg) in an open vessel with 3 mL 70% (w/w) ultrapure HNO₃ and 0.5 mL 30% (w/w) ultrapure H₂O₂. Unsurprisingly, results obtained by full ablation were in closer agreement (R² = 0.9655) with those achieved by the solution nebulisation method, than those determined by partial ablation (R² = 0.9179). However, apart from Li, which levels were below the LOD, measurements by LA-ICP-MS provided results higher than those obtained by the reference method. This bias was attributed to losses of the IS (sulfur) during the analytical steps. Christensen and LaBine³⁷ engaged in assessing reference points and time-related external contamination along the length of hair strands. They designated three hair regions, based on the distance from the root bulb (1 to 2 mm, 10 to 11 mm, 39 to 40 mm), representing, respectively, the reference point, short term exposure (less than 2 weeks) and long-term exposure (3.5 to 4 months), based on the assumption of an average hair growth rate of 1 cm per month. Analysis was carried out using LA-ICP-MS, which can cater for single hair strands and their sub-portions. Samples, collected from 61 adults, were washed with acetone and distilled water, air-dried and placed on glass microscope slides with adhesive tape. The levels of 26 elements (Al, As, B, Ba, Ca, Cd, Co, Cr, Cu, Fe, Hg, K, Mg, Mn, Mo, Na, Ni, P, Pb, S, Se, Sr, Ti, V, U and Zn) were determined by LA-ICP-MS. For calibration, the CRM (NRCC DORM-4 fish protein CRM for Trace Metals and Other Constituents) was analysed alongside each sample and S, that was present in all samples, was used as IS. The authors applied a first ablation, at 2 J cm⁻² power and 60 μm spot size, to obtain information on the element content of the external layer of the hair (cuticle), then a second one at higher power (10 J cm⁻²) with a smaller (30 μm) spot size, to determine the concentrations of the same elements in the inner layers (cortex). Accuracies, based on the CRM, were between 95% and 110% (cuticle) and from 101% to 120% (cortex). The observed precision (RSD%) was between 3.2% and 27% (cuticle) and from 5.2% to 25% (cortex). The worst precision values were observed for Ti (cuticle) and Cd (cortex), due to levels close to their LOD. The authors concluded that, for most elements, only hair samples collected below the scalp provided reliable information about the element status and reported, for the first time, reference ranges for this portion of the hair for most of the studied elements, except As, Cd, Co, Mo, Ni, U and V, which levels were below the LODs (see also Section 2.2).

Breast cancer is a widespread disease, with a high incidence and potential recurrence. Breast-conserving resection with free margins is now widely applied for early breast cancer therapy, however, research is in place to address how to define the tumour margin during surgery. Optical emission spectroscopy has emerged as a potentially valuable analytical technique for this purpose, based on its capability to detect chemical elements in the tissue vapour generated during electrosurgical sparking. A group of researchers²²⁰ investigated differences in the element profiles of malignant and healthy breast tissue that may be used in future for in situ differentiation and intraoperative tumour margin assessment. Paired samples of normal and abnormal breast tissue were obtained from 18 breast cancer patients (17 females and 1 males) aged from 25 to 85 years. Analyses were carried out using a prototype instrument containing an active radiofrequency needle electrode for spray coagulation, an optical fibre to transfer light emitted by the radiofrequency sparks to a high-resolution Echelle spectrometer and a lumen for CO₂ gas flow, to protect the optical fibre from contamination with tissue aerosols and smoke particles. A total of 972 spectra (480 from normal, 492 from abnormal tissue) were selected. Each spectrum was normalised to its total intensity, then integrated to obtain vectors focused on the emission lines of elements present in human tissues (C, Ca, K, Mg, Na, P and Zn). A support vector classifier was trained using these data, then applied to classify the tissue type, achieving an average classification accuracy of 96.9%, sensitivity and specificity of 94.8% and 99.0%, respectively, and positive and negative predictive values of 99.1% and 96.1%, respectively. The authors concluded that, notwithstanding the limited sample size of this study, their results supported the application of OES to discriminate healthy and malignant breast cancer tissue in a clinical setting.

Although sample size is usually not a cause of concern for urine analysis, it may be so when samples are collected as part of multi-purpose, long-term, epidemiological studies and therefore become a precious resource. In the Multi-Ethnic Study of Atherosclerosis (MESA), spot urine samples were collected from 6814 subjects, aged from 45 to 84 years, with no signs of clinical cardiovascular diseases. The impact of exposure to multi-element mixtures on human health is, however, not yet well understood or investigated. Therefore, as part of MESA, Schilling and co-workers²²¹ undertook to measure the baseline element concentrations of 18 elements in urine, using ICP-MS, although only a limited amount of sample (0.8 mL) could be made available for this purpose. To minimise matrix effects, they used a 50-fold dilution: 100 μL of urine were diluted to 5000 μL with an aqueous solution of 2% (v/v) HNO₃–0.02% (v/v) Triton X-100–500 μg L⁻¹ gold, containing the IS Ga, Ir and Rh each at a final concentration of 5 μg L⁻¹. Analysis were carried out on a DRC-ICP-MS instrument, fitted with platinum sampler and skimmer cones, using O₂ and NH₃ as the reaction gas. The levels of 18 elements (As, Ba, Cd, Cs, Co, Cu, Gd, Mn, Mo, Ni, Pb, Sb, Se, Sr, Tl, W, U and Zn) were measured simultaneously. The authors did not report measurements of Hg, or the reasons why, although the diluting solution included gold, usually meant to improve Hg measurements, Particular attention was paid to ensure the reliability of the results over a large dynamic range, using a nine point calibration curves. Method accuracy was established using a number of urine CRMs and EQA materials (QM-U-Q1822, 1823 and 1824, Quebec Multi-element External Quality Assessment Scheme; NIST SRM 2668 Level 1 and Level 2 and Recipe ClinChek Level 1). These CRMs covered the levels of all elements studied, except Gd, for which an in-house pooled urine, spiked at 3 concentration levels, was used as control. The method LODs (calculated as 3.33-fold the blank SD and multiplied by 50, to take into account the sample dilution), ranged from 0.001 μg L⁻¹ for U to 6.2 μg L⁻¹ for Zn, and were in-line with, or better than previously reported values. Precision, as RSD% was <8.5% for most elements, except for Gd (20%), Mn (19%) and U (16%) that were present at very low urinary concentrations. Recoveries, assessed on the five CRMs, fell within the interval between 90% to 128% for all elements but Ni at low concentrations, therefore Ni data were not included in the statistical analysis. The method was applied to a total of 7561 urine samples, of which 6618 were collected at the beginning of the MESA study (2000–2002) and 943 at a later date (2010–2011). The method LODs proved sufficient for the measurement of element concentrations in more than 99% of the samples, with the exception of Gd, Mn, U and W, where the concentrations were unsurprisingly below the method LODs in between 12% to 59% of the samples. Comparison of trace element concentrations corrected using either specific gravity or creatinine, with uncorrected data did not lead to definitive conclusions on the advantages of such corrections for each element. Medians and interquartile ranges of the unadjusted data for both the essential and non-essential trace elements were reported for the total set of data available (7652 samples) and for subsets classified by age, sex, ethnicity, location, education, smoking status and BMI, to address the influence of these covariates.

7.3 Progress for individual elements

7.3.1 Copper. Direct measurement of serum labile-bound Cu, also known as exchangeable or non-caeruloplasmin-bound Cu, has been an area of interest in this Update for some years. Bitzer et al.²²² employed a dual filtration-based approach to measure labile-bound Cu and derive reference ranges in a healthy adult population (110 males and 104 females). Serum (30 μL) was first passed through 100 kDa filters on a 96-well plate to remove high M_r Cu-binding proteins. The filtrate was then treated with EDTA before being subjected to 30 kDa filters to isolate the total labile-bound Cu. Detection of Cu was carried out using ICP-MS in KED mode. Figures of merit for the method were: LOD and LOQ, 7 ng mL⁻¹ and 19 ng mL⁻¹, respectively; inter-assay precision ranging from 38.2% at 25 ng mL⁻¹ to 6.2% at 723 ng mL⁻¹ and recoveries of Cu standards (10, 100 ad 200 ng mL⁻¹) diluted in a 1 + 1 ratio in water CRMs (NIST SRM 1640 and SRM 1643), prior to the labile-bound Cu isolation procedure, varying from 95% to 109%. The high level of imprecision at low labile-bound Cu concentrations was noted and attributed to a highly manual sample preparation process and the instrument LOD. Investigation of the stability of serum labile-bound Cu demonstrated no significant change in analyte concentration when samples were stored at −20 °C or −80 °C for up to 29 days. Selection of the individuals for the reference range study was robust and excluded those with Wilson Disease and renal, biliary, digestive and autoimmune disorders. Information was also collected around other potentially confounding factors, e.g., vitamin and mineral supplements, occupational metal exposure, hormone replacement therapy and use of contrast agents. The reference ranges derived for serum labile-bound Cu concentration (fraction of labile-bound Cu relative to total Cu) were 13–105 ng mL⁻¹ (1.0–8.1%) in females and 12–107 ng mL⁻¹ (1.2–10.5%) in males.

Two further studies reported associations between serum Cu concentrations and adverse health outcomes. The first reported a retrospective cohort study involving 183 patients from a single academic centre with advanced liver disease, in which serum Cu concentration below the reference interval were shown to be an independent risk factor for mortality.²²³ Within the cohort, 17% of individuals had Cu deficiency (defined as <80 μg dL⁻¹ in females and <70 μg dL⁻¹ in males) with the median serum Cu concentrations being 62 μg dL⁻¹ and 110 μg dL⁻¹ in the Cu-deficient and non-deficient groups respectively. There were statistically significantly higher infection rates in the Cu-deficient group compared with the non-deficient group (42% vs. 20%, p = 0.01). Over the follow-up period of 396 days, 22.6% of patients with Cu deficiency died compared with 10.5% in the non-deficient group, although liver transplant rates were similar between the two groups. When adjusted for age, sex and clinical disease severity scores, Cu deficiency was associated with a significantly higher risk of death before transplantation (hazard ratio 3.40, 95% CI, 1.18–9.82, p = 0.023). Other causes of Cu deficiency, e.g., weight loss surgery, intestinal disease or Zn supplementation, were excluded as influencing factors. Furthermore, there was no difference in nutritional status between the two groups and no statistically significant association between Cu deficiency and liver disease aetiologies. The authors concluded that cirrhosis may be an independent and previously under-recognised risk factor for Cu deficiency with the higher rates of infection in Cu-deficient individuals being an explanation for the higher mortality in this group. The second publication consisted of a meta-analysis of maternal antenatal serum Cu concentrations and risk of preterm delivery across 18 study cohorts covering five continents (10 449 singleton live births).²²⁴ Maternal serum Cu concentrations measured by ICP-MS were found to be normally distributed with a mean ± SD of 1.92 ± 0.43 μg mL⁻¹. Meta-analysis across all cohorts demonstrated that a 1 μg mL⁻¹ increase in maternal serum Cu was associated with a higher risk of preterm birth (odds ratio 1.30, 95% CI 1.08–1.57) and shorter gestational duration of 1.64 days (95% CI, 0.56–2.73). Further analyses in one cohort revealed a strong correlation between maternal serum Cu and acute phase reactant concentrations, as well as infection status, which is unsurprising as serum Cu is well known to increase during the acute phase response owing to the upregulation of caeruloplasmin. Inclusion of acute phase reactants and infection status as covariates reduced the effect of maternal serum Cu concentration on the risk of preterm birth substantially, however, a marginally significant association remained. Two notable limitations of the study were that it was not known whether the preterm births were spontaneous or iatrogenic and that serum caeruloplasmin was not measured.

7.3.2 Iron. A stable-isotope approach, involving ICP-MS determination of ⁵⁷Fe : ⁵⁶Fe isotopic ratios, was used to assess Fe absorption in patients with sickle cell anaemia.²²⁵ The patient cohort consisted of 13 sickle cell patients: three with ferritin concentrations ≥1000 ng mL⁻¹, considered to have Fe overload; ten with ferritin concentrations <1000 ng mL⁻¹, considered not to have Fe overload: and ten age, sex and ethnicity matched controls. The participants were given orally 5 mg of stable ⁵⁷Fe isotope in the form of FeSO₄ (prepared from elemental ⁵⁷Fe and H₂SO₄), and the ⁵⁷Fe : ⁵⁶Fe isotopic ratio was measured in acid-digested erythrocytes, isolated from blood samples collected 14 days after administration of the Fe isotope. The fractional Fe absorption was calculated using the measured erythrocyte incorporation of ⁵⁷Fe and an estimate of the total amount of circulating Fe in blood (based on haemoglobin concentration and estimated blood volume). The most notable finding was that the highest median Fe absorption (and the widest range of Fe absorption values) was seen in the sickle cell non-Fe overload group (median (range): 6.9% (0.3–64.2%)), although this did not reach statistical significance in the small study population. In contrast, the median Fe absorption results were similar in the sickle cell Fe overload group and the controls, with a greater range of absorption values observed in the latter (median (range): 0.5% (0.3–5.4%) and 0.9% (0.3–26.5%) respectively). The Fe absorption could not be determined in some participants due an inadequate LOD of the set-up, which reflected use of standard ICP-MS rather than e.g., HR-ICP-MS, for the measurement of the isotope ratios.

7.3.3 Gadolinium. Ruprecht et al.²²⁶ reported an investigation into the uptake of Gd-based contrast agents, used for contrast-enhanced MRI, by blood cells. In the in vitro part of the study, blood from two patients was incubated with 2 mmol L⁻¹ of a contrast agent, Gd-DOTA, for 15, 30 and 60 min before erythrocytes and white blood cells (WBCs) were isolated and purified. Following suspension of the cells in PBS at an approximate concentration of 4 × 10⁵ cells per mL, they were analysed by scICP-MS to assess the Gd content per cell. In the WBCs, Gd cell content increased with incubation time, approaching saturation at approximately 60 min, at which point the mean Gd content was 268 ± 24 ag Gd per cell (approximate intracellular concentration, 3.0 ± 0.3 μmol L⁻¹). No uptake was observed in erythrocytes over the 60 min period. Uptake of the DOTA complex by WBCs and the intracellular complex concentrations were confirmed through time-resolved fluorescence, following incubation of blood with Eu-DOTA. A pilot in vivo study was then undertaken, in which blood was collected from 42 patients immediately following a contrast-enhanced MRI examination. Erythrocytes and WBCs were isolated and purified from 0.2 mL blood, both at t₀ (mean 17 ± 3 min after injection of the Gd-based contrast agent) and then 60 min later (t₆₀). The mean whole blood Gd concentration in the patients was 1.5 mmol L⁻¹ (range: 1.2–1.9 mmol L⁻¹). At t₀, the Gd content per WBC, as measured by scICP-MS, was undetectable (<7 ag Gd per cell) in two patients, while it was surprisingly high, at between 21 and 444 ag Gd per cell (from 0.2 μmol L⁻¹to 5.5 μmol L⁻¹), in the remaining 40 patients. A pre-contrast injection sample would have been useful here to provide the baseline Gd content per cell and support the validity of the t₀ results. At t₆₀, an increase in Gd content per WBC was observed in all patients, ranging from 25 to 763 ag Gd per cell (0.4–9.7 μmol L⁻¹). Interestingly, there was no correlation between the intracellular WBC and whole blood Gd concentrations. A further finding was that the Gd uptake of individual WBCs was inhomogeneous and suggestive of a bimodal distribution, with one population of cells taking up more Gd than the other.

7.3.4 Mercury. A high-throughput method for selectively determining organic Hg species in human hair, which involved the coupling of frontal chromatography and ICP-MS, was developed by Spanu et al.²²⁷ The set-up avoided the use of expensive HPLC equipment by inserting a low pressure, homemade, anion exchange column, to separate MeHg from Hg^II, into a standard ICP-MS sample introduction system. Lipids were removed from the hair samples using acetone, however, there was no assessment reported of whether this step affected the Hg speciation results, with MeHg being highly lipid soluble. The subsequent sample preparation procedure was minimal, consisting of a 15 min UAE and filtration through a 0.45 μm syringe filter. Thiourea was added to the extractant and rinse solutions to reduce the well-known memory effects with Hg. The LOD (5.5 μg kg⁻¹) for a 10 mg hair sample, was favourable for a method with no sample preconcentration procedure. Results obtained for a CRM (NIMD-01 Hg in human hair) agreed with the certified values for both total Hg and MeHg. Over-recovery of MeHg was observed in spiked solutions with a high Hg^II to MeHg ratio and Hg^II content over 5 μg kg⁻¹, which may limit utility of the method in some situations. Application of the method to the analysis of scalp hair from 49 individuals, gave a mean (range) MeHg content of 0.77 mg kg⁻¹ (0.02–3.2 mg kg⁻¹) and MeHg : tHg ratios were 64.6% (5.2–123.5%). As expected, there was a significant correlation between hair MeHg content and dietary fish intake in the study population.

A second paper of interest for the analysis of Hg in biomedical samples, investigated the use of TDA-AAS (with a direct mercury analyser) to determine total Hg in dried blood spots.²²⁸ The approach afforded a number of advantages for screening in population studies, including a very low blood volume requirement, utilisation of simple, non-expensive equipment, as well as removing the need for sample pretreatment. Capillary blood (50 μL ± 0.25%) was collected into a calibrated microcapillary tube before being applied to the cellulose card, allowing careful control of the blood volume deposited, which was deemed to be critical for achieving adequate accuracy and precision. Presumably though, this would complicate self-sample collection. The performance characteristics of the method (as determined by analysis of spiked samples, the CRMs Seronorm™ Whole Blood Level 1 and 2, and test items from the Quebec Multielement EQA Scheme) were precision ≤5.6% and recoveries between 75% and 106%. The LOD and LOQ of 0.10 μg L⁻¹ and 0.40 μg L⁻¹, respectively, were comparable to other studies using this technique and considered to be fit for the purpose of biomonitoring in non-exposed adults. However, more sensitive ICP-MS methods are likely to be required for populations in which lower levels of Hg exposure are of concern, e.g., young children. A pilot study was undertaken, in which paired capillary samples, analysed by the proposed method, and venous blood, for conventional Hg determination by ICP-MS, were collected from 41 adults. No statistically significant differences (p = 0.4588) were found between the two methodologies, with means (95% CI) of 3.87 μg L⁻¹ (3.12–4.79 μg L⁻¹) vs. 3.46 μg L⁻¹ (2.80–4.27 μg L⁻¹) for the proposed method and ICP-MS, respectively.

7.3.5 Thorium. A high-throughput, “dilute and shoot”, method for the determination of low concentrations of radionuclide ²³²Th in urine using ICP-QMS was developed.²²⁹ Urine samples were diluted 1 + 9 with an aqueous solution containing 100 ng L^{−1 233}U, used as IS, in 2% (v/v) HNO₃. A rinse solution, containing 0.025 mol L⁻¹ oxalic acid and 5% (v/v) HNO₃, was employed in place of more commonly used HF, to eliminate the problematic memory effects of Th in the sample introduction system. The LOD was determined using Taylor’s method, which involved calculation of the SD of repeated measurements of a blank urine and urine spiked at low concentrations approaching zero (0.5, 1.0, 1.5 and 3.5 ng L⁻¹) and was comparable to SF-ICP-MS methods (0.77 ng L⁻¹). Acceptable accuracy and precision were demonstrated with recoveries from 93.5% to 98.0% and RSDs <6%, estimated from six urine samples from two separated urine pools spiked with known amounts of ²³²Th (from 15 ng L⁻¹ to 800 ng L⁻¹), and agreement with target values, typically within 4%, for urine-based EQA specimens.

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

The review article by Peng et al.²³⁰ summarised sample preparation and determination methods for Hg and its species (Hg^II, CH₃Hg, C₂H₅Hg) in traditional Chinese medicine. The current mainstream methodology for this task is the combination of separation and elemental detection, with HPLC-ICP-MS and LC-AFS being the prevalent techniques and the implementation of rarely available (due to high costs) instruments being also reported. With advanced SR light sources, the chemical forms of Hg can be determined directly in situ using microscopic analysis techniques, including SR-XANES and micro-zone XRF analysis. The review focused also on the exploration of the pathways of Hg, as an environmental pollutant, for reaching medicinal plants, the transformation mechanisms of its chemical state and the factors influencing these processes at a molecular level.

Two papers reported the analysis of radiopharmaceuticals by HPLC-ICP-MS. They did not focus on the determination of elemental impurities (EIs), but investigated the stability and separation of tracers. The aim of the study by Wallimann et al.²³¹ was to demonstrate that ICP-MS combined with RP chromatography and SEC can be used as a surrogate method for radio-HPLC to investigate the in vitro stability of selected metal conjugates. They assessed the stability of ¹⁷⁵Lu- and native Ga-labeled biomolecules, such as prostate specific membrane antigen, using these chromatographic separations and ICP-MS. The stabilities of the non-radioactive biomolecules were equal to those reported in the literature for the corresponding radioactive ones. For the in vitro serum stability assays, the analysis by SEC-ICP-MS required only a dilution of the mouse serum sample after 24 h incubation with the labelled biomolecules. Because no purification procedure such as immunoprecipitation was necessary, the risk of losing information about any metal-containing metabolites that might not be quantitatively extracted could be minimised. In another paper,¹¹⁹ speciation analysis was conducted for novel and established Tc radiopharmaceuticals by means of RP HPLC-ICP-MS and RP HPLC-ESI-MS, to enable direct identification, characterisation, and quantification. Due to a lack of Tc standards, on-line calibration of transient Tc signals was achieved via an isobaric dilution analysis (a calibration tool where the sample with the element of interest is spiked with a solution of a different element, which shares at least one isobaric isotope and shows similar chemical properties resulting in a similar response) in an on-line post-column dilution approach. In this study, four commercially available as well as two experimental tracers were characterised. The desired tracers could be separated from Tc-based impurities in the radiopharmaceuticals and their identity was confirmed.

The combination of element specific detectors with widely used separation techniques (HPLC) in pharma analyses can lead to results with significantly lower LOD and LOQ compared to HPLC-UV-VIS and provide suitable accuracy. To investigate the viability of employing HPLC-ICP-MS for quantifying drug-related compounds in the formulations employed in clinical treatments (injections), two drugs containing elements detectable by ICP-MS (oxaliplatin – Pt and ioversol – I) and their related compounds were chosen as models.²³² The chromatographic separation was carried out with a RP column. For the separation of oxaliplatin and its related compounds, an isocratic elution was applied using 0.1% (v/v) TFA in water as mobile phase. For the separation of ioversol and its related compounds, a gradient elution program with ultrapure water (eluent A) and ACN (eluent B) as mobile phase was applied. The developed HPLC-ICP-MS methods were validated according to the ICH Q2(R1) guidelines. The quantitation accuracy of HPLC-ICP-MS closely matched that of the standard HPLC-UV approach with LOQs lower by one or two orders of magnitude. Thanks to the structure-independent detection capability of ICP-MS, the quantification of target related compounds was readily accomplished through the utilisation of element standards, eliminating the necessity for standards containing the specific compound. However, it is crucial to note that achieving precise quantification required prior determination of the analyte of interest’s structure. Moreover, it is important to acknowledge that this study demonstrated the applicability of HPLC-ICP-MS solely to quantifying drug-related compounds containing elements detectable by ICP-MS. For related compounds lacking detectable elements, quantitative determination is not feasible.

In the area of elemental impurities in drugs the methodology of ICP-OES or ICP-MS determination and sample preparation by MAD with acid mixtures is an established standard. Only two papers are mentioned here: the first²³³ describing a new ultrasound-assisted microextraction method using a cup-horn sonoreactor for EIs determination in tablets by ICP-MS and the other⁷² focusing on greening MAD method for EIs measurements by ICP-OES in active pharmaceutical ingredients (APIs). Three different acid mixtures (HNO₃, 3 HNO₃ : 1 HCl, and 9 HNO₃ : 1 HF) were compared by Noronha et al.²³³ and addition of a small amount of HF provided recoveries closer to 100% for Pd. The developed procedure was validated with spike recoveries at three levels ranging from 85% to 120% and RSDs below 10% were obtained for 22 out of 24 EIs recommended in US Pharmacopeia Chapter 232 (except Ag and Pt) with satisfactory linearity (R² > 0.99) and good LODs and LOQs. Results from the new ultrasound-assisted microextraction and a reference procedure based on MAD were compared resulting in statistically equivalent concentrations. The method was applied to 37 samples of antidepressants. Many analytes were below the respective LOQs for both sample preparation procedures, except for As, Ba, Co, Cr, Cu, Mo, Ni, Pb, Se and V. The authors concluded that, since their procedure was performed at room temperature and pressure, it was relatively safer in comparison with MAD. A novel MAD method using only H₂O₂ was proposed for oxidising APIs in hypertension treatments,⁷² which involved sample masses from 100 mg to 500 mg mixed with 6 mL of 50% (w/w) H₂O₂. The efficiency of the digestion methods was evaluated by determining the RC by ICP-OES. The proposed methodology was suitable to be used in the QC of studied APIs and was in accordance with criteria in US Pharmacopeia Chapter 233. The MAD provided digests with low residual acidity in comparison with those obtained using conc. HNO₃. A key advantage of the procedure was that its waste primarily consists of water and oxygen. On the other hand, digestion only with H₂O₂ may not be sufficient for more complex matrices.

Papers published in this Update period reveal that EIs in food supplements attract almost similar attention than in drugs. In comparison with drugs, where methodologies are recommended by pharmacopoeias, several less common instrumental techniques were applied to food supplements. Total reflection XRF was applied to the quantification of Ca, Cr, Cu, Fe, K, Mn, P, Se and Zn in dietary supplement samples prepared by MAD and suspension preparation.¹⁰³ Two TXRF spectrometers were compared and the obtained results were collated to the conventional way of analysis by ICP-OES including MAD. Accuracy and precision were similar, LOQs by TXRF ranged from 0.01 mg per pill to 1 mg per pill for analysed elements. Three dietary supplement samples with different sets and concentrations of elements were analysed. It was shown that suspension preparation of samples using ethylene glycol might be a useful approach as it was less time consuming and cost efficient. Ratnaraju et al.²³⁴ implemented PIXE as a method for the elemental analysis of traditionally used medicinal plants in India. The mass fractions of 17 elements (Br, Ca, Cl, Cr, Cu, Fe, K, Mn, Ni, P, Rb, S, Se, Sr, Ti, V, Zn) were determined in most of the medicinal plants. The collected samples underwent a cleaning process, then they were dried at 70 °C, grounded in an agate mortar, sieved through a 100-mesh sieve and pressed into pellets. To ensure the accuracy of the experiment, a CRM (NIST SRM 1515 apple leaves) was analysed with acceptable results. The investigated medicinal plants were found to be abundant in one or more individual elements, which are related to their therapeutic value for treating various diseases and ailments. This methodology did not require sample digestion, on the other hand potentially toxic metals were not studied. The development and optimisation of an analytical method based on LA-ICP-MS for the analysis of 12 multivitamin dietary supplements²³⁵ compared two calibration strategies for As, B, Co, Cr, Cu, Mg, Ni, Na, Pb, S, Se and Zn: (1) using increasing mass fractions of cellulose powder fortified with multiple elements of interest and (2) matrix matching using a multivitamin CRM (NIST SRM 3280) in addition to the multi-element cellulose powder. Prior to analysis samples were ground and mixed with a cellulose powder, containing the elements Bi, Ge, Ir, Rh, Sc and Te as IS, and the resulting powder was pressed into pellets. Results were compared with ICP-MS data after MAD and two sets of dilutions (5-fold for trace elements and 500-fold for the major elements), which was time consuming. Laser ablation (193 nm excimer-based laser, optimum energy output 0.5 J cm⁻², repetition rate of 30 Hz and carrier gas flow rate of 800 mL min⁻¹) was considerably quicker and simpler. The matrix-matched calibration resulted in a mean recovery between 81% and 119% and better recoveries for Fe and S compared to fortified cellulose standards. Moreover, Pavlovich-Cristopulos et al.²³⁶ measured not only the total amount of As, Cr, Cu, Fe, Mn, Pb, V and Zn in 11 samples of zeolite clinoptilolite dietary supplements by portable XRF, following the EPA Method 6200; but also their oral bio-accessibility, using an in vitro test. The results showed that Fe was the most abundant essential (1.1% m m⁻¹) and As the most abundant toxic element (118.7 mg kg⁻¹). A physiologically based extraction test (PBET) was used to identify the bio-accessibility of metal(loid)s, that, in general, was low to intermediate. The mean bio-accessible fraction was highest for Mn (30.6%) and reached approximately 15% for As and Cu, whereas Cr, Fe, Pb, and V were less bio-accessible (<6%). Bio-accessibility was higher in the gastric phase than in the intestinal phase for As, Fe, Mn, Pb and V, whereas the contrary occurred for Cr and Cu.

9. Applications: foods and beverages

In this section, papers highlighting progresses in the area of food and beverages analysis are discussed. In addition, the technical details and most relevant findings of a number of other papers in this area are summarised in Table 3, complementing those reported in Tables 1 and 2. Continuing with the trend seen over the last few years, this subject area has seen an increase in the number of papers including chemometric studies investigating the origin, authenticity or provenance of foodstuffs using data generated by multielement spectroscopic methods. In last year’s Update,¹ 24 papers were included in a newly introduced table for this subject, whereas this year 71 papers appear in Table 4 with many others appearing in the main text of the Update and Table 3 having elements of chemometrics as part of the study. A significant number of papers looked at food relating to sustainable diets, including wild foods, milk alternatives and insect proteins, and how the impact of these diets can affect mineral contents of breast milk.

Table 3 Food and beverages

Analytes	Matrix	Technique	Study aim, procedure and comments	Ref.
As species	Fungi	ICP-MS	As speciation was carried out on Agaricus blazei Murrill and Tricholoma matsutake mushrooms. Speciation was performed using anion exchange and cation exchange chromatography. Using this separation technique, solvent use was reduced compared to that of other methods. AB, iAs, and MMA were the predominant species of As in A. blazei samples, accounting for 42.7–88.8%, 1.59–26.2%, and 0.74–1.12% of the extractable As, respectively. AB and DMA were the main forms of As in T. matsutake samples, which accounted for 88.9–97.9%, and 1.79–3.80% respectively. As a health risk from exposure to iAs, T. Mastusake did not pose a particular risk, but there was some concern about the A. Blazei samples	352
As species	Milk	HPLC-ICP-MS	Conventional cow’s milk and plant based “milks” underwent As speciation. The method was validated using spiked samples, with good recoveries between 85–108%, and CRM materials. Beside tAs, As species measured were As^III, As^V and iAs. High levels of all As species were seen in soy and rice based “milk”; surprisingly soy had higher As levels than those of rice. Vetch “milk” was also analysed and had high levels of tAs, iAs and As^V suggesting that this material needs further investigation	353
Hg & Se	Fish	HPLC-ICP-MS	Selenium and Hg speciation was performed on fish and fish products to evaluate Health Benefit Values (HBV) in these products. Analysis of fillets from four species of fish and seven processed products, such as fish roe, crabsticks, fish noodles and other processed fish products, were carried out. The Se species SeMet and SeMeSeCys were the most abundant Se compounds in all samples and MeHg was the most abundant Hg species. The HBVs were positive in all cases, showing the ratio of Se : Hg was such that in all cases the consumption of these products would be beneficial. The authors also noted that the HBVs in processed products and roe were significantly higher than those of fish fillets, indicating these would be safer to consume if Hg was of particular concern	182
K, Na	Coconut water	MIP-OES	Fresh and processed green and mature coconut waters were analysed following MAD. Recoveries were 95% & 103% and LODs were 1.77 mg kg^−1 and 2.31 mg kg^−1 respectively. Processed samples were found to have typically lower levels of these minerals, which for K is important as a nutritional mineral. A simple HCA analysis was able to show good classification of the sample types	354
Li	Whole diet	ICP-MS	In this extensive study 1071 food samples were tested for Li content. Approximately 10% of samples analysed had Li levels lower than the LOD; in samples with measurable Li the contest in different food groups were shown to be in the order vegetables (leafy > bulbous) > fructose > leguminous > egg whites > root vegetables > milk products > egg yolks > meats. For most sample groups the content was below the exposure limit (2 μg kg^−1 day^−1), though bulbous vegetables and fructose solano vegetables did present levels higher than this in some samples	355
Se	Mushrooms	HPLC-ICP-MS	In this study Se-enriched Lingzhi mushrooms were cultivated in substrate containing varying levels of sodium selenite. The fruit were shown to accumulate Se compounds, particularly SeMet, whereas the mycelium was shown to have lower levels of SeMet tending instead to accumulate elemental Se as NPs, marked out by a red colouration. The enrichment of Se was shown to be around 20 times in the fruit, but the mycelium showed a 2700 times enrichment of Se	356
Se	Plant based beverages	ICP-MS	Total Se and bioavailable Se compounds were assessed in a range of plant based beverages (PBB), including nut, oat and rice based products. Bioavailable Se was assessed by processing the samples with oral, gastric and intestinal digests. The undigested portion of these samples and the raw beverages were analysed by ICP-MS. The method was shown to have an LOQ of 4 μg kg^−1. The accuracy, assessed with spiked samples and CRMs, was between 95–111%. Se contents were between 4 and 226 μg kg^−1 and bioavailability was in the range of 64–96%. The authors suggest that only one sample of PBB produced from organic cashew nuts could provide sufficient Se to meet daily nutritional requirements	357
Se species	Fish	HPLC-ICP-QQQ-MS	A variety of commonly consumed fish in Thailand were prepared by enzymatic digestion and analysed by HPLC-ICP-QQQ-MS for Se speciation. Samples were cooked under different methods to understand the effect of different cooking techniques. The Se content of the fish was found to vary considerably across species, with organic Se most seen as SeMet and SeCys, but some species showing high levels of iSe. Boiling and frying appeared to increase levels of SeMet in some samples	358
TiO2	Foods	XRF	Used as a food additive and brightener, TiO2 is common in food manufacture, but its safety has been questioned, particularly when added as NPs. In this paper XRF is cited as a potential tool for detection and quantification of TiO2. It was suggested it was a useful measurement tool as it can dectect TiO2 at 0.25 mg kg^−1 as TiO2 as opposed to just Ti which would be the case with ICP-MS	359
Various	Coffee	Various	In this meta review of 23 papers the impact of elemental exposure from coffee was assessed. The key elements included were Cu, Zn & Fe, but it was concluded that the levels of these elements are with tolerable levels and show no risk to consumers	360
Various	Rice	LIBS	LIBS was used to assess degree of polishing on rice samples. The key spectral bands were shown to be Ca, K, Mg, Na, Rb and Si. PLS-R modelling was utilised to assess degree of polishing with promising results	361
Various	Wild fruits	ICP-OES	Raspberry (red & yellow), blackberry, wild strawberry, white-mulberry, apple, plum (red & green), currant, jujube, fig, cranberry, rosehip & medlar were analysed for a variety of macro, micro and toxic elements by ICP-OES. The study saw that all of these fruits were good sources of Ca, Cu, Fe, K, Mg, Mn, P & Zn. Levels of toxic elements were all found to be at safe levels in the fruits though the authors do call out higher levels of Al, As & B in strawberry, mulberry and currant	362
Various (14 elements: As, Cd, Co, Cr, Cu, Fe, Hg, Mn, Ni, Pb, Se, Tl, V, & Zn; 13 REE: Ce, Dy, Er, Eu, Gd, Tb, Ho, La, Nd, Pr, Sm, Tm & Yb)	Eggs	ICP-QMS	Exposure limits for 3 age groups were established from organic, barn and caged hens' eggs. The daily intake values were lower than the respective safety values for all elements suggesting that there is no risk from eating hens' eggs. It was also noted that there was not significant difference in contaminants between the different farming methods	363
Various (38)	Baby food	ICP-MS	159 samples of ready to eat baby foods from Gran Canaria. Fruit, meat and fish purees were included in the sample set. Samples underwent MAD followed by ICP-MS. As expected As and Hg were found to be highest in the fish-based samples. The team built a typical model diet and compared named brands against own label products. Consistently own brands had higher levels of metals, levels of Cr, Mn and Mo were found to be higher than the adequate intake levels and above the acute hazard levels for Mn and Mo for all intake assessments	364
Various (Ag, Al, As, Ba, Be, Cd, Ca, Co, Cr, Cs, Cu, Fe, K, Li, Mg, Mn, Mo, Na, Ni, P, Pb, Rb, S, Se, Sr, Ti, Tl, U, V and Zn)	Lentils	ICP-MS	This study was carried out on conventionally farmed as well as organic lentils. Total elemental composition samples underwent MAD followed by ICP-MS analysis. For bioavailability assessment, 1 g of cooked sample was placed into a synthetic digestive solution and incubated for 2 h at 37 °C followed by centrifugation. The resultant supernatant was diluted and analysed by ICP-MS. Mineral content varied between the two farming methods, with Ca, Cu, Mg and Zn found to be higher in organic lentils, but the study did not show significant differences bioavailable between the 2 types of production	365
Various (Al, As, Ba, Cd, Cr, Co, Hg, Mo, Ni, Pb & Sb)	Plant based drinks	ICP-MS	Milk alternatives composed of rice, oat, cashew, soybean, peanut, cocoa, coconut, almond, or mixes therein were prepared by water bath with sonication in HNO3 prior to analysis. A bio-accessibility study was also completed using salivary and intestinal digestion steps. Results were found to be similar in products with the same base materials, such as As in rice based milks and Ni in cashew based drinks. Aluminium, As, Cd, Co, Cr & Pb were found to have acceptably low levels, but Ba, Mo & Ni were found in high concentrations, and Ni also showed a bio-accessibility of >50% which may lead to over exposure	366
Various (Al, As, Cd, Cr, Cu, Fe, Ni, Pb, Se, Sn, V & Zn)	Insect protein	ICP-MS	In this study the content of minor and toxic elements in protein derived from black soldier fly was studied. With the growing demand for ecological foods and as the result of health scares, alternative proteins have garnered much interest. Samples were digested using MAD with HNO3 and H2O2 and analysed by ICP-MS. Micronutrients were found to be present in the order of Fe > Zn > Cu > Ni > Se > Cr. Toxic elements were found to be at levels lower than the EU limits	367
Various (Al, B, Ca, Cu, Fe, Mg, Mn, Ni, P & Zn)	Coconut water	MIP-OES	MIP-OES was chosen as the analytical technique, because it uses a nitrogen plasma, whose supplies were more reliable than those of argon. Samples underwent MAD prior to analysis. The method was validated using a CRM which returned recoveries of 87–116%. From a nutritional perspective, coconut water is a useful source of Mn and a significant source of Ca, Mg and P. More interestingly it was found that the consumption of 200 mL of water from mature coconut could give 79% of the daily tolerable intake of Ni, which warrants further investigation	368
Various (Al, Ba, Ca, Cd, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, Pb, Sr & Zn)	Yerba mate	ICP-OES	In this study a variety of “green” digestion methods were compared to a standard MAD preparation technique. It was found that a preparation containing 5% HNO3 which was treated with ultrasonic for 10 minutes at room temperature gave good recoveries of 97–105% measure against reference preparation and CRM recoveries. A bioavailability assessment was also performed which gave about 50% availability for most elements. Given the total content and bioavailability, it was concluded levels of nutritional elements in yerba mate were not significant toward RDI's	369
Various (Al, Ba, Ca, Co, Cr, Cu, Fe, K, Mg, Mo, Na, Ni, P, Se & Zn)	Plant based yoghurts	ICP-MS, ICP-OES	A variety of dairy-free yoghurt products underwent both wet digestion in HNO3 digestion and bio-accessibility digestion as per the INFOGEST method. Samples were then analysed by ICP-MS and ICP-OES. Statistical analysis (PCA) showed that categorisation of samples was possible. The overall conclusions were that total and bioaccessible levels of minerals vary widely and more work is required on this relative new product	370
Various (As, Ag, Ba, Co, Cu, Ni, Pb, Se, V & Zn)	Fruit juice	ICP-MS	Samples of clear fruit juices were analysed by ICP-MS for a range of elements in a simple dilute and shoot matrix. By adding ethanol to samples in the range of 1–5% many matrix effects were compensated for, either over or under expression of signal. The method was optimised to include 5% ethanol in the calibration and the method was validated by comparison against samples that had undergone MAD as well as spiked recovery studies which gave recoveries of 95–109%	371
Various (As, Cd, Co, Mn, Mo, Pb, Se, Sr, U, V & Zn)	Sea products	XRF, ICP-QQQ-MS	XRF and ICP-QQQ-MS were compared to establish the practicalities of handheld XRF for safety assessments of seafoods and seaweeds. The study had limited success, but some correlations were seen for some elements such as As, Mn, Sr & Zn, the authors suggest more work may show this to be a useful tool for in the field analysis	372
Various (As, Cd, Cu, Fe, Mn, Pb, Se & Zn)	Tree nuts and related oils	ICP-QQQ-MS	Whole samples and cold pressed oils of samples of tree nuts, namely, camellia seeds, walnuts, Acer truncatum, shiny leaf yellow horn seeds, pecan seeds, and Brazil nuts were analysed in this study to understand the migration of metals from seed to oil. Seeds and corresponding oils as well as sample residue from the pressing underwent MAD to solubilise the elements of interest. It was found that across all samples the elemental components were typically contained in the non-oil fraction of the extracts. Typically, in the oils the elemental components were found at <2% than that measured in the seed	373
Various (Ca, Mg, Na, K, P, B, Zn, Cu, Fe, Mn, Co, and Ni)	Teff	ICP-OES	Red, white and mixed teff grains were studied for nutritional and toxic elements. Levels of Cr and Co were below the LOD in all sample, though high levels of As, Hg and Pb were seen in all samples, which will warrant further investigation	374

---

Table 4 Applications related to origin and authenticity of food and beverages

Analytes	Matrix	Measurement technique	Statistical analysis	Purpose	Study aim, procedure and comments	Ref.
Ca, Fe, K, Mn & Zn	Teff flour	ED-XRF	Common Dimension Analysis (ComDim). Multiple Linear Regression (MLR)	Adulteration	Teff flour is a gluten free grain, so is highly valued, but often adulterated with cheaper cereals such as rye, wheat, oat and rice. By using ED-XRF and ComDim statistical analysis it was shown that adulteration with other grains could be seen with a LOQ of 6% adulteration	375
Ca, Mg, Na, K	Milk	LIBS	LDA, SVM, LR, GB, MLP	Species	LIBS analysis of 1296 raw milks and 683 lyophilized milks from different animal species (cow, goat & sheep) was carried out and a variety of algorithms applied to identify the origin of the milk. Prediction accuracies between 92.8% and 95.5% were obtained for the raw and dried milks, respectively, and classification accuracies of 87.6% and 92.9%	376
Ca, K, Mn – botanical indicators, Na, Mg, K – geographical indicators	Honey	ICP-MS	PCA	Origin	173 honeys of 13 floral types (acacia, fir, spruce, linden, chestnut, lavender, coriander, thistle, honeydew, rosemary, sage, euphorbia and ziziphus) from 5 regions (Slovenia, Croatia, Bulgaria, Turkey, and Morocco) were analysed by ICP-MS. PCA was able to identify 5 species (floral, linden, forest, acacia, and chestnut) from PCA profiles in Slovenian honeys as there were sufficient samples to have a usable dataset	377
Al, Ba, Bi, Cd, Co, Cr, Cu, Fe, Li, Mn, Mo, Ni, Pb, Rb, Se, Sn, Sr, V & Zn	Cereal bars	ICP-MS	PCA, CART (classification and regression trees), and LDA	Authentication/gluten free	120 cereal bars from the Argentinian market were analysed to assess the validity of gluten free label claims and of additions such as chocolate, yoghurt and fruit. MAD and ICP-MS were used for the analytical work. LDA was the most successful chemometric tool achieving 92% accuracy in determination of gluten free claim and type of addition	378
Al, As, B, Ba, Be, Ca, Cd, Co, Cr, Cs, Cu, Fe, K, Mg, Mn, Mo, Na, Ni, P, Pb, Rb, Se, Sn, Sr, V, & Zn, δ^13C & δ^15N	Ginseng	ICP-MS, EA-IRMS	PCA, LDA SVM, RF, FNN (feedforward neural network)	Origin	American ginseng from the USA, Canada and Northeast provinces and Shandong province in China were analysed to establish a model for identifying the region where the ginseng was grown. Discrimination of origin was established with C/N ratio and B, Ca & Fe levels being key identifiers. Using FNN modelling 100% accuracy was obtained for the identification of origin	379
Various (57)	Pig liver and diaphragm	ICP-MS TDA-AAS (direct Hg analyser)	HCA, SIMCA, NCA (neighbourhood component analysis)	Authenticity/GMO free/supplemented status	Analysis of liver and muscle tissue for 58 elements followed by chemometric analysis showed that pork tissues (liver and diaphragm) could be used to identify animals that had had conventional, GMO-free (with and without supplements) diets and non-PDO animals. SIMCA analysis gave 90–100% accuracy in identification. This tool could be useful for authenticity testing by use of by-products in the production of Parma ham	380
Various	Chinese Prickly Ash (CPA)	XRF	PCA, LDA, KNN, PLS-DA, SVM, RF	Origin/variety	Red and green CPA samples from 5 different provinces in China. XRF analysis was performed and the obtained data analysed by a variety of techniques. PCA showed overlap of groups, but with further analysis with the other listed techniques classification of 100% was obtained. Key identifiers were: C, Ca, Fe, K, Mg, O, P, S, Si & Sr	381
Various (36)	Cinnamon	ED-XRF	PCA, SIMCA	Adulteration	Ceylon cinnamon is highly valued but is often mislabelled or adulterated with cheaper cassia cinnamon or other plant materials. Al, Cl, Co, Cs, Cu, Fe, K, Mo, Nb, Nd, Ni, P, S, Se, Si, Ti, V & Zr were shown to be the key markers in Ceylon cinnamon. It was found that 58% of the market samples obtained for the study were found to adulterated in some form, which impacted the models. The team used HS-GCMS and TGA analysis to support findings from the XRF, which could be a promising tool in the screening of cinnamon samples	382
Al, As, Ca, Cd, Cu, Fe, K, Mg, Mn, Na, Ni & Zn	Honey	ICP-OES	PCA, PLS-DA	Origin	Fifteen honey samples from each of 3 regions on the Malaysian Peninsular were collected and analysed for 12 elements by ICP-OES. PCA and PLS-DA were used to determine whether these models could identify the region of the honey collected. The models gave a discrimination of 66% and 64% respectively, but the researched suggest the PLS-DA is a better model due to the limited sample numbers. The relevance of the markers was shown to be in the order of Mn > Fe > Cu > Na > Ca > Ni > As > Cd > Zn > Al > K > Mg	383
Ca, Cd, Cr, Cu, Fe, K, Mg, Mn, Na, Pb & Zn	Raspberry	FAAS	Factor Analysis (FA), Cluster Analysis (CA)	Species, origin	The team discussed the nutritional value of raspberries, and their importance in folk medicine and diet. The data analysis showed that both FA and CA could be used to identify species/cultivar as well as origin, but with 1 species (R. chamaemorus) neither model could give geographical separation	384
Ba, Be, Co, Cr, Cs, Cu, Fe, Ga, Li, Mn, Mo, Ni, Rb, Sb, Se, Sr, Ti, V & Zn, δ^13C, δ^2H, δ^15N & δ^18O	Salep orchid	ICP-MS, IRMS	PCA, HCA	Origin	Fifty-seven salep orchid samples were collected from 15 locations in 5 regions of Turkey. It was shown that the data from the IRMS was more significant than that of the elemental profile. Using the HCA and PCA modelling some discrimination was seen, with areas of overlap for some regions. The team propose further work to further define these regions	385
Rb, ^87Sr : ^86Sr	Saffron	MC-ICP-MS	Kruskal–Wallis test	Origin	The isotope ratio of Sr and analysis of Rb were measured in 27 saffron samples from 5 countries (India, Iran, Italy, Spain and Morocco). A simple statistical analysis showed that some differentiation was seen, with different regions in India being clearly defined as well as Moroccan and Indian samples	386
Al, Ba, Ca, Mg, Sr, Cu, Fe, Mn & Zn	Soybean pastes	ICP-OES	PLS-DA	Origin	Soybean pastes from S. Korea and China were studied to determine whether origin could be established using the listed techniques. Mg and Sr were found to be effective markers for this assessment and the model exhibited a classification accuracy of 93%. The authors suggest that use of Korean sea salts may be key in this discrimination	387
Ag, Al, B, Ba, Bi, Ca, Cd, Co, Cr, Cu, Fe, Mg, Mn, Mo, Ni, Pb, Sr, Ti, V & Zn	Sweeteners	ICP-OES, XRD	ANOVA	Adulteration	A variety of natural and manufactured sweeteners were assessed. In the study it was possible to see that key elements could be used as specific markers but more work is required to build this into a model	388
Various (up to 44)	Wine	ICP-MS	MFA (Multiple Factor Analysis)	Vintage	Wines from 15 US vineyards were analysed for comparison of cross-vintage variability. Most elements maintained strong correlations including Al, As, B, Ba, Ca, Cs, Co, Eu, Fe, Ga, K, Mg, Mn, Na, Ni, P, Rb, Sr, and V. It was also noted that between vineyard variability is more considerable that vintage to vintage	389
As, Ca, Cl, Cu, Fe, K, Mn, Ni, P, S, Sr, Th, Zn, Zr	Prawns	Handheld XRF	PCA, nonmetric multidimensional scaling, CDA, RF	Origin, provenance	Handheld XRF for establishing provenance and origin on cooked and raw prawns from 2 regions in Australia. Due to the wet nature of the matrix correction factors were required, but even so a good correlation was established and accuracy of >80% achieved. The authors discuss the need for further work to overcome the known issues of the technique	390
Various (53)	Honey	ICP-MS, INAA	HCA, PCA	Origin	Honey from North Dakota, South Dakota and Montana were assessed in this study. The key components were B, Ba, Ca, Cl, Cs, K, Mn, Rb & Sr and some discrimination between regions was established, indications are that Pearson correlations may be a useful approach to honey provenance	391
Al, Ba, Ca, Cd, Co, Cr, Cu, Fe, Fe, K, Li, Mg, Mn, Na, Ni, Sc, Sr, Ti, V & Zn	Fish	LA-ICP-MS	LDA, RF	Provenance	Wild and cultured fourfinger threadfin fish were analysed by LA-ICP-MS and the obtained elemental profiles processed using LDA and RF models to establish whether the fish came from wild or cultured stocks. The study showed 100% accuracy in discrimination of wild and cultured fish with a small number of analysed elements, namely K, Mg and Na	392
Ag, Al, As, Ba, Be, Bi, Ca, Cd, Co, Cr, Cu, Fe, Ga, K, Li, Mg, Mn, Mo, Na, Ni, Pb, Rb, Se, Sr, Te, Tl, U, V & Zn	Pepper	ICP-MS	PCA, HCA	Origin	Black, white and green pepper (Piper nigrum L.) and cayenne pepper (Capsicum annuum L.) underwent chemometric analysis following ICP-MS analysis. Unsurprisingly cayenne pepper was easily discernible from the other peppers as a different species. White pepper was clearly different from black and green pepper, which had similar profiles. The authors propose further work to include region, season and processing	393
Al, Ba, Ca, Cd, Co, Cr, Cu, Fe, K, Li, Mg, Mn, Na, Ni, Sc, Sr, Ti, V & Zn	Lotus root	ICP-MS, ICP-OES	ANOVA, PCA, stepwise LDA	Origin	The team gathered 147 lotus root samples from 9 Chinese provinces for this study. All elements analysed were shown to have significant differences under 1 way ANOVA analysis and a discrimination rate of 99.3% was obtained from stepwise LDA training set, thus demonstrating that the range of analytes and statistical analysis may be suitable for determination of origin of lotus root	394
Various (69)	Pu-erh tea	ICP-MS	OPLS-DA, PCA	Origin	Pu-erh teas (19 samples) from 3 regions were analysed for a wide range of elements. OPLS-DA analysis gave good discrimination of region with 7 elements, Ba, Ce, Mo, Nd, Sr, Tm & V being key markers	395
Various	Coffee beans	LIBS	PCA	Authentication	Eight varieties of coffee were analysed by LIBS and the spectra underwent PCA analysis. Nine primary wavelengths represented C(I), Ca(II), Fe(I), H(I), K(I), Mg(II,I), and O(I). Discrimination between sample types was established and suggests that LIBS may be a useful tool in coffee authentication studies	396
Various (45), δ^13C and δ^18O	Wines	ICP-MS, IRMS	Stepwise DA, ANN, PCA	Origin	Thirty-six wine and corresponding grape samples were collected from different vineyards in Deqing County, Yannan Province, China, with 18 samples collected over 2 years. Both wine and grape samples were able to show regionality of the samples with an 89% accuracy in grape samples and 95% accuracy with wine samples. Strontium was shown to be a key indicator and was consistent across vintages	397
Various (60)	Chicken meat	ICP-MS, ICP-OES	OPLS-DA	Origin	Domestic and imported chicken breast and drumstick meat was analysed in the study. Thirty samples were examined in each group and OPLS-DA analysis carried out on the obtained analytical results. Classification accuracy of 100% was obtained with both the breast and drumstick meat	398
Various (23)	Crab	ICP-MS	HCA, stepwise LDA, ANOVA	Origin	A health assessment was carried out on 19 of the analysed elements (Ag, As, Ba, Cd, Co, Cr, Cs, Cu, Fe, Ga, Hg, Mn, Ni, Pb, Rb, Se, U, V & Zn) and the authors concluded that no significant health risks exist in moderate consumption of wild mitten crabs. Samples from 3 regions were clearly distinguished with Ca, Co, Cu, Fe, Mn, Na, Pb & Zn being the key indicator elements with the models giving an accuracy and discrimination of 100%	399
Ca, Cu, Fe, K, Mg, Mn & Zn	Coffee	FAAS	Independent Component DA (IC-DA), PLS-DA, PCA	Species, origin	In this initial study Canephora and arabica coffees were analysed for 7 elements by FAAS to establish whether this simplified analysis could be used to identify the different species. Though some differentiation was seen, there was still some overlap of groups. The authors propose a larger sample set to improve the models	400
Various (29), δ^13C, δ^2H & δ^18O	Pork	ICP-MS, FAAS, IRMS	PCA, LDA, ANOVA	Origin	Pork loin meat from Romania studied to distinguish it from meat from other EU countries (Spain, Germany & Hungary). LDA model gave 91% identification with δ^2H, K, Pd & Rb being the key markers. It was noted that samples from the west of Romania had similar profiles to Hungarian samples due to similarities in geography	401
Al, Ca, Co, Cr, Cu, Fe, K, Mg, Mn, Mo, Na, Ni, Sr, and Zn	Mandarin	MP-AES	LDA, KNN, SVM, RF	Origin	Mandarin fruit from 3 regions and 2 cultivars were analysed in this study using MIP-AES. Seven elements (Al, Cr, Fe, K, Mg, Mo, and Zn) were shown to be the key characteristic elements though the initial PCA screening and SVM was shown to be the most effective model with classification rates of 95% correct	402
Various	Olive oil	LIBS	KNN, PCA, ANN, PLSR, SVM, CLPPS (Continuous Locality Preserving Projections), LR, GB	Authentication, adulteration, origin	Samples of Greek olive oils were adulterated with various cheaper oils, pomace, sunflower, soybean and corn. For discrimination of adulterated oil, almost 100% accuracy was achieved, and identification of the adulterant oil accuracies of between 92% and 99% were achieved	403
Various (60)	Onions	ICP-MS, ICP-OES	OPLS-DA, PCA	Origin	Determination of origin between Korean and Chinese onions was carried out. Key elements were identified as Al, As, Ba, Cu, Fe, Ge, K, Mn, Mo, Na, Nd, P, Rh, Sr, V, W & Zn. Classification accuracies were found to be 100% and 95% for the calibration and validation sets	404
Ag, Al, As, Cd, Co, Cr, Cu, Fe, Mn, Ni, Se, Sr, V & Zn	Rice	ICP-MS	LDA	Origin	Rice samples from USA, Thailand, India, Pakistan and Bangladesh were analysed for 14 elements and traceability of origin assessed by LDA. It was noted that levels of As were higher in US produced rice compared to the Asian harvested rice. The model gave 100% classification between USA and Thai rice though the separation was less clear between rice from South Asia, possibly due to similarities in soil composition, climate and farming techniques	405
Various	Beans	XRF	PLS-DA, PCA, LDA	Origin	Samples of 24 different genotypes of the species Phaseolus vulgaris L. were taken from 2 separate regions. Using XRF and statistical analysis it was possible to show clear differences in the region using 12 different elemental markers (Br, Ca, Cl, Cu, Fe, K, Mn, Ni, P, Rb, S & Zn)	406
Ag, Al, As, iAs, Ba, Be, Bi, Cd, Co, Cr, Cs, Cu, Fe, Ga, Hg, Li, Mn, Mo, Ni, Pb, Rb, Se, Sr, Tl, V & Zn	Rice	ICP-MS, LC-ICP-MS	SVM	Origin	A large study of 670 rice samples from 5 regions in Brazil was carried out and SVM analysis of the obtained results was carried out. Initially 16 elements were used for modelling as they gave results consistently above the LOQ. SVM analysis was performed on the full 16 element and a reduced 9 element data set. The reduction in analytes gave an improvement in classification, achieving 100% accuracy in prediction of region	407
Various	Milk powder	LIBS	PCA-MGD-LR (PCA-mini-batch gradient descent-LR)	Classification	LIBS was investigated as a tool to rapidly determine whether milk powder is a whole or skimmed milk. Various iterations of modelling eventually gave a robust test giving 99.3% accuracy in 3.42 seconds	408
Various (41)	Wine	ICP-MS, NMR	MWA (Multiway Analysis), DRL (deep reinforcement learning), ANN	Origin	Sparkling wines from 4 different growing slopes were analysed in this proof-of-concept study. The sample number was small (34) and initially the data seen with imbalances, the applied statistical analysis across data obtained from both techniques was able to elucidate good classification with 100% accuracy, thus demonstrating this as a potentially useful tool	409
Ce, Cu, Dy, Er, Eu, Fe, Gd, Ho, K, La, Lu, Mg, Mn, Na, Ni, Pr, Rb, Sc, Sm, Sr, Tb, Tm, V, Yb & Zn	Ginseng	ICP-MS, GC-MS, LC-MS	Multi-forest joint network (MFJN), extreme random forest (ExRF)	Origin	Multiple techniques, including ICP-MS were used to establish the origin of ginseng from 5 different areas in China. Using the ExRF model the team established a system that was able to classify with 97% accuracy in the test set and 95% accuracy in the prediction set	410
Various (37), δ^2H and δ^18O	Grapes	ICP-MS, IRMS	PCA, CA, PLS-DA	Origin, vintage	Thirty samples from 3 regions and 4 vintages (2017–2020) were collected for this study along with soil samples from each vineyard for each vintage. The PLS-DA model achieved 100% in prediction of grape sample origin and vintage	411
Various (71)	Wine	ICP-MS	RF, PCA	Origin, authenticity	Samples of grapes were collected from Crimea and Kuban regions on the Black Sea. These were fermented under laboratory conditions to generate 152 wine samples, which then underwent ICP-MS screening. PCA analysis did not provide sufficient discrimination by RF modelling allowed for 96% accuracy in prediction after training. The key markers were identified as Ag, Ba, Bi, Na, Ni, Rb, Re, Sb, Ti. U and Zn	412
As, Ba, Cd, Co, Cr, Cu, Ni, Pb, Sb, Tl, V, & Zn	Wine	ICP-MS, HS-SPME-GC-MS, UHPLC-QMS	OPLS-DA, PCA, KNN, SVM, RF	Authenticity	Thirty samples of wine were collected from 3 production areas in China and analysed for metals, volatiles and metabolites by a variety of techniques. Individually accuracies were 55–88% when modelling volatiles, 93–99% with metabolites and 71–97% with elemental profiling. Once combined the accuracy was found to be 100% for all algorithms except KNN	413
As, Ba, Be, Ca, Cd, Ce, Co, Cr, Cu, Dy, Er, Eu, Fe, Gd, Ho, K, La, Lu, Mg, Mn, Mo, Na, Nd, Ni, Pb, Pr, Sb, Se, Sm, Tb, Th, Tl, Tm, U, V, Y, Yb & Zn	Jujube	ICP-MS	OPLS-DA, PCA, LDA, MLP	Origin	Jujube (a date-like fruit) samples were collected from 4 regions in China. A total of 167 were analysed in this study. The MLP model gave a 91% prediction accuracy as to the origin of the jujube fruits	414
Various (37)	Fungus	ICP-MS, ICP-OES	OPLS-DA	Origin	Ophiocordyceps sinensis (OS), is a fungus valued in Chinese medicine. Samples came from 4 regions and 96 different batches were analysed. A health assessment found that levels of Cd, Cr, Cu, Fe, Mn, Ni, Pb & Zn were within safe levels. OPLS-DA modelling was able to predict origin with a 90% accuracy	415
Ca, H, K, N, Na & O, CN and C2 molecular bands	Fungus	LIBS	PLSR, SVR, LASSO (least absolute shrinkage and selection operator)	Adulteration	Fritillaria cirrhosa, is sometimes adulterated with cheaper Fritillaria thunbergia. In this study samples were prepared by mixing the 2 types of fungi in different ratios. The combined analysis and statistical operation rendered a correlation of r = 0.9983 across the range of samples demonstrating the ability of these tools to detect adulteration	416
Various (28)	Durum wheat	ICP-OES	PLS-DA	Organic status	Durum wheat samples from organic and conventional sources were analysed in various forms (flour, bran, semolina and seed). Discrimination was seen between the two cultivation types of samples, with an accuracy of between 86–100%. It was noted there was improved discrimination with a reduced range of elements (B, Cd Cu, K, and Se). The method was also able to identify processing type, likely due to distribution of elements in different parts of the grain	417
Al, Ca, Cd, Co, Cr, Cu, Fe, Mg, Mn, Ni, P, Pb & Zn	Wine	ICP-OES	PCA	Organic status	Czech wines from conventional and organic vines were assessed for safety and production assessment by PCA. All wines were shown to be with regulatory limits and therefor classified as safe. There was also a clear differentiation of the 2 types of production with PCA analysis, with organic wines with Al, Cd, Mg, Mn, Ni and Zn being key markers and typically lower in concentration in the organic wines	418
Various (60)	Chicken	ICP-MS, ICP-OES	OPLS-DA, CDA, VIP	Origin	As demand for chicken increases in Korea, there is a greater risk that imports will be sold as locally produced meats. The team used data from ICP-MS and ICP-OES analysis for 60 elements to model the origin of chicken meat. They found 20 elemental markers, and by using OPLS-DA they were able to get a discrimination model with 100% accuracy in identifying Korean chicken against other origins	398
Various (32)	Sausage	ICP-MS, FTIR	PCA	Authenticity	Six varieties of sausage common in the Vietnamese market were dried, digested using MAD and analysed by ICP-MS for a range of elements including REE. Data underwent PCA analysis evaluation, and this showed that the key elemental markers were Ba, Ca, K, Ni and Ti. The elemental data was complemented with the FTIR spectra which allowed for good definitions of individual brands	419
As, Cr, Cu, Fe, Mn, Mo, Ni, Rb, V & Y	Grape juice	ICP-MS	PCA, Cluster Analysis (CA), LDA	Origin	Grape juices from Argentina and Brazil were analysed by ICP-MS and various statistical analyses applied to the obtained data. The 2 origins had similar profiles, with some overlap, but utilising LDA allowed for an 81% accuracy in determination of origin	420
Al, Ba, Be, Ca, Co, Cu, Fe, Li, K, Mg, Mn, Mo, Na, Ni, P, Sr, V, & Zn	Anisodus tanguticus (Maxim.) Pascher	ICP-OES	PCA, LDA, OPLS-DA	Origin	Anisodus tanguticus is used in traditional Tibetan medicine. Samples were taken from 4 provinces in China. Samples underwent MAD and ICP-MS analysis. Discrimination accuracy of 92% was achieved with key elements being Al, Be, Co, Fe, Ca, Ni and Mo	421
Various	Coffee	ICP-MS	SIMCA, one class RF, one class PLS	Authenticity	Fifty good quality coffees, and one poor quality coffee were used in the study. Six randomly selected “good” coffees were mixed with 25, 50 of 75% “poor” coffee. Using the various modelling approaches the team were able to identify the adulterated coffees from the pure sample with a good classification of 98.6%. The key elemental markers were found to be As, Cd, Cr and Pb, potentially suggesting that poor quality coffees also pose a higher health risk from exposure to toxic metals	422
B, Ca, Fe, K, Mg, Mn, Na, P & Zn	Banana	ICP-OES	PCA, PCA-LDA, PCA-SVM, PCA-ANN	Species	Bananas have over 100 cultivars across 50 sub-species and genomic group, making classification tricky. In this study elemental analysis and chemometrics aimed to identify specific groups to aid in the classification. Bananas from 6 genomic groups were collected, giving 100 samples in total. Digested samples were analysed for 9 elements by ICP-OES. The elemental composition was found to be K > Mg > Ca > Na > Fe > Mn > P > Zn > B. The PCA-SVM model gave the best results with a 100% accuracy in identification of both genomic and sub-genomic group	423
B, Ba, Ca, Cu, Fe, K, Mg, Mn, Na, Ni, P, Sr, Zn	Pepper	ICP-OES	SIMCA, PCA	Authenticity	Bell peppers from Altino and Senise were analysed, with Senise peppers having PGI status. Fifty samples in total were used for this study. PCA analysis showed clear separation and further work with SIMCA allowing models to be built that gave 100% sensitivity and specificity in identifying the 2 types of peppers	424
Ba, Br, Ca, Cl, Cr, Cu, Fe, K, Mn, Na, Ni, P, Pb, Pr, Rb, Ru, S, Se, Sr, Ti, V, Y & Zn	Apples	XRF	VIP-PLS-DA	Authenticity	Maçã de Alcobaça apples are the oldest PGI in Portugal. In this study fruits grown under this PGI were compared to other varieties of apples and the same varieties of apples grown outside of the PGI jurisdiction. Apples grown in the same area showed moderate accuracy at 66% in different cultivars but showed 100% accuracy when comparing fruits from non-PGI area across different varieties	425
Various	Tomato and pepper	ED-XRF	PLS-DA	Authenticity	Fruits that had been grown organically or conventionally were collected from the south and south-east of Brazil, in total 272 were analysed (193 tomatoes, 79 peppers). With the combination of techniques in the study, the models were shown to have a 100% accuracy in prediction of region and cultivation	426
Various	Salmon	ICP-MS, Rapid Evaporative Ionisation Mass Spectrometry (REIMS)	OLPS-DA, PLS-DA, PCA-LDA	Origin/farming method	In this study a combination of methods was investigated to determine origin and farming method. Samples (522) of salmon from 5 regions both farmed and wild were tested. Using the combination of techniques a clear definition was seen between both region and type of salmon using 18 lipid markers and 9 elemental markers, with these a classification accuracy of 100% was attainable	427
Ag, Al, As, B, Ba, Be, Ca, Cd, Co, Cr, Cu, Fe, K, Li, Mg, Mn, Mo, Na, Ni, P, Pb, Sb, Se, Si, Sr, Ti, Tl, V & Zn	Saffron	ICP-OES	PCA	Origin	Saffron from 2 regions in India as well as other countries (Iran and Afghanistan) were the subject of this study. Twenty-nine elements were analysed by ICP-OES, and PCA was carried out. The key elements were Al, B, Ba, Fe, K, Na, Ni, Si, Sr, Ti, V & Zn	428
Various (31)	Prickly pear	ICP-MS, UHPLC-MS/MS	MANOVA (Multivariate ANOVA)	Origin	Prickly pear from Mediterranean countries were analysed for a range of elemental and molecular markers such as antioxidants, acids and vitamins. Forty-three samples from Greece, Cyprus, Italy and Spain. Different varieties of Cypriot fruits were also included in this study (red, yellow and spineless). The most abundant elements were Ca > P > Mg and Co was least abundant which was common across all samples. Using the statistical analysis accuracies of 87% for geographical origin and 91% for variety	429
REEs (16)	Lentils	ICP-MS	PCA, SIMCA	Origin	Using the content of REE and ratio of Sc/Y and Sc/Yb PCA models were built to determine origin of lentils from Greece from those from other countries and lentil specifically from the Eglouvi region in Greece. These lentils were also adulterated with either other Greek lentils or non-Greek lentils. The models were found to be robust and gave 100% accuracy, it was also determined that the limit of detection for adulteration was as low as 4.6% with Greek lentils and 6% with non-Greek lentils	430
Various	Clam shell	ICP-MS	RF, CAP (Canonical Analysis of Principal Components)	Origin	Mislabelling of Manila clams from Portugal was suspected, particularly those from the Targus Estuary. Analysis of the shells of clams from the west coast of Portugal and the Targus Estuary showed differences in the ratio of key elements namely: Ba/Ca, K/Ca, La/Ca, Mn/Ca, P/Ca, Pb/Ca, Sr/Ca, and Y/Ca. The models were successful at predicting the origin of the clams with 100% accuracy using both models	431
Various (31)	Pecorino cheese	ICP-MS, ICP-OES	LDA, PCA	Authenticity	Nutritional assessment and discrimination of cheeses (Pecorino Romano and Pecorino Sardo). Nutritional elements were present in similar concentrations Ca > P > K > S ≥ Mg, but higher Na is observed and expected in Pecorino Romano due to a salting step in its manufacture. LDA was found to have a discrimination of (7% and a prediction accuracy of 96% demonstrating the model could be useful for detecting counterfeit cheeses)	432
Al, Ba, Ca, Cd, Co, Cr, Cu, Fe, Ga, K, Li, Mg, Mn, Mo, Na, Ni, Pb, Rb, Sr, V & Zn	Wine	ICP-MS, IRMS	RF	Authenticity	Kyoho grapes from Puijang County are a PGI product and therefore susceptible to mislabelling or fraud. In this study IRMS and elemental analysis including Sr and Pb isotopic ratios were used to build models to assess grapes from the Puijang region and surrounding areas. The RF models were assessed and gave 100% accuracy for all parameters in the training set, and an overall accuracy of 96% in the test samples, the key markers were δ^2H, δ^15N & δ^18O and Ba, Cd, Co, Ga, Ni & Rb	433
Various	Tomatoes	XRF	PCA-DA	Authenticity	Tomato samples from SE Sicily were assessed using XRF to determine elemental fingerprints and assess whether PCA-DA could classify these fruits. Key elements were found to be Br, Ca, Cl, Cu, Fe, K, Rb, Sr & Zn. Tomatoes from PGI farms were distinguishable from those that weren't but also the different farms were distinguishable, likely due to varying soil types relating to volcanic activity. The team propose further work to build a database for rapid identification of tomato origin	434
Various (30)	Oolong teas	ICP-MS, UHPLC-MS-MS	OPLS-DA, PCA, SVM	Origin	Teas from 2 growing areas in China were assessed for organic compounds, 15 REE's and 15 non-REE's. The elemental profile for all teas was found to be the same order for both element groups, but with some key differences in concentrations, particularly with REE allowing for classification. Using the SVM and data from all the analytical testing a SVM model was able discriminate with 100% accuracy teas from Huang Guanyin	435
Various (50, including REEs), Sr ratio	Wine	ICP-MS, MC-ICP-MS	LDA, ANN	Origin	In this study 120 wines from 7 regions were analysed for Sr isotope ratio by MC-ICP-MS as well as a suite of elements by ICP-MS. The statistical analysis showed that Sr ratio provides a key marker, and when REE composition is added into the models a good prediction can be established. In this study the LDA model had an accuracy of 88% and ANN of 94%	436
Al, As, Ba, Bi, Ca, Cd, Co, Cr, Cu, Fe, Ga, In, K, Li, Mg, Mn, Na, Ni, P, Pb, Se, Sr & Zn	Pollen	ICP-OES	CDA	Origin	71 samples from 4 different apiaries were analysed by ICP-OES. The mineral contents of the samples were similar, with predominant elements being P, K, Ca, Mg and Na. CDA analysis was able to accurately assign apiary in 90% of cases, and season in 100%	437
Various	Meat and fish	Various	Various	Various	A review paper covering elemental analysis and chemometrics on meat and fish products to determine origin, authenticity, production system species and other attributes. This extensive review covers 135 papers in this area	438
Various (16)	Teff	ICP-MS	LDA	Origin	Teff from Ethiopia was the subject of this study. Samples from 4 regions were analysed, and it was found that only 5 elements could indicate origin namely, Cu, Mo, Se, Sr & Zn. LDA analysis gave a 90% accuracy for brown teff and 100% accuracy for white teff. The authors conclude the study was limited and could incorporate more regions and countries to expand the model	439
Al, Ca, Cd, Co, Cr, Cu, Fe, K, Mg, Mn, Na, Ni, Pb, Se & Sr	Egg	ICP-OES	SDE (Stahel–Donoho estimation), PLS-DA, LS-SVM	Production	In this study the elemental composition of eggshells was performed to determine whether a distinction between caged and free-range eggs could be established. The key markers were established to be Cd, K, Mg, Mn & Se with the PLS-DA model giving an accuracy of 91–93% and LS-SVM giving 95–96% accuracy	440
Various	Sea bass	LIBS	PCA, PLS-DA	Freshness	The freshness of sea bass was determined by use of analysis by LIBS followed by PCA and PLS-DA chemometrics. Determination of storage time of the fish was shown to have an accuracy of 91%, whereas the freshness grade was found to give 97% accuracy	441
Various	Rice flour	LIBS	PCA, non-negative matrix factorization, independent component analysis, multidimensional scaling, kernel PCA, local linear embedding, Laplacian eigenmaps, isometric mapping, SVM	Adulteration	In this paper, a comparison of different chemometric techniques was carried out on rice flour which had been adulterated. Rice flour was mixed with sorghum flour, talcum powder, wheat flour, buckwheat flour, gypsum powder, millet flour, or corn flour at a ratio of 1%, 3%, 5%, 10% or 15%. The analysed data was processed using many techniques. Overall, it was found that linear modelling gave the most acceptable results, and the authors suggest that selection of the modelling tools used is an important factor in these types of studies	442
Various	Wine	Various	Various	Various	A review paper covering elemental analysis and chemometrics on wines to determine origin, authenticity, production system species and other attributes. This extensive review covers 54 papers in this area	443
Various	Meat and animal-based foods	Various	Various	Various	A review paper covering elemental analysis and chemometrics on meat and animal products to determine origin, authenticity, production system species and other attributes. This extensive review covers 66 papers in this area	444

---

The paper by Kovalenko and team²⁴¹ is discussed here, as it covered several elements and food matrices. The authors reported on the addition of six elements to the current scope of the US Food and Drug Administration Elemental Analysis Manual (EAM) method 4.7. This method currently covers the analysis of 11 elements namely, As, Cd, Cr, Cu, Hg, Mn, Mo, Ni, Pb, Se and Zn, and this report covers the expansion of scope to include Co, Sn, Sr, Tl, U and V. The methodology set out in the EAM method 4.7 describes MAD (2 methods) followed by ICP-QMS or ICP-QQQ-MS analysis. In this study, both closed vessel and single reaction chamber MAD were used, to cover those described in the existing EAM. The sample (0.5 g) was digested with a 5 + 1 mixture of HNO₃ and H₂O₂, at 250 °C for 45 min, for reaction chamber microwave, whereas, for closed vessel digestion, an 8 + 1 mixture of HNO₃ and H₂O₂, at 200 °C for 40 min, was applied. A number of CRMs (NIST SRM 3280 multivitamin, SRM 1548a typical diet, SRM 1566b oyster tissue, SRM 1640a natural water; NRCC DOLT-5 dogfish liver and DORM-4 fish protein) and spiked sample matrices were selected to cover a range of sample types, including protein, fat and carbohydrate, as described by the AOAC food triangle. Spiked samples were mayonnaise, avocado, peanut butter, egg yolk, potatoes, milk, spinach, oyster and egg whites. Each matrix was spiked at between 50% and 300% of the endogenous elemental content. Analysis was performed by ICP-MS and ICP-QQQ-MS for all elements. In general, all recoveries were between 80% and 120%, with a few exceptions. The authors also noted difficulties in Sn analysis, such as lack of suitable CRMs and variable recoveries. A small intercomparison study was carried out between two different laboratories, using canned goods with an appreciable level of Sn, with acceptable results. Despite some issues within the study, the addition of these elements into the existing method was shown to be acceptable and practical.

9.1 Progress for individual elements

9.1.1 Arsenic. The metabolism of As^V in crickets was studied by researchers in Canada and Thailand.²³⁷ Crickets were reared together on standard diets of a commercial cricket feed product, then split into 4 groups of ∼100 crickets per group. Each group was fed diets containing different levels of As^V (1.3 ± 0.1 mg kg⁻¹, 5.1 ± 2.5 mg kg⁻¹ and 36.3 ± 5.6 mg kg⁻¹), as well as a control group with no additional dietary exposure. After 12 days, the samples were harvested, dried and homogenised for analysis. Total As was measured by ICP-QQQ-MS following MAD in HNO₃ : HCl at a 12 : 1 ratio. Speciation was carried out using HPLC-ICP-QQQ-MS following extraction in 1% H₂O₂ at 95 °C for 60 min. For cations, the separation was carried out using a Metrosep C₆ column (4.0 × 250 mm, 5 μm) with a gradient elution of ultrapure water and 50 mmol L⁻¹ pyridine. Anions were separated under isocratic elution with 50 mmol L⁻¹ carbonate buffer using an Hamilton PRP-X100 column (4.1 × 250 mm, 10 μm). Calibrations were constructed using AB, iAs, DMA and MMA. A bio-availability study was also carried out phasing simulated saliva, gastric and intestinal juices. It was shown that crickets were able to tolerate exposure to iAs and will methylate these species, with DMA being the most prevalent species, followed by MMA and TMAO. The AB naturally present in the feed was only observed in the control crickets, with the postulation that under As stress iAs is upregulated into methylated forms, and it appears AB is demethylated, the authors suggest this would need confirmation with further studies. In the bio-accessibility study, all As species were shown to be highly accessible, with uptakes in the range between 78% and 100%.

9.1.2 Chromium. Utilising alkali extraction and weak anion exchange HPLC-ICP-MS, Song and team²³⁸ investigated a rapid method for Cr speciation in foods. A wide range of matrices were assessed in this study, including yoghurt, milk powder, rice flour, orange juice, green tea, white vinegar, and whole wheat bread. The team also investigated the mechanism of conversion of Cr^VI to Cr^III in food components such as proteins, vitamins and sugars. Samples were extracted in a mixture of phosphate buffer, NaOH, Na₂CO₃ and MgCl₂ in an ultrasonic bath, before being centrifuged and filtered for analysis. The determination of Cr species was carried out by HPLC-ICP-MS, using an anion exchange column (50 mm × 4.6 mm, 5 μm) and a 70 mmol L⁻¹ NH₄NO₃ eluent, with a rapid run time of 1.5 min. A spiking study was also carried out. Samples were spiked at 5, 25, and 50 μg kg⁻¹ either into the extraction solution, or into the sample, which was then extracted after 2 or 60 min to investigate the conversion of Cr^VI to Cr^III. In the immediately analysed samples, recoveries were between 92% and 112%. Even after 2 min in the matrix, losses of Cr^VI were seen in some matrices, which were more pronounced after 60 min, when no Cr^VI was measured in orange juice and green tea, indicating complete conversion. Conversely, rice flour and bread had almost negligible losses. The ingredients study showed that the rate of conversion was dependent on the matrix type as well as temperature and pH, and compounds such as polyphenols and vitamin C also promoted the conversion. The authors concluded that the likelihood of Cr^VI being present in foods is small due to these strong conversions in the majority of food types.

9.1.3 Cobalt. Yang et al.²³⁹ developed a HPLC-ICP-MS method for the rapid determination of vitamin B₁₂ in powdered milk, based on the measurement of Co. Samples underwent a pre-treatment step where the proteins were precipitated out using potassium ferrocyanide and (CH₃COO)₂Zn. Solutions were sonicated at 30 °C for 40 min, then centrifuged. The supernatant was then filtered and injected onto the column. The separation was performed using a C₁₈ column (250 × 4.6 mm, 5 μm) and a mobile phase containing 1.6 mmol L⁻¹ EDTA and 0.4 mmol L⁻¹ KH₂PO₄ in 60% (v/v) MeOH. The spectra showed two main peaks, relating to free Co and vitamin B₁₂. For the determination of total Co, samples underwent MAD prior to ICP-MS analysis. To assess the accuracy of the method a CRM for pure cyanocobalamin (China NIM National Standard Substances GBW 10277) was analysed and a 98% recovery was reported. Since there are no CRMs for free Co in milk powder, the accuracy of the whole procedure was verified on milk powder samples spiked at three levels (25%, 50% and 100% of the endogenous content) with vitamin B₁₂ and elemental Co. Recoveries were between 91% and 103% and RSDs (n = 6) were <5%. The LOQs was found to be 0.15 μg kg⁻¹ for free Co and 2.1 μg kg⁻¹ for vitamin B₁₂, concluding that the method was suitable for the rapid assessment of vitamin B₁₂ in powdered milk samples.

9.1.4 Lead. Preparation and characterisation of NiCoFe-Layered Double Hydroxides (NiCoFe-LDH) for SPME extraction and preconcentration of Pb from juice samples was carried out by Arain et al.²⁴⁰ In this work the preparation of sorbent from nanodiamonds was explained and a full evaluation of the obtained NiCoFe-LDH material was undertaken to assess the ability of the material to aid in the quantification of Pb by FAAS. The sorbent was found to have optimal performance in mildly alkaline conditions with efficacy going from 20% at pH 7 to 100% at pH 8. For the test conditions 100% recovery was found with 5 mg of sorbent. The concentration and desorption time of the eluent was found to be almost 100% at the lowest tested concentration and time period, showing an efficient process. The material was tested using NIST SRM 1515 apple leaves and BCR-505 estuarine water, as no suitable juice CRM is widely available. These gave recoveries of 103% and 98%, respectively. The sorbent was then used to assess levels of Pb in 28 various juice samples, including mixed juices, single fruit and pulpy juices. The juices underwent MAD followed by the assessed extraction procedure. Spiked samples gave recoveries between 97% and 102%, the calculated LOQ was 2.1 ng mL⁻¹, giving a PF of 25. Given the 0.03 mg L⁻¹ and 0.05 mg L⁻¹ limits for Pb in fruit juices and berry juices, respectively, set out by the Codex Alimentarius, this method is shown to be suitable and efficient in achieving these requirements.

9.1.5 Selenium. A review of articles including total Se analysis in food was carried out by Schmitz et al.²⁰ In this comprehensive study (107 references) a meta-analysis of validated methods for Se determination in food matrices was undertaken. Unsurprisingly the most common analytical techniques utilised for Se analysis were found to be ICP-MS and hyphenated ICP-MS, predominantly HPLC-ICP-MS, covering about 50% of the work carried out. Other common techniques applied were ICP-OES and variants of atomic absorption technology, including HG, GF and hyphenated techniques. The authors called out the need for the preparation method to be tailored to the requirements of the matrix and the species under study, particularly to preserve the integrity of the species whilst achieving suitable extraction from the matrix. Commonly studied matrices were yeasts (15% of papers) and cereals (20%), the rest covered a wide range of food matrices including meats, mushrooms, dairy products, cereals and vegetables.

9.2 Single and multi-element applications in food and beverages

9.2.1 Human milk and infant formula. Iodine in human breast milk was determined by ICP-MS in a simple “dilute and shoot” approach. Ammonia was used as the diluent to avoid losses from the formation of HI. In this study small aliquots (50 μL) of samples were diluted in 4 mL of 0.5% NH₄OH, containing Te as IS, before being analysed by ICP-MS. The team²⁴² also looked at the performance of the analysis when samples underwent serial dilutions. The method accuracy was assessed using a CRM (NIST 1549a milk powder). Good recoveries were seen at all dilutions with a minimum accuracy of 96% and an RSD (n = 12) <4%. The method LOQ was 2.3 μg L⁻¹. A spiked recovery test was also carried out, with recoveries between 97% and 101%. The method was demonstrated to be a simple an efficient tool in the assessment of I in breast milk. In this study only four samples were used but the initial work suggests that the method may be suitable for wider studies.

In an extensive study of 232 samples, Nouzha et al.²⁴³ determined reference values for key nutritional elements in human breast milk. Samples were prepared by mineralisation with HNO₃ on a digestion block at 90 °C for 45 min, followed by a dilution step prior to analysis. Measurements were performed by ICP-MS for Co, Cu, I, Se and Zn, whereas ICP-OES was utilised for the determination of Fe. The method was validated using body fluid RMs (Utak® serum trace elements, normal and high ranges, and Seronorm™ urine levels 1 and 2), which showed good recoveries between 91% and 120%. As a full set the following mean values and reference intervals (2.5–97.5% percentiles) were obtained for trace elements in breast milk: Co, 12.28 (5.27–25.82) nmol L⁻¹; Cu, 6.02 (1.71–13.23) μmol L⁻¹; Fe, 4.72 (1.25–11.49) μmol L⁻¹; I, 0.29 (0.07–1.01) μmol L⁻¹; Se, 0.12 (0.07–0.24) μmol L⁻¹; and Zn, 43.86 (7.3–107.0) μmol L⁻¹. In general, a decrease in mineral content was seen as the milk came from later stages of lactation. The team proposed to expand the study, both as numbers of elements studied as well as participants, with a better understanding of patient history to establish patterns in the findings.

As many people choose to reduce meat and animal products in their diets, the occurrence of vegetarian and vegan lactating mothers becomes significant and the effect on the composition of breast milk becomes of interest. To assess the impact of diet on breast milk, Perrin and team²⁴⁴ utilised elemental analysis to screen breast milks from mothers with different diets. Lactating mothers following vegan, vegetarian and omnivore diets were studied. To evaluate the method, spiked samples were assessed and analysed by ICP-MS (As, Cd, Cr, Cu, Fe, I, Mn, Mo, Pb, Se and Zn) and ICP-OES (Ca, K, Mg, Na and P). The spiked samples showed good recoveries in the range between 93% and 108%. With standard statistical analysis, such as Dunn’s analysis, only Se was shown to have a significant difference based on diet, with vegan and vegetarian sourced milk being richer in this element compared to breast milk samples from women following omnivore diets. The team also used machine learning methods and found that Fe and I potentially may have some importance in the identification of diet types using chemometrics.

9.2.2 Fruits. The concentrations of essential elements in sycamore fruits (Ficus sycomorus L.) were investigated.²⁴⁵ These wild edible fruits have high mineral contents and may be a useful addition to the diet in Ethiopia where micronutrient deficiency can be a health problem affecting 30% of the population. Samples of the whole fruits were collected, washed and dried, then the seeds and fruits were separated. Dried seeds and fruits were ground and 0.5 g of homogenised material digested on a hot plate with a mixture of HNO₃ and H₂O₂, in numerous steps of heating and evaporation until a clear digestate was obtained. The digests were then analysed by FAAS for Ca, Cu, Fe, K, Mg, Mn and Zn, using LaCl₃ in the diluent to avoid refractory interferences and minimise the precipitation of Ca and Mg. Spiked samples gave recoveries between 80% and 120% for both seeds and fruits. The study found that the profile and concentrations of elements were Mg (227–410 mg kg⁻¹) > K 120–137 mg kg⁻¹) > Ca (30–34 mg kg⁻¹) > Fe (3–7 mg kg⁻¹) > Cu (1–4 mg kg⁻¹) > Mn (2–4 mg kg⁻¹) > Zn (∼3 mg kg⁻¹).

9.2.3 Beverages. The content of essential elements and PTEs in pure and blended coffee drinks was investigated by Kargarghomsheh et al.²⁴⁶ using ICP-OES. Samples of instant coffee, ground coffee and blended coffee drinks (milky, creamy, sugar free and chocolate) were purchased from the local market, homogenised and prepared by dissolving in hot water. Filtered samples then underwent MAD with a 9 HNO₃ : 1 H₂O₂ mixture. Samples were analysed for Al, As, Cd, Co, Cr, Fe, Mn, Ni, Pb, and Zn by ICP-OES. The method accuracy was established by preparing spiked samples, which gave recoveries between 93% and 103% with RSDs between 1.7% and 2.3%. The team found that typically coffee drinks had higher elemental concentrations, which the authors suggested came from ingredients, such as milks, creamers, chocolate and sugar, added to these samples as the simplest blended coffee had lowest overall element content following this order sugar-free mixed coffee < simple mixed coffee < chocolate mixed coffee < creamy mixed coffee. The profile of the elements was also different with the abundance in coffee samples being Fe > Mn > Zn > Al > Pb > Ni > Co > As > Cd > Cr whereas in coffee drinks it was found to be Fe > Al > Zn > Mn > Cr > Co > Pb > Cd > As > Ni, with mean Fe levels in coffees being 132 μg kg⁻¹ to 682 μg kg⁻¹ in mixed coffee drinks.

Though many studies on mineral content have been carried out on apple fruit and apple juice, less is understood about the profiles seen in apple ciders. Ageyeva and team²⁴⁷ investigated mineral contents in mono-varietal and commercial ciders and related this to sensory attributes of these ciders. Samples were analysed by a variety of techniques. Major elements (Ca, K, Mg and Na) were determined using high-performance CE with indirect UV measurement. Whereas Cs, Cu, Mn, Mo, Rb and Zn were quantified using direct analysis by GFAAS. The concentrations of As, Cd, Hg, Ni, Pb, Sn and Ti were measured by ICP-OES. In the single variety ciders, a wide range of elemental compositions were seen. Arsenic and Hg were not detected in any samples. As these fruits were grown and processed under the same controlled conditions, it demonstrated that the elemental composition is a genetic characteristic of the variety of apple. The levels of K and Na were shown to be the key elements driving sensory scores, with increased K and lower Na having the highest sensory score. Ciders with K levels from 850 mg L⁻¹ to 1900 mg L⁻¹and Na ranging from 10 mg L⁻¹ to 100 mg L⁻¹ had the most preferred flavour. The team proposed a general elemental profile of K > Na > Ca = Mg > Fe for macronutrients and Rb > Cu > Mn = Zn > Ni > Cs > Ti for microelements.

9.2.4 Honey. The geographical origin of honey was again the subject of investigation, based on TE profiles. In a relatively localised study in Northern Brazil, the team of dos Santos Silva²⁴⁸ applied ICP-OES and chemometric methods to batches of 36 samples of honey, collected from beekeepers in each of six regions. The 216 samples were digested in a hot block with HNO₃ and H₂O₂ and then analysed for ten elements (Al, Ca, Cu, Fe, Li, K, Mg, Mn, Na and Zn) by ICP-OES. As no suitable CRM was available, spiked samples were prepared to evaluate the method. Recoveries were between 97% and 104%, thus demonstrating the suitability of the preparation and analysis of these samples. From each region, 28 samples were used for training and 8 samples were used for model assessment. The team used RF and XGBoost as modelling tools on the obtained analytical data. Both methods gave an overall 93% prediction of the region of provenance. For two regions, the prediction was only 71–86% accurate. The researchers offered explanations behind these findings, linked to natural and anthropological reasons, considering that, given the relative proximity of the sampling sites, these had similar geology and fauna, which made it harder to discriminate between them on the basis of the elemental analysis. Origin from the other regions could be predicted with 100% accuracy, due to greater differences in mineral content, particularly K and Mn. One of the regions was in a costal mangrove region, leading to larger differences in Mg, Mn and Na content. In a similar study Mara et al.²⁴⁹ investigated both multi-floral and mono-floral honey from Spain and Italy. A total of 247 honey samples were digested using MAD with a mixture of 0.5 mL HNO₃–3 mL H₂O₂–4 mL H₂O and analysed by ICP-MS for trace and toxic 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), macro elements (Ca, K, Mg and Na) and lanthanide elements (Ce, Dy, Er, Eu, Gd, Ho, La, Lu, Nd, Pr, Sm, Tm and Yb). The method was evaluated using two CRMs materials (NIST SRM 1515 apple leaves, BCR-668 mussel tissue) which gave recoveries between 90% and 100%. Chemometric analysis was performed using PCA, LDA and RF models. It was found that prediction of the geographical origin was more successful, with an accuracy between 88% and 91%, whereas botanical origin was between 47% and 81% accurate. When assessment of geographical and botanical origin were combined the accuracy was further impacted in some cases to a mere 30%. The most effective markers for geographical origin were found to be Ce, Eu, Mg, Mn, Na, Nd, Sr, Tb and Zn, whereas the botanical origin was marked out by Ce, Ca, Eu, K, Lu, Mn, Na and Sr.

9.2.5 Edible salts. Edible salts of a particular region are often highly valued in the culinary world. The minor element composition of ten gourmet salts from all over the world were investigated for the first time.²⁵⁰ Samples (1 kg) of many different salt varieties, including Mozia, Atlantic grey sea, Persian blue, Smoked, Guerande Grey, Hawaii pink, Hawaii black, Himalayan pink, Maldon, and Baule Volante whole rock, were collected in plastic bags and then ground with a mortar and sieved to obtain homogeneous particle sizes. For each sample, an aliquot (5 g) was dissolved in 100 mL of 1% conc. HNO₃ and left to rest for three days in closed vessels, to minimise loss of Se volatile species. After filtration on 0.45 μm PTFE filters, the sample solutions were analysed using ICP-MS to determine the concentrations of 11 trace elements (Al, Ca, Co, Cr, Cu, Fe, Mn, Ni, Pb, Se and Zn). Measurements of the Hg content were carried out using TDA-AAS (with a direct mercury analyser). Recoveries, obtained on salt samples’ aliquots spiked with each of the elements under study, at two concentration levels, were between 91% and 101%, demonstrating the method was fit for purpose. Across the samples, some elements (i.e. Co, Cr, Cu, Hg, Pb and Se) showed relatively consistent concentrations, whereas the levels of others (Al, Ca, Fe, Mn, Ni, and Zn) varied greatly across samples, suggesting their potential use as chemical markers for authenticity. The authors noted the possible health benefits from consumption of some of these salts, due to high levels of useful nutrients, such as Ca and Zn, but this must be balanced against the detrimental effects of high levels of Na in the diet, whereas all samples analysed contained levels of Pb breaching the EU limits of 1 or 2 mg kg⁻¹ (depending on salt type) which warrant further investigation. The authors indicated that the elemental concentrations were specific to the salts described and varied among different brands and production methods.

9.2.6 Authenticity. Hur et al.²⁵¹ utilised fsLA-ICP-MS to determine the country of origin of dried chilli peppers, in particular to discriminate between South Korean or Chinese products. Fifty-one samples from each region were collected, dried and pelletised for analysis. Method optimisation was carried out on experimental parameters of repetition rates of 10 to 1000 Hz, spot sizes of 20 μm to 60 μm, laser energies of 60% to 100% and gas flow rates from 0.1 L min⁻¹ to 1.4 L min⁻¹. Optimised conditions which gave the best cps values and lowest RSD% were set as 200 Hz, a spot size of 50 μm, an energy level of 90%, as well as He and Ar gas flow rates of 1.2 L⁻¹ and 1.0 L⁻¹, respectively. The original dataset included measurements of the content of 61 elements then reduced to 33 analytes by removing elements with poor RSDs from replicate analysis and low intensities for the statistical analysis. Data was modelled using OPLS-DA (orthogonal partial least squares discriminant analysis) using a SIMCA autofit function, The accuracy of the model was assessed using R² and Q² values as well as numerous permutations, outlier removals and cross-checking operations for validation. R² and Q² values were 0.811 and 0.733 respectively, with the key elements in determination of origin being identified as Al, Cd, Cs, Cu, Ga, Na, Nb, V and Zn. Receiver operator curve (ROC) was used to assess selectivity and sensitivity, this returned a method accuracy of 98.7%, thus demonstrating that fsLA-ICP-MS could be a useful tool in the evaluation of origin in food samples without the need of MAD or other digestion techniques.

10. Abbreviations

AAS	atomic absorption spectrometry
AB	arsenobetaine
AC	arsenocholine
ACN	acetonitrile
AEC	anion-exchange chromatography
AES	atomic emission spectrometry
AF	atomic fluorescence
AF4	asymmetric flow-field flow fractionation
AFS	atomic fluorescence spectrometry
AGREE	Analytical GREEnness scale
AGREEprep	Analytical GREEnness metric for sample preparation
ANN	artificial neural network
ANOVA	analysis of variance
APDC	ammonium pyrrolidine dithiocarbamate
ASU	Atomic Spectrometry Update
BCR	Community Bureau of Reference
BMI	body mass index
BSE	backscattered electron
CA	cluster analysis
CAP	canonical analysis of principal components
CCD	charge coupled device
CDA	classical discriminatory analysis
CE	capillary electrophoresis
CEAEQ	Centre d’Expertises en Analyse Environnementale du Québec, Canada
CFU	colony-forming unit
ChCl	choline chloride
CI	confidence interval
CLSI	Clinical & Laboratory Standard Institute
CNN	convolutional neural network
CPE	cloud point extraction
cps	counts per second
CRC	collision/reaction cell
CRM	certified reference material
CS	continuum source
CSF	cerebrospinal fluid
CT	computed tomography
CTAB	cetyl trimethylammonium bromide
CV	cold vapour
CVG	chemical vapour generation
DA	discriminant analysis
DBS	dry blood spot
DDTC	diethyldithiocarbamate
DDTP	o,o-diethyldithiophosphate
DES	deep eutectic solvent
DESI	desorption electrospray ionisation
DGA	diglycolamide
DLLME	dispersive liquid–liquid microextraction
DLS	dynamic light scattering
DMA	dimethylarsenic
DNA	deoxyribonucleic acid
DRC	dynamic reaction cell
DSPE	dispersive solid phase extraction
DSPME	dispersive solid phase micro extraction
DTZ	dithizone
EC	European Commission
ED	energy dispersive
EDL	electrodeless lamp
EDTA	ethylenediaminetetraacetic acid
EDX	energy dispersive X-ray spectroscopy
EHF	enhancement factor (based on slope ratios)
ENM	engineered nanomaterial
EPA	US Environmental Protection Agency
EQA	External Quality Assessment
ERF	enrichment factor (based on concentration ratios)
ERM	European reference material
ESI	electrospray ionisation
ETAAS	electrothermal atomic absorption spectrometry
EtHg	ethylmercury
EtOH	ethanol
ETV	electrothermal vaporisation
EU	European Union
FAAS	flame atomic absorption spectrometry
FAO	UN Food and Agriculture Organization
FAPAS	Food Analysis Proficiency Assessment Scheme
FFPE	formalin-fixed paraffin-embedded
FI	flow-injection
FIA	flow-injection analysis
fs	femtosecond
GAPI	Green Analytical Procedure Index
GB	gradient boosting
GC	gas chromatography
GD	glow discharge
GF	graphite furnace
GFAAS	graphite furnace atomic absorption spectrometry
GMO	genetically modified organism
GO	graphene oxide
GPx	glutathione peroxidase
HCA	hierarchical cluster analysis
HEN	high efficiency nebuliser
HEPES	4-(2-hydroxyethyl)piperazine-1-ethanesulfonic acid
HG	hydride generation
HILIC	hydrophilic interaction liquid chromatography
HPAPEO	1-(2-hydroxy-5-p-tolylazo-phenyl)-ethanone
HPLC	high performance liquid chromatography
8-HQ	8-hydroxyquinoline
HR	high resolution
HS	head space
IAEA	International Atomic Energy Agency
iAs	inorganic arsenic
IBDA	isobaric dilution analysis
IC	ion chromatography
ICH	International Council on Harmonization (of Technical Requirements for Pharmaceuticals for Human Use)
ICP	inductively coupled plasma
id	internal diameter
ID	isotope dilution
IDA	isotope dilution analysis
iHg	inorganic mercury
IL	ionic liquid
ILC	interlaboratory comparison
INAA	instrumental neutron activation analysis
IR	Infrared
IRMM	Institute for Reference Materials and Measurements
IRMS	isotope-ratio mass spectrometry
IS	internal standard
iSb	inorganic antimony
iSe	inorganic selenium
ISO	International Organization for Standardisation
JRC	Joint Research Centre
KED	kinetic energy discrimination
KNN	k-nearest neighbour
LA	laser ablation
LC	liquid chromatography
LDA	linear discriminant analysis
LIBS	laser induced breakdown spectroscopy
LIF	laser-induced fluorescence
LLE	liquid liquid extraction
LLME	liquid liquid microextraction
LOD	limit of detection
LOQ	limit of quantification
LR	logistic regression
LS	least squares
Mr	relative molecular mass
m/z	mass-to-charge ratio
MAD	microwave-assisted digestion
MAE	microwave-assisted extraction
MAGBC	magnetic Alnus glutinosa sawdust biochar
MALDI	matrix-assisted laser desorption ionisation
MC	multicollector
ME	monochromatic excitation
MeHg	methylmercury
MeOH	methanol
MIC	microwave induced combustion
MICAP	microwave-sustained inductively coupled atmospheric-pressure plasma
MIL	magnetic ionic liquid
MIP	microwave induced plasma
miRNA	micro ribonucleic acid
MLP	multilayer perceptron
MMA	monomethylarsenic
MMMTA	monomethylmonothioarsonic acid
MNP	magnetic nanoparticle
MOF	metal–organic framework
MRI	magnetic resonance imaging
mRNA	messenger ribonucleic acid
MS	mass spectrometry
MS/MS	tandem MS
MWCNT	multiwall carbon nanotube
N2-MICAP-MS	N2-sustained microwave inductively coupled atmospheric pressure plasma mass spectrometry
NAA	neutron activation analysis
NC	nanocomposite
NCS	National Analysis Centre for Iron and Steel (China)
NEMI	US National Environment Methods Index
NIST	National Institute of Standards and Technology
NMIJ	National Metrology Institute of Japan
NMR	nuclear magnetic resonance
NP	nanoparticle
NRCC	National Research Council of Canada
ODS	octadecylsilane
OES	optical emission spectrometry
OPLS-DA	orthogonal partial least squares discriminant analyses
ORS	octopole reaction system
PAN	1-(2-pyridylazo) 2-naphthol
PAR	4-(pyridyl-2-azo)-resorcinol
PBS	phosphate-buffered saline
PCA	principal component analysis
PCR	polymerase chain reaction
PD	point discharge
PDO	protected designation of origin
PE	polyethylene
PET	polyethyleneterephthalate
PF	preconcentration factor (based on volume ratios)
PFA	perfluoroalkyl
PGI	protected geography indication
PIGE	particle-induced gamma-ray emission
PIXE	particle-induced X-ray emission
PLS	partial least squares
PLS-DA	partial least squares discriminant analysis
PLSR	partial least squares regression
PT	proficiency testing
PTE	potentially toxic element
PTFE	poly(tetrafluoroethylene)
PvbXa	polyvinyl benzyl xanthate
PVG	photochemical vapour generation
QC	quality control
QMS	quadrupole mass spectrometry
QQQ	triple quadrupole
RC	residual carbon
REE	rare earth element
RF	random forest
RM	reference material
RMSECV	root mean square error of cross validation
RNA	ribonucleic acid
ROC	receiver operator curve
RP	reversed phase
RSD	relative standard deviation
sc	single cell
SD	standard deviation
SDS	sodium dodecylsulfate
SeAlb	selenoalbumin
SEC	size exclusion chromatography
SeCys	selenocysteine
SelP	selenoprotein
SEM	scanning electron microscopy
SEM-EDS	scanning electron microscopy energy dispersive (X-ray) spectrometry
SeMeSeCys	Se-methylselenocysteine
SeMet	selenomethionine
SF	sector field
SI	Système International d’unités – International System of Units
SIMCA	soft independent modelling of class analogy
SIMS	secondary ion mass spectrometry
S/N	signal to noise ratio
sp	single particle
SPE	solid phase extraction
SPME	solid phase microextraction
SQT	slotted quartz tube
SR	synchrotron radiation
SRM	standard reference material
SSB	sample standard bracketing
SSID	species specific isotope dilution
SVM	support vector machine
SVR	support vector regression
TAN	1-(2-thiazolylazo)-2-naphthol
tAs	total arsenic
TBAH	tetrabutylammonium hydroxide
TCA	trichloroacetic acid
TDA	thermal decomposition amalgamation
TEM	transmission electron microscopy
TETRA	tetramethylarsonium
TFA	trifluoroacetic acid
THF	tetrahydrofuran
tHg	total mercury
TIMS	thermal ionisation mass spectrometry
TMAH	tetramethylammonium hydroxide
TMAO	trimethylarsine oxide
TOF	time-of-flight
TRIS	tris(hydroxymethyl)aminomethane
TXRF	total reflection XRF
UAE	ultrasound-assisted extraction
UHPLC	ultra high performance liquid chromatography
UN	United Nations
US	United States
USN	ultrasonic nebuliser
USP	US Pharmacopeia
UV	ultraviolet
UV-vis	ultraviolet-visible spectrophotometry
VG	vapour generation
VIP	variable importance in projection
vs.	versus
WAC	white analytical chemistry
WHO	UN World Health Organisation
XANES	X-ray absorption near-edge structure
XAS	X-ray absorption spectroscopy
XFM	X-ray fluorescence microscopy
XGBoost	eXtreme gradient boosting
XRD	X-ray diffraction
XRF	X-ray fluorescence

Conflicts of interest

There are no conflicts to declare.

References

1. M. Patriarca, N. Barlow, A. Cross, S. Hill, A. Robson and J. Tyson, J. Anal. At. Spectrom., 2024, 39(3), 624–698.
2. C. Vanhoof, J. R. Bacon, U. E. A. Fittschen and L. Vincze, J. Anal. At. Spectrom., 2023, 38(9), 1730–1743.
3. R. Clough, A. Fisher, B. Gibson and B. Russell, J. Anal. At. Spectrom., 2023, 38(11), 2215–2279.
4. 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., 2024, 39(1), 11–65.
5. E. H. Evans, J. Pisonero, C. M. M. Smith and R. N. Taylor, J. Anal. At. Spectrom., 2024, 39(5), 1188–1211.
6. R. Clough, C. F. Harrington, S. J. Hill, Y. Madrid and J. F. Tyson, J. Anal. At. Spectrom., 2024, 39(7), 1629–1664.
7. U. M. F. M. Cerqueira, M. A. Bezerra, V. A. Lemos, J. J. Coutinho, C. G. Novaes, S. A. Araújo and L. P. d. Nascimento, TrAC, Trends Anal. Chem., 2023, 167, 117277.
8. C. Douvris, T. Vaughan, D. Bussan, G. Bartzas and R. Thomas, Sci. Total Environ., 2023, 905, 167242.
9. A. V. Skalny, T. V. Korobeinikova, M. Aschner, O. V. Baranova, E. G. Barbounis, A. Tsatsakis and A. A. Tinkov, J. Trace Elem. Med. Biol., 2023, 79, 127241.
10. V. Bali, Y. Khajuria, V. Maniyar, P. K. Rai, U. Kumar, C. Ghany, M. A. Gondal and V. K. Singh, Biophys. Rev., 2023, 15(3), 401–417.
11. D. Kobylarz, A. Michalska and K. Jurowski, TrAC, Trends Anal. Chem., 2023, 169, 117355.
12. A. Frydrych and K. Jurowski, TrAC, Trends Anal. Chem., 2023, 166, 117165.
13. B. A. Sadee, Y. Galali and S. M. S. Zebari, RSC Adv., 2023, 13(44), 30959–30977.
14. K. S. Hoy, T. Davydiuk, X. Chen, C. Lau, J. R. M. Schofield, X. Lu, J. A. Graydon, R. Mitchell, M. Reichert and X. C. Le, Food Qual. Saf., 2023, 7, fyad032.
15. M. Morita and Y. Shibata, Chemosphere, 1988, 17, 1147–1152,  DOI:10.1016/0045-6535(88)90180-4.
16. D. Coniglio, G. Ventura, C. D. Calvano, I. Losito and T. R. I. Cataldi, J. Pharm. Biomed. Anal., 2023, 235, 115628.
17. T. Pasinszki, S. S. Prasad and M. Krebsz, Environ. Monit. Assess., 2023, 195(9), 1066.
18. S. L. C. Ferreira, I. S. A. Porto, C. S. Assis Felix, S. V. A. Dantas, A. V. Branco Chagas, T. Pereira Muller and A. C. Magalhaes Aleluia, Anal. Lett., 2024, 1–17,  DOI:10.1080/00032719.2024.2347455.
19. S. W. C. Chung, Food Addit. Contam.,:Part A, 2024, 41(7), 782–789.
20. C. Schmitz, I. M. Grambusch, D. Neutzling Lehn, L. Hoehne and C. F. Volken de Souza, Food Chem., 2023, 429, 136974.
21. International Vocabulary of Metrology – Basic and General Concepts and Associated Terms (VIM), 3rd edn, 2008, The version with minor corrections is available at: https://www.bipm.org/documents/20126/2071204/JCGM_200_2012.pdf/f0e1ad45-d337-bbeb-53a6-15fe649d0ff1.
22. Eurachem Guide: Terminology in Analytical Measurement – Introduction to VIM 3, ed. V. J. Barwick, 2nd edn, 2023, ISBN 978-0-948926-40-2. Available from https://www.eurachem.org.
23. L. J. J. Ashby, K. E. C. Mill, M. C. C. Arnold, J. L. L. Van Geest and A. M. H. de Bruyn, Environ. Toxicol. Chem., 2023, 42(10), 2119–2129.
24. C. Zhang, Y. Yan and C. Zhang, J. Trace Elem. Med. Biol., 2024, 84, 127451.
25. L. Rastogi, A. D. Prasad, D. Sai Krishna, S. Yadlapalli and K. Dash, Food Control, 2024, 158, 110241.
26. J. W. Lee, M. C. Lim, Y. Lim, G. B. Tegegn and K. S. Lee, Anal. Bioanal. Chem., 2024, 416(2), 475–486.
27. D. Wippermann, A. Zonderman, T. Zimmermann and D. Profrock, Anal. Bioanal. Chem., 2024, 416(11), 2797–2807.
28. R. Guo, H. M. Yu, S. B. Fang, Z. C. Xiao and F. Huang, J. Anal. At. Spectrom., 2023, 38(11), 2365–2377.
29. M. Perrais, B. Trachsel, S. Lenglet, M. Pruijm, B. Ponte, B. Vogt, M. Augsburger, V. Rousson, M. Bochud and A. Thomas, Clin. Chem. Lab. Med., 2024, 62(11), 2242–2255.
30. A. France Stiglic, I. Falnoga, A. S. Briski, M. Zavbi, J. Osredkar, M. Skitek and J. Marc, Clin. Chem. Lab. Med., 2024, 62(5), 946–957.
31. CLSI, Defining, Establishing, and Verifying Reference Intervals in the Clinical Laboratory; Approved Guideline, 3rd edn, CLSI document EP28-A3c, Clinical and Laboratory Standards Institute, Wayne, PA, 2008.
32. M. Gasull, J. Camargo, J. Pumarega, L. A. Henriquez-Hernandez, L. Campi, M. Zumbado, M. Contreras-Llanes, L. Oliveras, P. Gonzalez-Marin, O. P. Luzardo, A. Gomez-Gutierrez, J. Alguacil and M. Porta, Sci. Total Environ., 2024, 909, 168502.
33. Y. Zebbiche, A. Amziane, M. Y. Achouri, S. Aries, S. Abdennour, C. A. Islam, A. Lalaouna and B. Alamir, Bull. Environ. Contam. Toxicol., 2024, 112(6), 85.
34. E. I. Queirolo, K. Kordas, G. Martinez, Z. Ahmed, G. Barg and N. Manay, Environ. Pollut., 2024, 343, 123160.
35. C. Söderberg, L. Rodushkin, A. Johansson and F. C. C. Kugelberg, Int. J. Leg. Med., 2023, 137(3), 655–669.
36. M. H. J. Wiesen, D. Schneider, W. Hein, T. Streichert, M. Jubner and H. Andresen-Streichert, Drug Test. Anal., 2024, 1–8.
37. J. R. Christensen and G. O. LaBine, Biol. Trace Elem. Res., 2024, 202(9), 3910–3922.
38. J. M. C. Franco, R. A. Muldrow, W. Cieslak, P. Dearmond, C. Orahoske, D. Payto, D. N. Greene and D. Bunch, J. Mass Spectrom. Adv. Clin. Lab, 2023, 30, 45–50.
39. L. Schmidt, K. Peterson, T. S. Nunes, M. Knap, L. Petrick and J. A. Landero-Figueroa, Anal. Bioanal. Chem., 2024, 416(11), 2711–2724.
40. J. Silva, R. Castelhano, F. Silva, J. P. Santos, A. Félix, J. Cruz, J. Machado and S. Pessanha, Microchem. J., 2024, 200, 110281.
41. C. Faidutti, C. Doolette, L. Hair, K. R. van Daalen, A. Naheed, E. Lombi and J. Feldmann, Anal. Chem., 2024, 96(10), 4039–4047.
42. J. Lin, Y. Li, K. Ding, X. Lin and C. Che, Anal. Methods, 2024, 16(23), 3654–3662.
43. W. Xie, G. Fu, J. Xu, M. Zeng, Q. Wan, X. Yao, P. Yang and M. Yao, J. Appl. Spectrosc., 2024, 91(1), 209–217.
44. A. Sheehan and A. Furey, Talanta Open, 2024, 9, 100309.
45. R. S. Assis, U. N. Silva, L. B. Santos, A. Silva Melo, J. J. Coutinho, U. M. F. M. Cerqueira, V. A. Lemos and M. A. Bezerra, Microchem. J., 2024, 199, 110059.
46. S. Kumar, S. Satapathy, B. D. Kurmi, G. D. Gupta and P. Patel, Sep. Sci. plus, 2024, 7, 2400018.
47. A. R. Hussein, M. S. Gburi, N. M. Muslim and E. A. Azooz, Trends Environ. Anal. Chem., 2023, 37, e00194.
48. M.-M. Zhao, H.-Z. Wu, X.-K. Deng, R.-N. Yi and Y. Yang, Anal. Methods, 2024, 16(3), 333–343.
49. E. Kowa, A. Telk and M. Wieczorek, J. Anal. At. Spectrom., 2024, 39(4), 1004–1023.
50. M. N. Oviedo, C. E. Lujan, A. A. Lemos, M. B. Botella, M. Llaver and R. G. Wuilloud, Anal. Bioanal. Chem., 2024, 416(11), 2641–2656.
51. S. J. Shahtaheri, R. Khezri, M. N. Shahrak, M. Khadem and R. Mirzaei, Appl. Spectrosc. Rev., 2024, 59(4), 551–577.
52. S. S. Ferreira, T. A. Brito, A. P. R. Santana, T. G. S. Guimarães, R. S. Lamarca, K. C. Ferreira, P. C. F. L. Gomes, A. Oliveira, C. D. B. Amaral and M. H. Gonzalez, Talanta Open, 2022, 6, 100131.
53. Y. Zhang, M. Chen, L. Li, Y. Lv and Q. Ma, TrAC, Trends Anal. Chem., 2024, 170, 117412.
54. F. Cora Jofre, S. M. Azcarate, J. M. Camiña, P. Pacheco and M. Savio, Adv. Sample Prep., 2023, 5, 100054.
55. M. Sihlahla, A. Mpupa, M. Sojka, A. Saeid and P. N. Nomngongo, Adv. Sample Prep., 2023, 8, 100092.
56. J. Silva, V. Buhl, F. Iaquinta and M. Piston, Heliyon, 2023, 9(10), e20967.
57. J. B. Pereira Jr, V. S. Carvalho, W. Q. Ferreira, R. G. O. Araujo and S. L. C. Ferreira, J. Pharm. Biomed. Anal., 2024, 238, 115810.
58. F. L. Germer, E. H. C. Lacerda, E. da Silva Macedo and E. S. Chaves, ChemistrySelect, 2024, 9, e202303825.
59. M. B. Botella, M. N. Oviedo, P. Pacheco and R. G. Wuilloud, Spectrochim. Acta, Part B, 2024, 211, 106832.
60. D. L. Arbova, S. Y. Tolmachev and J. D. Brockman, Talanta, 2024, 270, 125607.
61. N. Kavasi and S. K. Sahoo, Food Chem., 2024, 434, 137421.
62. Z. Arslan and H. Lowers, Minerals, 2024, 14(3), 299.
63. M. F. Ayyildiz, M. Şaylan, E. Yazıcı, D. S. Chormey and S. Bakırdere, Anal. Lett., 2024, 57(13), 2170–2181.
64. A. L. Vieira, G. G. A. de Carvalho, J. A. Gomes Neto, P. V. Oliveira, M. Y. Kamogawa and A. Virgilio, J. Anal. At. Spectrom., 2024, 39(2), 356–363.
65. A. C. B. Luckow, D. H. Bonemann, A. O. De Souza, S. H. Scherdien, C. C. Pereira, M. A. Vieira, A. S. Ribeiro and A. M. Nunes, J. Food Compos. Anal., 2024, 127, 105974.
66. N. Zhang, T. Zhou, M. Ye, X. Yang, Z. Ding and H. Gao, Anal. Sci., 2024, 40(7), 1349–1356.
67. E. V. Shabanova, A. A. Zak and I. E. Vasil’eva, J. Anal. Chem., 2024, 79(3), 287–302.
68. V. Ramos do Nascimento, D. Pereira de Almeida, K. Giacobe, E. Marlon de Moraes Flores and C. Augusto Bizzi, Talanta, 2024, 270, 125575.
69. V. H. Cauduro, C. Alves, M. S. Nascimento, G. T. Druzian, F. P. Balbinot, M. F. Mesko and E. M. M. Flores, Anal. Bioanal. Chem., 2024, 416(11), 2859–2870.
70. O. Apine, S. Parit, V. Kanugade, S. Sutar, S. Mahajani and J. Jadhav, J. Food Compos. Anal., 2023, 124, 105662.
71. L. Givelet, H. Amlund, K. Loeschner and J. J. Sloth, Food Anal. Methods, 2024, 17(7), 1038–1045.
72. C. M. Senger, I. S. Dornelles, J. M. Queiroz, P. A. Mello, E. I. Muller and A. L. H. Muller, J. Pharm. Biomed. Anal., 2024, 238, 115802.
73. F. Hallwirth, H. Motter and H. Wiltsche, J. Anal. At. Spectrom., 2023, 38(11), 2378–2386.
74. S. M. Sorouraddin, A. Afrouziyan, M. A. Farajzadeh and K. Parvizzad, Int. J. Environ. Anal. Chem., 2024, 1–14,  DOI:10.1080/03067319.2024.2337233.
75. A. M. Trevelin, J. O. Vinhal, L. N. Viana, T. D. Saint’Pierre and R. J. Cassella, Food Chem., 2024, 442, 138492.
76. P. H. Saleem, S. Moinfar and I. A. Mohammed, J. Food Compos. Anal., 2023, 124, 105683.
77. F. S. Costa, L. S. Moreira, L. L. Ludovico, J. Volpe, A. C. de Oliveira, M. P. Dos Santos, E. G. P. da Silva, D. E. P. Souto, M. T. Grassi, M. H. Gonzalez and C. D. B. Amaral, Talanta, 2024, 266(Pt 2), 125108.
78. H.-w. Liu, L. Fu and L. Chen, Spectrosc. Spectral Anal., 2023, 43(10), 3111–3116.
79. F. Rovasi Adolfo, N. Balczareki Lucena, L. Trombe Do Valle, L. Machado De Carvalho and S. Noremberg Kunz, Food Chem., 2024, 448, 139139.
80. R. Ninayan, U. Markova, E. Nizov, M. Melesova, A. S. Novikov and A. Shishov, Microchem. J., 2024, 200, 110252.
81. N. Fattahi, H. L. Ahmed, Z. Nematifar, N. Hashemi, M. Moradi, S. Soltani and S. Akbari, Food Addit. Contam.,:Part A, 2024, 41(6), 617–628.
82. A. Patel, S. Chakraborty, S. K. Misra and B. Datta, ACS Food Sci. Technol., 2024, 4(5), 1152–1165.
83. V. Alves, J. K. de Andrade and M. L. Felsner, J. Food Compos. Anal., 2023, 123, 105495.
84. M. Welna, A. Szymczycha-Madeja and P. Pohl, Processes, 2023, 11(8), 2428.
85. P. Pohl, K. Greda, M. Welna, P. Jamroz, A. Dzimitrowicz and A. Szymczycha-Madeja, Anal. Methods, 2024, 16(25), 4187–4197.
86. G. F. R. Santana, R. S. Rodrigues, G. P. Brandão, V. C. Costa and M. T. W. D. Carneiro, Food Anal. Methods, 2024, 17(3), 475–485.
87. A. B. P. da Silva, L. M. L. Barreto, A. S. Araújo and R. J. Cassella, Anal. Lett., 2025, 58(2), 313–328.
88. H. Doh and M. H. Lee, Foods, 2024, 13(9), 1400.
89. J. Photi, K. Judprasong, S. Gowachirapant, P. Srisakda, J. Supanuwat and C. Zeder, Foods, 2023, 12(18), 3513.
90. M. Welna, A. Szymczycha-Madeja and P. Pohl, J. Pharm. Biomed. Anal., 2023, 236, 115682.
91. A. B. Viana, C. Pappis, V. L. Garcia, D. M. Hoffmann, A. F. Burg and V. L. Dressler, J. Braz. Chem. Soc., 2024, 35(6), e20230187.
92. M. Ribeiro, E. Galli, T. Guerin, J. A. L. Silva, I. Castanheira, A. Leufroy and P. Jitaru, Anal. Bioanal. Chem., 2023, 415(29–30), 7175–7186.
93. T. Chen, W. J. Cui, Z. W. Bi, Y. F. Guo, G. L. Zhu, X. Y. Pan, Z. Q. Cui, Z. Yang and T. L. Deng, J. Chem., 2023, 2023, 1–9.
94. F. Zhao, Y. Liu, F. Zhang, R. Dong, W. Yu, H. Zhang, X. Han, P. Gong, X. Zhang, Y. Liu, Z. Guo, X. Liang and J. Zhu, Microchem. J., 2023, 193, 109168.
95. J. C. Hu, X. M. Lv, H. Zhang, B. L. Zhu, D. Y. Liu and W. L. Ji, J. Liq. Chromatogr. Relat. Technol., 2023, 46(11–15), 218–224.
96. A. Sajnóg, K. Bierla, J. Szpunar and J. Jiménez-Lamana, J. Anal. At. Spectrom., 2023, 38(11), 2448–2457.
97. M. Klein, M. Menta, F. Auger, F. Ait-Ben-Ahmad, A. Papin, H. Biaudet, O. F. X. Donard and F. Séby, J. Food Compos. Anal., 2024, 125, 105836.
98. M. Zhang, H. Wang, Y. Wu and X. Yu, Anal. Chim. Acta, 2023, 1279, 341846.
99. S. Santini, B. Campanella, S. Giannarelli, V. Palleschi, F. Poggialini and S. Legnaioli, Spectrochim. Acta, Part B, 2024, 216, 106948.
100. A. Gianoncelli, K. V. Mikuš, M. Salomé, E. Pascotto, G. Ricci and L. Pascolo, J. Anal. At. Spectrom., 2023, 38(9), 1744–1750.
101. K. Lossow, W. Schlormann, M. Tuchtenhagen, M. Schwarz, T. Schwerdtle and A. P. Kipp, J. Trace Elem. Med. Biol., 2023, 78, 127167.
102. M. Musielak, E. Musielak and R. Sitko, Food Chem., 2024, 439, 138156.
103. N. V. Alov, D. V. Danilov, P. Y. Sharanov, V. G. Semenov, V. V. Panchuk, S. S. Savinov and M. M. Khaydukova, X-Ray Spectrom., 2024, 53(3), 181–189.
104. J. Karasinski, K. Tetfejer, P. Radzinski, A. Tupys, A. Gambin, E. Bulska and L. Halicz, Anal. Chem., 2024, 96(9), 3763–3771.
105. C.-F. Li, X.-C. Wang, Z.-Y. Chu and P. Peng, J. Anal. At. Spectrom., 2024, 39(1), 235–243.
106. Z. Macsik, S. P. LaMont, A. M. R. Cardon, L. A. Hudston, D. Poudel, A. M. Wende, J. D. Inglis and R. E. Steiner, J. Radioanal. Nucl. Chem., 2022, 331(12), 5359–5369.
107. M. Nian, B. Chen, M. He and B. Hu, Anal. Chem., 2024, 96(2), 766–774.
108. P. Menero-Valdes, M. I. Chronakis, B. Fernandez, J. C. D. Quarles, H. Gonzalez-Iglesias, B. Meermann and R. Pereiro, Anal. Chem., 2023, 95(35), 13322–13329.
109. L. Hendriks, V. M. Kissling, T. Buerki-Thurnherr and D. M. Mitrano, Environ. Sci.: Nano, 2023, 10(12), 3439–3449.
110. X. Zhang, X. Wei, C.-X. Wu, X. Men, J. Wang, J.-J. Bai, X.-Y. Sun, Y. Wang, T. Yang, C. T. Lim, M.-L. Chen and J.-H. Wang, ACS Nano, 2024, 18(8), 6612–6622.
111. M. Liu, X. Wei, C. Wu, J. Liu, Y. Wei, X. Wang, M.-L. Chen, T. Yang and J.-H. Wang, Anal. Chem., 2023, 95(38), 14447–14454.
112. L. Xu, A. Sigler, A. Chernatynskaya, L. Rasmussen, J. Lu, E. Sahle-Demessie, D. Westenberg, H. Yang and H. Shi, Anal. Bioanal. Chem., 2024, 416(2), 419–430.
113. T. Fernandez-Bautista, B. Gomez-Gomez, D. Vicente-Zurdo and Y. Madrid, Anal. Bioanal. Chem., 2024, 416(11), 2749–2759.
114. K. Loeschner, M. E. Johnson and A. R. Montoro Bustos, Nanomaterials, 2023, 13(18), 2547.
115. I. Bastardo-Fernandez, R. Chekri, J. Noireaux, L. Givelet, N. Lambeng, A. Delvallee, K. Loeschner, P. Fisicaro and P. Jitaru, Food Addit. Contam.,:Part A, 2024, 41(8), 867–884.
116. D. Ojeda, D. Bartczak, M. Singh, P. Hancock and H. Goenaga-Infante, J. Anal. At. Spectrom., 2024, 39(7), 1813–1823.
117. K. Hobin, M. Costas Rodriguez, E. Van Wonterghem, R. E. Vandenbroucke and F. Vanhaecke, Anal. Chim. Acta, 2024, 1315, 342812.
118. M. Horstmann, C. D. Quarles Jr, S. Happel, M. Sperling, A. Faust, K. Rahbar, D. Clases and U. Karst, Anal. Bioanal. Chem., 2024, 416(11), 2849–2858.
119. M. Horstmann, M. Austrup, F. Busch, A. Faust, M. Sperling, U. Karst and D. Clases, J. Anal. At. Spectrom., 2023, 38(10), 2038–2045.
120. G. Yang, H. Tazoe, E. Kim, J. Zheng, M. Kowatari and O. Kurihara, J. Anal. At. Spectrom., 2023, 38(12), 2562–2570.
121. C. Peng, J. Sun, F. Zhang, S. Xing, X. Liu, C. Chen, X. Hou, K. Shi and W. Wu, Anal. Chem., 2024, 96(6), 2514–2523.
122. Z. You, A. Winckelmann, J. Vogl, S. Recknagel and C. Abad, Anal. Bioanal. Chem., 2024, 416(13), 3117–3125.
123. S. Mukta and A. Gundlach-Graham, J. Anal. At. Spectrom., 2024, 39(2), 491–499.
124. D. Tukhmetova, J. Lisec, J. Vogl and B. Meermann, Anal. Chem., 2024, 96(8), 3276–3283.
125. T. Narukawa, S. Okabayashi, M. Tsuboi and K. Chiba, Anal. Methods, 2024, 16(10), 1523–1530.
126. F. Li, H. Cui, D. Zhang, H. Li, W. Guo, L. Jin and S. Hu, Anal. Bioanal. Chem., 2023, 415(24), 6051–6061.
127. M. Aramendía, A. L. M. de Souza, F. V. Nakadi and M. Resano, J. Anal. At. Spectrom., 2024, 39(3), 767–779.
128. L. A. Oliveira, J. L. O. Santos and L. S. G. Teixeira, Talanta, 2024, 266(Pt 1), 124945.
129. X. Guo, C. Y. Lin, S. Alavi, L. You and J. Mostaghimi, Anal. Chim. Acta, 2023, 1281, 341906.
130. J. Liu, J. Dong, S. Han, J. Zhang, X. Liu, H. Zheng and Z. Zhu, J. Anal. At. Spectrom., 2024, 39(5), 1343–1352.
131. R. Serrano, G. Grindlay, L. Gras and J. Mora, Talanta, 2024, 271, 125666.
132. J. Manrique, P. Garrido and J. Velasco, Atoms, 2024, 12(4), 21.
133. Z. Hou, L. Guo, W. Wang, Y. Chu and F. Lin, Laser Optoelectron. Prog., 2023, 60(24), 2400004.
134. E. J. Blanchette, E. A. Tracey, A. Baughan, G. E. Johnson, H. Malik, C. N. Alionte, I. G. Arthur, M. E. S. Pontoni and S. J. Rehse, Spectrochim. Acta, Part B, 2024, 215, 106911.
135. J. Lin, Y. Li, X. Lin and C. Che, Talanta, 2024, 275, 126194.
136. J. Lin, Y. Li, X. Lin and C. Che, Anal. Bioanal. Chem., 2024, 416(4), 993–1000.
137. A. Khumaeni, W. S. Budi, R. Hedwig, K. Kurihara, M. Tani and K. H. Kurniawan, Talanta Open, 2024, 9, 100322.
138. X. Zhang, X. Li, X. Chen, M. Shi and T. Ren, J. Anal. At. Spectrom., 2024, 39(5), 1332–1342.
139. J. Park, S. H. Han, S. H. Nam and Y. Lee, Appl. Spectrosc., 2023, 77(12), 1351–1361.
140. X. Wen, F. Cao, C. Yang, Z. Gao, H. Tian, X. Zhao, L. Guo, S. Ma and D. Dong, Food Chem., 2024, 448, 139210.
141. Q. Wang, A. Chen and X. Gao, J. Anal. At. Spectrom., 2024, 39(1), 261–268.
142. X. Yang, X. Wang, D. Li, X. Zhang, K. Li, H. Ren, Z. Zhou, Z. Qin and X. Zheng, J. Anal. At. Spectrom., 2024, 39(2), 433–438.
143. M. Cicero Ribeiro, J. Cabral, G. Nicolodelli, G. S. Senesi, A. R. L. Caires, D. A. Gonçalves, C. Menegatti, D. Milori, C. Cena and B. Marangoni, Microchem. J., 2024, 203, 110898.
144. A. Fayyaz, N. Ali, Z. A. Umar, H. Asghar, M. Waqas, R. Ahmed, R. Ali and M. A. Baig, Anal. Sci., 2024, 40(3), 413–427.
145. M. Atasoy, ChemistrySelect, 2023, 8(41), e202302370.
146. M. Atasoy and D. Yildiz, Anal. Lett., 2024, 57(12), 1892–1906.
147. J. Schlotthauer, L. Mántaras, L. Brusa and M. Sigrist, Spectrochim. Acta, Part B, 2024, 212, 106851.
148. N. Manousi and A. N. Anthemidis, Anal. Chim. Acta, 2024, 1290, 342208.
149. L. Dong, W. Wang, Y. Ning, X. Deng and Y. Gao, Anal. Chim. Acta, 2023, 1251, 341006.
150. G. Lan, X. Li, J. Yao, X. Yu, Q. Liu, C. Qiu and X. Mao, Front. Nutr., 2023, 10, 1201801.
151. G. R. Bitencourt, F. A. Duarte, V. L. Dressler, R. C. Bolzan, E. M. D. Flores and P. A. Mello, J. Anal. At. Spectrom., 2023, 38(11), 2342–2352.
152. E. Marguí, D. Eichert, J. Jablan, F. Bilo, L. E. Depero, A. Pejović-Milić, A. Gross, H. Stosnach, A. Kubala-Kukuś, D. Banaś and L. Borgese, J. Anal. At. Spectrom., 2024, 39(7), 1700–1719.
153. E. Visentin, G. Niero and M. De Marchi, Food Chem., 2023, 429, 136892.
154. M. Weinberger, I. Queralt, C. Streli, P. Wobrauschek, E. Besalú, J. Jablan and E. Marguí, Spectrochim. Acta, Part B, 2024, 215, 106898.
155. S. Wu, P. Zhao, A. Wang, L. Dong, X. Wu, J. Wang and Y. Zhang, Biol. Trace Elem. Res., 2024, 202(6), 2755–2763.
156. S. H. Wu, L. P. Dong, J. H. Ji, P. Zhao, G. Song, X. J. Wu, J. F. Wang, Y. F. Zhang and J. J. Wang, J. Anal. At. Spectrom., 2023, 38(10), 2127–2133.
157. S. Pandolfi, J. Machado, F. Silva, J. P. Santos, A. Felix and S. Pessanha, At. Spectrosc., 2024, 45(2), 99–105.
158. J. Labuda, J. Barek, Z. Gajdosechova, S. Jacob, L. Johnston, P. Krystek, Z. Mester, J. Moreira, V. Svitkova and K. J. Wilkinson, Pure Appl. Chem., 2023, 95(11), 1159–1196.
159. S. Fernandez-Trujillo, M. Jimenez-Moreno, N. Rodriguez-Farinas and R. C. Rodriguez Martin-Doimeadios, Anal. Bioanal. Chem., 2024, 416(11), 2657–2676.
160. C. Davison, D. Beste, M. Bailey and M. Felipe-Sotelo, Anal. Bioanal. Chem., 2023, 415(28), 6931–6950.
161. Y. Zhou, W. Du, Y. Chen, L. Li, X. Xiao, Y. Xu, W. Yang, X. Hu, B. Wang, J. Zhang, Q. Jiang and Y. Wang, Talanta, 2024, 277, 126325.
162. I. Bastardo-Fernández, R. Chekri, J. Noireaux, P. Fisicaro and P. Jitaru, Spectrochim. Acta, Part B, 2023, 208, 106782.
163. E. Espada-Bernabe, G. Moreno-Martin, B. Gomez-Gomez and Y. Madrid, Food Chem., 2024, 432, 137201.
164. A. Bruvold, S. Valdersnes, K. Loeschner and A. M. Bienfait, J. AOAC Int., 2024, 107(4), 608–616.
165. A. Chalifoux, M. Hadioui, N. Amiri and K. J. Wilkinson, Molecules, 2023, 28(11), 4442.
166. B. Li, S. L. Chua, D. Yu, S. H. Chan and A. Li, Molecules, 2024, 29(1), 248.
167. R. G. de Vega, T. E. Lockwood, L. Paton, L. Schlatt and D. Clases, J. Anal. At. Spectrom., 2023, 38(12), 2656–2663.
168. A. Sanchez-Cachero, M. Jimenez-Moreno, N. R. Farinas and R. C. R. Martin-Doimeadios, Microchim. Acta, 2023, 190(12), 476.
169. J. C. Garcia-Mesa, I. Morales-Benitez, P. Montoro-Leal, M. M. Lopez Guerrero and E. I. Vereda Alonso, Talanta, 2024, 268(Pt 1), 125360.
170. A. Justo-Vega, R. Dominguez-Gonzalez, P. Bermejo-Barrera and A. Moreda-Pineiro, J. Anal. At. Spectrom., 2024, 39(2), 573–581.
171. S. Zhou, H. Wang, J. Tang, H. Wang and J. Yan, Anal. Sci., 2024, 40(3), 555–562.
172. X. Zhang, X.-Y. Jiang, L. Zhao, S. Chen, Y.-L. Yu and J.-H. Wang, Anal. Chem., 2024, 96(4), 1742–1749.
173. X. Zhang, X.-Y. Jiang, J.-Y. Cai, S. Chen, Y.-L. Yu and J.-H. Wang, Anal. Chem., 2024, 96(18), 7155–7162.
174. M. E. Del Castillo Busto, C. Ward-Deitrich, S. O. Evans, M. P. Rayman, M. B. Jameson and H. Goenaga-Infante, Anal. Bioanal. Chem., 2024, 416(11), 2835–2848.
175. E. Matsumoto, S. Ito and T. Nishimura, Mar. Biotechnol., 2023, 25(5), 677–682.
176. R. Sim, M. Weyer and A. H. Petursdottir, Anal. Bioanal. Chem., 2024, 416(12), 3033–3044.
177. A. Morales-Rodriguez, A. Sahuquillo, C. Minguillon, J. F. Lopez-Sanchez and D. Barron, Anal. Chim. Acta, 2024, 1315, 342814.
178. E. Matsumoto and M. Matsumoto, J. AOAC Int., 2023, 106(6), 1525–1531.
179. D. Spanu, S. Fantinuoli, G. Binda, L. Rampazzi, D. Monticelli and S. Recchia, Spectrochim. Acta, Part B, 2023, 207, 106762.
180. A. S. Anil, S. Alam and L. K. Thakur, J. Food Compos. Anal., 2024, 129, 106092.
181. P. C. C. Ladeira, C. C. Augusto, B. A. Rocha, J. L. Rodrigues, G. d. F. M. Aguiar and B. L. Batista, Toxics, 2024, 12(6), 424.
182. T. Fernandez-Bautista, B. Gomez-Gomez, E. Gracia-Lor, T. Perez-Corona and Y. Madrid, Food Chem., 2024, 435, 137544.
183. A. Jerse, H. Amlund, S. L. Holdt and J. J. Sloth, Environ. Chem., 2023, 20(2), 95–104.
184. Y. Wu, X. Wang, Z. Pang, J. Zhang and C. Zheng, Microchem. J., 2024, 203, 110809.
185. I. Sowik, J. Zajda and L. Ruzik, Microchem. J., 2023, 191, 108857.
186. K. Mervič, M. Šala and S. Theiner, TrAC, Trends Anal. Chem., 2024, 172, 117574.
187. B. L. Gorman, S. V. Torti, F. M. Torti and C. R. Anderton, Biochim. Biophys. Acta, Gen. Subj., 2024, 1868(2), 130329.
188. M. E. Graziotto, C. J. Kidman, L. D. Adair, S. A. James, H. H. Harris and E. J. New, Chem. Soc. Rev., 2023, 52(23), 8295–8318.
189. Q. Shen, S. Wang, H. Wang, J. Liang, Q. Zhao, K. Cheng, M. Imran, J. Xue and Z. Mao, Compr. Rev. Food Sci. Food Saf., 2024, 23(4), e13398.
190. F. Zhou, Y. Liu, W. Xie, J. Huang, F. Liu, W. Kong, Z. Zhao and J. Peng, Crit. Rev. Food Sci. Nutr., 2023, 1–17.
191. T. Van Helden, K. Mervic, I. Nemet, J. T. van Elteren, F. Vanhaecke, S. Roncevic, M. Sala and T. Van Acker, Anal. Chim. Acta, 2024, 1287, 342089.
192. M. G. Mello, T. E. Lockwood, J. Wanagat, M. T. Westerhausen and D. P. Bishop, J. Anal. At. Spectrom., 2024, 39(7), 1720–1725.
193. A. Tisza, T. Klikovits, M. Benej, S. Torok, B. Szeitz, Z. Valko, M. A. Hoda, B. Hegedus, M. Bonta, W. Nischkauer, K. Hoetzenecker, A. Limbeck, K. Schelch, V. Laszlo, Z. Megyesfalvi and B. Dome, Br. J. Clin. Pharmacol., 2023, 89(11), 3364–3374.
194. R. Gonzalez de Vega, D. Clases, B. A. Cunningham, K. Ganio, S. L. Neville, C. A. McDevitt and P. A. Doble, Anal. Bioanal. Chem., 2024, 416(11), 2783–2796.
195. P. Niehaus, R. G. de Vega, M. T. Haindl, C. Birkl, M. Leoni, A. M. Birkl-Toeglhofer, J. Haybaeck, S. Ropele, M. Seeba, W. Goessler, U. Karst, C. Langkammer and D. Clases, Talanta, 2024, 270, 125518.
196. H. Koprivova, K. Kiss, L. Krbal, V. Stejskal, J. Buday, P. Porizka, M. Kaska, A. Ryska and J. Kaiser, Anal. Chim. Acta, 2024, 1310, 342663.
197. V. Gardette, L. Sancey, M. Leprince, L. Gate, F. Cosnier, C. Seidel, S. Valentino, F. Pelascini, J.-L. Coll, M. Peoc’h, V. Scolan, F. Paysant, V. Bonneterre, C. Dujardin, B. Busser and V. Motto-Ros, Small Sci., 2024, 4(5), 2300307.
198. V. H. C. Ferreira, V. Gardette, B. Busser, L. Sancey, S. Ronsmans, V. Bonneterre, V. Motto-Ros and L. Duponchel, Anal. Chem., 2024, 96(18), 7038–7046.
199. A. Graefenstein, D. Brueckner, C. Rumancev, J. Garrevoet, V. Galbierz, W. H. Schroeder, C. G. Schroer, G. Falkenberg and A. Rosenhahn, Anal. Chem., 2023, 95(27), 10186–10195.
200. A. N. Webb, O. Antipova, S. Shughoury, J. M. Farfel, D. A. Bennett, Y. Du, W. Zheng and L. H. Nie, Metallomics, 2024, 16(5), mfae018.
201. J. Hu, X. Yan and X. Chris Le, Anal. Bioanal. Chem., 2024, 416(11), 2625–2640.
202. H. Du, P. Yang, L. Xia, J. Chen and X. Hou, TrAC, Trends Anal. Chem., 2024, 172, 117552.
203. B. Billet, B. Chovelon, E. M. McConnell, D. Andre, L. Puillet-Anselme, E. Fiore, P. Faure, C. Ravelet, M. C. DeRosa and E. Peyrin, Talanta, 2024, 267, 125107.
204. A. Rodriguez-Penedo, E. Costa-Rama, B. Fernandez, C. Garcia-Cabo, L. Benavente, S. Calleja, M. T. Fernandez-Abedul and R. Pereiro, Microchim. Acta, 2023, 190(12), 493.
205. Y. He, J. Hu, S. Wei, C. Li and X. Hou, Microchem. J., 2024, 200, 110342.
206. C. Liu, P. Li, X. Yan, L. Yang, P. Liu and Q. Wang, Anal. Bioanal. Chem., 2024, 416(11), 2691–2697.
207. X. Li, X. Li, J. Chen, Z. Tan and C. Wang, Anal. Biochem., 2024, 692, 115559.
208. Y. Wang, X. Chen, X. Shen, Y. He, Z. Zhan, C. Liu, Y. Xie, F. Lin, K. Huang and P. Chen, Anal. Chem., 2023, 95(38), 14244–14252.
209. Y. Cheng, Q. Xie, R. Xu, M. He, B. Chen, G. Chen and B. Hu, Sens. Actuators, B, 2024, 416, 136013.
210. G. Sun, W. Liu, J. Liu, L. Sheng, Z. Xing, S. Zhang and X. Zhang, At. Spectrosc., 2023, 44(6), 392–400.
211. H. Tang, G. Sun, Y. Xu, S. Men, W. Jiang and C. Wang, J. Immunol. Methods, 2024, 528, 113666.
212. CLSI, Evaluation of Detection Capability for Clinical Laboratory Measurement Procedures; Approved Guideline-second Edition: CLSI Document EP17-A2, Clinical and Laboratory Standards Institute, Wayne, PA, 2012.
213. CLSI, Measurement Procedure Comparison and Bias Estimation Using Patient Samples; Approved Guideline-third Edition: CLSI Document EP09c, Clinical and Laboratory Standards Institute, Wayne, PA, 2013.
214. CLSI, Evaluation of Precision of Quantitative Measurement Procedures; Approved Guideline-third Edition: CLSI Document EP05-A3, Clinical and Laboratory Standards Institute, Wayne, PA, 2014.
215. Y. Wang, L. Shao, Z. Zhao, C. Huang, Y. Jiao, D. Sun, R. Liu, D. Jiang and X. Gao, Anal. Chim. Acta, 2024, 1294, 342272.
216. X. Chen, X. Zhao, S. Pu, R. Liu and Y. Lv, At. Spectrosc., 2023, 44(6), 427–433.
217. D. Torregrosa, M. Jauset-Rubio, R. Serrano, M. Svobodova, G. Grindlay, C. K. O’Sullivan and J. Mora, Anal. Chim. Acta, 2023, 1252, 341042.
218. F. Gregar, J. Gallo, D. Milde, J. Hegrova, P. Kucerova, J. Grepl and T. Pluhacek, Anal. Bioanal. Chem., 2024, 416(16), 3785–3796.
219. A. Fernandes, P. Yang, D. Armstrong and R. Franca, Talanta, 2023, 265, 124909.
220. S. Guergan, B. Boeer, R. Fugunt, G. Helms, C. Roehm, A. Solomianik, A. Neugebauer, D. Nuessle, M. Schuermann, K. Brunecker, O. Jurjut, K. A. Boehme, S. Dammeier, M. D. Enderle, S. Bettio, I. Gonzalez-Menendez, A. Staebler, S. Y. Brucker, B. Kraemer, D. Wallwiener, F. Fend and M. Hahn, Diagnostics, 2024, 14(3), 338.
221. K. Schilling, R. A. Glabonjat, O. Balac, M. Galvez-Fernandez, A. Domingo-Relloso, V. Slavkovich, J. Goldsmith, M. R. Jones, T. R. Sanchez and A. Navas-Acien, Anal. Methods, 2024, 16(2), 214–226.
222. A. C. Bitzer, J. Fox, P. L. Day, V. K. Pazdernik, C. Y. Smith, M. Wermers, P. J. Jannetto and J. A. Bornhorst, Arch. Pathol. Lab. Med., 2024, 148(7), 818–827.
223. L. Yu, S. Yousuf, S. Yousuf, J. Yeh, S. W. Biggins, C. Morishima, I. Shyu, G. O’Shea-Stone, B. Eilers, A. Waldum, V. Copie and J. Burkhead, Hepatol. Commun., 2023, 7(3), e0076.
224. N. K. Monangi, H. Xu, Y. M. Fan, R. Khanam, W. Khan, S. Deb, J. Pervin, J. T. Price, L. Kaur, INTERBIO-21st Study Consortium, A. Al Mahmud, L. Q. Thanh, A. Care, J. A. Landero, G. F. Combs, E. Belling, J. Chappell, J. Chen, F. Kong, C. Lacher, S. Ahmed, N. H. Chowdhury, S. Rahman, F. Kabir, I. Nisar, A. Hotwani, U. Mehmood, A. Nizar, J. Khalid, U. Dhingra, A. Dutta, S. M. Ali, F. Aftab, M. H. Juma, M. Rahman, T. Ahmed, M. M. Islam, B. Vwalika, P. Musonda, U. Ashorn, K. Maleta, M. Hallman, L. Goodfellow, J. K. Gupta, A. Alfirevic, S. K. Murphy, L. Rand, K. K. Ryckman, J. C. Murray, R. Bahl, J. A. Litch, C. Baruch-Gravett, S. Sopory, U. Chandra Mouli Natchu, P. V. Kumar, N. Kumari, R. Thiruvengadam, A. K. Singh, P. Kumar, GARBH-Ini study team, Z. Alfirevic, A. H. Baqui, S. Bhatnagar, J. E. Hirst, C. Hoyo, F. Jehan, L. Jelliffe-Pawlowski, A. Rahman, D. E. Roth, S. Sazawal, J. S. A. Stringer, P. Ashorn, G. Zhang and L. J. Muglia, Am. J. Clin. Nutr., 2024, 119(1), 221–231.
225. J. Omena, F. F. Bezerra, V. M. Voll, B. F. Braz, R. E. Santelli, C. M. Donangelo, G. F. Jauregui, A. S. Ribeiro, C. Dos Santos Cople Rodrigues and M. Citelli, Eur. J. Nutr., 2024, 63(6), 2163–2172.
226. N. Ruprecht, D. Parakkattel, L. Hofmann, P. Broekmann, N. Ludi, C. Kempf, J. T. Heverhagen and H. von Tengg-Kobligk, Invest. Radiol., 2024, 59(5), 372–378.
227. D. Spanu, L. Butti, S. Recchia, C. Dossi and D. Monticelli, Talanta, 2024, 270, 125612.
228. J. M. Gonzalez-Rubio, N. Dominguez-Morueco, S. Pedraza-Diaz, A. Canas Portilla, M. A. Lucena, A. Rodriguez, A. Castano and M. Esteban-Lopez, Environ. Int., 2023, 177, 107958.
229. Y. Liu, G. Xiao and R. L. Jones, J. Radioanal. Nucl. Chem., 2022, 331(9), 3957–3964.
230. C. Peng, L. Kang, X. Yuan, J. Qiao, Y. Fan, L. Yao, K. Qi, Y. Sun, X. Dai, Y. Zhang and Q. Huo, Processes, 2023, 11(10), 2821.
231. R. H. Wallimann, H. Hensinger, C. Muller, R. Schibli, R. Kneuer and P. Schindler, Pharmaceutics, 2024, 16(3), 299.
232. Q. Wu, J. Lv, F. Xie and S. Li, J. Pharm. Biomed. Anal., 2024, 242, 116026.
233. I. F. P. C. Noronha, F. C. Pinheiro and C. C. Nascentes, Talanta, 2024, 271, 125604.
234. M. Ratnaraju, P. V. M. Rao, K. B. Dasari, A. Tejaswi, M. K. Raju and B. C. Sekhar, J. Radioanal. Nucl. Chem., 2023, 332(12), 5247–5252.
235. C. Martinez Lopez, J. A. Carter, R. Corzo and T. I. Todorov, Food Anal. Methods, 2022, 15(11), 3079–3091.
236. G. Pavlovich-Cristopulos, B. Schiavo, F. M. Romero, E. Hernandez-Mendiola, A. Angulo-Molina and D. Meza-Figueroa, J. Food Compos. Anal., 2023, 115, 104990.
237. Z. Gajdosechova, C. H. Palmer, W. Sukhaket, P. Kumkrong, K. Busarakam, P. Khetnon, S. Deawtong and Z. Mester, Chemosphere, 2023, 141032.
238. G. Song, H. Tan, C. Cheng, P. Li, X. Sun, Y. Zhou and Y. Fang, Toxics, 2024, 12(5), 325.
239. Y. Yang, B. Zhou and C. Zheng, Molecules, 2024, 29(8), 1795.
240. M. Balal Arain, H. Elzain Hassan Ahmed and M. Soylak, Microchem. J., 2024, 199, 109922.
241. A. Kovalenko, D. Stutts and P. J. Gray, J. AOAC Int., 2024, 107(4), 632–640.
242. C. J. Huang, J. Z. Li, C. M. Hwu, H. S. Chen, C. C. Yeh, F. F. Wang and C. C. Yang, Biol. Trace Elem. Res., 2024, 202(4), 1517–1523.
243. D.-O. Nouzha, G. Louise, S. B. Claudia, F.-M. Heloise, L. Laurence, P. Joel and R. Virginie, J. Trace Elem. Med. Biol., 2023, 80, 127317.
244. M. T. Perrin, R. Pawlak, N. Judd, J. Cooper and G. L. Donati, Br. J. Nutr., 2023, 130(6), 1005–1012.
245. D. A. Tafere and M. A. Teshager, J. Chem., 2023, 2023, 1–11.
246. P. Kargarghomsheh, F. Tooryan, G. Sharifiarab, M. Moazzen, N. Shariatifar and M. Arabameri, Biol. Trace Elem. Res., 2024, 202(5), 2338–2346.
247. N. M. Ageyeva, A. A. Khrapov, A. A. Shirshova, L. E. Chemisova, E. V. Ulyanovskaya and E. A. Chernutskaya, Foods Raw Mater., 2024, 12(2), 273–282.
248. A. dos Santos Silva, M. C. Maciel, A. A. F. de Oliveira and T. F. de Oliveira, J. Food Compos. Anal., 2024, 128, 106050.
249. A. Mara, M. Migliorini, M. Ciulu, R. Chignola, C. Egido, O. Nunez, S. Sentellas, J. Saurina, M. Caredda, M. A. Deroma, S. Deidda, I. Langasco, M. I. Pilo, N. Spano and G. Sanna, Foods, 2024, 13(2), 243.
250. E. Di Salvo, R. Tardugno, V. Nava, C. Naccari, A. Virga, A. Salvo, F. Corbo, M. L. Clodoveo and N. Cicero, Toxics, 2023, 11(8), 705.
251. S. H. Hur, S. Kim, H. Kim, S. Jeong, H. Chung, Y. K. Kim and H. J. Kim, Curr. Res. Food Sci., 2023, 6, 100532.
252. D. Lledo, G. Grindlay, R. Serrano, L. Gras and J. M. Sansano, Spectrochim. Acta, Part B, 2023, 204, 106672.
253. A. A. Unar, T. G. Kazi, H. I. Afridi, J. A. Baig and A. A. Lashari, Talanta, 2024, 273, 125847.
254. N. M. Abdulhussein, N. M. Muslim, M. A. Hussien, E. A. Azooz and E. A. J. Al-Mulla, J. Iran. Chem. Soc., 2024, 21(5), 1203–1212.
255. A. Shishov, I. Timofeeva, A. Gerasimov, D. Israelyan and A. Bulatov, Talanta, 2024, 266(Pt 2), 125078.
256. F. A. Semysim, R. K. Ridha, E. A. Azooz and D. Snigur, Talanta, 2024, 272, 125782.
257. N. Rahman and I. Ahmad, Chemosphere, 2024, 351, 141272.
258. S. Bodur, S. Erarpat, Ö. F. Tutar and S. Bakırdere, J. Food Compos. Anal., 2023, 123, 105531.
259. Y. Cao, J. Qin, Z. Su, L. Cai, G. Fang and S. Wang, Food Chem.:X, 2023, 20, 100926.
260. N. Altunay, B. Hazer, M. F. Lanjwani and M. Tuzen, Foods, 2023, 12(19), 3620.
261. E. Yilmaz, A. Boztug, H. Ul Haq, G. Boczkaj and N. Altunay, Food Chem., 2024, 435, 137590.
262. B. Dogan and N. Altunay, Food Addit. Contam.,:Part A, 2023, 40(12), 1600–1613.
263. Y. Surme, G. K. Yildirim, M. Ucan and B. Narin, Microchem. J., 2024, 196, 109648.
264. M. Rouhi, J. Abolhasani, M. R. Afshar Mogaddam and M. T. Vardini, Chem. Pap., 2023, 77(12), 7421–7429.
265. W.-H. Tsai and C.-K. Su, Anal. Chem., 2024, 96(11), 4469–4478.
266. N. Ng, A. S. Abdul Keyon, W. A. Wan Ibrahim, M. M. Sanagi, Z. A. Sutirman and F. Mohd Marsin, J. Food Compos. Anal., 2023, 123, 105561.
267. Z. Shakourzadeh, S. M. Sorouraddin, M. A. Farajzadeh and A. A. Fathi, Chem. Pap., 2024, 78(9), 5501–5512.
268. M. Morvaridi, S. M. Sorouraddin, M. R. A. Mogaddam and M. A. Farajzadeh, Int. J. Environ. Anal. Chem., 2023, 1–15.
269. S. Yilmaz, N. Ullah, D. Citak, B. Hazer and M. Tuzen, J. Food Compos. Anal., 2023, 123, 105487.
270. V. Christoforou, N. Manousi, C. K. Zacharis and A. Anthemidis, Sustainable Chem. Pharm., 2024, 39, 101567.
271. A. A. Zendehdel, S. M. Sorouraddin and M. A. Farajzadeh, Anal. Methods, 2024, 16(11), 1593–1602.
272. A. Elik and N. Altunay, Sustainable Chem. Pharm., 2024, 39, 101602.
273. H. Ul Haq, A. Elik, G. Isci, M. Ekici, N. Gursoy, G. Boczkaj and N. Altunay, Food Chem., 2024, 447, 139024.
274. X. Ou, M. He, B. Chen and B. Hu, J. Hazard. Mater., 2024, 462, 132680.
275. A. Changhizi, S. M. Sorouraddin, M. A. Farajzadeh and A. A. Fathi, Anal. Bioanal. Chem. Res., 2024, 11(1), 41–50.
276. U. Haseen, S. G. Ali, R. A. Khan, H. Ahmad and B. H. Koo, J. Food Compos. Anal., 2024, 127, 105920.
277. H. Zolfaghari, S. M. Daryanavard and A. Gholamshahzadeh, New J. Chem., 2024, 48(20), 9119–9128.
278. M. R. Jamali, P. Khodayar and R. Rahnama, J. Dispersion Sci. Technol., 2024, 45(11), 2103–2111.
279. M. R. Jamali, M. R. Kardgar, R. Rahnama and M. Babazadeh, Anal. Lett., 2024, 57(17), 2840–2854.
280. C. E. Caliskan, Iran. J. Chem. Chem. Eng., 2023, 42(4), 1282–1293.
281. K. Abdi, M. Ezoddin and N. Pirooznia, Int. J. Environ. Anal. Chem., 2023, 103(16), 4806–4819.
282. A. G. Siqueira, B. E. d. S. Costa, L. L. Romualdo and V. N. Alves, Anal. Chem. Lett., 2023, 13(4), 355–368.
283. J. Ali, M. Tuzen, W. Bux Jatoi, M. Ahmed Jakhrani and B. Hazer, Food Chem., 2024, 446, 138871.
284. J. Yan, C. Zhang, C. Wang, D. Lu and S. Chen, Food Chem., 2023, 406, 134276.
285. L. Trzonkowska, B. Lesniewska and B. Godlewska-Zylkiewicz, Water, 2022, 14(4), 529.
286. K. Turan, R. Saygili Canlidinc and O. M. Kalfa, Turk. J. Chem., 2022, 46(2), 550–566.
287. D. F. Moreira Naves, C. Dos Reis Feliciano and M. G. Santos, Anal. Chim. Acta, 2024, 1296, 342308.
288. C. Zapater, M. A. Aguirre, N. Gonzalez-Gallardo, D. J. Ramon, L. Vidal and A. Canals, Talanta, 2024, 274, 125939.
289. P. Sun, C. Wang, S. Li, N. Li and Y. Gao, Anal. Bioanal. Chem., 2024, 416(15), 3533–3542.
290. A. Elik, H. Ul Haq, G. Boczkaj, S. Fesliyan, O. Ablak and N. Altunay, J. Food Compos. Anal., 2024, 125, 105843.
291. S. Jabbari, S. M. Sorouraddin, M. A. Farajzadeh and A. A. Fathi, Anal. Methods, 2023, 15(34), 4321–4330.
292. C. Caner, M. S. Dundar, S. Arpaözü, H. Altundag, M. Arslan and Ö. Güleç, J. Chem. Metrol., 2024,(1), 60–70.
293. W. I. Mortada, S. Shaaban, H. A. Althikrallah, M. Alaasar, H. A. Alshwyeh and A. H. Ragab, J. Food Compos. Anal., 2024, 132, 106358.
294. J. Yan, S. Zhang, C. Wang, D. Lu and S. Chen, Anal. Lett., 2024, 57(1), 17–29.
295. A. Elik, N. Gursoy and N. Altunay, Chem. Pap., 2024, 78(5), 3261–3272.
296. S. R. Garmarodi, A. Nilchi, M. R. Yaftian and J. Rafiei, Int. J. Environ. Sci. Technol., 2023, 20(3), 2829–2838.
297. T. Ahmad, F. Shah, R. A. Khan and A. Y. Ahmed, Water, 2023, 15(21), 3750.
298. S. Tokalioglu, S. Shahir, E. T. Akgul and B. F. Senkal, Biol. Trace Elem. Res., 2024, 202(9), 4314–4323.
299. P. Y. Quintas, E. Agostini, A. L. W. Oller and L. B. Escudero, J. AOAC Int., 2023, 106(6), 1542–1549.
300. A. Darroudi, J. Fadaee Kakhki and F. Dousti, Int. J. Environ. Anal. Chem., 2023, 103(19), 8519–8531.
301. F. T. Abdelwahed, W. I. Mortada, M. M. El-Defrawy and R. M. Eltabey, J. Food Compos. Anal., 2024, 129, 106119.
302. S. Yilmaz, B. Hazer and M. Tuzen, Food Chem., 2024, 457, 140114.
303. P. Chaikhan, Y. Udnan, R. J. Ampiah-Bonney and W. C. Chaiyasith, J. Food Compos. Anal., 2023, 118, 105214.
304. S. Duman and M. Soylak, J. Food Compos. Anal., 2024, 129, 106099.
305. M. J. Muñoz-Sandoval, Y. Vicente-Martínez, M. Hernández-Córdoba and I. López-García, Spectrochim. Acta, Part B, 2024, 216, 106947.
306. M. Roulier, P. A. Baya, S. Roberge and D. Lariviere, J. Mass Spectrom., 2024, 59(2), e5005.
307. N. Altunay, B. Hazer, M. Farooque Lanjwani, M. Tuzen, H. Ul Haq and G. Boczkaj, Food Chem., 2023, 425, 136523.
308. N. M. Muslim, B. K. Hussain, N. M. Abdulhussein and E. A. Azooz, Anal. Bioanal. Chem. Res., 2024, 11(1), 11–22.
309. Q. Yang, C. Li, J. Hu and X. Hou, Anal. Chim. Acta, 2023, 1283, 341901.
310. N. Ullah, B. Hazer, M. Tuzen and T. A. Saleh, J. Mol. Liq., 2024, 394, 123703.
311. N. Altunay, M. Farooque Lanjwani, M. Tuzen and G. Boczkaj, J. Mol. Liq., 2024, 396, 124000.
312. J. Ali, M. Tuzen, W. B. Jatoi and B. Hazer, Food Chem., 2024, 437(Pt 2), 137908.
313. Y. Wang, J. Yan, C. Wang, D. Lu and S. Chen, J. Food Meas. Charact., 2024, 18(1), 756–763.
314. M. Soylak, A. H. Kori and H. E. H. Ahmed, At. Spectrosc., 2023, 44(5), 326–335.
315. K. Greda, M. Welna, A. Szymczycha-Madeja and P. Pohl, Microchem. J., 2024, 196, 109715.
316. S. Tokalioglu, M. S. Demirisler, H. Sahan and S. Patat, Food Chem., 2024, 454, 139780.
317. M. Soylak, F. Uzcan, O. Goktas and Z. P. Gumus, Food Chem., 2023, 429, 136855.
318. M. Soylak, A. N. Coban and H. E. H. Ahmed, Food Chem., 2024, 442, 138426.
319. D.-B. Zhou, H.-T. Zhu, Y.-Z. Hou, G.-L. Shen, C. Shi, Y.-J. Zhu, Y.-B. Xiao, W. Song, Y.-X. Liu, F. Han, P. Zheng and X.-F. Yu, J. Food Compos. Anal., 2024, 126, 105921.
320. H. E. H. Ahmed and M. Soylak, J. Food Compos. Anal., 2023, 121, 105396.
321. M. Soylak, H. E. H. Ahmed and A. N. Coban, J. Food Compos. Anal., 2024, 125, 105716.
322. A. A. Gouda, R. El Sheikh, A. M. Khedr, S. Abo Al Ezz, W. Gamil, M. M. El-Gabry and E. H. Youssef, Int. J. Environ. Anal. Chem., 2023, 103(16), 4515–4529.
323. M. Musielak, M. Serda, A. Gagor, E. Talik and R. Sitko, Spectrochim. Acta, Part B, 2024, 213, 106863.
324. D. Beauchemin, At. Spectrosc., 2024, 45(02), 115–122.
325. J. Wu, S. Lan, J. Sun, H. She, G. Wang, X. Wen, S. Zhou, B. Ying, X. Wang and H. Wang, Food Anal. Methods, 2023, 16(11–12), 1655–1672.
326. A. H. Kori, F. Uzcan and M. Soylak, J. Food Compos. Anal., 2023, 122, 105474.
327. S. Duman and M. Soylak, At. Spectrosc., 2023, 44(3), 153–159.
328. Q. Salamat, A. Elci and S. G. Elci, Anal. Lett., 2024, 57(18), 3164–3179.
329. S. Tokalioglu, S. Shahir, Y. Yilmaz and S. Patat, Talanta, 2024, 266(Pt 1), 125002.
330. S. Tokalioglu, M. S. Demirisler, H. Sahan and S. Patat, Anal. Methods, 2023, 15(43), 5846–5854.
331. P. Ali Mohammadzadeh Baghaei, M. R. Afshar Mogaddam, M. A. Farajzadeh, A. Mohebbi and S. M. Sorouraddin, J. Food Compos. Anal., 2023, 117, 105125.
332. J. Zhou, L. Qi, X. Song, Z. Yu, S. Wang, M. Zhang, X. Yuan and K. Huang, Food Chem., 2023, 426, 136638.
333. E. B. Ingrassia, E. F. Fiorentini, R. G. Wuilloud, E. Agostini, A. L. Wevar Oller and L. B. Escudero, J. Food Compos. Anal., 2024, 127, 105967.
334. H. Shirkhanloo, M. Habibnia, A. Rashidi, A. Faghihi Zarandi and M. Dehghani Mobarake, Food Chem., 2023, 403, 134394.
335. P. Montoro-Leal, J. C. Garcia-Mesa, I. Morales-Benitez, L. Vazquez-Palomo, M. D. M. Lopez Guerrero and E. I. Vereda Alonso, Microchim. Acta, 2023, 190(8), 296.
336. A. F. Zarandi, H. Shirkhanloo and J. Rakhtshah, J. Pharm. Biomed. Anal., 2024, 240, 115941.
337. Ş. Tokalıoğlu, S. T. Hosseini Moghaddam, Y. Yılmaz and Ş. Patat, Microchem. J., 2023, 195, 109515.
338. P. M. Dos Santos, M. Zanetti Corazza and C. Ricardo Teixeira Tarley, Food Chem., 2024, 440, 138238.
339. M. Soylak, B. Aksu and H. Elzain Hassan Ahmed, Food Chem., 2024, 445, 138733.
340. S. Ozdemir, A. Dundar, N. Dizge, E. Kilinc, D. Balakrishnan, K. S. Prasad and N. Senthilkumar, Chemosphere, 2023, 317, 137840.
341. T. Havasi, E. Ezzatzadeh and A. Taheri, J. Food Compos. Anal., 2024, 128, 106014.
342. H. E. H. Ahmed, O. Ozalp and M. Soylak, J. Food Compos. Anal., 2023, 118, 105163.
343. H. E. H. Ahmed, A. M. A. Mohammed and M. Soylak, Food Chem., 2023, 428, 136794.
344. M. Balal Arain, H. Elzain Hassan Ahmed and M. Soylak, Microchem. J., 2023, 195, 109495.
345. E. Cakmak and T. Oymak, Anal. Sci., 2022, 38(4), 703–709.
346. L. M. Costa, F. A. Borges, M. H. da Silva Cavalcanti, A. C. do Lago, C. R. T. Tarley, G. de Fatima Lima Martins and E. C. Figueiredo, Anal. Chim. Acta, 2023, 1251, 340709.
347. S. T. H. Moghaddam, Ş. Tokalıoğlu, Ş. Patat, Ş. Dinçer and H. Şahan, J. Food Compos. Anal., 2024, 131, 106207.
348. F. Uzcan, A. H. Kori and M. Soylak, J. Food Compos. Anal., 2024, 130, 106166.
349. Z. D. Firouzabadi, S. Dadfarnia and A. M. H. Shabani, Int. J. Environ. Anal. Chem., 2024, 1–14.
350. B. Feist, Molecules, 2023, 28(10), 4140.
351. K. Umaz, A. Umaz, I. Aydin and F. Aydin, Instrum. Sci. Technol., 2025, 53(1), 109–125.
352. S. Chen and L. Liu, J. Food Compos. Anal., 2023, 115, 105001.
353. I. S. Permigiani, N. K. Vallejo, P. E. Hasuoka, R. A. Gil and M. C. Romero, J. Food Compos. Anal., 2024, 126, 105898.
354. L. Menezes de, R. Lorençatto, R. Peixoto, C. Duyck and A. Rocha, J. Braz. Chem. Soc., 2023, 34(3), 461–469.
355. A. M. Iordache, C. Voica, C. Roba and C. Nechita, Foods, 2024, 13(4), 592.
356. K. L. LeBlanc, T. Kumlung, A. Suarez Priede, P. Kumkrong, T. Junvee, S. Deawtong, J. Bettmer, M. Montes-Bayon and Z. Mester, Anal. Bioanal. Chem., 2024, 416(11), 2761–2772.
357. J. L. d. P. Teixeira, A. P. Rebellato, M. I. A. Fioravanti, R. F. Milani and M. A. Morgano, J. Trace Elem. Med. Biol., 2024, 81, 127329.
358. A. Singhato, K. Judprasong, P. Sridonpai, N. Laitip, N. Ornthai and C. Yafa, J. Food Compos. Anal., 2023, 120, 105303.
359. I. Djekic, S. Levic, N. Smigic, A. Bouleau, K. Ilijevic, J. Roganovic and V. Rakic, J. Sci. Food Agric., 2024, 104(9), 5031–5041.
360. A. Rahimi, E. Talebi-Ghane, A. Heshmati, A. Ranjbar and F. Mehri, Drug Chem. Toxicol., 2024, 47(3), 356–364.
361. Y. Lee, J. Anal. Sci. Technol., 2024, 37(1), 12–24.
362. F. Al-Juhaimi, D. A. Kulluk, I. A. Mohamed Ahmed, M. M. Ozcan and O. Adiamo, Environ. Monit. Assess., 2023, 195(11), 1370.
363. M. Bellocci, M. Leva, V. Melai, S. Milone, G. A. Romeo, R. Rosato, G. Scortichini, G. Tammaro and G. Diletti, Environ. Sci. Pollut. Res., 2023, 30(43), 97353–97362.
364. L. A. Henriquez-Hernandez, A. C. Acosta-Dacal, L. D. D. Boada, M. Zumbado, L. Serra-Majem and O. P. P. Luzardo, Nutrients, 2023, 15(14), 3251.
365. V. Mauer, A. Smalley and M. Menon, J. Food Compos. Anal., 2024, 132, 106372.
366. M. I. A. Fioravanti, A. P. Rebellato, R. F. Milani, M. A. Morgano and A. P. A. Bragotto, J. Food Compos. Anal., 2023, 123, 105565.
367. S. Pederiva, R. Avolio, D. Marchis, M. C. Abete and S. Squadrone, Biol. Trace Elem. Res., 2023, 201(8), 4133–4142.
368. R. R. A. Peixoto, C. Duyck, R. Lorençatto, F. V. Durão, M. da Silva Teixeira, E. A. dos Santos and A. A. Rocha, J. Food Compos. Anal., 2023, 118, 105190.
369. M. Welna, A. Szymczycha-Madeja and P. Pohl, Molecules, 2023, 28(8), 3374.
370. A. P. Rebellato, M. I. A. Fioravanti, R. F. Milani and M. A. Morgano, J. Food Compos. Anal., 2023, 123, 105639.
371. G. Chen and B. Lai, Anal. Chim. Acta, 2024, 1309, 342675.
372. A. A. Roberts, D. Guimarães, M. W. Tehrani, S. Lin and P. J. Parsons, X-Ray Spectrom., 2024, 53(6), 506–519.
373. Z. Ni, L. Yang, M. Qu, Z. Li and F. Tang, Anal. Lett., 2025, 58(1), 79–90.
374. M. K. Seyoum, G. R. Fayissa and G. S. Geleta, Food Sci. Nutr., 2024, 12(6), 4151–4159.
375. P. Casarin, L. D. D. Santos, F. L. G. Viell, F. L. Melquiades and E. Bona, Anal. Chim. Acta, 2023, 1276, 341639.
376. E. Nanou, N. Pliatsika, D. Stefas and S. Couris, Food Control, 2023, 154, 110007.
377. A. Pavlin, D. Kocar, J. Imperl, M. Kolar, G. Marolt and P. Petrova, Foods, 2023, 12(15), 2826.
378. M. Perez-Rodriguez, M. Jazmin Hidalgo, A. Mendoza, L. T. Gonzalez, F. Longoria Rodriguez, H. Casimiro Goicoechea and R. Gerardo Pellerano, Food Chem.:X, 2023, 18, 100744.
379. M. Shuai, C. Peng, Y. Yang, Y. Ren, R. Hou, L. Cao, J. Ning and H. Cai, J. Food Compos. Anal., 2023, 122, 105417.
380. M. O. Varrà, L. Husakova, E. Zanardi, G. L. Alborali, J. Patocka, A. Ianieri and S. Ghidini, Meat Sci., 2023, 204, 109285.
381. M. Zhang, Q. Xiong, C. Li, J. Hu and X. Hou, LWT–Food Sci. Technol., 2023, 183, 114947.
382. M. Ghidotti, S. Papoci, D. Pietretti, T. Zdiniakova and M. B. l. C. Guntinas, Anal. Bioanal. Chem., 2023, 415(22), 5437–5449.
383. G. Manickavasagam, M. Saaid, V. Lim, Z. J. Tey and R. Osman, Chem. Pap., 2023, 77(10), 5859–5871.
384. N. Adamczuk, J. Osko, M. Grembecka, P. Konieczynski, P. Migas, A. Orzel, B. Baj-Wojtowicz and M. Krauze-Baranowska, Nutrients, 2023, 15(17), 3782.
385. A. K. Bilgin, M. F. Cengiz, B. Karakaş-Budak, C. Gümüş, S. A. Kılıç, F. Perinçek, O. Basançelebi, E. Sezik and M. Certel, J. Appl. Res. Med. Aromat. Plants, 2023, 37, 100505.
386. M. Horacek, M. Lage and J. Vakhlu, Foods, 2023, 12(15), 2830.
387. H. Kim, S. Jeong, H. Chung, S.-H. Nam and Y. Lee, J. Food Compos. Anal., 2023, 123, 105501.
388. A. Lesniewicz, M. Welna, A. Szymczycha-Madeja and P. Pohl, Molecules, 2023, 28(18), 6618.
389. M. M. M. Lima, D. Hernandez and R. C. Runnebaum, ACS Food Sci. Technol., 2023, 3(10), 1646–1653.
390. N. Malo, D. Mazumder, J. Crawford, P. Gadd, K. Gopi, N. Saintilan and J. Sammut, Foods, 2023, 12(15), 2874.
391. T. M. Weilert, C. L. Ray, J. A. Gawenis and J. D. Brockman, J. Radioanal. Nucl. Chem., 2022, 331(12), 4971–4981.
392. Z. Xuan and W.-X. Wang, J. Food Compos. Anal., 2023, 123, 105597.
393. M. Zeiner, H. Fiedler, I. J. Cindric, I. Nemet, D. Toma and I. Habinovec, Foods, 2023, 12(16), 3132.
394. L.-H. Zhou, W.-H. Huang, L.-F. Ma, Y.-J. Jin, M. Li, W.-H. Wang and X.-L. Cao, Spectrosc. Lett., 2023, 56(9–10), 479–487.
395. J. Zhu, L. Chen, Y. Chen, Y. Rong, Y. Jiang, F. Liu, Q. Zhou, X. Wei, H. Yuan, J. Zhang and J. Li, Food Control, 2023, 154, 109978.
396. F. H. Alkallas, A. M. Mostafa, E. A. Rashed, A. B. G. Trabelsi, M. A. I. Essawy and R. A. Rezk, Coatings, 2023, 13(10), 1790.
397. Y. Su, J. Zhang, L. Wang, L. Dias Araujo, D. Tan, C. Yuan and A. Zhang, J. Food Compos. Anal., 2024, 125, 105671.
398. J.-M. An, S. H. Hur, H. Kim, J. H. Lee, Y.-K. Kim, K. S. Sim, S.-E. Lee and H. J. Kim, Food Chem., 2024, 437, 137836.
399. S. Bai, S. Wang, D. Qin, L. Gao, Q. Hao, N. Du and P. Wang, Qual. Assur. Saf. Crops Foods, 2023, 15(4), 133–142.
400. M. R. Baqueta, A. C. Costa-Santos, A. P. Rebellato, G. M. Luz, J. A. L. Pallone, F. Marini, A. L. Teixeira, D. N. Rutledge and P. Valderrama, Microchem. J., 2024, 196, 109603.
401. A. Dehelean, I. Feher, P. Romulus, D. A. Magdas, F. D. Covaciu, A. M. Kasza, V. Curean and G. Cristea, Foods, 2023, 12(23), 4271.
402. M. Jazmin Hidalgo, J. Emilio Gaiad, H. Casimiro Goicoechea, A. Mendoza, M. Perez-Rodriguez and R. Gerardo Pellerano, Food Chem.:X, 2023, 20, 101040.
403. E. Nanou, N. Pliatsika and S. Couris, Molecules, 2023, 28(24), 7960.
404. K. S. Sim, H. Kim, S. H. Hur, T. W. Na, J. H. Lee and H. J. Kim, Food Res. Int., 2024, 175, 113676.
405. D. Woldetsadik, D. B. Sims, M. C. Garner, A. C. Hudson, J. Monk, B. Braunersrither, W. N. Adepa Sunshine, L. Warner-McRoy and S. Vasani, Biol. Trace Elem. Res., 2024, 202(8), 3829–3839.
406. I. Allegretta, G. Squeo, C. E. Gattullo, C. Porfido, A. Cicchetti, F. Caponio, S. Cesco, C. Nicoletto and R. Terzano, Food Chem., 2023, 401, 134124.
407. L. Suchecki Barnet, G. Harumi Yamashita, M. J. Anzanello and D. Pozebon, Food Chem., 2023, 402, 134208.
408. Y. Ding, W.-j. Chen, J. Chen, L.-y. Yang, Y.-f. Wang, X.-q. Zhao, A. Hu, Y. Shu and M.-l. Zhao, Anal. Sci., 2024, 40(3), 399–411.
409. A. M. Jagatic Korenika, A. Jeromel, I. Tomaz, T. Jednacak, S. Roncevic, I. Nemet, I. Primozic, T. Hrenar and P. Novak, Food Chem.:X, 2024, 21, 101162.
410. Y. Tang, H.-Q. Zhao, X.-Y. Zhang, X.-Z. Wang, C. Du, S. Chen, Y. Chen and T. Wang, Anal. Lett., 2025, 58(2), 272–284.
411. G. Zeng, X. Hao, H. Wang, H. Li and F. Gao, Food Chem., 2024, 435, 137646.
412. L. A. Oganesyants, A. L. Panasyuk, D. A. Sviridov, O. S. Egorova, D. R. Akbulatova, M. Y. Ganin, A. A. Shilkin and A. A. Il’in, Separations, 2024, 11(5), 148.
413. K. Chen, H. Xue, Q. Shi, F. Zhang, Q. Ma, J. Sun, Y. Liu, Y. Tang and W. Wang, Food Chem.:X, 2024, 22, 101412.
414. X. Kong, Q. Chen, M. Xu, Y. Liu, X. Li, L. Han, Q. Zhang, H. Wan, L. Liu, X. Zhao and J. Nie, J. Integr. Agric., 2024, 23(5), 1749–1762.
415. F. Wang, J. Fan, Y. An, G. Meng, B. Ji, Y. Li and C. Dong, Microchem. J., 2024, 201, 110512.
416. K. Wei, G. Teng, Q. Wang, X. Xu, Z. Zhao, H. Liu, M. Bao, Y. Zheng, T. Luo and B. Lu, Foods, 2023, 12(8), 1710.
417. M. Fattobene, F. Liu, P. Conti, S. Zamponi, C. Governatori, S. Nardi, R. E. Russo and M. Berrettoni, Foods, 2024, 13(12), 1924.
418. Z. Slavíková, P. Jaromír and P. Diviš, J. Microbiol., Biotechnol. Food Sci., 2024, 13(5), e10219.
419. Q. M. Bui, Q. T. Nguyen, T. T. Nguyen, H. M. Nguyen, T. T. Phung, V. A. Le, N. M. Truong, T. V. Mac, T. D. Nguyen, L. T. A. Hoang, H. M. D. Tran, V. N. Le and M. D. Nguyen, J. Anal. Methods Chem., 2024, 2024, 1329212.
420. B. V. Canizo, A. L. Diedrichs, A. Londonio, P. Smichowski, R. G. Pellerano and R. G. Wuilloud, J. Food Meas. Charact., 2024, 18(3), 2409–2419.
421. C. Chen, B. Wang, Y. Xiao, F. Xiong and G. Zhou, Arabian J. Chem., 2024, 17(5), 105730.
422. T. de Oliveira Costa, J. Rangel Botelho, M. Helena Cassago Nascimento, M. Krause, M. Tereza Weitzel Dias Carneiro, D. Coelho Ferreira, P. Roberto Filgueiras and M. de Oliveira Souza, Food Chem., 2024, 442, 138268.
423. R. Devarajan, S. R. Dibakoane, O. C. Wokadala, B. Meiring, V. Mlambo, F. R. Kutu, J. J. Sibanyoni and J. K. Jayaraman, J. Food Compos. Anal., 2024, 131, 106205.
424. F. Di Donato, A. Biancolillo, M. Foschi, V. Di Cecco, L. Di Martino and A. A. D’Archivio, J. Food Compos. Anal., 2023, 115, 104948.
425. B. Duarte, J. Melo, R. Mamede, J. Carreiras, A. Figueiredo, V. F. Fonseca, M. L. de Sousa and A. B. Silva, J. Food Compos. Anal., 2023, 115, 104976.
426. D. Galvan, J. C. de Andrade, L. Effting, C. A. Lelis, F. L. Melquiades, E. Bona and C. A. Conte-Junior, Food Control, 2023, 143, 109326.
427. Y. Hong, N. Birse, B. Quinn, Y. Li, W. Jia, P. McCarron, D. Wu, G. R. da Silva, L. Vanhaecke, S. van Ruth and C. T. Elliott, Nat. Commun., 2023, 14(1), 3309.
428. L. Kumari and S. S. Tripathy, Food Addit. Contam.,:Part A, 2024, 41(7), 715–729.
429. A. P. Louppis, M. S. Constantinou, M. G. Kontominas, F. Blando and G. Stamatakos, J. Food Compos. Anal., 2023, 119, 105219.
430. T. Lymperopoulou, K. Balta-Brouma, L. A. Tsakanika, C. Tzia, A. Tsantili-Kakoulidou and F. Tsopelas, Food Chem., 2024, 447, 138965.
431. R. Mamede, A. Santos, E. Ferreira da Silva, C. Patinha, R. Calado and F. Ricardo, Food Control, 2024, 163, 110501.
432. A. Mara, M. Caredda, M. Addis, F. Sanna, M. Deroma, C. A. Georgiou, I. Langasco, M. I. Pilo, N. Spano and G. Sanna, Molecules, 2024, 29(4), 869.
433. H. Mei, J. Nie, M. Pan, K. M. Rogers, Z. Guo, C. Li, S. Shao, Y. Zhang and Y. Yuan, Food Control, 2024, 162, 110454.
434. S. Panebianco, M. G. Pellegriti, C. Finocchiaro, A. Musumarra, G. Barone, M. C. Caggiani, G. Cirvilleri, G. Lanzafame, A. Pulvirenti, A. Scordino and P. Mazzoleni, Sci. Rep., 2023, 13(1), 13739.
435. S. Shao, M. Xu, X. Liao, Q. Luo, Y. Lin, P. Wang, D. Fang, Y. Huang, S. Jin and N. Ye, Food Res. Int., 2023, 165, 112522.
436. Y. Su, J. Zhang, L. Wang, G. Jin and A. Zhang, Food Chem., 2024, 446, 138812.
437. S. Valverde, J. A. Tapia, A. Pérez-Sanz, A. V. González-Porto, M. Higes, J. J. Lucena, R. Martín-Hernández and J. Bernal, J. Food Compos. Anal., 2023, 119, 105235.
438. M. O. Varrà, E. Zanardi, M. Serra, M. Conter, A. Ianieri and S. Ghidini, Molecules, 2023, 28(11), 4300.
439. D. Woldetsadik, D. B. Sims, M. C. C. Garner and H. Hailu, Biol. Trace Elem. Res., 2024, 202(3), 1305–1315.
440. S. Xie, C. Hai, S. He, H. Lu, L. Xu and H. Fu, J. Anal. Methods Chem., 2023, 2023, 1271409.
441. Z. Yue, H. Lin, X. Fang, Y. Tian, W. Yan, J. Yao, X. Chen and K. Wang, Food Control, 2023, 147, 109594.
442. D. Zhang, X. Yang and L. Guo, Microchem. J., 2024, 201, 110664.
443. Z. A. Temerdashev, A. G. Abakumov, A. A. Kaunova, O. N. Shelud’ko and T. G. Tsyupko, J. Anal. Chem., 2023, 78(12), 1724–1740.
444. H. Ye, J. Yang, G. Xiao, Y. Zhao, Z. Li, W. Bai, X. Zeng and H. Dong, Food Chem., 2023, 402, 134216.

---