Atomic spectrometry update: review of advances in the analysis of metals, chemicals and materials
Simon
Carter
a,
Robert
Clough
b,
Andy
Fisher
*b,
Bridget
Gibson
c and
Ben
Russell
d
aINEOs Acetyls, Saltend, Hull, UK
bSchool of Geography, Earth and Environmental Science, University of Plymouth, PL4 8AA, Plymouth, UK. E-mail: afisherplymouth@gmail.com
cIntertek Sunbury Technology Centre, Shears Way, Sunbury, Middlesex, UK
dNational Physical Laboratory, Nuclear Metrology Group, Teddington, Middlesex, UK
First published on 26th October 2021
Abstract
This update covers the literature published between June 2020 and May 2021 and is the latest part of a series of annual reviews. It is designed to provide the reader with an overview of the current state of the art with respect to the atomic spectrometric analysis of various metals, chemicals and materials. It is not designed to be a comprehensive review of all papers that have used an atomic spectrometric technique. Instead, it includes only those papers that have some novelty through making an advance in either knowledge or the application of atomic spectrometry. Also included, especially for forensic-based applications, for recycling applications and for cultural heritage samples, are those papers that have treated the analytical data produced with chemometric packages to provide invaluable information on provenance or to aid identification. Other research has focussed on calibration strategies for LIBS analyses and improving the resolution of depth and lateral profiling to facilitate the elucidation of mechanistic aspects of electronic components. An additional topic of interest includes the identification of oxidation states and neighbouring atoms using techniques such as XPS and XANES, again attempting to elucidate mechanisms for electronic components or to elicit speciation data.
1 Introduction
This is the latest review covering the topic of advances in the analysis of metals, chemicals and materials. It follows on from last year’s review1 and is part of the Atomic Spectrometry Updates series.2–6The ever-expanding repertoire of LIBS analysis has continued in this review period. Its on-line/in situ/standoff abilities continues to make it an attractive proposition in many areas. Research in the metals industry has used it for several years, but it has been plagued by the fact that calibration can be problematic. Methods to improve the accuracy of calibration have continued and many different models have been produced. In this review period, this has been especially true for non-ferrous metals, where it may be used during manufacture or during the re-cycling of scrap. It has also found use for sorting different polymers during re-cycling, estimating the calorific values of coal, and use in the cultural heritage sector. The standoff ability of LIBS has also found significant use for nuclear applications. The drive to expand the number of radionuclides measurable to contribute to safe, cost-effective decommissioning continues. To this end, several reports of new techniques were developed. The analysis of fuels has continued to be of great interest, although, somewhat surprisingly, the analysis of alternative fuels has received less attraction during this review period. The analysis of cultural heritage artifacts has also received huge interest. Since these samples are precious, the majority of these applications use non-destructive or minimally destructive methods of analysis. Therefore X-ray-based techniques, LIBS or laser ablation sample introduction feature heavily. Often, these are used in combination with one or more chemometric techniques to try and elucidate manufacturing processes, provenance etc. Similar strategies are often adopted for the forensic analysis of materials, where sample destruction should be avoided and chemometric methods may aid classification/identification. The analysis of electronic components, wafers, thin films and solar cells often makes use of surface analysis tools such as SIMS, TOF-SIMS and assorted X-ray-based techniques. Some of the X-ray techniques, e.g. XPS and XANES give information on the chemical surroundings or speciation of the analyte whereas others give bulk surface concentrations. Other techniques may be used for depth-profiling studies, e.g. LA, LIBS or SIMS. The requirement for better resolution (both lateral and depth) is driving the research for depth profiling/surface analysis techniques. The analysis of nanostructures has also matured, with fundamental studies of nanoparticle sizing being fewer in number than previous years. Instead, a large number of reviews have covered varying aspects of nanostructure analysis. Other papers have concentrated on improving figures of merit and validating existing procedures. The number of application-based papers for nanomaterials has increased markedly. This indicative of the methodology developed and reported over the last few years reaching maturity.
1.1 Reviews of instrumental techniques
A brief section has been provided in this year’s review because several review pieces have been published that cover numerous topic areas. It was therefore deemed sensible to cover these in this one section rather than writing about them all in four or five different sections. Other, more focused reviews, are given in the relevant individual sections.The majority of the review papers covering general industrial applications have discussed the use of laser induced breakdown spectrometry (LIBS). The use of LIBS has certainly increased significantly over the last five years. This is because of its ability for rapid and cheap analysis as well as its potential for on-line use at manufacturing plants. It is certainly a money saver for industry if their product can be analysed as it is made rather than collecting a sub-sample and then having it analysed off-line in a laboratory. Alterations to the product recipe can be made in real-time, saving time and expense. The stand-off ability of LIBS also offers numerous advantages where dangerous or unpleasant conditions occur. Examples include the nuclear industry where the use of LIBS enables analysis of components from a safe distance for the operators. Similarly, it may prevent the need for a reactor shutdown so that sample can be collected for laboratory analysis. The standoff analysis of potential explosive devices also has clear beneficial health implications. Despite its many obvious advantages, LIBS does have a couple of drawbacks. It is not completely sample non-destructive, but its main drawback is that it can be very difficult to use quantitatively. This is because the sensitivity of analysis can be affected by numerous parameters, e.g. the matrix, temperature, pressure, the atmospheric medium, etc. Numerous research papers have been published over the years describing assorted different calibration strategies. These have been summarised in a review with 93 references presented by Costa et al.7 It covered methods such as matrix matched calibration, internal standardization, standard additions, multi-energy calibration, one-point gravimetric standard addition, one point multi-line calibration, slope ratio calibration, two point calibration transfer, single sample calibration, multiple linear regression, principal component regression (PGR), partial least squares and artificial neural network. Examples of each were given and the advantages, disadvantages and limitations of each discussed. A review by Zhang et al. (with 70 references) covered the use of machine learning algorithms as a method for overcoming calibration problems and for gleaning the maximum amount of information from the data.8 Other review papers covering LIBS were prepared by Guo et al.9 which contained 210 references and covered developments in the application of LIBS in recent years and another by Senesi et al. who discussed field portable LIBS, giving a historical review, current status and future prospects (118 references).10 In the latter review, two types of instrument were compared: hand-held instruments and those that are modular and joined by an umbilical. The performance of both was also compared with portable XRF. The industrial applications of LIBS were reviewed (with 190 references) by Legnaioli et al.11 The subject areas covered in particular were applications in the steel and coal industries. Also covered was LIBS applications in the re-cycling industry, e.g. of plastics for recycling. A further review, of remote analysis LIBS, was presented by Li et al.12 The review covered 123 references and concentrated on methods of signal enhancement, its use in a variety of fields and some of the problems encountered. A review of methods and applications of LIBS containing 230 references was published by Shah et al.13 In one section, the sensitivity, speed of analysis, accuracy and sample preparation methodology were compared with other techniques. A second part focussed on the instrumentation used and the third part on quantitative and qualitative analysis. Different analysis parameters and their effects as well as the fields of application were also discussed. A final review of LIBS analysis covering its use for mapping studies of samples was prepared by Limbeck et al.14 This review cited 188 references and covered sample types such as cultural heritage samples, thin films, catalysts, biological materials and concrete. The review was split into instrumentation types (different lasers, spectrometers, collection optics etc.), sub-sections for assorted chemometric data handling methods and calibration. Overall, it is a good read for a novice just starting in the area.
The related technique of laser ablation assisted spark induced breakdown spectrometry (LA-SIBS) and its applications was reviewed (with 42 references) by Zhou et al.15 This methodology also needs no sample preparation and hence, is quick and likely to gain in popularity. The fundamentals were described in the review, which also covered applications – especially those involving conducting materials.
Synchrotron-based X-ray techniques are known to have several advantages over some techniques because they have a high intensity, monochromatic and collimated beam. The use of techniques such as small/wide angle X-ray scattering (SAXS and WAXS), X-ray photoelectron spectroscopy (XPS), X-ray absorption spectroscopy (XAS), grazing incidence X-ray diffraction (GI-XRD) and different methods of X-ray imaging were reviewed (with 131 references) by Rahimabadi et al.16 Each technique was discussed in its own sub-section with both theory and applications being present.
2 Metals
2.1 Ferrous metals
This has been an even more active area of research that in previous years. There has, however, been a change of emphasis. In previous years, a lot of the research was based on the calibration of laser induced breakdown spectrometry (LIBS) or on methods to minimise interferences during LIBS analysis. This review period still has a high number of LIBS-based applications, but these are more application-based and include several papers describing other techniques used. It should be noted that those applications describing the analysis of cultural heritage samples can now be found in Section 5.One of the advantages of LIBS is that it can be portable and can hence be taken to the field or production site. This helps speed of the analysis since samples do not have to be returned to a laboratory. There have therefore been several applications reported of the LIBS analysis of steel at the production line. An example of this was a paper entitled “feasibility investigation for on-line elemental monitoring of iron and steel manufacturing processes using LIBS”, published by Wang et al.17 The paper studied the effects of different sample temperatures, lens to target distance and the angle of analysis. The Q-switched Nd:YAG laser operating conditions and the spectrometer used were discussed in detail with the aid of a schematic diagram and the method validated by the use of numerous standard steel samples. The sample temperature effects were studied by undertaking analysis at room temperature, 300, 500 and 700 °C. The emission intensity increased with increasing temperature. This was attributed to lower laser energy being required for sample ablation and vaporization meaning that more can be used for plasma evolution processes to enhance the plasma emission intensity. Similarly, the ablation and vaporization of sample also increased causing the increase of plasma emission intensity. Consequently, the emission intensity of measured spectra at higher sample temperature is higher than that at lower sample temperature. This study therefore indicated that calibration of the LIBS system was sample temperature dependent. Although this finding is not completely unexpected, it does highlight the potential pitfalls of simply using an instrument and assuming correct answers will be produced. Lens to target distance was varied between 3.5 and 4.1 m, with 3.9 providing highest intensity. Spectra obtained using different sample angles of 0°, 15°, 30° and 75° were obtained with results indicating that an increased sample angle (meaning the sample surface was far away from the vertical direction of laser beam), led to the signal intensity decreasing. A 3D profile measurement system was also described in the paper. A paper by Zeng et al. described the production of a prototype instrument capable of the rapid on-line analysis of steels through the use of a mobile fiber optic LIBS system.18 A schematic of the prototype instrument was provided to aid the reader. Numerous standard steel samples were used for calibration purposes, including the six carbon steels (GBW01211–GBW01216), seven standard samples of microalloy steel (GSB 03-2453-2008-1 to GSB 03-2453-2008-7) and 11 samples of low alloy steel purchased from NIST. A polynomial fitting was required to obtain the best calibration functions for the analytes Cr, Cu, Mn, Mo, Ni and V, since linear fitting led, in some cases, to poor regression coefficients (with 0.861 for Cu being the poorest). Once polynomial fitting was used, regression coefficients of the calibration curves improved to greater than 0.99. Average absolute errors of Cr, Cu, Mn, Mo, Ni and V were 0.440, 0.003, 0.039, 0.07, 0.033 and 0.057 wt%, respectively. Their average relative errors fluctuated from 2.9% to 15.7%. The instrument provided LOD at the mg kg−1 level, with the best being 31 (for Cr) and worst being 290 (for Mo). As mentioned previously, the production of instruments capable of working on-line or in-line provide a huge benefit in terms of time- and cost-saving since samples do not have to be returned to the laboratory for analysis. Once they are reliable and easy to use, even relatively untrained workers may use them successfully.
An application was reported by Lednev et al. in which LIBS was used on-line to detect defective laser welding.19 A description of the system was provided along with schematic diagrams and a photograph. Briefly, the diode pumped pulsed solid state Nd:YAG laser (1064 nm, 5 ns, 130 mJ per pulse, 10 Hz, M2 = 90) beam was focused by quartz lens (F = 280 mm) through a pierced aluminium mirror at a 45° angle to the sample surface. Analysis of the melt pool enabled both atomic and ionic lines for major components (Cr, Fe or Ni) to be compared as well as lines for alloying elements (Mn, Si). The spectral region from 275 to 290 nm enabled this to be undertaken. If the melt pool becomes too hot, then this results in a defective weld. By measuring the intensity of the atomic Fe 278.81 nm and ionic Fe 274.67 nm iron lines signals as well as their ratio, it was possible to use hypothesis testing by paired sample t-test to identify defects. Two other sensing procedures were attempted. These were: thermal emission spectroscopy of the welding melt pool induced by a continuous wave laser (passive sensing) and LIBS analysis of a solidified hot weld. Both of these methods failed to detect defects. Another paper to use LIBS for the analysis of welds was presented by Taparli et al. who used it for the in situ chemical composition analysis of tungsten inert gas (TIG) austenitic stainless steel welds.20 These authors used a similar setup to that of Lednev et al., but employed an analysis angle of 20° rather than 45°. The Mn signal was enhanced during the welding process. This was attributed to Mn vapour forming over the weld pool and then on-cooling, it re-condenses and forms a layer on the surface of the weld. As well as the on-line LIBS, the authors also used LIBS to study the effects of differing welding currents, finding that local increases in Mn and Ni concentration occurred at higher current. Again, this was attributed to higher Mn concentrations entering the vapour phase, later to re-condense on the surface. The advantages of these systems is that they can detect faults in real time, enabling corrective action to be undertaken and hence decreasing the number of failures.
Another paper that discussed LIBS measurements was presented by Petersson et al. who used the technique for the rapid analysis of steel slag.21 The experimental setup was described in detail and entailed a YAG laser operating at 10 Hz and with a pulse energy of 300 mJ at a wavelength of 1064 nm. Also incorporated into the experimental setup was a spherical aluminium mirror in a near-colinear configuration to collect the light emitted, a CCD detector and a rotating sample holder to ensure that data could be averaged from a sufficiently high number of laser shots covering a large area of sample. This last attribute was necessary because of the inhomogeneous nature of the slag samples. Averaging of several hundred laser pulses was required to obtain a good overall estimate of bulk composition. A large number of samples were also analysed using XRF and the concentration data from these helped calibration of the LIBS system. Samples analysed using XRF were also used for a comparison of data between the two techniques, with good agreement being obtained. Despite the large number of laser shots used, precision for the LIBS system was still 2–4 times worse than that from XRF. However, it was concluded that they were still adequate for on-line or near the line applications and, with further optimization, could potentially be improved. These authors also indicated that slag at 1200 °C gave a very different LIBS intensity response to that from cold slag. Further work was required to develop “transfer functions” from room temperature calibrations to those at elevated temperature.
The concept of transfer learning that enables LIBS calibration obtained at room temperature to be transferred to analysis at elevated temperature was also addressed by Chang et al.22 Spectral data of calibration samples obtained at room temperature (12 CRMs) and spectral data of uncalibrated samples (8 samples) at high temperature were transferred to a high dimensional feature space using kernel function mapping. One in this space a model is trained and established. Results for Cr indicated that the use of the feature-based transfer learning method decreased the mean relative error from 32.31% to only 6.08%.
A paper by Lednev et al. investigated the feasibility of using on-line LIBS for the analysis of compositionally graded alloy parts during their fabrication.23 Chromium bronze and stainless steel 410 powders were deposited using laser cladding, forming a metal part with a compositional gradient. The LIBS measurements were undertaken on the melt pool, the finished material as well as the powder starter materials. Results produced from the melt pool and the finished product were comparable because of a homogeneous mixture of analytes being present. However, very poor precision degraded the data from the start powders to the extent where it was concluded that the LIBS analysis could not be performed reliably. Despite this, the transition zone between two pure components was quantified using LIBS for the first time and so the experiment could be regarded as a success.
Some papers have used chemometrics approaches to aid either the calibration of LIBS analysis of steels or to distinguish between different steel grades in a scrapyard. The determination of C in steels can be problematic because the C line at 193.09 nm is in the vacuum UV region and light emission is easily absorbed, leading to a huge drop in sensitivity. Another C line (at 247.856 nm) is interfered with by Fe lines. Zhang et al. employed the chemometric method of least absolute shrinkage and selection operator for spectral feature selection in combination with back-propagation neural networks to correlate LIBS spectra with C concentration without explicitly measuring C spectral lines.24 This machine learning approach succeeded by first using 12 certified materials with a C content ranging from 0.013 to 3.73% and with a widely varying concentration of other metals to help train the calibration model. A further three CRMs were used to validate the method developed. The multivariate chemometrics approach was summarised nicely in a flow chart. Once optimized, the method had a LOD of 0.0366% and produced regression coefficients of 0.9999. Poggialini et al. used multiple artificial neural networks (ANN) to greatly improve the precision and accuracy of a commercial hand-held LIBS instrument.25 The ANN algorithm links a set of inputs (the spectral intensities of some lines in the LIBS spectrum) to a set of outputs (the concentration of the elements of interest) through a non-linear relation that can be determined by minimizing the deviation between the predicted and nominal outputs on a set of known samples. The relation obtained between inputs and outputs is then used for obtaining the outputs associated with unknown samples from the measured inputs. Several ANN were prepared including: one for the major elements and one for minor elements. The performance with the calibration built into the instrument were compared in terms of linearity of calibration and relative mean standard error (RMSE). Performance with and without the ANN were comparable for standards/known samples, but the ANN assisted calibrations were significantly improved for unknown samples, with RMSE improving from 1.83 to 0.70% for Cr and from 2.98 to 0.62% for Ni. Kashiwakura and Wagatsuma developed a method using a LIBS system made in-house followed by partial least squares regression (PLSR) for the rapid determination of components of austenitic stainless steels.26 Eight CRMs of the JSM M 200 series (materials 21–28) were used to train the model for the analytes Cr, Mo, Nb, Ni and Ti. Spectral data could be obtained very rapidly (approx. 13 s), but more than 10 minutes was required for the calculation of regressions when all the relevant wavelengths were utilised. However, the calculation time could be reduced if single wavelengths that had higher excitation levels were chosen for each analyte. The method was applied to the analysis of five stainless steel types: SUS304, SUS310, SUS316, SUS321 and SUS347. The range of recovery rates were 94.3–111% and precision was better than 10% RSD. Another paper that determined C in both solid and molten steels was presented by Cui et al.27 These workers employed a UV long-short double pulse LIBS system that they both discussed in detail and provided a schematic diagram of. Six CRMs were used for calibration at room temperature and then one of them was heated to 1800 °C. Spectra were taken at three different periods after the melting process. After 0.5 min, a clear C peak was observable at 193.091 nm. However, this peak diminished significantly after 12 min and had disappeared completely after 15 min. This is a well known effect in the steel industry and is termed the carbon escape phenomenon. The influence of the presence of oxygen was studied and had a significant effect on the presence of C, with much of it boiling off as carbon dioxide. It was concluded that this was unavoidable in molten steel. In a second paper by the same research group, sample temperature effects were compared for both single pulse LIBS and colinear long-short double pulse LIBS.28 The authors used the emission intensity ratio of Mn 404.136 nm/Fe 402.187 nm and the Fe 402.187 nm/Fe 400.524 nm for the comparison. The effect of the delay time between pulses was also studied. As would be expected, a temperature rise from 20 to 700 °C had a very adverse effect on accuracy of analysis when single pulse LIBS was used. This effect was very much diminished in the long-short pulse version. This was attributed to it maintaining a higher plasma temperature and for longer. An attempt to correct for plasma temperature using Fe 402.187 nm/Fe 400.524 nm resulted in a marginal improvement in precision from 22.56 to 21.86% RSD for the single pulse LIBS. However, the long-short double pulse LIBS provided a much improved precision of 6.46% anyway, which improved further to 1.78% once plasma temperature had been corrected for. Improvements to calibration are obviously welcome since it betters the chance of obtaining accurate data. These papers therefore add to the increasing volume of literature that will enable more accurate data to be provided by instruments in the future.
Several other papers reporting the LIBS analysis of steels have been reported. In one, by Cai et al. a portable LIBS instrument built in-house was used to study the aging of the heat resistant steel T91.29 The chemometric tools of wavelet threshold de-noising followed by multiple scattering correction were used to clean up the data. Then, both support vector machine (SVM) and k-fold support vector machine-recursive feature elimination (K-SVM-RFE) were used on the data from slices of pipe that had undergone different levels of aging to establish the model. This was then applied to artificial samples. The performance of the wavelet threshold de-noising and multiple scattering correction on the performance of the model were studied. Once optimized, the process was capable of determining the age of the material with an accuracy of 85.17%. Although clearly not perfect, it does give a very good indication of how old and hence, how likely to fail, a steel is.
Electrical steel (steels containing between 1.1 and 4% Si) is used for things such as electric motors, transformers, generators and high frequency convertors. It requires analysis to ensure that the correct grade is used for the right job. Silicon is known to be problematic to detect using LIBS because of self-absorption and matrix effects. A combination of LIBS and laser induced molecular fluorescence was used by Zhang et al. to measure Si as the molecule SiO.30 The paper showed a convenient schematic of the experimental setup, which comprised two lasers one of which (a Nd:YAG operating at 1064 nm; pulse duration: 6 ns; beam diameter: 5 mm; repetition rate: 10 Hz; pulse energy: 60 mJ) was used to ablate the sample and the second (OPOTek, VIBRANT™ HE 355 LD; wavelength range: 225–2400 nm; pulse duration: 10 ns; operating at 10 Hz and 1 mJ) was used for SiO molecular excitation. An XYZ translation stage, collection optics and a Czerny–Turner spectrometer equipped with an intensified CCD were also part of the setup. Seven certified materials (GSB 03-2453-2008-1 to GSB 03-2453-2008-7, with Si contents of between 0.021 and 1.370%) were used as the samples. The atomic Si 288.16 nm line was interfered with by the ionic Fe 288.08 nm line at a low silicon content. This obviously greatly limited the analytical accuracy and sensitivity of LIBS for silicon in steels. Two types of molecular excitations were therefore evaluated: vibrational ground state excitation and vibrational excited state excitation. Both showed a huge improvement in Si detection capacity, with the excited state excitation (with detection of SiO at 248.68 nm) showing the best results of all. For that, regressions (R2) were 0.988, the LOD was 187 μg g−1 and the root-mean-square error of cross-validation (RMSECV) was 0.046%. The method is therefore a significant advance on other LIBS methods.
The rapid analysis of the content and size of aluminium inclusions in middle alloy steels using LIBS was reported by Wang et al.31 The instrument comprised a Q-switched Nd:YAG laser operating at 1064 nm, a laser beam shaping system and a vacuum Paschen–Runge spectrometer. A total of 15 reference materials were analysed including GBW01238, GBW01241, GBW01242, and GBW (E) 010224, which are all certified for total Al and acid soluble Al, were analysed. One laser pulse per spot was used for the analysis, with the excited region being 10 mm × 10 mm, and the distance between the centres of two adjacent spots being 250 μm. A total of 1681 sets of intensity data were obtained for each analysis. Several series of mathematical functions were derived enabling a relationship between intensity of light emitted and the particle size to be deduced. Results obtained using the method developed were compared with those obtained using SEM-EDS. According to the authors, general agreement was obtained, but there was still room for improvement. However, the inclusion content and particle sizes can be analysed simultaneously using LIBS, which therefore, offers a simple sample preparation and rapid analysis.
Wang et al. used LIBS employing a picosecond laser to analyse molybdenum–copper stainless steels containing Fe within the range 40–70%.32 The micro-LIBS experimental setup mainly consisted of a combination of a laser, a timing trigger, an energy attenuator, a camera and two integrated spectrometers. One spectrometer had a wavelength range of 244–360 nm and the other 382–469 nm. The setup was shown in a schematic diagram. The inter relationship between the different parts of the unit were described. When the laser power was 0.044 W, a crater with an inner diameter of approximately 3 μm was achieved. After background correction of the spectra, a quadratic function was used to correlate the intensity and concentration of each of the eight analytes (Al, Cr, Cu, Mn, Nb, Ni, Ti and V). Regression coefficients of better than 0.99 were obtained for Al, Cu, Nb, Ni and V, whereas correlations for the other analytes were better than 0.95. Detection limits varied significantly, with Ti providing the lowest at 0.04%, despite the fact it produced the worst regression. The highest LOD (1.77%) was for Ni.
A commercial hand-held LIBS instrument was used by Senesi et al. to determine C, Mn, P, Si and Ti in pig iron from blast furnaces in the steel industry.33 Data obtained from the instrument underwent a three-stage pre-treatment. These steps were: employing a standard normal variate procedure, applying a Savitzky–Golay smoothing procedure and then cutting the spectra into fragments to cover only the ranges required for the analytes. Once these steps had been performed, both univariate and multivariate (partial least square) calibration and validation protocols were tested. The predictive capability of each calibration model was evaluated by calculating the Pearson’s correlation coefficient (R), root mean square error (RMSE), and mean absolute prediction error (MAPE) and LODs, whereas the leave one out cross validation (LOO-CV) method was used to perform external validation. Correlation between LIBS and XRF data were plotted. The multivariate approach was significantly better than the univariate with R2 values of 0.93–0.99. These were particularly good for Si and Ti (both 0.99) compared with the univariate approach (0.95 and 0.85, respectively). In addition, the MAPE values improved from 18.04 and 19.4% to 14.07 and 11.61%, respectively.
Techniques other than LIBS have also been used to analyse ferrous materials. An example by Hu et al. used glow discharge-mass spectrometry (GD-MS) to determine metal impurities in iron–nickel-based superalloys.34 High resolution of the mass spectrometer was required to overcome some polyatomic interferences. The GD-MS operating conditions, in terms of argon flow rate and discharge current, were also optimized. Relative sensitivity factors of the calibrations of 12 analytes were determined and then applied to the analysis of the CRM IARM Ni909-18. Results were in good agreement with certified values. In addition, results obtained using alternative techniques, namely ICP-OES and ICP-MS following acid dissolution procedures, were also compared with those from GD-MS. Semiquantitative data were also obtained for some analytes. The agreement between semiquantitative GD-MS and ICP data was less impressive, with data for some analytes agreeing reasonably well whereas others differing by a factor of three. Despite this, the direct analysis of a solid sample is attractive because it is quicker (and therefore cheaper) and uses no corrosive substances.
Two publications by Kuptsov et al. discussed the optimization of operational parameters35 and the analysis of steels36 using a two-jet arc plasmatron using spark ablation. The technique uses much higher power than an ICP (15 kW) and is used to analyse powdered solid samples directly, without recourse to dissolution procedures. The principle of operation is based on the action of a spark discharge created between the surface of a conductive sample and a tungsten electrode. The solid aerosol formed as a result of exposure is transported by a flow of argon to the excitation source. Parameters optimized in the first of these papers included the operating conditions of the plasma torch (argon gas flow rate, angle between plasma jests and torch height), the sparking time, discharge frequency and pulse duration.35 The samples used throughout the study were a high purity copper and standard reference samples GSO 130 alloyed structural steel. Under optimal conditions, detection limits varied significantly between different analytes and were 10−5% for Mn and V, 10−4% for Cr and Cu, and 10−3% for Mo and Si. The second paper applied the technique to the analysis of steel samples including the reference materials UG-5K and UG-7K.36 Agreement with certified values was very good. However, it was noticeable that precision was relatively poor. As an example, the experimental Cu result in UG-5K was 0.53 ± 0.06% (n = 4) compared with the certified value of 0.490 ± 0.003%.
A 3D-printed electrochemical flow cell coupled with ICP-OES was used by Klaes et al. to characterise hot-dip zinc/zinc alloy coated steel.37 Samples were cut into 3 × 10 cm rectangles, cleaned with acetone and lint-free cloth to remove the anti-corrosion oil coating and then subjected to analysis. Sample was dissolved in 20% hydrochloric acid, a qualitative electrochemical measurement potential is measured against a Ag/AgCl reference electrode was made to identify the elements present and then the resulting solution transported to the ICP-OES instrument for fully quantitative analysis. Results were compared with those obtained using GD-OES and were in reasonable agreement. Linearity, repeatability and reproducibility were all tested. Repeatability of ten measurements was variable, with the Fe content ranging from 0.23–0.27%, the Al between 0.46 and 0.49% and the Zn being most variable, ranging from 42.75 to 50.13 g m2. Reproducibility of 20 different values determined on three different days yielded similar, but slightly more variable results.
2.2 Non-ferrous metals
Many of the overall themes for the analysis of non-ferrous materials are the same as for the ferrous ones, although the sample types are more diverse. Strangely, there are still a large number of papers reporting methods for LIBS calibration. The number of these were significantly diminished in the Ferrous section during this review period. Several papers discussed the use of chemometric methods to aid the calibration, assist with classification or undertake data pre-treatment or reduction. The use of chemometrics is still a very popular area of research for several reasons. Data reduction is required because LIBS can produce an enormous amount of data. Reducing this to the relevant wavelengths facilitates calculations and enables it to be accomplished with a more basic computer. Calibration is very dependent on matrix, temperature and pressure. For on-line analysis, chemometrics may be used to calculate correction factors or isolate data that are least affected by changes in operating conditions. The biggest advantage of LIBS over more traditional atomic spectrometric methods is that it (as well as portable XRF) may be used in the field. This has a particular advantage in the recycling of scrap, since different alloys may be identified rapidly, minimising the potential of cross-contamination and, in some cases, facilitating an estimate of worth to a customer. Portable LIBS instruments are available commercially. If the calibration models can be improved, then the accuracy of these instruments may be improved.A two jet arc plasma with spark sampling and calibration was discussed at length in the Ferrous section.35,36 A related paper, by Kuptsov et al., discussed its use for the analysis of dissolved solutions of brass samples.39 Operating conditions were optimized and the effect of the concentration of matrix elements on the signal to background ratio of the analytes was determined. Calibration was achieved through the use of standard solutions and the coefficients of relative sensitivity factors. Analysis of the reference materials MCH-7 yielded results within 15% of certified values. This was thought to be fit for purpose.
Relative sensitivity factors for elements in a copper matrix were also studied by Zhang et al. who compared continuous direct current glow discharge mass spectrometry and micro-second pulsed GD-MS for the analysis.40 High purity copper powders (99.999%) were spiked with a solution containing 72 analytes and the relative sensitivity factors obtained using a high resolution double-focusing sector-field GD-MS instrument equipped with a Grimm-type fast-flow dc source. The precision was comparable to that of data obtained using ICP-MS and ICP-OES following a dissolution procedure. The GD-MS method therefore offered the clear advantages of time-saving and of not requiring the use of corrosive chemicals and the associated fume extraction. The technique had a limited linear range, with a regression of 0.9993 when additions of 1, 5 and 10 mg kg−1 were used, but this reduced to 0.9177 when the calibration was extended to 50 and 100 mg kg−1. Recovery of 10 mg kg−1 spikes ranged between 80 and 120%. Comparison of relative sensitivity factors obtained using the two modes of GD-MS indicated that those for the pulsed mode were between 2 and 40% lower. Repeatability was, however, better for the pulsed mode, with a value of <10% being obtained compared with <15% for the continuous mode. The method was validated through the analysis of the certified reference materials M-376a, ERM-EB375 and SRM-1117. Results were mixed, with the pulsed mode providing better accuracy for some analytes (e.g. Al, As, Cd, Mn, P and Zn) and the continuous mode providing better accuracy for others (e.g. Bi, Pb and Si).
Collinear long-short double pulse LIBS was described previously in the Ferrous section. Cui et al. discussed its use for the underwater analysis of copper materials in two papers.41,42 Underwater analysis obviously has huge benefits in the field of archaeology where in situ measurements can prevent the expense of raising articles to the surface only to find that it is not made of the material expected. Water is known to have quenching problems though and so operating conditions tend to be quite different to normal analyses. Both papers provided a schematic diagram of the setup. The first of the papers found that the plasma formed underwater had a cold shell with a very low emission intensity and a hot core with much greater emission intensity.41 The early-stage plasma formed produced continuum emission from the hot core and this was absorbed by the Cu atoms in the colder plasma edge, causing Fraunhofer-type absorption at both 324.754 nm and 327.396 nm. An inter-pulse delay of 10 ns provided an absorption at 324.754 nm with a precision of 1.5% RSD, which increased to 5.3% RSD when the inter-pulse delay was increased to 30 ns. Precision at 327.396 nm was 2.7% and 3.3% for 10 and 30 ns inter-pulse delays, respectively. The reason for poorer precision at higher inter-pulse delay was that the amount of absorption decreased. Underwater LIBS can suffer from light transmission losses back through the fibre optic to the detector. This, when added to the quenching effect by the water has a deleterious effect on the sensitivity. The experiments undertaken in these experiments help optimise signal so that some of the losses can be minimised.
An interesting application was reported by Bonin et al. who reported the ICP-MS detection of Ra at the attogram levels in copper samples.43 After acid dissolution of the copper matrix, the Ra was retained on an AG50W-X8 BV column. A 5 mL bed volume had to be used to retain all of the Ra and to allow the copper matrix to be removed. Sample was flushed onto the column which was then rinsed with dilute hydrochloric acid to remove the copper. A further rinse with an ammonium EDTA solution completed the copper removal process. The Ra was then eluted with a solution of the ammonium salt of nitrilotriacetic acid. Using a sample volume of between 200 and 500 mL and an elution volume of 10 mL, significant preconcentration factors were achieved. The figures of merit of the method were conveniently tabulated and showed that an excellent instrumental LOD of 2 mBq L−1 (53 fg L−1) was obtained. When 100 g of copper was dissolved, the method LOD was 0.3 μBq g−1 (equating to 9 ag g−1). Recovery was 100 ± 3%. The chromatographic separation was fairly rapid, enabling a sample to be prepared in 20 minutes. The rate limiting step was the acid dissolution, which took over three hours per sample.
Machine learning applied to LIBS data from aluminium alloys has been reported in other papers.46,47 The paper by Dai et al. identified different aluminium alloys (060, 6061, 5052, 2024, and 7075) with the prospect of re-cycling them from scrap.47 The authors first used PCA to help with the identification and then followed this by least squares support vector machine to classify five different aluminium alloys. The results were compared with those obtained from data analysis using support vector machine alone. As is the norm for this type of paper, a training set of spectra was required from samples of known composition and this was used to develop the model. Other samples of known composition are then used to verify the accuracy of the model before it can be used on unknown samples. Both classification methods performed well, with the support vector machine alone classifying 98.3% of samples correctly and the PCA – least squares support vector machine 100%. The results for the support vector machine showed that of the 60 spectra of the test set, one 7075 aluminium alloy spectrum was misclassified as 5052, with the remaining 59 spectra being predicted correctly. Therefore, the prediction accuracy of 7075 is 91.67%, and the prediction accuracy rate of the other four aluminium alloys was 100%. It is known that LIBS spectra can vary through changes in laser fluence. Yue et al. used machine learning to correct for this efficiently.46 These authors deliberately changed the fluence of the Nd:YAG laser used between 7.9 and 71.1 mJ and collected the LIBS spectra from aluminium alloys, with Mg being the analyte. Comparison of the data obtained using the proposed method was made with other methods, e.g. through correction for crater volume, electron number density and plasma temperature and a classical univariate correction method (normalising with laser pulse energy and total spectral intensity). The authors noted that improvements over no correction were made for the crater volume, electron number density and plasma temperature method. However, they pointed out that measurement of crater volume is not always easy and can be time-consuming, especially for an industrial application. The machine learning method, however, first used the back propagation neural network to train and validate the model. This led to an average precision of 6.3% RSD being achieved for the test samples; a significant improvement over the other methods. This could potentially allow it to be used in an on-line industrial setting.
A paper by Wang et al. described a feature selection method called recursive feature elimination (RFE) and combined that with ridge regression and then applied this combination to data obtained from the LIBS analysis of aluminium alloys.48 The paper described the model in full mathematical detail. This method uses a base model for multi-round training. After each round of training, several features of weight coefficients are removed, and the next round of training is based on a new feature subset. The RFE method recursively reduces the size of feature subsets by sequential backward selection. This greatly reduces the randomness of spectral lines. In this paper, the ridge regression is used as the basic model of RFE to solve the quantitative problem. The ridge regression model is used to fit the calibration set data, and the regression coefficients are recorded. The spectral features are sorted according to the absolute value of the regression coefficients. The lowest ranking spectral feature is deleted from the feature set in each cycle. Once the weakest of the data had been removed, the remaining subset was input to a partial least squares regression model (PLS). The analytes determined were Cu, Fe, Mg, Mn, Si and Zn. Comparison was made between partial least squares regression using the full data set, a genetic algorithm partial least squares model and the proposed model. The calibration results of the analytes and root mean square error of prediction were improved significantly using the proposed model.
Sorting of different aluminium alloys using LIBS was a subject tackled by Fugane et al., who attempted the sorting through the determination of seven analytes (Cr, Cu, Fe, Mg, Mn, Si and Zn).49 After careful optimization of the gate delay (500 ns) and gate width (50 μs) to ensure optimal signal to background ratio, calibration curves were prepared using certified aluminium alloys containing different concentrations of the analytes. Several commercial alloys (A1050, A1100, A2017, A2024, A5052, A5083 and A6061) were then analysed using the system. Overall, the method was deemed successful, although several could not be classified into the alloy type. This though was not a failing of the LIBS system. Instead, it stemmed from the different alloys having extremely similar chemical fingerprints. To prove their point, the authors used a dissolution method followed by ICP-OES and had similar results. It was noted that the LIBS analysis was rapid and could be undertaken on-line, whereas the ICP-OES analysis could not and required a sample dissolution procedure, making it far slower.
A second paper by Fugane et al. used LIBS for the rapid detection and characterization of inclusions in re-cycled aluminium materials.50 The Nd:YAG laser system, optics, detection system and motorized X,Y,Z sample plate were all discussed in detail. Again, the gate delay and gate width were optimized prior to analysis. The inclusions contained significant amounts of Fe and Si. The number of inclusions was estimated by analysing the surface of the 4 mm × 5 mm sample and then, assuming homogeneity throughout the sample, the number of inclusions on the surface was extrapolated to the number in the whole sample. Lateral resolution of the LIBS system was restricted by the size of the ablated crater and was approximately 100 μm. Analysis of the entire surface took approximately 40 minutes to complete. The authors did specify that this time could be reduced significantly if spectrum analysis was automated using a high-performance computer. The LIBS system was clearly well-suited to such an analysis, since a standard procedure would require the dissolution of the sample which would also probably dissolve the inclusion.
The on-line monitoring of nickel alloy smelting processes was discussed by Li et al. who described an iterative multi-energy calibration (IMEC) method for LIBS analysis.52 This process is an advancement of the commonly used multi-energy calibration method. The IMEC method is a two-stage process. The first is the preparation of a standard. This is undertaken in the same furnace, under the same conditions and should have as close to the same composition as the samples to be formed as possible. A portion of this is sent away for analysis to confirm the composition. The second stage is to use this standard during the LIBS analysis along with a quantitative analysis algorithm. This algorithm was discussed in detail in the paper. The performance of the methodology was compared with that of an internal standardisation method during the analysis of 13 batches of sample. Average values of both the root mean square error (RMSE) and the RSD were improved by a factor of approximately two.
A paper by Medvedev et al. described the ICP-MS and ETV-ICP-MS determination of trace analytes in high-purity tungsten55 and molybdenum.56 For ICP-MS analysis, sample (0.3 g with 99.999% purity) was dissolved and the resulting solution analysed directly. A total of 54 analytes were determined with LOD ranging from 0.4 to 3000 ng g−1. Unfortunately, there was quite a sizeable matrix effect for many of the analytes, with their signal dropping by up to 90% when the tungsten reached 3 g L−1. The worst affected analytes were those with a mass of between 40 and 130 a.m.u. A bigger dilution of sample (e.g. a tungsten concentration of <700 mg L−1) went some way to overcoming this problem, but led to many analytes being diluted to below their LOD. Subsequent ICP-MS analyses were undertaken with a tungsten concentration of 700 mg L−1. The operating conditions for ETV-ICP-MS had to be optimized. Factors considered were the plasma power, ETV temperature program, the transfer gas flow rate and the ion lens settings. The sample was introduced as a 5000 mg L−1 tungsten solution, i.e. far more concentrated than for pneumatic nebulisation. A total of 26 analytes were chosen. Their LOD ranged from 1 to 200 ng g−1, representing improvements of between 3 and 70 compared with nebulisation. Method validation for the ETV-ICP-MS was achieved through comparison with ICP-MS and ETAAS data and through spike/recovery experiments. The ETV-ICP-MS method could use a lesser dilution, thereby improving LOD whilst not suffering the problems associated with viscosity affecting nebulisation efficiency and signal suppression.
The same research group has also undertaken the analysis of high-purity cadmium.57 A novel method of vacuum preconcentration with matrix volatilization was described. Briefly, 1.5 g of sample was placed in a quartz cup that was placed in a vacuum distillation device. This was then heated to 390 °C. When the first sign of melting occurred, a tiny bleed of filtered air was admitted so that an oxide layer formed on the surface. Vacuum distillation then occurred for a period of 3 hours, leaving a thin film of cadmium oxide rich in impurities in the quartz container and the cadmium coated elsewhere. The cadmium oxide film containing the analytes was then dissolved in nitric acid and diluted to 3 mL for ICP-MS analysis or to 5 mL for ICP-OES analysis. The LOD for 41 analytes ranged from 0.002 to 10 ng g−1 for ICP-MS and from 0.1 to 40 ng g−1 for 43 analytes using ICP-OES. Method validation was through spike/recovery experiments and through comparison of data obtained using the two instrumental techniques. The detection limits obtained were therefore very impressive. However, at three hours per sample, the method was clearly quite slow.
Quantitative analysis of zirconium alloys using borate fusion and WDXRF analysis of the glasses formed was described by Sieber et al.58 The alloys were prepared for analysis through the formation of lathe turnings that were 0.5–2 mm wide and 2–10 mm long. Approximately 0.5 g of these “chips” were weighed into 10 mL of water. Then, hydrofluoric acid (1 mL) was added very slowly (20 min) to minimise splashing. Then, 5 mL aliquots of the solution were placed in a clean platinum crucible and dried at 107 °C. Further aliquots were added and dried until the residue from the whole solution was present. The lithium borate flux was added and the mixture fused using a complex temperature program (provided in the paper). Overall, the zirconium and 17 analytes were diluted by a factor of 16. The methodology was applied to the analysis of assorted zirconium alloy reference materials, including SRM 360 c, 360 b and the discontinued materials Zr alloy SRM: 360, 360a, 1210, 1211, 1212a, 1213, 1214, 1215, 1234, 1235, 1236, 1237, 1238 and 1239. Results from the procedure were in sufficiently good agreement with those obtained using ICP-MS for the method to be deemed successful for many analytes. There were some elements, however, that showed particularly poor agreement, e.g. P, where WDXRF indicated there was 21.6 mg kg−1 whereas ICP-MS gave a value of 85.7 mg kg−1; which is closer, but still far away from the certified value of 62 mg kg−1. Hafnium was also noted to provide a biased result.
The analysis of platinum group elements has also been undertaken by several workers. An example, by Harouaka et al. determined Th-229 and Th-232 as well as U-233 and U-238 in gold, iridium, platinum and tungsten matrices using a triple quadrupole ICP-MS instrument with oxygen as the reaction gas.59 A thorough investigation was undertaken of the polyatomic interferences that may arise through the combination of matrix ions and either ions from the dissolution media or with the oxygen. In single quadrupole mode, numerous possibilities were identified over the mass range 227–239. The full list was tabulated, however the most important were: 194Pt35Cl+ on Th-229, 198Pt40Ar+ on U-238, 193Ir40Ar+, 197Au36Ar+ and 184W16O3H+ on U-233 and 195Pt37Cl+, 197Au35Cl+ and 192Pt40Ar+ on Th-232. Scanning at a higher mass range (227–270 a.m.u.) showed that higher oxides of matrix elements did not form or, if they did, they did not interfere with the analytes. The mass shift method of measuring Th-229 (i.e. measuring ThO at m/z 245) provided data with 96% accuracy at a concentration of 100 fg g−1, but that this decreased to 83% accuracy at 10 fg g−1 and to 73% at 1 fg g−1. The corresponding figures for U-233 were 95%, 94% and then sharply decreasing to 51% at the 1 fg g−1 level. Absolute detection limits were 2.74 fg and 12.9 fg for Th and U, respectively which, depending on sample type analysed, corresponded to LOD at the pg g−1 range. The method was applied successfully to the analysis of electronic components.
Habibpour et al. constructed a single-shot, spark-assisted LIBS instrument for the soul purpose of measuring the purity of gold.60 The details of the instrument were discussed in detail, but briefly, a Nd:YAG laser combined with a spark generator and a single channel CCD spectrometer were used. The spark assists the LIBS in forming a more energetic plasma, even at low energy single shots. The more energetic plasma increases the plasma temperature by up to 20% and that, combined with the elongated lifetime of the plasma, leads to an increase in sensitivity by an order of magnitude. Using the method, the grade of gold could be determined with an analytical error of less than 0.5%. An added advantage was that stylus profilometry demonstrated that the damage caused by the laser shot was less than that caused during normal LIBS.
The analysis of silver alloys or silver has been reported in two papers. One, by Brocchieri et al. used a number of methods, all employing a hand-held EDXRF instrument, to estimate the silver coating thickness on materials.61 Calibration standards were prepared by covering metal plates (copper, iron and lead) with silver foils of known thickness, ranging from 0.2 to 15.7 μm. Methods of determining the thickness included a partial least squares regression, and ratios of the Kα, Kβ, Lα and Lβ intensities of both the silver and the base metal. The partial least squares method provided the best thickness estimates, although it was noted that the other methods also provided good data. The method was then applied to the analysis of a Roman coin and to a modern coin. The other paper was prepared by Gao et al. who applied a calibration-free, high repetitive rate laser ablation spark induced breakdown spectroscopy method for the quantitative elemental analysis of a silver alloy.62 Sample was ablated by a fibre laser operating at a repetition rate of 30 kHz and then a spark was applied to enhance the plasma emission. The usual plasma diagnostics were used, i.e. the Stark broadening of spectral lines was used to calculate electron number density (1.11 × 1017 cm−3) and a Saha–Boltzmann plot was used to estimate plasma temperature (7355 K). Analytical data obtained from the system for the analysis of the alloy were compared with those obtained using a dissolution procedure followed by ICP-OES analysis. Agreement was within 15%, which was deemed acceptable, especially as a standard sample was not required. The method described therefore had the advantage of speed since dissolution was not required.
3 Organic chemicals and materials
3.1 Organic chemicals
There have been two main research topics in this area over the last review period. One is the forensic analysis of materials and the other is the analysis of pharmaceutical materials and/or cosmetics. Both areas have had a number of good applications presented.The preparation of certified reference materials is always welcome as it continues to be the gold standard for validating methodology. The preparation of certified materials that are pure organic compounds is especially welcome given that there are so few. The preparation of natural and Se-82 labelled selenomethionine reference materials (SENS-1 and SEES-1, respectively) was reported by LeBlanc et al.63 The high purity natural isotopic abundance SENS-1 was available as a powder, whereas the SEES-1 was a solution. The SENS-1 material was purchased from a commercial manufacturer and bottled without further sample manipulation. This was produced as a primary standard (for traceability) by the certifying body (NRC Canada). The SEES-1 was made using an in-house method that was described fully in the paper. For certification purposes, 10 mg aliquots from different bottles of SENS-1 were weighed into microwave bombs and then digested with microwave assistance using a mixture of nitric acid and hydrogen peroxide. The resulting solutions were spiked with Se-82 and then an isotope dilution analysis undertaken using a triple quadrupole ICP-MS instrument employing oxygen as the reaction gas to quantify the Se present. The selenium-containing impurities were quantified by coupling HPLC with the ICP-MS instrument. These and other impurities were identified using LC-MS. An assessment of purity was also made using NMR. The purity of SENS-1 was 0.961 ± 0.007 g g−1 (k = 2). Having established the purity of SENS-1, it was used as a primary standard for the reverse isotope dilution analysis of SEES-1. This contains 1556 ± 60 μg g−1 (k = 2).
Both copper and zinc have been ingredients of anti-fouling paints for many years. At present, there are two standard methods used to determine their leaching rate from paint films, both of which have been questioned recently. Lagerstrom and Ytreberg developed a new method for determining Cu and Zn based on XRF analysis of the paint films.64 Unfortunately, the paint formulations vary significantly, with different concentrations of Cu or Zn, different fillers, etc. and will also vary in thickness. This means that the X-rays could potentially have different amounts of self-absorption, leading to potential errors. The work described the preparation of standards made from seven different paints and with different thickness coated on Mylar® film. A calibration curve was therefore prepared for each of the seven paints. Two different portable EDXRF spectrometers were compared to test the transferability of the method between instruments. Both instruments produced data that were within 5% of the figures produced through wet chemical analysis; indicating good transferability. The study also showed that universal calibration curves could be created that had a prediction uncertainty of approximately ±130 μg cm−2 for both analytes.
Simultaneous compound-specific isotope analysis of S-33 and of S-34 in organic compounds using GC-MC-ICP-MS was described by Kummel et al.65 The isotopic reference standards IAEA S-1, IAEA S-2 and IAEA S-3 were transformed into SF6 so that they could be used as standards. The transformation reaction and the instrumental operating conditions were described in full. The determination of S is potentially fraught with problems arising from polyatomic interferences. The instrument was therefore operated in both low resolution (m/Δm = 300) and medium resolution (m/Δm = 4000). Although small amounts of the interferences occurred, the low resolution data were not significantly different to those obtained using medium resolution. The low resolution mode was therefore adopted since it offered greater sensitivity. The method developed was applied to the analysis of the organic compounds thiophene, tetrahydrothiophene, diethyl sulfide and dimethyl disulfide. These were dissolved in pentane and then aliquots with concentrations in the range 170 to 580 pmol for each analyte were injected into the GC-ICP-MS system. The simultaneous and precise determination of δ33S and δ34S in individual organic compounds by GC-MC-ICPMS was demonstrated. The method provides a simpler and faster tool than existing techniques to obtain Δ33S values for the evaluation of mass-independent isotope effects in organosulfur compounds.
Glow discharge-mass spectrometry is conventionally used for the analysis of solid samples. A new methodology was reported by Gubal et al. who used microsecond pulsed glow discharge in a copper hollow cathode followed by TOF-MS detection for the ionization and subsequent determination of volatile organic compounds.66 The discharge cell of the system was modified slightly to incorporate a quartz capillary into the argon input channel. This enabled the analysis of the vapours of the compounds in air to be determined. The method was applied to the determination of vapours of acetonitrile, benzene, toluene, o-xylene, p-xylene, m-xylene and n-heptane. The ionization mechanism was studied at length and new mechanisms of chemical and electron ionization resulting in the formation of associate ions with the copper cathode material were identified. The Penning process and proton transfer reactions also played a part. Operating conditions such as the effect of repelling pulse delay, the effect of pulse duration and discharge power and the effect of air humidity were all studied and helped elucidate ionization mechanisms. Calibration graphs were prepared and yielded R2 values of greater than 0.99. Detection limits were not magnificent (typically just sub-ppm or single figure ppm, depending on the ionization mechanism), but the calibrations spanned four orders of magnitude.
Inks from 17 pens were analysed using LIBS followed by PCA treatment of the data in an attempt to distinguish the inks.68 The LIBS operating conditions such as the delay time, laser pulse energy and number of pulses per point, were optimized using a factorial design and a second order regression model. The data obtained from LIBS was corroborated by acid digestion of the inks followed by ICP-OES analysis. It was noted that the ICP-OES instrument was capable of determining P, which acted as a very good discriminating element. Unfortunately, the LIBS instrument used was not capable of this. Data from only 20 spectral lines were collected using LIBS and subjected to PCA. This was sufficient to be able to distinguish between two inks of the same colour. The method was applied to the analysis of a cheque in which a 10 had been converted to 100. The last zero was identified as being added by a different pen because it contained large concentrations of Cu whereas the one and the first zero did not. Such a measurement clearly could not be undertaken using a standard ICP-OES experiment.
Amorim et al. used both ICP-MS and ICP-OES to study the inorganic profiles of street cocaine samples.69 A total of 52 samples were collected from three regions of the Espirito Santo state, Brazil and these were acid digested in a closed vessel microwave system. The digests were then analysed using ICP-OES (for Al, Ca, Cu, Fe, Mg, Mn and Zn) and ICP-MS (for Co, Mo, P and Pb). Analytical data obtained were interrogated using hierarchical cluster analysis (HCA) and PCA. Although some analytes were found to form strong clusters, it was impossible to discriminate between samples from the different sites. The authors suggested that future work could perhaps try to separate impurities from the cocaine prior to the analysis. Spike tests indicated good recoveries for the majority of analytes (typically 90–115%). However, there were examples of extremely poor recovery, e.g. a Ca result of 186% and a Mg recovery of 71%. It was also possible to identify that region three diluted the drug more than the other two regions.
Another paper to span both cosmetic and forensic analysis was presented by Horiguchi and Kunimura, who used both surface enhanced Raman scattering (SERS) and XRF to characterise hair dye products on single hairs.70 Hog hair was used for the experiment in which five hair dye products were used. The SERS identified clear differences between them when analysing solutions of the dyes, but sometimes produced very similar spectra when analysing the treated hair. Occasionally different dyes on the hair could be differentiated, but those containing similar ingredients or types of dye produced similar spectra. However, the XRF identified clear differences in the metallic signatures for many of the dyes, even on a single hair. For instance, product A yielded a Fe signal, product B both K and Mn and product C Ti. A combination of SERS and XRF was more efficient at differentiating the dyes than either technique alone.
The speciation of Sb in injectable leishmanicidal drugs using hydride generation (HG)-AAS was reported by Fabrino et al.72 Total Sb was determined by reducing the SbV to SbIII using 0.1 M hydrochloric acid and potassium iodide. Then, HG-AAS analysis took place using hydrochloric acid and a mixture of sodium tetrahydroborate in sodium hydroxide. To optimise the detection capability of SbIII, a factorial design followed by a central composite design was used. The presence of citric acid in the system to stabilise the SbV and to prevent it from forming hydrides resulted in the blocking of the capillary at concentrations commonly used in other studies. The concentration was therefore decreased by a factor of between 4 and 20 and was still capable of the task, but did not cause blockages. The LOD for total Sb and SbIII were 0.15 and 0.05 μg L−1, respectively. Precision was, at best 3.1% rising to 19.6% closer to the LOD. Recovery from spike experiments were 95.6–102.3% for total Sb and 89.1–108.1% for SbIII. The total Sb content of the drugs was between 0.08 and 0.21% m m−1, with the SbIII comprising only a very small proportion.
Three types of surgical thread (absorbable polydioxanone and polyglycolic acid as well as non-absorbable polypropylene) were analysed for total Al, Cu, Pb and Zn as well as the amounts of the same analytes extracted into 0.1 M hydrochloric acid.73 For total metal analysis, sample (0.25 g) was acid digested using nitric acid with microwave assistance and then the digests analysed using ETAAS. For the extractable concentrations, 1 g of sample was used. Studies of the effects of temperature (25, 37, 50, 75 and 100 °C), time (15, 30, 60, 90, 120 and 240 min) and of the pH (2, 4, 7, 8 and 10) on the amounts extracted were also undertaken. In general, the temperature had little effect until 50 °C, whereupon the analytes were extracted much more efficiently. The pH had a big effect, with the highest extraction occurring at pH 2 and then decreasing rapidly at pH 4 and then decreasing further at higher pH. The effect of time was more variable, with some analytes increasing up to 60 min and then decreasing again, whereas others, e.g. Al remaining vaguely constant. The LOD were 2.8, 1.5, 2.9 and 1.5 μg L−1 for Al, Cu, Pb and Zn, respectively. The only method validation was through the analysis of QC-026, an aqueous standard which, although not ideal, is better than nothing.
A method for discriminating between ZnO nanoparticles and Zn2+in cosmetic samples (eye shadow) was described by Garcia-Mesa et al.74 The method was based on solid sampling high resolution continuum source GFAAS (SS-HR-CS-GFAAS). Sample was first dried in an oven and then vortex mixed to ensure homogeneity. Then, between 0.1 and 0.3 mg was transferred to a platform which was then inserted to a GFAAS atomiser and a temperature program, that had been optimized using central composite design, run. Calibration was against aqueous standards. The absorbance from the less sensitive Zn line at 307.590 nm was exported as a CSV file and then the data were converted to a standard data transfer format for surface chemical analysis. It was now that the deconvolution was made using Multipak software, resulting in a peak for the Zn2+ and a second for the ZnO nanoparticles. The LOD were 0.06 μg for Zn2+ and 0.13 μg for the ZnO. These, of course, had the potential to be improved had the more sensitive Zn line at 213.857 nm been used. Precision for 10 replicates of Zn2+ at 0.6 μg in an aqueous standard was 2.5% RSD. The corresponding value for ZnO nanoparticles was 6.9% RSD. Recovery values were between 98 and 115% for aqueous standards of Zn2+ and between 80 and 100% for the nanoparticles. The total Zn result was verified by an acid dissolution prior to GFAAS. In general, good agreement between the sum of Zn2+ and ZnO and the “total” Zn was achieved.
3.2 Fuels and lubricants
In this section the number of papers seemed similar to last year but there was an increase in papers relating to crude oil and coal with the majority of papers on coal coming from China and employing LIBS for the analysis. A decrease in papers relating to alternative fuels was also noticed which seems at odds with the current environmental trend for alternative fuels.Four other papers were of interest in this section this year. One, by Andreu et al.76 described a comparison of nebulisers using ICP-OES for the direct determination of trace levels of P in petroleum products with high particulate content. The purpose of the study was to develop a method to determine P below 0.1 mg kg−1 in petroleum products with a high particulate content in the shortest analysis time. Seaspray™, TSP OptiMist®, and OptiMist XL® nebulizers were compared in order to determine which provided the best performance for the method. The OptiMist XL® nebulizer was the only device that provided a quantification limit lower than 0.1 mg kg−1 with a 1![[thin space (1/6-em)]](https://www.rsc.org/images/entities/char_2009.gif)
:3 sample dilution ratio without clogging and had the shortest analysis time. The time period for the sample preparation and analysis was approximately 60 min for a batch of four samples. Identifying the nebuliser type least likely to cause problems whilst providing the relevant sensitivity is clearly of use to workers in the field.
A paper by Ferreira et al.77 described a method for the determination of Cu, Mn, Ni and Pb in diesel samples using reversed-phase vortex-assisted liquid–liquid microextraction with EDXRF spectrometry. In this method a nitric acid solution was used to extract and isolate analytes from the samples. After a centrifugation step, the aqueous phase was added dropwise to a filter paper disc for EDXRF determination. The following variables were optimized: type of extraction phase solution, concentration of the extraction phase, stirring time and sample volume. The instrumental parameters of atmospheric conditions, irradiation energy and irradiation time were also evaluated. Using 100 μL of a 0.075 mol L−1 nitric acid solution as the extraction phase and a sample volume of 5.0 mL with a stirring time of 45 s, the limits of detection were 14, 8, 10 and 7 μg L−1 for Cu, Mn, Ni, and Pb, respectively. The LOD were impressive thanks to the enrichment factors obtained (34 (Cu), 62 (Mn), 59 (Ni) and 64 (Pb)). The precision expressed as RSD was calculated from ten replicate analyses under optimized conditions using standard solutions containing 200 μg L−1 and 400 μg L−1 of the four analytes and ranged between 2.1 and 6.4%. The results of recovery tests ranged from 87 to 112%. The procedure was applied to the determination of the four analytes in diesel oil samples. The results were compared with those obtained using ICP-OES after sample digestion. A paired t-test p > 0.05 indicated that no significant differences were found.
The determination of Cd in used engine oil, gasoline and diesel using ETAAS after employing magnetic ionic liquid-based dispersive liquid–liquid microextraction was investigated by Aguirre et al.78 The solvent used for the microextraction procedure was a magnetic ionic liquid which presents a paramagnetic property and allows easy phase separation using a magnet. A back-extraction procedure was then performed to transfer the analyte into an aqueous phase. The main experimental factors affecting the extraction of Cd were optimized using a multivariate analysis consisting of two steps, a Plackett–Burman design followed by a circumscribed central composite design. The optimal conditions were: 6.2 g of sample 119 mg of magnetic ionic liquid, 1 min extraction time, 200 mg of 1 mol L−1 nitric acid and 1 min back-extraction time. The proposed analytical method was validated using three samples: used engine oil, gasoline and diesel. The three samples were spiked at two levels 10 and 20 μg kg−1 of Cd for used engine oil and 1 and 3 μg kg−1 of Cd for gasoline and diesel. Precision and recovery values were produced within the range of 6–11% RSD and 95–110%, respectively.
The paper by Sanchez et al.79 described a method for Si speciation in light petroleum products using GC-ICP-MS/MS. Silicon in petroleum products is a ‘hot topic’ following historic contamination issues and its presence in re-processed fuels. The only useable isotope for Si, 28Si suffers from interferences mainly related to the polyatomic ions 12C16O and 14N14N. These interferences were reduced by adding hydrogen as a reactant gas in the octopole reaction cell and the optimization of the operating parameters such as carrier gas, optional gas and hydrogen gas flow rates. A test mixture of 11 Si compounds was used for this work and included cyclic and linear siloxanes, silanols and silanes. Calibration curves ranging from 0 to 500 μg kg−1 were produced with correlation coefficients up to 0.9999. The LOQ ranged from 8–60 μg kg−1. The results were compared with those obtained using ICP-OES and XRF and no significant differences in total Si concentrations were found. The light petroleum products contained cyclic siloxanes, D3–D6 mainly hexamethylcyclotrisiloxane (D3) and octamethylcyclotetrasiloxane (D4). The paper is a useful contribution to this evolving topic.
A contribution by Li et al.80 was in the field of calorific value measurement and described a method of determining light elements associated with the organic fraction using LIBS and inorganic ash forming elements using XRF. The combined approach greatly improved the measurement repeatability of the coal calorific value. The prediction model used PCA and multiple regression. The experimental results showed that the SD of calorific values predicted is 72 J g−1 which is an order of magnitude lower than the 700 J g−1 of traditional LIBS. This method could be applied in power plants and other fields that need to pay close attention to coal quality. The quick, reliable and on-site analysis afforded by LIBS is therefore an ideal solution to the problem of obtaining best combustion from coal.
The next contribution in this section, from Xie et al.,81 described a method for As and Se distribution and speciation in coal and combustion by-products from coal fired power plants. In this study, a mixture of 1 M H3PO4 and 0.1 M ascorbic acid was used to release As from coal and related solid wastes. The Se in coal and fly ash samples was extracted using 0.1 M NaOH, and gypsum was extracted using 0.2% (v/v) HCl. After extraction by sonication, the extracts were diluted and then analysed using LC-ICP-MS. Stability experiments showed that inorganic As and Se species were almost unchanged during the extraction procedure with the recoveries ranging from 89.7% to 90.6% for inorganic AsIII and AsV and from 98.5% to 112.9% for inorganic SeIV and SeVI. The recovery was calculated by dividing the sum of different species of As and Se by the total As and Se concentrations measured directly using ICP-MS after microwave digestion. The dominating species was AsV, however, the more toxic AsIII was detected in the gypsum by-product most of which ends up in landfills or stockpiled in open areas. The results suggest that proper disposal of these wastes is needed.
The next three papers are on the subject of LIBS. The first, by Zhang et al.82 described a method for the determination of Fe, Ni and V in petroleum coke using LIBS. This paper established a novel calibration model by using a genetic algorithm optimized support vector regression. The traditional AAS and ICP-AES methods of analysis for these elements have some disadvantages such as use of reagents, long turnaround time and possible contamination. Although LIBS technology has a gap in measurement precision and detection limits compared with AAS and ICP-AES it has unique advantages in rapid detection and simplicity which meets the demands of industrial production monitoring. The characteristic wavelengths of Fe, Ni and V were found by comparing the LIBS spectrum of petroleum coke with the NIST database. Feature selection was then carried out by modelling and sorting root means square error (RMSE) after cyclically removing one feature wavelength. After feature selection, support vector regression modelling was performed again. The prediction accuracies of V and Ni with feature selection were close to that of standard ASTM D6376. It was therefore concluded that it is feasible to determine Fe, Ni, and V in petroleum coke using LIBS in an industrial environment.
A paper by Dong et al.83 described coal discrimination analysis using tandem LIBS and LA-ICP-TOF-MS. This tandem approach allows simultaneous determination of major and minor elements in coal. The research focused on coal-classification strategies based on PCA combined with K-means clustering, partial least-squares discrimination analysis (PLSDA) and support vector machine. Correlation analyses performed using TOF-MS and LIBS spectra from the coal samples showed that most major, minor and trace element emissions had negative correlation with the volatile content. Suitable variables for the classification models were determined from these data. The individual TOF data, LIBS data and combined data from TOF and LIBS as the inputs for different models were analysed and compared. In all cases the results obtained with the combined TOF and LIBS data were found to be superior to those obtained with the individual TOF or LIBS data. The nonlinear support vector machine model combined with TOF and LIBS data provided the best coal-classification performance, with a classification accuracy of up to 98%.
Deng et al.84 described a method for N and S determination in coal using LIBS. The detection of N and S in coal is essential for the evaluation of its quality. This study employed LIBS to test coal quality and combined two variable selection algorithms, competitive adaptive reweighted sampling and the successive projections algorithm to establish the corresponding partial least square model. The partial least square modelled with the full spectrum of 27620 variables had poor accuracy. The coefficient of determination and RMSE of the test set for N were 0.5172 and 0.2263, respectively. For S they were 0.5784 and 0.5811, respectively. The competitive adaptive reweighted sampling and the successive projections algorithm screened 37 and 25 variables respectively in the detection of N and S but the prediction ability of the model did not improve significantly. Successive projections algorithm partial least square finally screened 14 and 11 variables respectively through successive projections and obtained the best prediction effect among the three methods. The coefficient of determination of the test set and root mean square error of N were 0.9873 and 0.0208 and those of S were 0.9451 and 0.2082 respectively. In general, the predictive results of the two elements increased by about 90% for root mean square error and 60% for coefficient of determination of the test set compared with partial least square treatment. The results showed that LIBS combined with successive projections algorithm-partial least square has good potential for detecting N and S content in coal and is a very promising technology for industrial applications.
A method developed by Mketo et al.,85 described an improved microwave assisted sequential extraction method followed by ICP-OES/MS detection. The study aimed to develop a rapid and eco-friendly microwave-based sequential extraction procedure for heavy metal distribution determination and mobility in coal samples. The steps for the sequential extraction were: (1) water, (2) 5 M HCl, (3) 2 M HNO3, (4) 7 M HNO3–3 M H2O2. The most influential extraction parameters, i.e. coal amount (0.1 g), extraction time (5 min) and temperature (200 °C) affecting metal sequential extraction efficiencies were successfully optimized by using 23 full factorial design and response surface methodology. The optimized method was then applied to three coal CRMs (SARM 18, 19 and 20) and three metal ions Ga, Sr and Ba showed solubility towards water irrespective of the CRM. These three metals were regarded as highly mobile metal ions and are therefore expected to leach out into water bodies. The rest of the metals were only mobile in acidic conditions; therefore, these metals can be regarded as immobile.
A review, by Mdluli et al.86 contained 113 references and provided a critical review of extraction methods published between 2005 and 2020 for the determination of trace metals in oily matrices with spectrometric determination. Information gathered from the literature showed that liquid/liquid extraction was the more favoured extraction method over solid phase extraction and extraction induced emulsion breaking was the most studied method over the past 16 years.
The first paper in this section by Li et al.87 described a procedure for the direct determination of Ni in crude oil samples using a miniaturized electrothermal AAS system. A fast and direct analysis was achieved using dilution in cyclohexane followed by the pipetting of a 20 μL aliquot onto the tungsten coil of the atomizer. The detection limit was 0.01 mg kg−1, similar to that of other comparable methods. The portable nature of this method means it could be useful in the field, for exploration analysis, environmental monitoring or emergency assessment.
The next contribution, by Vinic et al.,88 described a method using LIBS for the quantification of heavy metals in oils. Thin oil films were produced on silica wafer substrates by rotation. The more rotation the thinner the film and the fewer the matrix effects encountered. Different film thicknesses were studied with the optimum being 0.74 μm. The oil volume sampled by each laser pulse was 0.3 nL. Detection limits for Cd, Cr, Cu and Zn were 0.49, 0.082, 0.16 and 3.9 ppm, respectively. Although this is an interesting paper, all work was done on standards produced in base oil and no CRMs or ‘real world’ samples were analysed. Methods tend to work well on very controlled samples, and it would have been useful to see some analysis of ‘real samples’ included.
A paper by Hasme et al.89 described an integrated method for the analysis of in-service lubricants for particle size and metal content. This article discussed adding a liquid particle counter in-line with an ICP-OES system to analyse wear metals and particles in lubricant samples in one run. This approach streamlines the monitoring of in-service lubricants and meets the ASTM D5185 standard.
A paper discussing the analysis of asphalts and bitumens was presented by Adolfo et al.90 It described a method for the determination of Co, Fe and Ni in petroleum asphalt cement using high-resolution continuum source AAS after extraction using emulsion breaking. Optimized conditions were: 0.08 g of the sample dissolved in 1.5 mL of toluene with 3.5 mL of mineral oil, this was then vigorously mixed with the extracting solution containing 40% (v/v) of HNO3 and 40% (v/v) of Triton-X-100. The emulsions were then broken by heating just after their formation and the aqueous extracts collected for analysis. Simultaneous determination was carried out using the secondary lines of Co (352.685 nm), Fe (352.604 nm) and Ni (352.454 nm). For this purpose, the best pyrolysis and atomization temperatures were 700 and 2400 °C, respectively. Calibration was performed with standard aqueous solutions and no matrix effects were observed. The LOD was 0.003 mg L−1 for Co, 0.018 mg L−1 for Fe and 0.027 mg L−1 for Ni. This method was then compared with conventional acid digestion. Recoveries ranged from 100 to 107%, 103 to 109% and 95.5 to 98.9% for Co, Fe and Ni, respectively.
Chauhan et al.91 investigated dissolution methods for the quantification of metals in oil sands bitumen. Seven different dissolution methods were investigated: direct dilution, dry ashing, sulfated ashing, ultrasound assisted extraction, extraction induced emulsion breaking, detergentless micro-emulsification and acid decomposition in closed vessels. Of these methods only direct dilution and sulfated ashing were recommended for the determination of Ni and V in bitumen and of these two methods, only sulfated ashing produced samples with good storage stability. The analysis of bitumen diluted in toluene found that 40–50% of the V-containing species separated from the bulk solution after a storage period of 28 days. It was not established whether instability of dilute bitumen in toluene was due to toluene insolubility or aggregation leading to separation, but this behaviour could have implications for the removal of metals from bitumen.
A paper by Gonzalez et al.92 studied very high molecular weight asphaltene clusters in THF solution using GPC-ICP-HRMS. Three Venezuelan heavy crudes (Hamaca, Cerro Negro and Boscan) were investigated. The results showed very clearly the presence of asphaltene clusters in the THF solution. As expected these clusters occur at very low retention times indicative of cohesive long structures resulting from aggregation of species. These results provide strong support to current models regarding cluster formation. Isotopes monitored were 58Ni, 32S and 51V, at a medium resolution of 4000. The spectrometer was fitted with a commercial micro-flow total consumption nebulizer mounted with a laboratory-made jacketed spray chamber whose temperature was set at 60 °C. Large quantities of clusters were detected in Boscan crude and its toluene soluble and toluene insoluble subfractions both at room temperature and after heating. Dilution experiments showed no significant change of these profiles suggesting that these clusters, as well as the other lower molecular weight aggregates and molecules, are not in equilibrium with each other and behave as independent units. A temperature cycle was then used to monitor possible changes of GPC-ICP-HRMS profiles with temperature. Here Boscan crude samples were heated from 25 to 200 °C, left standing for 24 h and then cooled back to 25 °C before measurement. Small changes were found for Boscan crude asphaltenes however vast changes in the profiles were detected for toluene soluble and insoluble subfractions indicating interchange of material between the cluster and the other components of the solution. These results were discussed in terms of different arrays that may result when asphaltenes are separated into the two subfractions, and nanoaggregates are relocated within the cluster before and after heating.
Another paper that utilised GPC-ICP-HRMS was presented by Garcia-Montoto et al.93 These authors also monitored 58Ni, 32S and 51V to understand their removal in crude oil from atmospheric residues during hydrodemetallization and hydrodesulfurization. The Ni, S and V species size distributions were studied to track their evolution during the catalytic processes. For the Ni and V aggregates, it was found that high molecular weight compounds are more refractory compared with low molecular weight and medium molecular weight compounds which were removed at low temperatures. For the S aggregates a slightly different behaviour was observed, the low molecular weight and medium molecular weight aggregates are not as easily hydrotreated compared with the high molecular weight aggregates. This phenomenon illustrated the inherent difference in chemical composition of the metallo-organic compounds compared with the S-containing compounds and their reactivity.
The sole paper, by Arthur et al.94 investigated the monitoring of dissolved active trace elements in biogas plants using TXRF. Natural plant biomass or biowaste plants in many cases show deficiencies in Fe, Co and Ni and they are supplemented with trace elements to guarantee stable gas and energy production. In this study a procedure was developed to monitor dissolved elements in these processes to prevent an overdose or underdose of these elements. The digester liquid is comparable to sludge with a high solids content and a high salt concentration. A series of two centrifugation steps followed by microwave assisted digestion was used prior to analysis using TXRF. The centrifugation steps also significantly separated carbonates and sulfides which are deemed to be not bio-available for microbial uptake. The calculated lowest LOD was 2–4 μg L−1. The developed method was first tested in laboratory scale biogas digesters with wheat straw as substrate. After that, it was tested in a full-scale biogas plant for several months to estimate the dynamics of “active” trace elements in the range of 2–1000 μg L−1. This demonstrated that cost-effective biogas plant management is possible through the application of TXRF spectrometry to monitor trace elements.
3.3 Polymers and composites
Research into the atomic spectrometric analysis of polymers and composites has followed two main themes: the analysis of micro or nano-plastics in the environment and classification for recycling purposes. The second category has made significant use of the varying forms of LIBS; usually in combination with a chemometrics method. There are, however, several other un-related but interesting applications to be discussed.For some time now there has been concern that micro/nanoparticles of polymers that enter the environment may act as sinks for toxic organic compounds and for heavy metals. This has led to a huge increase in the number of papers undertaking their analysis. An example was presented by El Hadri et al. who employed LA-ICP-MS to determine the trace element distribution in marine microplastics.95 The operating conditions of the Nd:YAG laser used were optimized in terms of signal intensity and precision through the analysis of the reference material NIST 612 glass. It was possible to distinguish analytes that were additives (Cu, Pb and Sb) from those that were adsorbed to the surface (As, Cd, Cu, Fe, Pb, Sb, Sn, U and Zn). The authors proposed a spatiotemporal correlation of the sorption pattern comparing microplastics in terms of relative exposure time and time-weighted average exposure concentrations.
Single particle ICP-MS analysis has normally been applied to the analysis of engineered nanoparticles. However, in a paper by Jimenez-Lamana et al.96 it was employed for the detection and quantification of nanoplastics. The polymer particles were attached to functionalized gold-containing nanoparticles. This was achieved by using the negatively charged carboxylate groups on the plastics surface with a positively charged gelatin moiety attached to the gold nanoparticle. The Au was then detected using ICP-MS operating in single particle mode enabling the nanopolymers to be quantified with <5% error. The model was developed using polystyrene particles of nominal diameter of 759, 990 and 3030 nm which mimic the natural breakdown of plastics in the environment. The quantification limit was 8.4 × 105 nanoparticles per litre and the calibration was linear up to 3.5 × 108 nanoparticles per litre. Once developed, the method was applied to the detection of nanoparticles in drinking, tap and river waters. Assuming a fully functionalized nanoplastic, the theoretical limit of detection for size of the proposed strategy was 135 nm and the maximum size that could be quantified was 1000 nm.
A paper by Chen et al. described the use of LIBS for the analysis of single microplastics.97 The LIBS system used was discussed and a schematic diagram provided. The methodology was optimized using pure polypropylene particles of 150 μm diameter that had been treated with different concentrations of lead acetate and cadmium sulfate. Analytical signals were normalised to the C signal at 247.9 nm to compensate for any laser energy fluctuation or particle morphology effect. The Cd and Pb signals were proportional to the concentrations of analyte the polymers were exposed to. The methodology developed was applied to particles collected from seawater through filtration, the polymers identified using Raman and then the LIBS system to determine trace metal content. The advantages of the system were described as: its ability for multi-elemental determination, its speed (100 ms acquisition time – but this was limited by the speed of the laser used and could probably be reduced further), not having the requirement of chemical treatment of the samples (unlike other techniques) and its ability to analyse individual particles of micron size. The size LOD was 70 μm, but that was restricted because of the need to transfer the particle from the Raman microscope to the LIBS setup. It was thought that the particle size LOD could be improved further if a hybrid system was developed. In that case, size would be determined by the resolution of the LIBS system and could be as low as 1 μm.
An interesting application of microplastic analysis was reported by Holmes et al. who developed a method for the determination of the in vitro avian bioaccessibility of metals adsorbed to the pellets.98 Polyethylene pellets were collected from beaches and then exposed to an acidified (pH 2) saline solution of pepsin for 168 h at 40 °C. The extracted Co, Fe, Mn and Pb as well as their residual concentrations, i.e. those that could be extracted using dilute aqua regia, were quantified using ICP-MS. Time-based studies indicated that the Co, Fe and Mn were mobilised quickly at the beginning of the pepsin leach, but slowed to a quasi-equilibrium. The Pb leaching was more complex, with evidence of secondary maxima and re-adsorption. After the full 168 hours, the concentrations of metals extracted from the solid material were 0.014, 0.81, 38.9 and 0.10 μg g−1 for Co, Fe, Mn and Pb, respectively representing an extraction efficiency of 50, 80, 60 and 80%, respectively.
A metrologically traceable protocol for the quantification of trace metals in different types of micro-plastics was described by Hildebrandt et al.99 Three different microwave digestion systems, that differed in their temperature and pressure regulation as well as the number of samples that could be processed simultaneously, were compared. Also compared were different acid digestion mixtures as well as different temperatures and hold times of the digestion procedure. The digestion and analysis stages were validated by the use of six CRMs: two polyethylene CRMs (ERM1-EC680m and ERM1-EC681m), a PVC (NMIJ CRM 8123-a), two polypropylene (NMIJ CRM 8133-a and lead in plastic-QC) and an acrylonitrile butadiene styrene (ABS) CRM (BAM-H010). Determination of a total of 56 analytes (although the certified ones were: As, Cd, Cr, Hg, Pb, Sb, Sn and Zn) was by ICP-MS/MS. Recovery in the certified materials fell in the acceptable range of 95.9 ± 2.7% to 112 ± 7% for all analytes except for Cr in ABS. After optimization of the procedure, these figures improved marginally to be between 103 ± 5% and 107 ± 4% for all analytes. However, this further optimization also brought the Cr result in ABS to much more acceptable levels.
The other main topic of research has been the analysis, characterisation and identification of waste plastics. An example by Junjuri and Gundawar used a low-cost LIBS system to obtain analytical data and then the chemometric tools of principal component analysis (PCA) and artificial neural network (ANN) to analyse the data and to characterise 10 types of polymer.100 The two LIBS setups used were described in detail. One comprised a Nd:YAG laser, an Echelle spectrometer, an intensified CCD, optics and an X–Y translational stage. The other comprised a low cost, compact, and portable non-gated Czerny–Turner CCD spectrometer. Data were recorded in two modes: single shot using both CCD and intensified CCD and an accumulation of 10 shots using only the intensified CCD. Spectra were collected that obtained data for Ca at 393.33 nm, H at the 656.36 nm line, K at 766.69 nm, Mg at 279.51 and 280.33 nm, N at 742.42, 744.35 and 746.90 nm, Na at 589.52 and 589.61 nm and for O at the 777.23 nm line. In addition, the C2 Swan band and the band for CN were also measured. Data analysis using PCA was described as having “excellent discrimination”. However, use of ANN demonstrated even better performance, with discrimination rates of up to 99% being achieved. The Echelle-based instrument equipped with the intensified CCD and using 10 laser shots provided data that had 97% discrimination. However, the Czerny–Turner-based instrument equipped with the ordinary CCD and using only one laser shot provided a similar performance. This latter instrument was therefore much lighter, smaller, was more portable and less expensive than the Echelle instrument. It also provided data much more quickly (by a factor of 15) because of a reduced amount of time required to train and test the ANN model.
A paper by Brunnbauer et al. described the use of LIBS combined with PCA and K means clustering to classify bulk polymers as well as individual layers of layered polymer composites, e.g. two-sided sticky tape.101 The LIBS parameters (laser energy, gate delay, laser spot size and the gaseous atmosphere) were all optimized. A short gate delay resulted, in increased emission intensities of the elements C, H and O and longer gate delays improved the signal-to-noise ratio especially in the wavelength range of the C2 swan band. The compromise conditions were tabulated. Bulk polymer sample types from a 3-D printer included polyethylene, polyacrylate, acrylonitrile butadiene styrene, polyvinyl chloride (PVC) and polylactic acid. The two-sided sticky tape was known to comprise four different layers: a polyethylene liner, polyacrylate adhesive layer, PVC support and a second polyacrylate adhesive layer providing a total thickness of 250 μm. The PCA was incapable of discriminating between PVC and polyacrylate. However, the K means clustering discriminated between all polymer types. The obtained 3D distribution of the different polymers present in the multilayer sample was compared with data obtained from a microscopic cross-section of the investigated sample. The sequence of the different layers present in the sample was correctly classified and also the thickness of each layer is in good agreement with the microscopic cross-section.
An array of different spectroscopic techniques (attenuated total reflection (ATR)-FTIR, near IR reflectance spectroscopy, LIBS and XRF) and seven different classification methods were used by Michel et al. to classify different polymers.102 Classification methods included: naive spectral matching and three machine learning classifiers (support vector machine (SVM), k nearest neighbours (kNN), and linear discriminant analysis (LDA)), each of which was also coupled with PCA: PCA-SVM, PCA-kNN, and PCA-LDA. Polymer types tested were: polyethylene terephthalate (PET), high density polyethylene, PVC, low density polyethylene, polypropylene and polystyrene. The success rates for the different techniques using the different classifiers were tabulated. Linear discriminant analysis proved to be the best classifying method for the ATR-FTIR, near IR-reflectance and LIBS analyses with a success rate of 99 ± 1, 91 ± 1 and 97 ± 2%, respectively. The best classification success rate for XRF analysis was obtained using k nearest neighbour; but was much lower at 70 ± 4%. The last paper of interest in this little sub-section was provided by Liu et al. who described the development and use of a software system to be used in conjunction with LIBS analysis to analyse different plastics.103 Partial least squares discrimination analysis (PLS-DA) and partial least squares regression (PLSR) were used for qualitative and quantitative analysis of plastics. The LIBS devices, including laser, spectrograph, delay generator and sample holder, were automatically controlled by the software. The plastic database was created with the capability of input, query, modify, delete and storage. It included a basic element database based on NIST and a dedicated plastic database which contains industrial plastics, plastic bottles and standard samples. A de-noising method, a baseline correction method and a peak detection method enabled material identification based on PLS-DA and quantitative analysis based on PLSR were all realized by the software. The software performance was excellent with an identification accuracy of 99.80% for industrial plastics being achieved.
Several other applications of interest have been published. One by Costa described LIBS calibration strategies for the determination of Pb in polypropylene recycled from car batteries.104 The LIBS parameters were optimized using central composite design, a multivariate approach to optimization and the calibration strategies evaluated were: partial least squares (PLS), principal component regression (PCR) and two point calibration transfer. The “percent trueness” was calculated by comparing the data obtained using the different calibration strategies with the Pb concentration in the samples obtained by using a microwave-assisted acid decomposition method followed by ICP-OES analysis. This reference method was validated through the analysis of the certified reference material ERM-EC681k polyethylene. Results for the calibration strategies ranged between 74 and 139% for PLS, 74 and 123% for PCR and 100 and 127% for two point calibration transfer. Precision was also very variable, ranging between 3 and 37% for PLS, 5 and 30% for PCR and 6 and 21% for two point calibration transfer. Although the results and precision were very variable, the authors still concluded that they were fit for purpose.
A forensics application of polymer analysis was reported by Nishiwaki et al. who described the non-destructive analysis of individual white polyester clothing fibres using synchrotron radiation microbeam XRF with vertical focussing.105 A Kirkpatrick–Baez mirror was used to focus the 20 keV X-ray beam onto individual fibres from clothing commonly sold in Japan. A focussed beam of dimensions 2 μm × 300 μm was approximately 12.8 times more sensitive than synchrotron radiation XRF with an un-focussed 300 μm × 300 μm beam. Scatter plots of the Ti kβ/Sb Lαβ against the Zr kα/Nb kα X-ray intensities were used to discriminate between individual fibres. These elements originate from catalysts used during the manufacture of the material, e.g. polymerization catalysts, transesterification catalysts, delustering agents (titanium dioxide) and their contaminants (e.g. Br, Co, Cu and Zn). The fibres can therefore be classified according to the methods used to synthesise them. Precision for the Ti kβ/Sb Lαβ and Zr kα/Nb kα intensity ratios was in the range 1.5–7% RSD and 1–8% RSD, respectively. Using these ratios as well as the extra elements for classification, 98% of the fibres were differentiated.
Although XRF is a powerful analytical tool, most instruments cannot quantify analytes with very low atomic number (i.e. less than 11, Na). Aidene et al. published a paper reporting a method whereby scattering of monochromatic X-rays at different incident radiation angles provided quantitative information on both physical and chemical properties of the plastics.106 The XRF instrument contained both a copper and a molybdenum tube and both were used at radiation incident angles of 0.5, 1 and 2°, yielding six different measuring modes. Such radiation incidence angles provided scattered radiation angles of 90.5, 91 and 92°. The polymer blocks were placed in the cell holder without undergoing any treatment and were analysed using a spectral accumulation time of 300 s. The chemometric tool of partial least squares (PLS) regression was used to relate the scattered spectral radiation with the concentrations of C and H, hence enabling physical sample properties such as density, mass per atom and water absorption to be elucidated. Root mean square error of cross validation was used as the performance metric and provided values of approximately 4% for C.
Calibration of LA-ICP-MS measurements is known to be problematic and has been tackled in a paper by Kuczelinis et al. who employed a method of standard addition using dried picoliter droplets of aqueous standards.107 A dosing device based on a modified commercial ink cartridge and dosing interface was developed to deliver a droplet on demand. Once dried, the residue had dimensions in the low μm range, yielded LOD at the fg level and provided a calibration that spanned six orders of magnitude. Since it is a standard additions method, there is a necessity to measure the sample mass ablated. This was achieved using microscopical analysis of the crater diameter and depth, i.e. its volume, and then using its density to calculate a mass. The methodology was applied to the successful analysis of polymeric thin films and to a glass cover slip.
Rodrigues et al. described a method for the determination of analytes (Cr, Fe, Mg, Mn, Na and Ni) in samples of diphenylmethane diisocyanate, a material used during the formation of polyurethane polymers.108 Samples (0.5–23 mg) were weighed using a balance capable of 0.01 mg resolution into a solid sample introduction device and then analysed using GFAAS. Calibration was achieved by pipetting aqueous standards onto a solid sampling platform and then running the normal furnace program. Zeeman background correction was used in all cases. Extremely low LOD were obtained with values ranging from 0.06 ng g−1 for Fe up to 3 ng g−1 for Ni. The results obtained using the method were compared with those obtained using a microwave assisted acid digestion in which 400 mg of sample was digested using a mixture of nitric and sulfuric acids. A second alternative method was also employed in which 200 mg of sample underwent a microwave induced combustion. Both alternative methods used ICP-OES and ICP-MS to obtain the data. No significant difference was observed between the data obtained from the proposed method and those from the alternatives.
The measurement of polymeric thin film thickness using EDXRF and multivariate analysis of the raw data was reported by de Almeida et al.109 Nine polymeric films with thickness ranging from 1.5 to 7.6 μm and of differing composition were used for calibration and validation processes. A univariate approach was first adopted, but failed to determine thickness adequately with the coefficient of determination being, at best, 0.873 (for the Rh-kα Compton counting rate against film thickness). Other spectral regions yielded even poorer coefficients. The multivariate regression was achieved by employing PLS, with three different regions of the X-ray spectrum being used. Eight of the films were used for training the model and one for validation. The coefficients of determination ranged between 0.993 and 0.998, indicating excellent ability and this was achieved with an acceptable precision of 4% RSD. Scanning electron microscopy was used as an alternative technique to validate the results.
4 Inorganic chemicals and materials
4.1 Catalysts
The review period has seen an uptick in papers reporting the coupling of electrochemical flow cells with ICP-MS for the tracking of electrocatalyst stability under operating conditions. The configuration enables potential- and time-resolved dissolution monitoring of individual metal counterparts with extremely high sensitivity. It has been effectively employed to monitor Pt dissolution in gas diffusion electrode (GDE) half-cells,110 carbon supported single atom catalysts,111 bimetallic PtRu/C alloy112 and PtNi alloy113 catalysts and proton-exchange membrane fuel cells.114,115 The technique was also employed to assess the stability of IrO2 coated TiO2 core–shell microparticles used as oxygen evolution reaction catalysts in water electrolysers.116 In each case, the mechanistic insights gained improved understanding of the operation and limitations of the electrochemical fuel cells tested. It also highlights how atomic spectrometry can play a role in sustainable development.The accurate and precise determination of PGMs in catalytic convertors remains an area of interest, likely driven by their sustained high market price. This is a very mature topic, and it is difficult to find truly novel work. Rather, papers are often minor extensions of well-established procedures. For example, an ‘optimized’ microwave extraction procedure for determination of Pd in Pt in spent auto catalysts by ICP-OES and GFAAS failed to enthuse this author.117 Preconcentration methods continue to be reported to allow reduced limits of detection. Zhang et al. described a bismuth fire assay method for the preconcentration of precious metals for analysis using ICP-OES.118 Ground samples were mixed with 48 g Bi2O3, 25 g Na2CO3, 5.5 g wheat flour and 80 g slag silicate. The mixture was transferred into fire-clay crucibles and fused at 1080 °C for 25 min followed by dissolution of the cupel in aqua regia. Measurement of multiple preparations indicated a repeatability better than 0.21% at ∼2000 ppm concentration for Pt. As well as reporting an optimized method, the group discussed at length the thermodynamic mechanism of the process. Anthemidis and Tzili reported a quick and simple online flow injection technique for the preconcentration of Pd prior to determination using FAAS.119 The method was based on the in-line formation of palladium diethyldiphosphate and sorption on a PTFE turning column packing. Elution with MIBK prior to atomisation and quantification achieved a preconcentration faction of 114 for a 60 s run time. The detection limit was 0.9 μg L−1 with an RSD of 2.8% at 20.0 μg L−1.
An interesting collection of papers reporting the use of ICP-MS/MS as a screening reactor to monitor intermediate species formed during gas phase reactions was reported by Zhu et al.120–122 Catalytic metal ions of interest were generated in the plasma and isolated using the first quadrupole mass filter. These were then introduced into the collision-reaction cell with gaseous reactants where reactions took place. The product ions and newly formed metal adducts were then immediately analysed via mass spectrometry using the second quadrupole, allowing rapid assessment of the possible formation pathways. The technique was used to understand reaction pathways for the conversion of ethanol to butanol catalysed by a Ce, Ir, Ru, Zr, and TiO catalyst and the TaO catalysed coupling of methane with ethane and ethylene.
A quick, flexible and reliable method was developed, based on LA-ICP-MS, for accurate assessment of nanoparticle catalyst materials with sample amounts in the picogram to nanogram range.123 The calibration and quantification was a variation of a micro-dried droplet approach, where self-aliquoting wells were used to avoid common drawbacks, such as the “coffee stain” effect and size discrimination due to increased evaporation rate towards the rim of the droplet. Wells were prepared as 20 × 5 arrays using a NWR213 nm laser to cut 50 μm holes in polycarbonate plate. Plates were loaded with 10 μL of sample and any excess removed with a glass slide. The small volume of liquid in each well evaporated within seconds allowing almost immediate analysis. Accuracy of the method was assessed by the analysis of AUROlite reference material, with results for Au (0.74% wt ± 0.13% wt) in good agreement with the certified value (0.8% wt). The study highlighted the difficulties in obtaining a genuinely homogenous sample when significantly reduced sample sizes, as low as 3 pg, are used for analysis. In the reported case, a minimum 300 pg of material was required to cancel out any micro-heterogeneity.
4.2 Forensic applications
The interrogation of gunshot residue (GSR) continues to be a major inorganic forensic application of atomic spectrometry, serving as alternatives to the established SEM-EDX analysis. Madeira et al. reported the successful discrimination of GSR collected from the hands of shooters of four different firearms using WDXRF.124 Statistical analysis of trace component concentrations and Rh tube Compton scattering arising from the organic components in the sample allowed identification of the GSR examined, including those that were Ba, Pb and Sb free. Analysis was performed using a standard laboratory sequential spectrometer equipped with a 4 kW Rh target X-ray tube and a range of diffracting crystals. Spectral data were processed using smoothing, baseline correction, normalisation and scaling prior to analysis by HCA, PCA and nearest known neighbour techniques. Detailed elemental mapping of tape-collected GSR was achieved using LA-ICP-MS equipped with a 213 nm Nd:YAG pulsed laser.125 A 4 mm2 area was scanned in 33 minutes with the ablation of 100 lines, 2000 μm in length, and a spot size of 20 μm. A constant flow of He was used to carry ablated material to the plasma. Accurate determination of 137Ba, 121Sb and 208Pb allowed for the visualisation of collected samples, whilst ternary distribution plots allowed discrimination of collected GSR particles from other sources of characteristic elements such as brake dust and fireworks. The mapping of 15 elements associated with GSR using LA-ICP-MS was also reported by Aliste et al.126 Sampling was performed using swab devices impregnated with EDTA, with the material collected from the nostrils of shooters where GSR persisted for longer periods in the nasal mucus than on the hands. Particles were detected 6 h after shooting occurred.Trace elemental profiling coupled with chemometric and other statistical techniques was utilized to characterise and discriminate a variety of pyrotechnic sparklers used in the production of homemade explosive devices. The concentration of 50 elements in 48 sparkler samples from eight brands available in Australia were determined using ICP-MS. Ground up samples were extracted in 10% nitric acid for 24 prior to analysis and a seven-element profile, consisting of Co, Ni, Sb, Sn, Sr, V and W, developed to identify the correct brand. The paper demonstrated the potential of using trace element data and chemometrics to correctly identify the origin of materials for forensic evaluation.
Several papers have reported forensic applications of glass analysis. Akmeemana et al. discussed the standard method ASTM E2927-16e, which is a consensus-based approach to sampling, sample preparation, quantitative analysis and “match” comparison of chemical properties.127 Using glass from different manufacturers prepared days, weeks and months apart and analysing using LA-ICP-MS enabled the analytical data to be studied using chemometrics. Likelihood ratios were established that range between 10−3, where the glass comes from different manufacturers or from the same manufacturer months apart, to 103 where the glass samples are from the same manufacturer and prepared on the same day. The paper was very light regarding the analytical chemistry detail and concentrates more on the chemometric side of the work. However, is worth mentioning simply because it is a standard method. The method requires a glass fragment with dimensions of approximately 400 × 200 × 100 μm. A paper by Becker et al. reported attempts to reduce the size of the glass particle required.128 Ten small fragments of glass along with the standard reference materials FG1, FG2, NIST SRM 610 and NIST SRM 612 were mounted on a 3D moveable stage. Float glass standard FGS 2 was used as external standard, FGS 1 was used as a quality control sample and NIST SRM 612 served as a standard to quantify elements not certified in FGS 2. The samples were then analysed using LA-ICP-TOF-MS using an ArF excimer laser, the operating conditions for which were tabulated. The spot size of the laser was only 90 μm so the dimensions of the glass particles required for analysis could be reduced to 100 × 100 × 33 μm. This represented a sample mass reduction from 20 μg to only 0.8 μg. A total of 110 laser shots were made on each sample, with data from the first 10 being ignored because of potential surface effects. Data for 18 analytes were obtained and a modified 5-sigma criterion was used to identify match and no match samples from different sources.
A total of 17 laboratories participated in three inter-laboratory comparisons for the forensic analysis of glass fragments from car windshields.129 The report, by Corzo et al. reported the assessment of the data obtained using refractive index as well as from LIBS and μ-XRF. Every laboratory was sent three fragments of each of several samples, some of which were known samples and others were “unknown”. The exact details of each of the comparisons were given in the paper. The most interesting was in comparison three, where the participants were given a case scenario where three questioned glass fragments (Q1) from suspect one and three fragments (Q2) from suspect two were sent to the participants. Participants were asked to follow ASTM E1967 (for RI) and ASTM E2926 (for μXRF). In this example, all laboratories using refractive index correctly associated Q1 to K1, resulting in no false exclusions. All labs also correctly excluded all K1/Q1 and K1/Q2 pairs, resulting in no false inclusions. Similar results were obtained for those laboratories using μ-XRF and the labs using LIBS. Overall, the correct association rate was greater than 92% for all three techniques for samples that originated from the same source. For samples that originated from different vehicles, the correct exclusion rate is 82, 96 and 87% for refractive index, μ-XRF and LIBS, respectively.
Almirall et al. published a paper reporting the results of 17 major and trace elements in new float glass materials intended to be used as matrix matched standards for LA-ICP-MS forensic analysis of glass.130 Three Corning glasses, one each at low, medium and high concentrations of analytes (CFGS1–CFGS3), were distributed to seven laboratories who then analysed them using the standard ASTM E2927-16e1 method. A range of ICP-MS instruments and LA units were used during the study, with the diameter of the laser sampling spots being in the range 50–100 μm. A total of eight sets of data were obtained using the standard LA-ICP-MS methodology along with one set obtained using the standard ASTM E2926-17 μ-XRF method. Precision was typically better than 3% RSD within each laboratory and better than 5% RSD between labs. Given the laser spot size and the precision achieved, it was concluded that the glasses were homogeneous. Results were checked against the existing SRMs NIST 1831 and FGS2. At present, the materials developed in this study are not yet regarded as being SRMs. However, work is on-going with NIST to make them so.
Scatter is a problem associated with the forensic μ-XRF analysis of glass fragments, a problem exacerbated by their small size (typically <1 mm). A study by Corzo and Steel attempted to improve the signal to noise ratio of μ-XRF measurements by reducing the scatter produced by the sample stage.131 This was achieved by using 3D printing to produce a plastic mount that enabled the sample to be as raised as possible from the sample stage. Results for LOD from the analysis of NIST SRM 1831 were compared between a simple sample cup, a stage and two different mounts as well as for two different instruments. In both instrument cases, the LOD improved for the cup and the mount when compared with the standard stage, although the improvement was not uniform across the analyte mass range. Instead, the light elements, e.g. Mg and Na as well as the heavy elements Sr and Zr, were improved by a factor of approximately two using the mount. Other analytes, e.g. Ca, Fe, K, Mn and Tl showed no improvement in LOD. Raising a specimen above the stage can improve the analysis of any type of sample that is partially transparent to the primary beam or small relative to the primary beam, as this will ensure that the background scatter is predominantly from the specimen itself and not the instrument. As particles approach bulk thickness (i.e. they become opaque to the primary beam), the stage plays an insignificant role and raising the specimen above the stage is not necessary. A second method to improve the LOD, that of using primary beam filters, was also tested during the study. Elements with lines in the high energy range (e.g. Rb, Sr and Zr) showed the greatest improvement in LOD. Unfortunately, different elements demonstrated an improvement whilst employing different filters.
A standardized method for the in-air particle induced gamma-ray emission (PIGE) analysis of glass fragments that uses a tantalum window material as a current normaliser was reported in a paper by Sharma et al.132 Glass fragments were irradiated in air using a 3.5 MeV proton beam and simultaneous measurement of the prompt gamma rays from proton-induced reactions of isotopes of low atomic number elements e.g. Al, B, Mg, Na and Si as well as those at 135 or 165 keV from the tantalum window were measured. The results were compared with those obtained using a conventional (in vacuum) PIGE method in which sample in pellets of cellulose were analysed. Both methods were validated through the use of soda lime and borosilicate glass reference materials. The external (in air) method was both rapid and non-destructive and was able to discriminate between the two different glass types qualitatively and quantitatively. The only analyte capable of discriminating soda lime (automobile) glasses was Al. Therefore, an instrumental neutron activation analysis (INAA) method was also developed that enabled 13 analytes (including some transition and REE) at both major and trace levels, to be determined. Statistical analysis of the data using dendrograms and cluster analysis indicated that the glasses used in the study fell into three distinct groups. This gave the methods developed the potential to be used for forensic analysis.
4.3 Building materials
The application of LIBS in the field of civil engineering continues to receive attention due to its portability, speed of analysis, minimal sample preparation and multi-element capability. Bryukhova et al. reported its used for the evaluation of aging of reinforced concrete structures.133 The paper focused on the detection of the corrosion products Cr, Fe and Mn, emerging on the structures surface. However, this was limited to surface areas painted with corrosion products. The application of LIBS for the identification of concrete type was reported by Volker et al.134 Measurements were performed with a diode-pumped low energy laser (1064 nm, 3 mJ, 1.5 ns, 100 Hz) in combination with two spectrometers covering the UV and NIR spectral range. A linear discrimination model effectively allowed classification of 10 different cement types. However, the analysis was found to be sensitive to moisture and sample matrix and thus required further work.Sulfur can occur in concrete aggregates in the form of sulfate (gypsum and anhydrite), as well as sulfides (pyrites, marcasite and pyrrhotite), the latter being linked to concrete degradation. Cruz-Hernandez and colleagues exploited the variation in intensity of the S Kα, S Kβ and S Kβ′ peaks due to changes in the electronic environment of S in different oxidation states to determine sulfate/sulfite ratios in cement using WDXRF.135 Measurements were performed using a basic laboratory WDXRF spectrometer calibrated using mixtures of CaSO4 and FeS blended with synthetic cement matrix. Since S should only be present as S6+, sulfate, and S2−, sulfide, the impact of matrix effects was investigated by comparing the summed ratio of each species to total S as determined by an independent method. Any significant deviation from 1 indicated an error. This method was validated for S concentrations between 0.1 and 2 wt%, so long as the lead content was below 8 wt% since the Pb Mβ and Pb Mα lines overlap with S Kβ, S Kβ′.
A detailed overview of the optimization and validation of an EDXRF method for the determination of metals in cement leachates by means of chemometric evaluation of the analytical variables was reported.136 The method enabled the determination of As, Ba, Cd, Co, Cr, Cu, Mo, Ni, Pb and Zn in liquid leachates, with uncertainty ranging from 9.1% for Cr to 19.1% for Zn and LOD from 0.022 mg L−1 for Pb and 0.080 mg L−1 for Cu. The researchers concluded that the mass of liquid samples presented to the spectrometer significantly affect the quality of the analysis. Given the extensive fundamental knowledge of critical depth, this is hardly ground-breaking.
4.4 Fertilizers
The rapid, standoff analysis capability of LIBS lends itself well to the application of quality monitoring on production conveyer belts. One such example was reported by Zou et al. for the online monitoring of H2O and KCl in potash fertilizers.137 The research focussed on the development and application of machine learning models, including back-propagation neural networks and convolutional neural networks, to overcome the challenges caused by water on the collected LIBS spectrum. A compact LIBS setup was installed on a potash production line, continuously collecting spectra at a rate of approximately 3 per second. Offline samples were collected at hourly intervals with H2O and K concentrations determined by gravimetric and ICP-OES methods. In total 119 samples were collected, and the corresponding LIBS spectra combined to train the data models. With the rapid evolution of data science and machine learning, this looks like an area that could see major developments in the near future.In a more conventional approach to the analysis of fertilizers, Patidar et al. detailed the development and validation of an ICP-MS method for the determination of 11 trace elements.138 The work focused on the utilisation of an Argon Dilution Kit, available with most modern instruments, to address the issues associated with high dissolved solids. The optimized method had LODs ranging between 1 and 7 ppb with typical repeatability of 2–5% RSD at 80 ppb. A method of microwave assisted digestion for the determination of 19 elements in phosphate fertilizers and related materials by MIP-OES was reported.139 Whilst not particularly novel in approach, the paper does document the capability of MIP-OES for the determination of trace analytes in these complex matrices.
4.5 Other inorganic materials
The diversity of the applications of atomic spectrometry was once again evident in the review period, with studies ranging from the analysis of high-purity germanium oxide semiconductor material140 to a review of heavy metals in children’s chalk sticks in China.141 The former utilised the reactivity, and subsequent volatility, of Ge and chlorine gas to remove matrix interference by distillation of GeCl4 at 120 °C. Nguyen et al. reported a simple separation system for the elimination of polyatomic interferences encountered during the determination of ultra-trace rare earth elements (REE) in high-purity Eu2O3 and Yb2O3 using ICP-MS.142 Whilst many REE impurities could be determined with a standard ICP-MS setup, monoisotopic 169Tm+ and the most abundant 175Lu+ isotope suffered from significant interference from 153Eu16O+ and 174YbH+ ions formed in the plasma. Interference removal was achieved using chromatographic separation of the high concentration of matrix elements by reverse-phase HPLC using a C18-silica column impregnated with 2-ethylhexyl, 2-ethylhexyl phosphonic acid and a dilute HNO3 mobile phase. This separation method was ideal for coupling with ICP-MS. The optimized method allowed for complete separation of analyte and interferents within 3.5 minutes.The matrix plays a crucial role in trace determination of analytes by all atomic spectrometry techniques and the skill of the analytical chemist is knowing how to overcome matrix effects. A simple approach, based on the standard dilution analysis method, was presented by Sloop et al.143 Unlike conventional standard dilution analysis, where two solutions are mixed during measurement, the team’s approach consisted of a 50% sample 50% standard solution and a 50% standard 50% blank solution to produce a matrix matched two-point calibration curve. The method was used to determine Cd, Co, Cr, Cu, Fe, Mn, Ni, Pb and Zn in four matrices: 1% v/v HNO3, 20% v/v HNO3, 0.1% m/v Na, and 0.1% m/v Ca solutions using MIP-OES, ICP-OES and ICP-MS. Results were shown to be consistent with those obtained using other calibration methodologies such as external standardisation, internal standardisation and standard additions. In many cases, improved recovery was obtained for the proposed method compared with the conventional calibration methods. The role of the matrix in trace analysis using TXRF, often marketed as ‘matrix free’, was studied by Dhara.144 Dilute multielement ICP standards were measured with varying concentrations of the matrix elements Al, U and Zr. Recovery rates were calculated against theoretical concentrations and were within 7% for analytes up to 20 μg mL−1 and within 12% for higher concentrations. Of the three matrix elements studied, the most pronounced effects were observed with increasing atomic mass and concentration, i.e. 1000 μg mL−1 uranium.
Accurate determination of impurities in a bulk matrix using glow discharge mass spectrometry (GD-MS) requires a sample and standard of identical matrix to be measured. Alternatively, relative sensitivity factors, RSFs, can be calculated from reference materials of similar matrix. However, these are sensitive to variations in the discharge conditions (amongst other things). Paudel and Sabatino studied the impact of changes in current, voltage and argon flow on the determination of 16 elements in silicon pressed onto an indium sheet.145 The variation in argon flow was quantified along with impurity elements to identify a range of discharge conditions where the deviation is minimum. Further, the discharge settings leading to an optimum crater in a flat silicon sample were also considered before finalizing the optimum discharge parameters for RSF determination. The RSFs were estimated and verified by subjecting the CRMs NIST 57b and 195 to the discharge condition of 2.3 mA and 1.2 kV and comparing with certified values. The estimated RSFs were further verified by ICP-MS analysis, where comparable data were obtained with errors below 20% for most elements.
The varied applications of LIBS continued to feature during the review period. The approach of sampling salt solutions by drying on filter paper prior to LIBS analysis was used for the determination of Ba, La and Sr in LiCl–KCl salt mixtures.146 Using Yb (369.42 nm) as an internal standard, linear calibration curves were obtained from 50–800 ppm for Sr (421.55 nm), 50–1000 ppm for Ba (455.40 nm) and 300–4000 ppm for La (394.91 nm) with a detection limit of 21, 19 and 68 ppm, respectively. The same approach was used by Kumar et al. to determine Sr in sea salt samples using a weak Na I line at 330.2 nm for signal normalisation.147 The team found that preparation and analysis of samples in triplicate significantly improved the quality of analysis and optimized conditions enabled the determination of Sr with an LOD of 2 ppm and precision of 5%. The direct analysis of liquid boric acid solutions was achieved by creating a laminar liquid jet of sample using a nebuliser and peristaltic pump.148 A double pulse LIBS setup was used to excite the sample, with molecular emission bands for BO and BO2 interrogated by laser ablation molecular isotopic spectrometry. The use of a double pass laser increased the intensity of molecular emissions tenfold compared with the single pass configuration, enabling determination of B with an LOD of 72 ppm.
An interesting and novel technique for the suppression of self-absorption effects in LIBS measurements was reported.149 The experimental setup consisted of a 1064 nm YAG laser focussed laterally 6 mm above the sample surface generating a He gas plasma with flowing helium. A second 355 nm YAG laser focused directly on the sample surface created the target plasma. Ablated sample ions arrive at the secondary He plasma in the ground state, where they were excited through a Penning-like energy transfer with excited He ions. The approach was demonstrated for the determination of K and Na in salt samples, where concentrations would normally be beyond the linear range of standard LIBS.
Yang et al. reported the use of reactive gas LA-ICP-MS/MS to overcome the isobaric interferences encountered when determining trace level Rh and Pd in copper minerals.150 Single quadrupole ICP-MS suffered from 63Cu40Ar+ and 65Cu40Ar+ overlaps on the 103Rh and 105Pd analyte masses that were only partially removed by kinetic energy discrimination. Ammonia reactive gas used in the analysis by MS/MS mode allowed quantitative removal of interference through reaction with Cu ions to form Cu(NH3)+ ions. The optimized method enabled the accurate determination of Rh and Pd with LODs 1.7 ng g−1 and 7.0 ng g−1, respectively.
4.6 Ceramics and refractories
In truth, there has been relatively few papers of interest in this section published during the current review period. Following on from last year, archaeological ceramics or those of cultural heritage have been discussed in their own section (Section 5.3). Several papers have been published describing diffusion experiments with data being obtained using SIMS or TOF-SIMS. However, the novelty arises in the ceramic type rather than the atomic spectrometry for which diffusion analysis using these techniques is routine. Those papers of interest that have been published are of diverse subject areas.A dried droplet calibration approach for the LA-ICP-MS analysis of glass and ceramic was described by Guo et al.151 A dispersant had to be used to ensure uniform distribution of analyte in the droplet and its residue once dry. The optimized conditions were a Pluronic F-127 (dispersant) concentration of 2% and a drying temperature of 85 °C. Precision was better than 20%. Analysis of the dried droplets did not affect ablation of the sample material and the residue from the droplet could be ablated in one raster of the laser. Regression coefficients of calibrations were excellent, ranging between 0.9936 and 0.9999 without the requirement of an internal standard. The method was applied to a YAG crystal and the NIST SRM 612 glass. Data in good agreement with the certified values were obtained for the glass sample. As a further method validation step, the samples were also analysed using ICP-OES and ICP-MS. Results from the proposed method were again in good agreement with those from the alternative methods.
An interesting application presented by Yahiaoui et al. correlated the hardness of a sample with its spectrographic properties.152 During the study, XRD and Raman spectroscopy determined structural properties, such as the grain size and phase of crystallization, of the alumina samples. A morphological study was also undertaken using SEM. In addition, LIBS with a Nd:YAG laser and XRF were used to analyse three alumina samples which had purities of 99.990, 88.869 and 73.842%. A commercial microhardness tester and a nano-indenter were used to calculate the Vickers hardness and Young’s modulus of the three samples. After optimization of the LIBS operating conditions based on the best signal to noise ratio, a gate width of 4 μs and a delay time of 2 μs, were used for all samples. After accruing several LIBS spectra for each sample, correlation curves of the surface hardness were plotted against the Al, Ca and Mg ionic to atomic line intensity ratios, the plasma temperature, the electron density and the vibrational temperature of AlO radical. Linear relationships with correlation coefficients of greater than 0.98 were achieved for the ion to atom line intensity ratios, of 0.92 for the electronic temperature, 0.82 for the electron number density and 0.99 for the vibrational temperature of the AlO radical.
Ceramics are generally regarded as being inert. However, a study by Reclaru et al. questioned whether zirconia bioceramics and ceramics intended to come into contact with skin are inert.153 Zirconia ceramics can change from a stable tetragonal phase to a less stable monoclinic form over time. Different preparation methods using various additives containing different contaminants, grain boundaries and porosities may all have an influence on the corrosion behaviour leading to the samples potentially not being as inert as first thought. To this end, the authors studied the ceramics using several different methodologies. In one, the ceramics were dissolved in a mixture of nitric and hydrofluoric acids (60:40). Other methods looked for leaching from the bioceramics using various media designed to mimic biological fluids. These included 0.07 M hydrochloric acid (for dental bioceramics and ceramics used in jewellery and in watches), 0.1% sodium fluoride and 0.1% potassium fluoride, and an artificial sweat medium (undertaken according to the standard EN 1811-2011). For all cases, a suite of elements (24) was determined using either ICP-OES or ICP-MS after a 10 or 100-fold dilution, respectively. Numerous ceramic types were tested including some cermets (composites of ceramics and metals), in all cases, the “inert” ceramics were found actually to have numerous toxic elements leached from them. The elements and their leached concentrations varied, but in some cases, the concentrations were extremely high.
Zirconia-based ceramics were also studied by Prymak et al. who synthesised and then analysed them, magnesium oxide ceramics and mixtures of magnesia/zirconia.154 Methods of analysis to determine the incorporation of Mg into the zirconia included SEM, XRD, EDXRF and, after dissolution in sulfuric acid, AAS. The hardness and Young’s modulus were also determined. Although the AAS analysis demonstrated incomplete dissolution and hence, lower Mg results than expected, the samples were homogeneous. Porosity was between 30 and 37%. Once thoroughly characterised, the samples were tested for their osteogenic potential in vivo. The mechanical analysis, alkaline phosphatase activity and calcium and collagen production indicated that the zirconia/magnesia ceramics hold a great potential for bone tissue engineering.
4.7 Glass
Various applications of the analysis of glass have been prepared during this review period, but the most common theme has been of a forensic nature. For those applications, minimally damaging methods of analysis are required to preserve the evidence. Therefore, techniques such as XRF, LIBS or LA-ICP-MS are the most common. The forensic applications of glass analysis have been placed in Section 4.2, the reader is referred there for those.The production of reference materials is always useful as it helps validate methodology. A paper by Chen et al. reported the production of two silicate glass standards containing precious group elements (Au, Ir, Os, Pd, Pt, Re, Rh and Ru) at relatively high concentrations.155 The materials were prepared at very high oxygen potentials, i.e. at elevated pressure. Experimental details of their preparation were provided. The samples were then analysed using SEM, electron probe microanalysis (EPMA) and LA-ICP-MS. Although EPMA is not a sensitive technique, it could be used because the concentration of the analytes were sufficiently high. This is important because the correction procedures needed with this analytical method are well understood and so good accuracy is achievable using only pure metal or similar readily available substances as standards. The LA-ICP-MS was undertaken using an ArF laser operating at 193 nm and at a frequency of 5–10 Hz with an energy of 60–80 mJ. This provided craters of between 47 and 100 μm diameter. Calibration was achieved through the analysis of the SRMs NIST 610 and 612. The SEM data indicated that there were no remaining metal nano-nuggets, although this was limited by the resolution of the instrument (10 nm). It was declared that the glasses were homogeneous down to at least the micron scale. Chemical analysis indicated that Pt was present in both materials at between 400 and 5000 mg kg−1, with the other analytes being less. The data obtained using EPMA and LA-ICP-MS were in good agreement.
Another paper to discuss classification of glasses was published by Devangad et al.156 These authors used a LIBS system (described and shown schematically in the paper) to obtain analytical data for iron phosphate simulated nuclear waste glasses and then used assorted chemometric tools to analyse the data. The chemometric tools were PCA, partial least squares-discriminant analysis (PLS-DA), partial least squares regression (PLSR) and the support vector machines support vector classification (SVC) and support vector regression (SVR). The paper provided a brief but useful introduction to each of the chemometric methods that a new worker in the area could find useful. The main elements of interest were Cr, Sr and Ti. Classification studies were made using PCA, PLS-DA and SVC. The PCA was unsuccessful in classifying all groups of glasses. The supervised techniques of PLS-DA and SVC were more successful, with PLS-DA classifying 90% successfully, and the SVC 100% (for both calibration and validation sets of samples). For quantitative analysis, SVR and PLSR were used. For PLSR, the root mean squared error of prediction (RMSEP) was 0.16, 0.20 and 0.08% for Cr, Sr and Ti, respectively. This was marginally better than those obtained using SVR. It was concluded that such an analytical approach could usefully be employed for process control of nuclear waste, nuclear forensics etc.
An interesting application of LIBS was reported by Stefanuk et al. who increased the sensitivity of the analysis of glass by coating it with a few layers of graphene prior to analysis.157 This is an extension to the now accepted nanoparticle enhanced LIBS analyses reported in previous years. The graphene layers were deposited onto copper foils using cold vapour deposition by CVD using methane and hydrogen as precursors. All depositions were performed at atmospheric pressure at 1000 °C for a duration of 30 min. After deposition, the copper foils were dissolved in a chemical solution of iron nitrate and, after repetitive washing cycles, the graphene layers remained floating in distilled water. They were then transferred to the sample to be analysed (Corning microscope slides). Each round of deposition resulted in six or seven layers of graphene being deposited. Using a Nd:YAG laser operating at 355 nm, the sensitivity of the LIBS determination of several analytes including Ca, Mg, Na and Si was enhanced significantly. Six-seven layers provided a much higher intensity than the plain glass surface and 12–14 layers enhanced sensitivity even further. The method was described as being cheap, rapid and easy to implement.
The surface contamination on optical glass was determined by Gerhard et al. using a sensitivity-improved calibration-free LIBS system.158 The LIBS system, employing a Nd:YAG laser operating at 266 nm was described in full with the aid of a schematic diagram. Calibration-free LIBS can suffer from reduced sensitivity. The authors overcame this by recording two spectra with different delays between the laser pulse and the detector gate. The first measurement was performed in full local thermal equilibrium conditions, when the electron density is large enough to ensure Boltzmann equilibrium distributions of all plasma species. Since this has a low signal-to-noise ratio, this spectrum enables the quantification of major and minor elements only. To obtain data for trace analytes, a second spectrum was recorded with a longer delay. The lower electron density means that the equilibrium condition is not fulfilled for all elements. Atoms such as oxygen are characterised by large energy gaps between their electronic excitation levels. They are therefore out of equilibrium. Contrarily, the equilibrium condition is still fulfilled for metallic atoms because their energy levels lie closer to each other. Since the number of charged particles are reduced, the continuum emission intensity is lower and hence low-intensity emission lines are observable with an improved signal to noise ratio. The system enabled depth-resolved measurements to be made. Method validation was achieved by combining the depth-resolved measurements to give a bulk concentration, this was then compared with data obtained using a nitric/hydrofluoric acid etching followed by ICP-OES analysis.
A method of standard addition calibration for LA-ICP-MS analysis of glass and polymers films was reported by Kuczelinis et al.107 In brief, it employed deposition of pL volumes of aqueous standard onto the sample, drying and then LA-ICP-MS of the residue and sample surface. The paper was discussed in detail in the Polymers section so will not be repeated here. Readers are referred to the Polymers section for a lengthier discussion.
4.8 Nuclear materials
The properties of materials used in nuclear fusion reactors remains a popular topic, with LIBS a dominant technique, either used alone or in combination with other techniques. The safe and efficient operation of existing nuclear reactors was also a key topic, including assessing the condition of reactor components and quantifying the impurities in fuel. The drive to expand the number of measurable radionuclides to contribute to safe, cost-effective decommissioning continues, with destructive mass spectrometric techniques commonly used, in some cases supported by rapid, online radiochemical separation of the radioanalytes. Finally, isotopic ratio measurement was a popular area for identifying the source of nuclear contamination and for the safeguarding of nuclear materials. Whilst TIMS and MC-ICP-MS were dominant analytical techniques for isotope ratio measurements, several destructive and non-destructive alternatives were presented.Tungsten-based material properties were the focus of multiple publications. Picosecond LIBS was investigated by Zhao et al. for high depth resolution diagnosis of W.164 Three ablation regimes in laser fluence were identified based on average ablation rate changes and surface morphology variations, with SEM used to investigate crater morphologies and microstructures from the three regimes. One regime was identified as superior based on the limited average ablation rate of <40 nm per pulse, the small thermal effect and therefore a potentially high depth resolution capacity. Further investigation showed the average ablation rate to be weakly dependent on crater diameter at the same laser fluence, with spectral intensities increasing with crater diameter. The formation of a rectangle-like cross section in the crater achieved by the first regime was considered beneficial for depth analysis. Veis et al. quantified H/D content in beryllium/tungsten mixture (67/33) coatings on molybdenum substrates using calibration-free LIBS, which was not known to have been performed previously.165 Different pressures were used to enhance resolution to distinguish H and D, with Be and W lines free from interference and self-absorption selected for evaluation of the electron temperature of the plasma using Boltzmann plots. The electron density was extracted from the Saha equation using the average electron temperatures obtained from the Stark broadening of the H(alpha) line. The D content of the sample (4.7 ± 2.9%) agreed with thermal desorption spectroscopy measurements (4–5%), with the depth profile in agreement with SIMS measurements.
Deuterium retention in compact, porous W–O coatings was successfully assessed using LIBS.166 The coatings were exposed to a D plasma, with strong signals measured in the laser shot corresponding to the surface layer, reducing to noise levels deeper in the coating, with higher signals in the central region. The results were supported by SIMS measurement of the lateral surface, while SEM and XRD data also supported the measured change in surface morphology. The LIBS was used as an in situ monitoring tool by Sattar et al. to assess tungsten heavy alloy (97% W, 2% Ni, 1% Fe) hardness after exposure to different plasma irradiations.167 The results achieved suggested in situ LIBS could be applied to estimate the hardness of such materials. Calibration curves were generated using the ratio of ionic to atomic line intensities of W versus the Vickers hardness. The ratio of ionic to atomic spectral emission line intensity and plasma electron temperature increased with hardness. The obtained Pearson correlation coefficient values in measuring hardness of all investigated samples had good approximation, which was a positive indicator of the LIBS approach.
Jin et al. used several techniques (focussed ion beam-SEM, TEM, GD-OES and thermal desorption mass spectrometry (TDS)) to analyse He pre-implantation on deuterium (fuel) retention in tungsten from microstructure and concentration depth profiles.168 The GD-OES showed He ion pre-implantation increased the diffusion depth of D in tungsten, whilst TDS showed the pre-implantation increased D retention due to defects produced providing new trap sites. Deposition of Be and D retention in tungsten divertors was studied by Catarino et al. following several exposure campaigns in JET with International Thermonuclear Experimental Reactor (ITER)-like wall divertors.169 Ion beam analysis techniques including proton induced X-ray emission and nuclear reaction analysis were used during the study. The results showed D retention not to be cumulative and was determined mainly by the most recent campaign, whilst total Be deposition differed to the sum of the deposits from individual campaigns for different tiles. Moon et al. investigated the fuel retention on tungsten tiles with a castellated structure, which are present for thermo-mechanical durability and integrity under high heat flux loads.170 Carbon deposition was modelled with the impurity transport code 3D-GAPS, with different concentrations at the tile groove entrance depending on the shaping, and an exponential decrease inside the gap. The highest C and D densities were measured at the plasma-exposed side of the flat tile and aligned gap, with modelled deposition profiles reproducing experimentally observed trends.
The D retention and permeation characteristics of dispersion-strengthened tungsten containing 1–10 wt% TaC, TiC or ZrC was investigated using GD-OES by Lang et al.171 The D depth profile was dependent on the second phase added, but was observed beyond the implantation depth in many samples. Reduced retention with increased O was observed using XPS. A study by Pisarev et al. measured tungsten tile surfaces after decommissioning using several techniques (LIBS, SEM/EDA, XRD, TDS and NRA).172 A deposited film was measured in 5 mm and 1 mm gaps to their full depths of 22 mm and 15 mm, respectively. The films were mainly formed from the Li limiter and C from sputtered volatile hydrocarbons from the Tokamak walls. Chemical interactions led to the formation of W2C, WC, WO2, and Li2WO4, while trapping of D and He was also found.
Molybdenum is a second element used in PFC materials. Quantitative analysis of Mo–Zr–Ti alloy by CF-LIBS was carried out by Miskovicova et al.,173 with Mo acting as a substrate and support for W thin layers. The LIBS spectra were measured under air, Ar and N2 flow at the sample laser spot, with the evolution of plasma parameters (electron density and electron temperature) observed by choosing different gate windows and delays relative to the laser pulse. The elemental concentration calculated by CF-LIBS showed the most suitable conditions for elemental analysis to be with gate delay and gate width equal to 750, 1000, 1500 ns respectively, where similar elementary concentrations in all three conditions were observed.
Wu et al. focused on molybdenum as a PFC in the Experimental Advanced Superconducting Tokamak (EAST), using TF-MS to investigate the characteristics of multi-charged ions emitted from a Mo plasma produced by a nanosecond laser at various power densities.174 The density ranges (0.85–7.9 GW cm−2) were commensurate with that used by LIBS and laser induced ablation spectrometry (LIAS), with the charge state increasing with laser power density up to a maximum charge of seven. Greater velocities of higher charged ions due to separation between charged ions during plasma expansion were also observed. Plasma shielding from strong laser absorption and reduced ablation rates with increased laser power density were the explanations given for multi-charged ion origins. Picosecond laser-based techniques were used to study the ablation features of Mo by Zhao et al.175 Confocal microscopy was used to determine the ablation depth and morphology, with FIB-SEM assessing the craters structural features, leading to information on ablation threshold, depth resolution and minimum depth of bulk Mo. The paper went on to discuss the application of picosecond laser-based techniques for current and future fusion devices.
Liquid metals offer the potential for PFCs to overcome issues with material erosion inside a reactor. Veis et al. analysed the performance of a liquid lithium divertor module based on capillary porous system technology in the COMPASS Tokamak.176 Lithium-coated Ni–Cr based screws located around the vacuum vessel were analysed to assess the migration of liquid elements, using laser ablation in air at atmospheric pressure, focusing on Li, Ca and Fe spectral lines using CF-LIBS. Zaplotnik et al. assessed D retention in liquid Sn (250 °C) following D exposure177 using thermal desorption spectroscopy and ToF-SIMS. Deuterium was not detected in pure tin, but up to 165 μg g−1 D/Sn was detected in the SnO layer, peaking with a D fluence of approximately 5 × 1025 m−2. At larger fluences the D concentration decreased below the detection limit (∼50 ng g−1), which was explained by the reduced oxide film. The impact of the plasma in reducing the SnO layer therefore means the retention of H isotopes in a liquid Sn divertor is unlikely. Tin and lithium were combined as Sn70Li30 alloy as a potential liquid metal PFC by de Castro et al.178 Wetting temperatures from 360 °C to 405 °C were determined for fresh alloy on stainless steel, Mo and W, with alloy conditions strongly affecting wetting characteristics. Several techniques (ICP-OES, TOF-SIMS, SEM, EDX and 3D laser microscopy) were used for characterization of deposited droplets, including composition, depth profile and the nature and composition of boundaries between the substrate and alloy. The results showed good compatibility of alloy with Mo and W following <3 hour exposure at 550 °C, whilst under the same conditions, corrosion through Li–Cr association and Fe–Sn mixing was observed on 316 stainless steel.
The corrosion behaviour of steel materials in fusion reactors was investigated in several studies. Two studies by Cao et al. characterized Li-induced corrosion behaviour of China Low activation Ferritic steel (CLF-1), one using calibration-free LIBS for depth profile analysis,179 and the second using an improved internal standard LIBS method.180 Following 500 hours exposure at 500 °C liquid Li, CF-LIBS effectively showed the non-uniformity of the corrosion layer and gradient distribution of elements, including the weight% of elements along the longitudinal corrosion depth for the corroded sample. In the second study by Cao et al., the same steel sample and exposure conditions showed that the matrix effect caused by a non-uniform surface microstructure could not be neglected, causing a fluctuation of spectral intensity. To eliminate this, neutral and ionised Fe atoms were selected as an internal standard, with the trends of relative concentrations between elements used to deduce the relative mass loss during the matrix elements to be Fe > Cr > V and Mn. Wang et al. also measured CLF-1, focusing on H isotope permeation reduction in steel corroded by Li4SiO4 following long-term high temperature exposure.181 A combination of XRD, SIMS and spherical aberration corrected TEM showed a three-layered corrosion product, with a middle LiFeO2 layer and inner LiCrO2. Deuterium permeation tests showed a decrease in permeability with an increase in corrosion period based on results from 15 and 30 day periods, with a cracking and reoxidation process proposed to explain reduced permeability during Li4SiO4 corrosion. Ye et al. used a combination of SEM, ED-XRF and self-designed LIBS to measure Type 316L stainless steel in stagnating liquid lithium under elevated temperatures.182 After 500 hours exposed to 350 °C liquid lithium a porous and rugged corroded layer was formed on the substrate surface. The Cr on the corroded sample decreased when the depth reached ∼2.8 μm, gradually restoring at ∼5.6 μm, compared with Li intensity that reduced to zero at 4 μm. The stability of the corrosion layer of low activation steel by contact with a ceramic breeder was studied by Hernandez et al.183 The layer was characterized using SEM, XRD, SIMS and Mössbauer spectroscopy to verify oxide formation and modification following 164 hours exposure at 550 °C in a He/H2O gas purge simulation mixture (99.8/0.2% volume). The study also identified different Fe states and H isotope traps that could be generated in the corrosion layer, with results showing no additional defects to those generated by hematite transformation to wüstite, the fundamental compound of the corrosion layer by environmental D.
A D.C. arc carrier distillation atomic emission spectrometric (D.C. Arc AES) was used for determination of multiple trace metallic impurities in uranium silicide dispersion fuel.186 A 5% AgCl carrier gas was used to sweep away trace constituents into the arc, leaving the major matrix as a refractory material. A charge coupled device detector improved the analytical performance, in particular by enabling additional analytical lines that are valuable for trace level measurements, to be monitored. The method was validated using synthetic samples, with EDXRF being used for comparative evaluations. Wavelength dispersive XRF was applied by Pandey et al. for non-destructive trace U determination in plutonium oxide.187 Following pelletisation of PuO2 powder, the U Lα X-ray was used as the analytical line, which provided a reproducibility of better than 5% at U concentrations of 200–1000 μg g−1. There was good agreement between results from WDXRF and ICP-OES.
Namitha et al. focused on the detection and identification of failed fuel pins and sodium-water reaction in fast reactors using LIBS.188 The presence of the gaseous fission products Kr and Xe, or He (for He-bonded fuel pins) in the core cover gas indicates fuel pin failure, with H2 indicating possible sodium–water reactions. Detection of these gases is therefore an important part of safe fast reactor operation. LIBS was applied to the measurement of H2, He, Kr and Xe in ultra-high purity Ar (similar to fast reactor cover gas) using their characteristic emission lines. Good linearity and correlation coefficients were reported for all elements. However, an interference from moisture in the LIBS sample chamber was observed during the determination of H2.
Tandem ICP-MS/MS was successfully used for detection of several radionuclides. Xing et al. combined ICP-MS/MS with the testing of different sample preparation techniques for measurement of Pu in soil.191 Acid leaching, aqua regia and lithium metaborate fusion were tested for different soil samples, focusing on the influence of the major interference elements (U, Hg, Pb, Tl, and Bi) that can prevent accurate measurement of Pu by ICP-MS. Combined with chemical purification, the tandem quadrupole setup with NH3 and He gases in the collision/reaction cell reduced the contribution of interferences, with values of <0.01 counts per second for Hg, Tl and Bi across the digestion methods tested. Russell et al. tested the capabilities of ICP-MS/MS for 41Ca detection, which suffers from significant interferences, some of which cannot be removed by offline chemical separation (stable Ca, Ar plasma gas).192 Concrete samples were digested using lithium borate fusion followed by multi-stage chemical separation prior to measurement. Using a combination of NH3, He and H2 in the collision/reaction cell, successful detection was achieved at activity concentrations down to 0.67 Bq g−1 (0.21 ng g−1) at 41Ca/40Ca ratios of 10−6 to 10−7. This represented the first known measurement of 41Ca using any ICP-MS design. Tandem ICP-MS/MS was also tested for the rapid assessment of 93Zr and 93Mo in concrete rubble following solid phase extraction.193 Zirconium and Mo recoveries of >90% were achieved, with method detection limits of 1.7 mBq g−1 and 0.2 Bq g−1 for 93Zr and 93Mo, respectively, based on measurement of inactive concrete. A decontamination factor of the Nb interference that affects both 93Zr and 93Mo was of the order of 10−5, which was deemed sufficient for rapid, low-level measurement in real samples.
Long-lived 135 Cs and 135 Cs/ 137 Cs isotopic ratios for source identification were measured using ICP-MS/MS in different waste samples (steel, zirconium alloy, reactor coolant, ion exchange filter paper and spent ion exchange resin) by Zhu et al.194 A multi-stage separation scheme achieved decontamination factors ranging from 2.1 × 105 (Sn) to 7 × 105 (Co). Yields of greater than 85% and a 135Cs detection limit of 3.1 × 10−14 g g−1 for 0.2 g stainless steel or spent resin were obtained. Caesium-135 and 135Cs/137Cs were the focus of other studies using alternative mass spectrometric techniques. Dion et al. employed TIMS for the determination of radioactive Cs and Sr isotopes.195 Based on stable Sr and Cs measurements, a minimum 90Zr decontamination factor of 2 × 108 was achieved, compared with 5 × 107, 5 × 106 and 8 × 106 for Y, 135Ba and 137Ba, respectively. Sample dissolution and chemical separation techniques were tested on a soil matrix, whilst the 89Sr/90Sr and 135Cs/137Cs values from measurement of an irradiated target agreed with published Evaluated Nuclear Data Files (an IAEA database), with a 5–10 times reduction in isotopic ratio uncertainties using TIMS. The same measurement technique was used by Shibahara et al. for neutron capture reaction studies of 135Cs.196 Several additive agents were investigated for their impact on Cs ionization, with silicotungstic acid the most effective of those tested, at the expense of higher polyatomic ion interference levels. The optimized setup was validated through testing of two radioactive Cs samples. Asai et al. measured 135Cs directly in spent adsorbent used for water decontamination using LA-ICP-MS.197 The aim was to address the challenges of irreversible sorption and the relatively high 137Cs radiation dose. A crushed sample was coated with a nitrocellulose based curing agent to provide a flat surface for stable sampling using laser ablation. This was combined with gamma spectrometry measurement of 137Cs. The method achieved a 135Cs/137Cs ratio in good agreement with the results obtained using solution nebulisation whilst having the advantage of only requiring small (<10 mg) samples.
The measurement of difficult to measure radionuclides using AMS was successfully demonstrated in several papers. Strontium-90 was determined by Sasa et al. as a rapid, sensitive alternative to traditional decay counting techniques.198 Strontium was recovered as SrF2 to provide sufficient negative ions. As well as ion exchange separation prior to measurement, a five-anode gas ionization detector was used to avoid isobaric 90Zr interference. The 90Sr/Sr atomic ratio background of approximately 6 × 10−13 was comparable to other AMS facilities, with good linearity from 1.75 × 10−10 to 3.38 × 10−9. Schiffer et al. investigated projective X-ray AMS (PXAMS) for measurement of 63Ni and 90Sr through their characteristic X-rays, studying production yields in a projectile ion energy range of 0.35 MeV u−1 to 1.80 MeV u−1.199 A paper by Gong et al. focused on 129I measurement using AMS at the China Institute of Atomic Energy.200 The measurement efficiency of the system was calculated to be 28% following optimization of ion transport, gas stripping, charge state selection and interference removal, with a 129I/127I sensitivity of 1 × 10−14. Pacesila et al. outlined the extended capability of the AMS facility based on the 1 MV Tandetron at the Romanian institute IFIN-HH, with preliminary results for plutonium isotopic ratios (239Pu/242Pu and 240Pu/242Pu) being obtained.201
Plutonium isotopes were also the focus of a study investigating solution loaded laser ablation modified sources for TIMS analysis.202 A Parafilm coating was applied to the filament surface and then ablated through the film to the metal beneath creating a central point source. A Pu solution can then be applied to this area and slowly dried. This simplifies the sample loading process for TIMS, with a 10% uncertainty (2σ) for 240Pu/239Pu from a 5–7 fg CRM sample, with an average sample utilisation efficiency of 1.3%.
Gadolinium, Nd, Pu and U were quantified in uranium oxide and Gd-enriched spent fuel using isotope dilution sector field ICP-MS following high performance ion chromatography separation.205 Uranium was present at higher concentrations and required off-line dilution and measurement, whilst an online approach was used for the other analytes. The final method was compared against routinely used techniques, namely isotope dilution TIMS and alpha spectrometry. In a paper by Manard et al., U was the analyte of interest for an automated separation method prior to ICP-OES and ICP-MS/MS measurement of uranium ore concentrate samples.206 Seventeen samples, two quality control reference material samples and 26 process blanks were sequentially run through a single extraction chromatography column, with the resin separating trace elemental impurities. The column was successfully regenerated even after a total of 50 mg U had been processed, with the results showing good correlation to previously reported data. Liu et al. also studied U separation, specifically electroseparation from lanthanides on a liquid gallium electrode.207 The motivation for this work was the identical reduction potentials of U and lanthanides in multi-component melt systems that could not be resolved on the Ga cathode. The study applied different electroseparation techniques, followed by ICP-OES measurement in molten salts and alloy deposits. Separation factors of up to 1000 was achieved for U/La and U/Sm, compared with approximately 350 for U/Nd. Current pulse electrolysis achieved a larger separation factor than galvanostatic and potentiostatic methods.
Green radiochemistry was highlighted in some studies as an alternative to more established separation materials. Mortada et al. investigated an eco-friendly alternative for the extraction and separation of Th from U and lanthanides in wastewater and phosphate rocks prior to ICP-OES measurement.208 Morin formed a complex with Th4+ at pH 2–4 that quantitatively extracted it into the surfactant-rich phase in the presence of 0.1% (v/v) Triton X-114 and 0.1% (w/v) KI at room temperature. The analytical range and detection limit compared favourably to other cloud point extraction procedures for Th, with an enrichment factor of 49.5 for a 50 mL sample and a recovery of at least 95% in Th-spiked samples. Phillip et al. investigated palm oil fly ash as a green supplementary material for cementitious backfill in disused sealed radioactive source borehole disposal, focusing on 226Ra confinement.209 Multiple techniques were used to measure the performance of ordinary Portland cement with and without the palm oil fly ash, with 30% ash replacement found to improve geochemical conditions by reducing competitive Ca release into the disposal environment. Samples with up to 50% fly ash replacement met the mechanical and hydraulic performance requirements, and 226Ra confinement was improved owing to the higher fraction of active sites. This suggested that the use of palm oil fly ash may reduce the radiological hazard.
Measurements of Cs-bearing microparticles emitted from Fukushima were the subject of studies by Miura et al. and Kurihara et al. The study by Miura et al. focused on measuring a larger number of particles emitted from Unit 2 or 3 (0.1–10 μm diameter particles termed Type-A) and Unit 1 (50–400 μm diameter particles termed Type-B).212 Micro X-ray computed tomography and XRF revealed multiple voids and Fe-rich parts of Type-B particles, whilst the 137Cs activity was approximately 10000 times higher in Type-A particles compared with Type-B. The differences suggested different formation routes for the two particle types, which contributes to improved understanding for decommissioning and environmental assessment. Kurihara et al. used SIMS and LA-ICP-MS to measure U and Cs isotopic ratios in five spherical Cs microparticles (approximately 2 μm diameter) sampled 50 km west of the Fukushima plant.213 The 235U/238U ratios were close to the estimated values for Units 2 and 3 fuel cores and identical in the case of Cs isotopic ratios. The results were combined with atmospheric release information to conclude that the particles originated from the Unit 2 reactor on the night of March 14th and the morning of March 15th 2011.
Nuclear forensics remains a very popular topic, with a large number of publications on measurement of U isotopic composition in fuel and particle samples. This was reflected in a publication by Jovanovic et al. who reported the results from the fifth collaborative materials exercise of The Nuclear Forensics International Technical Working Group.214 Two uranium dioxide fuel pellets were prepared and sent to participating laboratories to independently analyse their characteristics using different techniques.
Varga et al. published two papers on the measurement of U using mass spectrometric techniques. One used LA-MC-ICP-MS to study heterogeneity of solid UO2 pellets at micrometre levels as part of the Nuclear Forensics International Technical Working Group.215 Despite the identical source material of the two pellets, different production routes could be observed at micron-scale and the U isotopic ratios determined, with the imaging helping to assess the lateral dimensions of individual grains which can be used as an indicator of the production route. A second study by Varga used MC-ICP-MS to determine the U production date from a swipe sample of a highly enriched U seizure.216 Measurements following acid leaching were used to calculate the ‘bulk’ composition (72.51 ± 0.03 wt% 235U) and the production date, which was similar to other cases seized earlier in Europe. A separate study by Krachler et al. used LA-MC-ICP-MS for the measurement of U isotopic composition in five contaminated scrap metal samples found in the EU.217 Two U isotopic reference materials (CRM U-020 and CRM U-030) were also measured for quality assurance, ion counter gains and mass bias correction. Similar to the study by Varga et al.,215 LA-MC-ICP-MS was highlighted for its ability to identify inhomogeneous distribution of U within samples, with the large range of 235U enrichment confirmed using SIMS, and four of the five samples tested containing 236U, indicative of the presence of reprocessed U.
The technique of LA-MC-ICP-MS with an extended dynamic range was used for U isotopic analysis by Yamamoto et al.218 Three Faraday detectors with 10−11 and 10−13 ohm resistors were tested for a series of solid materials on transient signals with LA durations of <3 seconds to minimise sampling volume. Using a tau correction technique, the 235U/238U values were approximately 3% higher than the literature values, which was ascribed to incomplete correction of the slow response of the Faraday amplifiers. When continuous ion monitoring was applied, precision for 235U/238U improved to 1 part per thousand and for 234U/238U to 7% (2σ), showing good agreement with literature values. It demonstrated the suitability of the technique for measuring transient signals in high-gain amplifiers. Craig et al. also investigated transient signal isotope ratio analysis, in particular the limitations of ‘blind time’ on LA-MC-ICP-MS.219 A high speed integrated ablation cell and dual concentric injector design was tested for U isotopic ratios. When compared with a slower washout, established low-volume cell design, the highly transient signals of the new design introduced a new bias, a major component of which was referred to as the ‘blind time’. The impact was modelled on UOX particles, with average precisions in sub-micron particles of 3% (1σ) and 8% (1σ) for 235U/238U and 234U/238U, respectively.
The application of TIMS for U isotope ratio measurement was demonstrated by Bhatia et al. through the design and development of a compact instrument.220 The sector radius of the magnetic sector analyser in the compact instrument was 20 cm compared with 30 and 27 cm in the conventional and commercial geometries, respectively. A U200 standard was measured to compare the conventional and compact instruments. A sensitivity improvement of 1.5 for the compact instrument and an improved precision of 0.05% for 235U/238U for the compact instrument were obtained. Lee et al. assessed the signal-detection performance of TIMS with continuous heating for isotopic U measurements in the fg to pg range for nuclear safeguards.221 Peak jumping mode with a single detector was compared with static mode using multiple detectors, with 235U/238U measured down to 1 fg and 3 fg in the two modes, respectively. Both modes could only measure down to 100 fg for 236U/238U. The similar capabilities demonstrated that continuous heating during TIMS measurement offers the advantage of exhibiting the properties of multiple detectors within a single detector.
Both TIMS and MC-ICP-MS remain popular techniques for isotopic analysis, but a number of alternative analytical methods were applied to isotopic analysis. Rinaldi et al. carried out U isotopic analysis using LIBS, citing the speed of detection and identification of elements in a sample using this approach.222 The combination of LIBS with Chemometrics to determine isotopic ratios at atmospheric pressure could achieve precisions of 4–10% depending on the enrichment level, which is applicable in the measurement of certain special nuclear materials. Fast, quantitative analysis of U using LIBS was also employed by Rollin et al.223 The detection of trace and minor elements is challenging owing to the large number of actinide emission lines in the UV–visible region. Rollin et al. therefore investigated the vacuum UV (VUV) range below 200 nm as an alternative, estimating the detection limits of C and V. The results agreed with findings from non-nuclear samples, with the technique not suitable for metal impurities with useful spectra in the VUV, whilst offering improved detection limits for non-metals that have intense VUV lines.
In situ TOF-SIMS analysis of uranium oxide was presented by Yu et al. as a method for surface analysis of liquid and solid–liquid interactions as a neat, non-destructive alternative or support to destructive techniques such as ICP-MS.224 Initial results were presented for liquid uranium oxide standard solutions. Paing et al. presented rapid U isotopic measurement in cotton wipes using a planar surface reader coupled with a liquid sampling-atmospheric pressure glow discharge microplasma ionization source employing an Orbitrap mass spectrometer.225 A 30 second extraction step achieved quantitative recovery with approximately 100 ng sampled per extraction. The intra sample variability was around 30% RSD, which the authors claim was offset by <10% RSD for 235U/238U and (for enriched samples) 234U and 236U ratios. An additional benefit was that it is a convenient, versatile approach for swipe samples in a range of applications. A final alternative technique published by Hoegg et al. was liquid sampling-atmospheric pressure glow discharge with Orbitrap mass spectrometers, aimed at meeting the international target values for U uncertainty.226 The system was tested on a series of uranium CRMs, with measurement of 234U and 236U related species that had not been achieved previously. In the case of CRM U-800, the measured 235U/238U value of 4.266922 compared favourably to the expected value of 4.265622, with the instrument showing promise for forensics and safeguard applications.
4.9 Electronic materials
An enormous amount of research has been put into electronic materials recently. Many of the papers describe the development of new lithium ion batteries, their alternatives and their components and so most do not describe the analytical chemistry in huge detail. Many of the papers use the atomic spectroscopy as a tool to characterise their product. These papers are therefore not included because they do not represent an advance in the analytical aspect of the work. Papers that are included are those that describe new or improvements to existing methods of depth-profiling.Several review papers have been published over the last year or so. An excellent overview entitled “Laser ionization mass spectrometry at 55: Quo Vadis?” prepared by Azov et al.227 The paper was extremely long (>50 pages) and contains 274 references, but covered the historical aspects as well as more modern instrumental designs (Orbitrap etc.). Also covered were recent applications and more theoretical aspects, for example, the effects of different lasers and their powers. The review would be a good place to start for anybody entering the research area.
A review (212 references) by Giurlani et al. covered the different methods of measuring the thickness of metal coatings.228 The review was split into two main sections: destructive and non-destructive methods, with each of those being split into numerous sub-sections covering the techniques available. The destructive methods were: mechanical cross-sectioning, focused ion beam cross-sectioning, angle lapping, TEM, Calo tester, optical microscopy, scanning ion microscopy and chemical dissolution. The non-destructive techniques were: XRF, EPMA, ellipsometry, RBS and resonant methods. Each of the methods were described in terms of their pros and cons, their accuracy, simplicity, time required etc.
Research into the analysis of wafers impregnated with substances is still on-going. An example was a paper presented by Czyzycki et al. who used GIXRF to analyse three silicon (111) wafers that had been deeply implanted with 200 keV argon ions of nominal dose 1015–1016 cm−2.229 The technique is usually used to analyse implants with a depth of <25 nm. However, this paper reported the analysis of depths far greater than that. This was achieved using a new mathematical model that did not rely on X-ray standing wave. The calculated concentrations were in agreement with nominal values and were further validated using an ion beam analysis method. The authors concluded that their new model would expand the use of GIXRF into areas not previously explored.
Several papers have reported the analysis of thin films. Included in this number was a paper by Huang et al. who determined residual dimethyl sulfoxide (DMSO) in methylammonium lead iodide perovskite thin films using high resolution continuum source graphite furnace molecular absorption spectrometry.230 The sample was placed in the graphite furnace which was then heated to 160 °C converting the DMSO to the molecule CS, which was then detected at 258.055 nm. Calibration was against aqueous standards of DMSO and Pd was used as the matrix modifier. The characteristic mass was 17 ng. The DMSO content was normalized to the Pb content. For this, the Pb in the perovskite was reduced to metallic Pb using zinc powder and then measuring the signal at the weak line 261.418 nm using HR-CSAAS. The Pb content remained constant with increasing annealing time but the S/Pb ratio decreased, indicating better removal of the DMSO.
Refractory transition metal nitrides have many uses, e.g. as protective coatings for industrial machine parts. However, they are also gaining increasing interest in the microelectronics world as diffusion barrier layers. Their applications and desired performance require precise control of the film thickness and elemental depth profile. Torrengo et al. used a combination of XRR and GIXRF for the non-destructive analysis of titanium tungsten nitride films that had been prepared using physical vapour deposition.231 This approach is normally undertaken using a reference-free approach which requires knowledge of the whole experimental system. This can be problematic in a laboratory or workplace where parameters may be difficult to control precisely or to access. The authors developed a reference-based method in which known standard samples close in composition and structure to the samples are analysed so that key parameters of the instrumental setup are identified. These parameters are then used for the combined XRR/GIXRF analysis of the samples of interest. The method was very successful and met the requirements for both qualitative and quantitative depth-profiling analysis while being easier to implement in laboratories than the reference-free method.
Copper oxide thin films are used in many types of solar cells. Analysis of Al or Co-doped copper oxide films using LIBS was reported by Wu et al.232 The films, prepared by radiofrequency magnetron sputtering deposition, were analysed using LIBS in a rapid manner enabling qualitative and quantitative determination of the doping elements. The emission intensity ratios of Al/Cu in the copper oxide–aluminium films and of Co/Cu in the copper oxide–cobalt films were used for the analysis.
Aluminium–indium-tin oxide thin films are formed by doping Al into the well known indium tin oxide material and are promising materials for deep ultra-violet optoelectronic devices. A paper by Liu et al. described the preparation and subsequent analysis of such films prepared under different magnetron sputtering powers, pressures and times.233 The concentration ratios of Al/Al + In + Sn and Sn/Al + In + Sn were determined using LIBS. The accuracy of a comprehensive calibration protocol was determined using samples prepared under random sputtering parameters and with data obtained using EDXRF. There was less than 5% error in the data between the two techniques, indicating the accuracy of the LIBS methodology.
The work into laser ablation of solids in liquid (LASIL) followed by either ICP-OES or ICP-MS detection has continued to be a subject of interest for researchers in the field of thin films. A paper by Herzig et al. used a LASIL system built in-house coupled with ICP-MS detection for the analysis of complex metal oxide thin films.234 The system required sample transport with as little particle dispersion as possible so that a spatial map of the film could be made. This 2D mapping ability was applied to films of varying composition with a thickness of 220 nm that had been deposited using pulsed laser deposition. The workers then extended the study so that a simultaneous on-line quantification using standard additions was developed. This helped correction for instrumental drift during long-term measurements.
The analysis of layered structures and depth-profiling studies have, as always, featured heavily in the literature. The techniques capable of depth profiling are mature now and therefore many of the applications simply use them as characterising tools. As such, they offer no advances in the atomic spectroscopy and will not be discussed. A useful contribution was presented by Lobo et al. who used pulsed GD-TOF-MS for the depth-profile analysis of GexNi100−x/Ni/NdyNi100−ythin films.235 The technique of GD-TOF-MS is ideally suited to the task because it offers rapid elemental determination whilst also specialising in the capture of transient signals. The thin films, of total thickness between 15 and 75 nm were deposited on two different substrates: non-conducting glasses and conducting silicon wafers. The operating conditions for the glow discharge were optimized for each substrate to ensure best instrumental performance. Short pulse width (e.g. 0.16 ms) and periods (1.32 ms) enabled the best depth-profile resolution for the layers deposited on glass. However, the pulse width was not such an important parameter when analysing films on wafers.
The technique of TOF-SIMS is gaining increased attention for depth-profiling studies. An example, by Smentkowski and Goswami, reported the Al depth-profile in silicon carbide wafers.236 The paper indicated that TOF-SIMS was capable of replicating the quantitative data obtained using dynamic SIMS, although at the lowest concentrations, a greater number of spectra needed to be collected. Until now, TOF-SIMS had been regarded as being excellent for the analysis of thin films, but only in a qualitative manner. Applications utilising TOF-SIMS were also reported by Miyayama and Iida237 and by Priebe et al.238 In the former paper, low energy (2 keV) bismuth cluster ions (Bi3+) at an incident angle of 65° were used to generate bismuth adducted secondary molecular ions (M + Bi)+, which are not observed under normal TOF-SIMS operating conditions.237 The method was applied to multi-layers of Irganox 1010/Irganox 1098 thin films and provided a better dynamic range for both layers. It was concluded that the low energy bismuth cluster is more suitable for organic films than the normal 54 keV bismuth cluster. The paper by Priebe et al. investigated the matrix effects exhibited during the TOF-SIMS analysis of two element inorganic thin films.238 Matrix effects can be severe in TOF-SIMS, with the sensitivity of analyte signals changing depending on their surroundings. This is the main reason it has hitherto been regarded as being capable of qualitative analysis. The work undertaken in this study found that Zr ionization efficiency was four times higher in the silicon matrix compared with a copper matrix and two times higher than an aluminium matrix, the topography of the film can also affect TOF-SIMS measurements. An interesting paper by Spampinato et al. combined the use of a scanning probe microscope and TOF-SIMS in an attempt to overcome this problem.239 The scanning probe microscope provided high resolution topographical information whilst the TOF-SIMS analysed the thin layers. Rapid alternation between the scanning probe microscope and the TOF-SIMS instrument enabled far better accuracy to be obtained during the analysis of 3D structures.
Hybrid organic–inorganic perovskite devices are known to be unstable and hence have not, as yet, been commercialised. An additional problem is that their complexity makes it difficult to analyse them fully, so mechanisms of failure can be hard to elucidate. A paper by Higgins et al. described the depth-profile analysis of the interface between methylammonium lead bromide perovskite and the gold electrode.240 The authors used TOF-SIMS in conjunction with a machine learning workflow that combined the Hough transform and non-negative matrix factorization and non-negative tensor decomposition. The TOF-SIMS enabled chemical composition characterisation whereas the chemometric approach enabled interpretation of the multi-dimensional data obtained while avoiding some of the limitations. This enabled salient features of the chemical changes to be extracted whilst separating the voltage- and light-dependent dynamics.
Oxygen diffusion behaviour through a commercial A-site deficient lanthanum strontium chromium iron oxide delta perovskite was studied by Sha et al.241 These materials can be used as solid oxide fuel cells. It had been suggested that atmospheric parameters such as humidity can affect the chemistry and hence performance and durability of such devices. The technique used for the study was isotopic exchange depth-profiling SIMS. An O partial pressure of 200 mbar was used with a constant water pressure of 30 mbar. A temperature range of 600–900 °C was also used. Only very limited O diffusion was observed above 800 °C. To study this phenomenon in more depth, the authors used angle resolved XPS, SEM and scanning TEM. Strontium segregation was found to be one of the causes for suppressing the water surface exchange.
A paper by Yi et al. reported the use of picosecond LIBS for depth-profiling studies in an attempt to improve the depth resolution which, for many LIBS applications, can be poor.242 The ablation depth was measured and then correlated with the laser pulse number. A series of layered graphite samples was analysed under a residual pressure of 1 × 10−5 Pascals and a resolution of 24 nm and 102 nm was obtained for Mo and C, respectively. This represented a significant improvement over standard LIBS analysis.
A variant of calibration-free double pulse LIBS was used by Shakeel et al. to analyse gadolinium/germanium/silicon alloys and solar cells.244 The method was both simpler and faster than conventional calibration-free LIBS. The inter-pulse delay was optimized and a Boltzmann plot used to calculate plasma temperature and subsequently the concentration of the analytes. The results obtained were compared with those obtained using conventional calibration-free LIBS and were in agreement. The method was applied to the analysis of three unknown polycrystalline silicon solar cells. Traces of Al, C, Ca, Fe, In, Sb, Sn and Ti (at the mg kg−1 level) were observed. The authors stressed the possibility of their method being used in real time during the manufacture of the materials.
Perovskite quantum wells (a bulk 3D perovskite layer combined with a reduced dimensional perovskite layer) have been shown to improve the efficiency and stability of perovskite-based solar cells significantly. A study by Hu et al. used time-resolved grazing incidence WAXS to study the formation mechanisms of these quantum wells.245 The technique enabled real-time insights into phase-transition during the organic cation (in this case hexyltrimethylammonium bromide) and the perovskite assembly formation process. An ionic reaction between the 3D perovskite and the long-chain organic cation occurs rapidly forming an intermediate phase in just a few seconds. The optimal conditions can be elucidated from evidence obtained using TOF-SIMS. By controlling the second ionic reaction during the long-chain cation coating process as well as the fluorinated poly(triarylamine) as a hole-transport layer, the efficiency of the perovskite solar cells exceeded 22%.
Lithium ion batteries and their components have received a huge amount of research. This is because assorted different anode and cathode materials have been developed in an on-going attempt to improve efficiency. However, the vast majority of this work does not involve atomic spectrometry and, of those papers that do use some form of atomic spectrometry to characterise the material, most use basic techniques and hence, do not further knowledge. Consequently, those papers will not be discussed. Instead, those few papers that do describe an interesting application or modification of a method will be discussed at greater length, the temperature-dependent chemical and physical micro-structure of lithium metal anodes was investigated by Adair et al. who used synchrotron-based techniques for the task.250 X-ray computed tomography and energy-dependent X-ray fluorescence mapping were utilised to assess the surface morphology. Another technique employed was μ-XANES. This was used to determine the chemical composition of the solid electrolyte interphase in quasi-3D space combined use of the techniques enabled information to be gleaned for the origin of cycling performance at different temperatures through analysis of coulombic efficiencies, surface morphology and the chemical composition of the solid electrolyte interphase in quasi-3D space.
Lithium lanthanum zirconium oxide garnets are regarded as being promising solid state electrolytes for next generation lithium ion batteries. Their analysis using LA-ICP-OES was reported by Smetaczek et al. who first prepared matrix-matched standards and then characterised them using an acid dissolution procedure followed by conventional ICP-OES analysis.251 Once characterised, the standards were used for calibration of the LA-ICP-OES protocol. An internal standard-independent calibration strategy based on 100% normalization was applied, enabling the quantification of all cations within the material. Results from LA-ICP-OES were compared with those obtained using LA-ICP-MS and were found to be superior. Results from the analysis of an Al-doped material using the method were compared with those from a fusion-based dissolution ICP-OES method and were in good agreement.
A paper by Yamagishi et al. reported the use of operando TOF-SIMS for visualizing the Li distribution and degradation of composite electrodes in sulfide-based all-solid-state batteries.252 Two types of electrode materials were analysed during cell operation: lithium nickel cobalt aluminium oxide (LiNi0.8Co0.15Al0.05O2) and lithium phosphorus sulfide (75Li2S·25P2S5). Evolution of the non-uniform reaction of the lithium cobalt aluminium oxide particles during charge and discharge cycles was successfully visualized by mapping fragments containing Li. In addition, degradation of the interface between the materials was investigated by mapping POx− and SO(x)− fragments. Results indicate that there was a redox active component at the interface. The paper is a good example of how atomic spectrometry can be used to elucidate mechanistic aspects of processes.
Another paper to explore mechanistic aspects of processes was presented by Cheng et al., who used in situ synchrotron XRD and small angle X-ray scattering (SAXS) to investigate copper oxide/graphene oxide anode materials.253 Different copper oxide morphology could be obtained by reacting at 83 °C for different times. The analytical techniques were then used to characterise the materials during the first cycle. As the Li becomes incorporated into the material the copper oxide Bragg peak shifted indicating a higher angle in the crystal lattice. In addition, the growth of the solid electrolyte interphase increases with increased lithiation dose to the extent where cracks form. Another example was presented by Haridas et al.254 who examined the effects of the electrolyte additives tris(trimethylsilyl) phosphite and fluoroethylene carbonate on lithium nickel manganese cobalt oxide–silicon lithium ion batteries. The analytical techniques of TOF-SIMS and XPS were used to study the electrode–electrolyte interface and to investigate the depth profiles of the species. Results indicated adequate hydrogen fluoride scavenging. Different ratios of the two electrolytes were tested. A three-fold excess of fluoroethylene carbonate over the other electrolyte led to a two-fold increase in life cycle compared with an even greater proportion. Lithium titanium oxide (Li4Ti5O12) has also been proposed as an alternative for the negatively charged electrode of lithium ion batteries. The different states of lithiation were investigated using energy dispersive inelastic X-ray scattering spectroscopy in a paper by Robledo et al.255 The technique was used to differentiate between the lithiated and de-lithiated forms of the compound with good sensitivity and was regarded as being a reliable tool for undertaking such speciation work. The results were compared with those obtained using XANES and were in good agreement.
Alternative battery types to lithium ion ones exist and studies have also been undertaken on these. A battery type that has very benign components is zinc/alpha-manganese dioxide. This battery type has been investigated by Wu et al. who used operando spatiotemporal resolved synchrotron X-ray fluorescence mapping measurements to provide direct evidence of a Mn dissolution–deposition faradaic mechanism that governs the electrochemistry.256 Analysis of the electrolyte found an aqueous form of Mn during battery discharge and a depletion when charged. The results were confirmed by an assortment of other analytical techniques that included TEM, XANES and EXAFS. The mechanism was therefore different to the Zn insertion or conversion reactions that had previously been assumed.
Several other applications have been published that are worthy of mention. One by Miner and Beauchemin described the ICP-MS determination of Ni in an alkaline electrolyte solution.257 This is not a straightforward analysis because the presence of high concentrations of concomitant elements leads to signal suppression of the analyte. An additional problem is that the alkaline nature tends to de-vitrify the glassware of the instrument sample introduction system. The method developed utilised an on-line system comprising a column of Dionex™ IonPac™ CS-10 cation exchange column to retain the Ni whilst allowing the interfering ion (K, Na etc.) to elute to waste. The system was first tested on standards of Ni in sodium and potassium hydroxide solutions. The Ni was eluted using 6 M nitric acid. The whole process lasted only 260 s yielding a detection limit of 0.1–0.2 μg L−1. Strangely, there did not seem to be much quality control.
Secondary ion mass spectrometry has been used for diffusion studies several times in this review period. An example, by Wojcik et al., determined the depth profiles of hydrogen-passivated and active Zn in gallium arsenide/aluminium gallium arsenide.258 This work used ultra-low impact energy SIMS for the task of measuring the zinc-arsenic signal as a marker for active Zn and the H signal to provide information about the hydrogen-passivated impurity profile. The method could potentially be used for optimizing the growth process and controlling the level of impurity activation in such materials. Another example by Duan used TOF-SIMS for the analysis of the boehmite layer of resistive switching random access memory.259 A 3D map of the H ion distribution was obtained during the high/low resistance states. This enabled the mechanism of H ion diffusion to be elucidated. A third paper used TOF-SIMS to elucidate degradation mechanisms of organic light emitting diodes.260 Although popular, the organic light emitting diodes are renowned for failure and so information into the mechanisms of failure is required. The paper, by Iida et al., used depth-profiling analysis of degraded materials. The MS/MS mode was used to try and minimise the mass spectral interferences often observed during analysis of this type of material. The results identified a number of small hydrocarbon molecules that were degradation products of either the starting material or contaminants at the indium tin oxide interface.
The purity of the room temperature semiconductor radiation detector material caesium lead bromide was determined using ICP-MS by Makanda et al.261 The analysis for this sample type is more frequently undertaken using GD-MS and so a quick, easy and reliable alternative is welcomed. The sample was ground using a mortar and pestle and then dissolved in 2% nitric acid, first at room temperature overnight and then at 95 °C on a hotplate to ensure complete dissolution. After dilution the sample was analysed using ICP-MS. Method validation was through spike recovery experiments with recoveries obtained for the 12 analytes being generally close to 100%. Limits of detection were 0.0004 (Tl) to 0.377 (Cr) μg kg−1 and regression coefficients were better than 0.99 for all analytes. The method developed was applied to a high temperature zone-refined sample. The concentration of contaminants varied depending on the part of the ingot, with the heel being the most contaminated. In all cases, the chief contaminants were Rb and U. Strangely, U was by far the highest in the tip of the ingot, whereas Rb was far higher in the middle and the heel.
The composition of contamination deposited on transmission line insulators can affect their surface flashover voltage, currently, to estimate the extent of the contamination requires an electrical shutdown while the insulators are sampled and taken to a laboratory for analysis. An on-line method that does not require power shutdown would therefore be of huge benefit. The technique of LIBS could possibly provide such an analysis capability, but can potentially run into problems from the different forms of salts present giving different sensitivity. Lu et al. described a LIBS method of analysis that first tested whether the different forms of elements, e.g. Na2CO3, NaHCO3, CaSO4 and CaCO3 affects the sensitivity of the analysis.262 Samples containing different proportions of Na and Ca were prepared and the linear correlation coefficients for the calibration curves calculated. No significant change of the linearity of the calibration was observed, indicating that matrix effects were minimal. However, the laser energy density had a very large effect. Partial least squares regression helped improve the accuracy of the analysis significantly.
4.10 Nanostructures
Atomic spectrometry, through techniques such as XRD, XPS, XRF, single nanoparticle (sNP) ICP-MS and ICP-OES has a key role in the characterization and detection of nanoparticles (NPs) with over 200 papers published in the period covered by this ASU. However, many of these articles only mention the technique(s) used without any further analytical detail and as such are not reported on here.Further reviews on the use of atomic spectroscopy for NP characterization and detection have also been published this year. One of these, by Semenova and Silina, covers the various techniques that can be used to investigate the manufacture of organic–inorganic nanohybrids.267 The review, which cites 93 papers, gives an overview of a number of techniques that can be used for this purpose, namely FT-IR, Raman spectroscopy, surface plasmon resonance, ICP-MS, sNP-ICP-MS, LA-ICP-MS, TOF-SIMS, LC-MS and laser desorption ionization mass spectrometry. The authors concluded that the nanoanalytical techniques mentioned in the review require a fundamental understanding of the physical processes between nano-objects and these detection techniques. Developments in this field will help the extraction of a maximum of analytical information from a single run. To address this, developments in both nanoanalysis tools (instrumentation), as well as new nanoanalytical methodologies, are to be expected. Two reviews covered different aspects of X-ray-based techniques for NP characterization and detection. The first of these, by Bleiner et al. with 128 articles covered, was a tutorial review of soft X-ray LA for nano-scale chemical mapping microanalysis.268 The review gave a good overview of the topic, including differentiating between imaging and mapping, spatial resolution and the trade off with chemical resolution, mechanisms of laser sampling, laser-plasma emission spectrometry and laser MS with a wealth of information included. Detailed conclusions on the benefits and drawbacks of the techniques discussed were given. The authors concluded that the results presented show that XUV/SXR-laser desorption spectrometry shows promise as an alternative to XUV/SXR-laser ablation spectrometry although ‘engineering robustness’ of the instrumentation needs to improve for this to be achieved. The other aspect of X-ray-based techniques that has been reviewed is angle resolved XRF, as given by GE-XRF and GI-XRF.269 This review by Baumann et al., which cited 158 papers, gave a good introduction to the fundamentals of these two allied techniques, with sections on excitation sources including X-ray tubes, synchrotron radiation, laser produced plasmas, ion beams and the various detectors in current use. The review also had an extensive section on applications of GI-XRF and a smaller section on coupling of GE-XRF with other techniques such as microfocus excitation. In separate reviews, the preparation of nanoparticles for TOF-SIMS and XPS analysis was presented by Bennett et al. (54 references)270. The detection of TiO2 in aquatic samples by Heilgeist et al.271 and a historical perspective on the synthesis, investigation techniques and properties of Ag NPs, by Pryshchepa et al.272 (with 159 and 255 articles cited, respectively) have also been published in the period covered by this update.
An alternative approach to improving sensitivity for single NP analysis, using a newly developed conical torch as opposed to the parallel path of the conventional Fassel type torch fitted to most modern instruments, has been reported by Alavi et al.274 The torch was fitted into what appears to be an in-house built ICP-OES instrument, with axial monitoring of the produced emission signals. A microdrop generator was used to produce monodisperse 50 μm Ø droplets containing NPs from one of eight elements, Ag, Al, Be, Ca, Fe, Mg, Na or Sr, which were desolvated and introduced into the plasma via an Ar gas carrier stream. A detailed description of the conical torch, which operates at a reduced RF power and gas flow rates compared with a Fassel type torch, is given in a cited reference and the sample introduction system is described in ESI. After optimization of the system for both conical and Fassel torch types, it was found that the conical torch gave a 1.5 to 8 times larger peak intensity, a two to four times greater peak area and higher precision (on average a 1.5 times lower RSD for peak intensity and a 1.8 times lower RSD for peak width) than the Fassel type torch. The authors concluded that these improvements observed using the conical torch were due to a much higher electron density in the plasma, a higher excitation temperature and plasma robustness and an improved particle trajectory which led to more rapid vaporization, atomization and ionization of particles with a lower atom/ion cloud diffusion. The capability of analysing single particles at a rate of at least 2000 particles per second was also demonstrated. It remains to be seen if any of the major instrument manufacturers or component suppliers will offer this conical torch design commercially. A microdrop generator has also been used to produce monodisperse droplets, containing different amounts of a dissolved element, to expand the versatility and the general understanding of sNP-ICP-TOF-MS measurements.275 In this work by Gundlach-Graham and Mehrabi, the temporal durations and elemental sensitivities obtained from the produced microdroplets and NPs were very similar such that microdroplets were used as proxies for NPs by controlling the absolute mass of elements injected into the plasma in each droplet. The paper described how microdroplets can be used to calibrate sNP-ICP-MS, in terms of both NP mass and particle number concentration and shows that, when combined with sNP-ICP-TOF-MS, online microdroplet calibration can be a powerful method for untargeted quantitative single-particle analysis. The use of microdroplets as a tool to study approaches for split event correction and single-particle detection with ICP-TOF-MS was also demonstrated. The authors concluded that microdroplet-based sample introduction, giving total consumption and quantitative transport of analyte in each microdroplet, could reduce sampling bias and thus improve quantification of element-mass distributions in analyte particles or cells. They also pointed out that the low volumetric throughput for microdroplet sample introduction remains a challenge for the detection of anthropogenic NPs at environmentally relevant particle number concentrations (102 to 106) but anticipate improvements in this direction.
A general factorial design, with nebuliser gas flow, plasma RF-power and sampling depth as the independent variables and the intensity of 197Au as the response variable, was used by Kinnunen et al. to improve the sensitivity of sNP-ICP-MS for Au NPs.276 The optimization was performed using both Au NPs and solutions of ionic Au and the instrumental parameter values after optimization were found to be in a good agreement for both analyte types, indicating a similar behaviour of the NPs in the plasma when compared with the dissolved analyte. The results obtained showed that significant interactions occur between the main instrument parameters under study hence consecutive single parameter optimization approaches may not be suitable for optimizations of sNP-ICP-MS instrumental conditions. The authors also point out that for more challenging analytes, such as Fe or Ti NPs, the potential for polyatomic interferences and doubly charged ions also needs to be included in the optimization routine. In sNP-ICP-MS the ion cloud produced from each particle is detected as a transient signal and, due to the size range of NPs that can be detected, this can span many orders of magnitude of the detector response. Non-linear detector responses can therefore cause biased results and this has recently been investigated by Strenge et al. using an in-house constructed data acquisition system fitted to a commercially available ICP-MS instrument.277 The device was used to assess the suitability of applying dead time correction to time resolved signals from gold NPs of both milli- and microsecond durations generated by a microdroplet sample introduction device. The results were compared with those obtained from the analysis of 11 Au NP suspensions, covering a size range of 10–100 nm, introduced via regular solution nebulisation. Applying a dead time correction routine to μs time resolved data increased the maximum number of counts tolerated per particle by four to fifteen-fold which resulted in the linear dynamic range increasing from 10–40 nm to 10–60 nm for Au NPs. It was also found that an accurate cross-calibration between pulse and analogue counting detector modes could extend the linear range up to 100 nm. The authors concluded that the results obtained support the theory of dead time related count losses being the main reason for nonlinear response in pulse counting sNP-ICP-MS. However, applying a dead time correction can lead to slightly distorted measured particle size distributions. In addition, despite the fact that counting electronics with ultra-low dead times could further improve the linear dynamic range, other ways must be explored to extend the top end of the tolerated particle mass range and, thereby, the methods’ applicability and usability. The five papers covered in this paragraph provide a wealth of information and interest for those who wish to explore the fundamentals of sNP-ICP-MS.
A number of reports of attempts to improve the figures of merit achievable using sNP-ICP-MS have been published this year. The use of IR energy to heat the sample introduction system, with the aim of reducing the LOD values obtainable, has previously been described in conjunction with ICP-OES. Williams et al. have now investigated whether this approach can improve the sample transport efficiency for sNP-ICP-MS.278 It was found that the number of detected Au NPs per min increased from 540 at room temperature to 600 at 60 °C and 840 at 80 °C. However, as it was only possible to use IR heating outside of the torch box, condensation could subsequently occur as the aerosol cooled on the way to the torch and no significant change in the LODsize resulted at 60 °C and degradation of this parameter occurred at 80 °C. The authors then investigated mono-segmented flow analysis (MSFA), in which a discrete sample volume is injected into a flow of air, in conjunction with sNP-ICP-MS for the same purpose. It was found that, with IR heating the transport efficiencies rose from 3.7 ± 0.3% to 10 ± 1% at 20 °C and 4.1 ± 0.2% to 12 ± 2% at 60 °C for sNP-ICP-MS and MSFA-sNP-ICP-MS, respectively. In all cases the measured size of the Au NPs, which ranged from 50 ± 13 to 54 ± 6 nm, were in statistical agreement with the certified size of 49.9 ± 2.2 nm. The same research group have also published a paper which described a different approach for calculating NP masses from sNP-ICP-MS data.279 If external calibration with standard solutions is used to find the mass of NPs, the average signal intensity for each standard solution is typically used, which requires measurement of the transport efficiency of solutions through the spray chamber. In this new approach the signal is integrated over a constant time period for all standards and samples and the mass of analyte aspirated can be calculated from the analyte concentration in each standard, the gravimetrically determined sample uptake rate and the integration time used. The line of best fit through the calibration curve of integrated signal versus sample mass can then be used to find the total mass of NPs nebulised during the integration time, and the mass of each NP then corresponding to the fraction of the total integrated signal caused by the NP. Measurement of the transport efficiency is only required if the concentration of NPs is desired. An alternative approach to improving sample transport efficiency is to directly nebulise NP suspension into the plasma, which obviates the need for transport efficiency measurements. These two papers contain a good explanation of the theory of IR heating and the calculations involved for the new approach for estimating transport efficiency and are well worth a read. A demountable direct injection high-efficiency nebulizer introduction system hyphenated to a flow-injection valve and a gas displacement pump was used by Tharaud et al. for this purpose with sample flow rates as low as 8 μL min−1.280 Measurements made using this system were described as accurate and repeatable and the average measured diameters of Ag, Au and Pt NPs were in agreement with the manufacturers’ reference values. The system was also used to analyse for Au NPs in surface waters.
A study by Yamashita et al. reported on a new approach to extend the quantifiable size range of Au NPs in sNP-ICP-MS, which is typically limited by detector saturation.281 To achieve this, the signal intensities of Au related polyatomic ions such as 197Au12C+, 197Au14N+, 197Au16O+ and 197Au40Ar+, instead of the 197Au+ signal, were used for the size calibration of Au NPs. Using the 197Au40Ar+ for example, the signal intensities emanating from the Au NPs were reduced to the 105 level for 400 nm Au NPs. This allowed the 197Au+ signal to be used for the analysis of samples containing Au NPs in the 10 to 100 nm size range and the 197Au40Ar+ signal for the 200 to 400 nm size range. The use of different collision/reaction cell gases to allow the reliable detection of Fe, which suffers from a number of polyatomic interferences, was reported by Rua-Ibarz et al.282 Based on the figures-of-merit obtained for the different approaches evaluated, an on-mass approach, i.e. measuring the Fe isotope directly rather than as an adduct at higher mass, using NH3 as the reaction gas in single quadrupole-ICP-MS, an on-mass approach using H2 as the collision/reaction gas in ICP-MS/MS and pseudo-medium resolution in SF-ICP-MS were found to be the best-suited approaches for fast interference-free monitoring of the ion signals generated by Fe3O4 NPs in single particle mode. The authors stated that the use of H2 as a reaction gas provided accurate and precise results for custom-made 50 nm Fe3O4 NPs whilst the use of NH3 was less successful, due to a lower signal-to-background ratio.
000 to 100000 cps for high purity water with a typical sensitivity of 1000 cps per 1 mg L−1 13C. After careful optimization of the system for both particle transport and atomisation efficiency, a LOD of 0.8 mg L−1 was achieved, illustrating the challenges involved. The authors stated that acidification and inert gas purging of the samples was not considered appropriate. The reasons for this are not given and would be interesting to know but an LOD of 0.1 mg L−1 for dissolved organic carbon, using combustion/IR based instruments, can be readily achieved. It would seem that further work on this aspect could reduce the background to more manageable levels. The transport efficiency for 2.22 μm polystyrene microparticles was 29.9 ± 0.9% which was in fair agreement with the efficiency calculated using 50 nm Au NPs of 28.2 ± 0.4%. This fell to 12.5 ± 0.6% for 4.82 μm polystyrene μPs, showing that transport efficiency is dependent on size for polystyrene microparticles over ca. 3–5 μm. For size measurements of the polystyrene microparticles recoveries ranged from 90 to 95% for particles ranging from 1.98 to 5 μm but were 163% for 1.04 μm particles. Particle number concentrations were accurately measured up to 2 × 109 L−1 for 3 μm particles and a good discussion of the steps needed to achieve accuracy in both particle size and number measurements was given in the paper. The size of the particles detected in the personal care products and teabag extracts ranged between 2 and 3 μm with 107 to 1011 particles per g released from the former and 104 particles per g released from the latter. Although, as the authors stated, the technique has limitations, particularly for environmental samples where particle number concentrations are often much lower than those measured here, it also shows promise if the challenges of high background levels of C can be reduced.
A multi-method approach to characterize metal nanoparticles contained in nanoplastic colloids, which were described as composite particles, was reported by Barber et al.288 The techniques FFF-sNP-ICP-MS, TEM, Asymmetric Flow Field-Flow Fractionation-UV–Vis (AF4) and centrifugal field-flow fractionation (CFFF) were employed. The in-house synthesised target analytes were Au polystyrene-block-polyacrylic acid (Au-PS-b-PAA), Au-PS and Au-citrate NPs. The metal NP size and particle number concentration were obtained using sNP-ICP-MS whereby AF4 and CFFF were used to separate and size the composite particles based on their hydrodynamic diameter and buoyant mass, respectively. Off-line sNP-ICP-MS analysis of fractions obtained by the FFF separations facilitated measurement of the mass and number of Au-PS NPs (from 1 to >8) contained in the composite particles. Differences or similarities in the NP size results obtained by AF4 and sNP-ICP-MS were used to further characterize composite particles, with the use of AF4 also allowing the detection of organic material associated with the inorganic NPs. In CFF, using a carrier that was density-matched to the polymer component of the composite particles allowed the measurement of multiple gold NPs in the composites. The authors concluded that to effectively apply the developed methodology to the environmental behaviour of nanoplastics in model systems, the mass of metal NPs in the composite particles should be kept small compared with the overall mass of the polymer in the nanoplastic.
Various analytical techniques, FFFF, sNP-ICP-MS, dynamic light scattering (DLS) and TEM were used by Maknun et al. for the size analysis of SeNPs coated with either proteins, giving positively charged NPs, or sodium dodecyl sulfate, giving negatively charged NPs.289 A commercially available detergent, FL-70, was used as the symmetrical channel carrier in FFFF with a 10 kDa regenerated cellulose membrane to minimise particle–membrane interactions. The synthesised SDS coated SeNPs ranged in size from 30 to 190 nm whilst the protein coated Se NPs were in the 30 to 75 nm size range. The measured hydrodynamic sizes by both asymmetric and symmetric FFFF were always statistically lower (as tested using a t-test), than those obtained using DLS whilst the core diameters estimated using sNP-ICP-MS were always in agreement with the TEM method used. Due to the slow response time of Faraday detectors, MC-ICP-MS is rarely considered for sNP-ICP-MS. However, if the instrument is fitted with ion counting detectors, which are much faster to respond, then using MC-ICP-MS for the determination of the isotopic ratio of a NP becomes possible and this has been described for Pt NPs.290 In this work the instrument was fitted with three electron multipliers and three Daly detectors, which allowed four isotopic signals to be monitored in any combination, and high-time resolution amplifiers which allowed dwell times as short as 10 μs to be used to monitor the four Pt isotopes, m/z 194, 196, 197 and 198. Data presented for the 195Pt:194Pt isotope ratio, which is naturally close to unity thus minimising the effects of mass bias, was always in agreement with, but lower than, the natural ratio. The uncertainties (2 SD) on the isotope ratios were 35% for 30 nm NPs, 17% for 50 nm NPs and 10% for 70 nm NPs with the precision obtained presumably improving due to the increasing ion signal with increasing NPs size for a given particle number concentration. The measured sizes for 30 and 50 nm Pt NPs were 30 nm ± 3 nm and 47 nm ± 5 nm which was in agreement with the data acquired by TEM. The authors suggested that the effect of dead time correction can account for part of discrepancy in the measured 195Pt:194Pt ratio from the natural value. The dead time correction factor was calculated using a solution of dissolved Pt and it may be that alternatives to this, using NPs, needs investigating. The aim of the work was to develop a method which could be applied to the measurement of isotope ratios in the small metallic grains (<500 nm) found in meteorites and the authors concluded that the approach described shows promise in this direction.
An alkaline extraction procedure was reported by Ojeda et al. for the extraction of TiO2 NPs from photocatalytic products and crab sticks.293 The procedure consisted of the suspension of the sample in TMAH (10% v/v) with UAE at 37 kHz for 2 hours followed by centrifugation and then dilution of the supernatant (1:10) in the AF4 carrier (Novachem, 0.05%) or storage at 4 °C for analysis by sNP-ICP-MS. Particle sizes in the range of 50–90 nm and 160–170 nm were estimated in the different photocatalytic products by AF4-DLS, which were in good agreement with the sizes predicted by calibration versus SiO2 and polystyrene standards. In the crab sticks, particle sizes of between 140 and 350 nm were estimated by AF4-DLS. These results were also compared with those obtained using sNP-ICP-MS. The size value obtained using sNP-ICP-MS was about a quarter of that for AF4-DLS. This difference is explainable by the different parameter measured by the two techniques, mass spherical equivalent diameter vs. hydrodynamic diameter, respectively and the higher LODsize obtainable with AF4-DLS. Silicon dioxide is a permissible food additive and a method for the extraction of SiO2 NPs from coffee creamer and their subsequent characterization using AF4-ICP-MS was reported by Li et al.294 The dominant primary size of the extracted SiO2 NPs was 36.5 nm and the NP fraction, compared with the total SiO2 content, was 18.6% for the dominant primary nano-silica particles by pre-channel calibration and 35.7% for total SiO2 NPs (<100 nm) by post-channel calibration, with recoveries of 90% for the former and 75% for the latter. Possible reagents to assist in the extraction of Ag and CeO2 NPs spiked into river waters were investigated by Loosli et al.295 Of the extractants investigated, which included NaOH, Na2C2O4 and Na4P2O7, the greatest recovery was achieved using Na4P2O7 which the authors attributed to the greater efficiency in the break-up of natural and engineered nanomaterial heteroaggregates by this reagent. The size distributions of the extracted suspensions, assessed using AF4-ICP-MS, showed that the spiked Ag NPs were extracted from the river water as both primary particles and small (<100 nm) aggregates, whilst the spiked CeO2 NPs produced aggregated particles in the size range 0–200 nm.
An analytical methodology based on AF4-ICP-MS was developed by Sanchez-Cachero et al. for monitoring citrate coated PtNPs (5, 30 and 50 nm) in water samples.296 After optimization of the carrier composition, AF4 separation program, the focusing step and cross flow values, the PtNPs were fractionated in less than 30 min with quantitative recoveries of 100 ± 7%, n = 5. The optimized method was then used to study transformations, in terms of size and surface modifications, of PtNPs in synthetic and natural water samples over time and in the presence of organic matter. This was found to be a critical parameter in this regard. In AF4 the dissolved fraction passes with the cross flow through the membrane of the separation channel and is usually not analysed which is a disadvantage when compared with centrifugal field flow fractionation and chromatographic techniques. Thus, an AF4-ICP-MS system which allows the conversion of the discontinuous cross flow into a continuous flow to the ICP-MS for analysis was developed by Tan et al.297 Quantification was performed both for aqueous standard solutions and for environmental water samples using ultrafiltration as a reference method. In addition, a combined method including monitoring of the dissolved fraction in the cross flow and the particulate fractions in the detector flow was also established. Comparison of the dissolved fraction data obtained with the new method and via ultrafiltration gave recoveries in the range of 87% to 120% for Ca, Mg and Si. For Al and P the dissolved fraction, which was in the low μg L−1 range, elevated recoveries were observed, presumably due to contamination from the AF4 system, but their values obtained still agreed with the ultrafiltration values within confidence intervals. This fraction is relevant to determine the ratio of free and particle bound nutrients or toxicologically relevant elemental species, to establish the dissolution rate of non-stabilised (nano)particles and to establish a mass balance for quality control of the fractionation results. Fraction collection of the cross flow followed by off-line elemental analysis is rarely performed to determine the dissolved fraction because of delayed elution through the frit of the channel, dilution and elevated blank levels for some elements. The authors concluded that the system could be used to determine the ratio of free and particle bound nutrients or toxicologically relevant elemental species, for NP dissolution rate studies and for determining a mass balance for quality control purposes.
A number of studies have used X-ray-based techniques to characterize and/or detect NPs in different matrices. The fungicide activity of chitosan–silver nanocomposites (Ag–Chit-NCs) against Penicillium expansum from feed samples was evaluated by Alghuthaymi et al. using XRF, SAXS, XPS and TEM.298 It was shown that Ag–Chit-NCs, in sizes ranging from 4 to 10 nm, were internalised into the P. expansum in cells, formed agglomerates in the cytoplasm, and were bound to cell organelles. The ability of the Ag–Chit-NCs to influence protein and DNA fragmentation in P. expansum was also assessed using sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) which explained the apparent cellular protein response to the presence of the Ag–Chit-NCs. The intensity of P. expansum hyphal cell protein lines treated with Ag–Chit-NCs was very thin. This indicated that high molecular weight proteins were largely prevented from entering the electrophoretic gel, reflecting cellular protein modification and the possible damage caused by the binding of protein fragments to Ag–Chit-NCs. The in vivo percutaneous permeation and dermal safety of cosmetic cream containing Au nanosheets, and Au nanosheets extracted from cosmetic creams were investigated by Cao et al.299 Imaging by SR-XRF showed that the Au nanosheets in cosmetics penetrate mainly through hair follicles in a time-dependent manner. Cosmetic creams, rather than the extracted Au nanosheets applied separately, decreased the cell viability of keratinocytes and slightly induced apoptosis/necrosis of keratinocytes and skin dermal fibroblasts. Hair growth was also inhibited by the cosmetic cream and the extracted Au nanosheets. A soft XRF-based methodology that allowed both a dimensional reconstruction of nanostructures and a composition characterization using GI-XRF was presented by Honicke et al.300 The capability of this technique for this purpose was demonstrated by performing an element-sensitive reconstruction of a lamellar grating made of Si3N4, with GI-XRF data for the O-K alpha and N-K alpha fluorescence emissions allowing a thin oxide layer to be reconstructed on the surface of the grating structure. In addition, the technique was also applied to three dimensional nanostructures to derive both dimensional and quantitative compositional parameters. A semi-analytical approach for the characterization of ordered 3D nanostructures using GI-XRF has been reported by Nikolaev et al.301 A computational scheme based on the dynamical diffraction theory in many-beam approximation, which allows a semi-analytical solution to the Sherman equation to be derived in a linear-algebraic form was used for this purpose as well as recently published GI-XRF data measured on 2D Si3N4 lamellar gratings, and on periodically structured 3D Cr nanopillars. Both the dimensional and structural parameters of these nanostructures were reconstructed by fitting numerical simulations to the experimental GI-XRF data and the results of this showed good agreement with the nominal parameters used in the manufacturing of the structures, as well as with reconstructed parameters based on the previously published finite-element method simulations.
The application of a convolutional neural network to in vivo XRF images of Au NPs obtained using a benchtop instrument to eliminate Compton-scattered photons was reported by Jung et al.302 The XRF imaging system comprised a 2D CdZnTe gamma camera, a pinhole collimator and fan-beam polychromatic X-rays. An architecture of the 2D convolutional neural network model for Compton background elimination was optimally designed and trained with data sets obtained by the measurements of water only and Au NP-embedded imaging phantoms. The XRF images generated by the trained 2D convolutional neural network were compared with those generated by a direct subtraction method. The developed 2D convolutional neural network was also applied to generate in vivo XRF images of Au in living mice exposed to Au NPs. The in vivo XRF images generated by the two methods were then compared in terms of the difference in Au NP concentration and it was shown that the 2D convolutional neural network model could be successfully applied to in vivo XRF images for detecting the concentration and location of Au NPs in living mice. Pulsed laser ablation in liquids (PLAL) is a hierarchical multi-step process to produce pure inorganic nanoparticle colloids. Controlling this process is hampered by the partial understanding of individual steps and structure formation. In situ X-ray methods were employed by Reich et al. to resolve macroscopic dynamics of nanosecond PLAL as well to analyse the distribution and speciation of ablated species with a microsecond time resolution.303 High time resolution was achieved using ‘single shot’ synchrotron-based methods. X-ray multi-contrast imaging by a Shack–Hartmann setup and small angle X-ray scattering (SAXS) resolved evolving NPs inside the transient cavitation bubble, while the total material yield and the chemical state of the ejecta was monitored by dispersive mode XAS. It was observed that nanoparticles are produced directly during ablation and that reactive material was also detected. This was identified as Zn atoms. The NPs within the cavitation bubble exhibited a metal signature for a few milliseconds before gradual oxidation occurs.
A range of other techniques has also been used for investigations involving nanomaterials. One of these is GF-AAS which was employed by Brandt et al. to determine activation energies for ionic Au and Au NPs, by constructing Arrhenius-type plots from absorbance signals.304 Ionic Au atomisation is governed by a two-precursor mechanism. For the first precursor no release order can be determined, but an estimated activation energy of 211 ± 26 kJ mol−1, corresponding to atomisation from small atomic clusters was found. A release order of 0.3 showed that larger in situ formed aggregates are the second precursor. By varying the temperature ramp programme, the ratio of the precursors was altered resulting in different shapes of the absorbance signals. In contrast, the atomisation of Au NPs followed a one precursor atomisation mechanism. A “pseudo-first” atom release order was found for all NP sizes and atomisation behaviour was found to be similar for each heating rate. The activation energy of Au NPs smaller than 20 nm increased with increasing NP size, while for larger sized NPs the activation energy was constant and approached the heat of sublimation of the bulk material. On the basis of the different atomisation mechanisms, discrimination of ionic Au and Au NPs, as well as sizing of nanoparticles seemed to be independent of the used atomisation temperature program, i.e. primarily the heating rate. A second paper by the same research group reported the use of GF-AAS for examining non-spherical Au and Ag NPs, such as rod and prism structures, as well as Fe, Pd and Pt NPs.305 Using the methodology developed for Au and Ag NPs, the Fe NPs could be distinguished from ionic FeIII and rod-shaped Au NPs and Ag nanoplates could be differentiated from their ionic forms. Discrimination between rods and plates was not possible as the rods and plates were converted to spheres during the temperature ramp program prior to atomisation. The determined diameters of spheres formed in situ were in good agreement with the theoretical values and diameters determined by SEM. For Pd and Pt neither a differentiation between ions and NPs nor a size evaluation was possible. A report on single-cell imaging of Au and Ag NPs by near-field desorption ionization mass spectrometry (NDI-MS) was published by Cheng et al.306 A mean crater diameter of 310 nm on the cellular surface and single cell imaging with a 250 nm pixel size were achieved. Some compounds originating from surface ligands or endogenous substances were simultaneously imaged with Au NPs. Additionally, the location of the Au NPs was compared with the images from confocal laser scanning microscopy, which were in good agreement. The NDI-MS data showed a unique advantage of simultaneously imaging NPs and ligands at subcellular lateral resolution and the authors proposed that it can be extended to other NP systems and thus used as a general method for the exploration of the use of engineered NPs in biomedicine.
A CE-ICP-MS/MS method for the determination of superparamagnetic iron oxide nanoparticles (SPIONs) was reported by Kruszewska et al. with the aim of assessing changes in these NPs after incubation with proteins under simulated physiological conditions.307 The SPIONs were added to a 10 mmol L−1 phosphate buffer, pH 7.4, containing 100 mmol L−1 NaCl and 1 mg mL−1 albumin, to give a final concentration of 30 μg mL−1 and the mixture incubated at 37 °C. A polyimide coated fused silica capillary (i.d. 75 μm; o.d. 375 μm; length 70 cm) was used for the CE separations with ammonium bicarbonate (20 mmol L−1, pH 7.4) as the running buffer and an applied voltage of +18 kV. Oxygen was used as the ICP-MS cell gas to allow 32S16O to be monitored as a marker for proteins and V was added to the sheath liquid for use as an internal standard. Under these conditions the repeatability of migration times and peak areas ranged from 0.23–4.98% RSD, spike recoveries were 93 to 97% and the LOD values were 54 ng mL−1 and 101 ng mL−1 for SPIONs with carboxyl and amino terminal groups, respectively. The carboxyl coated SPIONs were found to be stable in the buffer solution for 24 hours whilst some of the amino coated SPIONs transformed into a form with higher electrophoretic mobility which matched the retention time of FeCl2. Monitoring of the 32S16O signal led the authors to conclude that the majority of the carboxyl SPIONS and albumin present in the sample formed a conjugate. A CE-ICP-MS application was developed by Wroblewska et al. for monitoring the formation of cisplatin targeting delivery systems with Au NPs.308 In this work cisplatin was bound to Au NP with 11-mercaptoundecanoic acid as a nanomaterial surface modifier, the electrolyte was 40 mmol L−1 HEPES at pH 7.4, the applied voltage was +15 kV, the sample injection pressure was 30 mbar for 5 s and the capillary length was 70 cm (i.d. 75 μm) with a total run time of 40 minutes. The results showed that an activated form of Au-cisplatin could be produced. Details of this procedure were given in a cited reference. All of the Au NPs were incorporated in the Au-cisplatin produced by this method. Taylor dispersion analysis is a microcapillary flow-based technique whereby a nanolitre-scale sample pulse is injected into the laminar flow of running buffer, which then undergoes axial dispersion. Detection of the equilibrium concentration profile of the dispersed sample pulse allows the molecular diffusion coefficient and hence hydrodynamic radius of solute molecules to be calculated. This technique has been coupled with ICP-MS to allow the determination of the hydrodynamic radius of Gd containing NPs, including sub 5 nm particles, in urine, cerebrospinal fluid, and undiluted serum. Precision of the measurements (RSD) was less than 10%. The authors stated that the specificity of the method provided the opportunity to perform measurements in complex biological media.
The physicochemical properties and applications of C based nanomaterials are dependent on the amount of carboxyl group on their surfaces, and it remains a challenge to determine the amount of carboxyl groups on the surface of these structures. A strategy to achieve this, comprising coupling carbon dioxide vapour generation to a microplasma OES, was developed by Yang et al.309 The carboxyl group on multiwall carbon nanotubes (MWCNTs), graphene or its oxide was converted to carboxylic acid with HCl, the generated carboxylic acid was then purified, reacted with NaHCO3 to generate CO2, which was swept into a miniaturized point discharge OES for the detection of C atomic emission lines. Potassium hydrogen phthalate was used as a calibration standard and the LOD was 0.1 μmol g−1 for the carboxyl group based on a sample mass of 10 mg. Quantitative imaging of amyloid beta (A beta) in brain tissues is important in the development of Alzheimer’s disease treatments. A method using antibody-conjugated Au NPs for quantitative imaging of A beta peptide in the brain of a mouse suffering from Alzheimer’s disease using LA-ICP-MS was presented by Gao et al.310 A beta antibody (anti-A beta) was labelled with Au NPs to form the conjugate Au NP–anti-A beta which was immunoreactive with A beta in the brain slice of mouse. Quantitative imaging of Au was acquired with homogenised brain slice matrix-matched standards as external calibrants. The stoichiometric ratios between metal conjugates and A beta were optimized and the immunoreaction efficiency after labelling was also investigated. According to the molar relationship between Au NPs and anti-A beta (1:4.3) and the ratio of anti-A beta to A beta (1:1), quantitative imaging of A beta in brain was accomplished. The method displayed the location and concentration of A beta aggregation in the brain tissues, and this was consistent with traditional immune-histochemical staining methods. The proposed methodology was stated to have potential for investigating the quantitative imaging of biomarker heterogeneity, and could also be useful for understanding complex brain mechanisms in the future. Analysis for NPs directly in solid biomatrices by LA-sNP-ICP-MS can produce erroneous results if non-optimal operational conditions are used. A report offered insights into the analysis for NPs by LA-SP-ICPMS, based on modelling the chain of events from NP release to LA cell washout to ICP-MS detection as a function of the biomatrix characteristics, instrumental specifications and variable operational settings was published by Metarapi and Van Elteren.311 By processing the computational data via an “outlier” filter to differentiate between NPs and dissolved ions, guidelines have been established to achieve the optimal instrumental setup and operation. This assumes the measurement of homogeneously distributed metal NPs and dissolved metal and uses a laser fluence ≤ 1 J cm−2 to circumvent NP degradation.
Table 1 shows other applications of nanomaterial characterization and/or detection presented in the literature during the time period covered by this review. For those applications where a sample matrix was analysed, the material has been inserted to the table. However, for those applications where nanoparticles were analysed directly, this field has been left blank.
| Analyte | Matrix | Technique | Comments | Reference |
|---|---|---|---|---|
| Au NPs | sNP-ICP-MS | Use of various NP sizes to determine membrane pore size. A mixture of 20, 30, 40 and 60 nm gold particles, at a concentration of approximately 7 × 107 particles per mL for each nominal size or 2, 8, 19, 66 μg L−1 gold, respectively was used to determine the filtration performance over a wide size range | 312 | |
| Au NPs | sNP-ICP-MS | Assessment of NP aggregation and colloidal stability under physiological conditions. The technique was used to quantify aggregate size and aggregation kinetics at the individual aggregate level. Potential applications may include nanoparticle aggregation in environmental samples and the preparation of colloidally stable nanoparticle formulations for bioanalytical assays and nanomedicine | 313 | |
| Au NPs | Cell culture medium | sNP-ICP-MS | Characterization and quantification of AuNPs in only one minute of analysis at concentrations at the ng L−1 range. Sample preparation involved only a dilution step. Precise determinations of particle size and concentration (1% RSD) were reported with 22.4 nm and 2.1 × 105 particles per L as limit of detection for particle size (LODsize) and for particle-number concentration (LODNP), respectively | 314 |
| Ag NPs | Nano-textiles | sNP-ICP-MS | Study of the release behaviour of silver from nanotextiles. Simulated sweat and water were the extracting media. Different extraction behaviour was observed between the two extractants, with the sweat releasing larger nanoparticles. However, the particle size extracted was related to the dilution factor. Hence, a standardized method was required to obtain reliable data | 315 |
| TiO2 NPs | Confectionery products | sNP-ICP-MS, TEM | Inter-laboratory test using seven experienced food analysis laboratories, employing different instruments and software packages for particle size analysis of TiO2 NPs confectionery products. Bias of the data was estimated using TEM | 316 |
| Te NPs | Bacteria | sNP-ICP-MS, single cell ICP-MS | Single cell and single particle analysis evaluating uptake and biotransformation of Te NPs in bacteria S. aureus and E. coli. Both strains converted nanospheres to nanorods. The amount of Te taken up was at the fg range, with E. coli taking up marginally less | 317 |
| ZnO NPs | Food packaging materials | sNP-ICP-MS, TEM | Study of effect of different types of food (orange juice and chicken) on the fate of ZnO NP migration from food packaging materials. Nanoparticle diameter LOD were 26 nm, 95 nm, 108 nm and 129 nm for aqueous solution, chicken breast extract and for oral and intestinal extracts, respectively | 318 |
| Au, Ag, ZnO, and CeO2 NPs | Simulated gastric fluid | sNP-ICP-MS, SAXS, TEM | Study on the fate of NPs in simulated gastric fluid. Size detection limits for Ag-NP, Au-NP, ZnO-NP, and CeO2-NP ranged from 15 to 35 nm. The particle concentration detection limit was 135 particles per mL. The ZnO-NPs dissolved completely and rapidly, whereas Au-NPs and CeO2-NPs showed apparent aggregation and did not dissolve significantly | 319 |
| CeO2 | Natural waters | sNP-ICP-MS | Analysis of natural waters for CeO2 NPs using a sector field instrument with a dwell time of 50 μs. CeO2 NPs in rainwater at a concentration of 2.2 ± 0.1 × 108 L−1 with a mean diameter of 10.8 ± 0.2 nm; and in a river water at a concentration of 1.6 ± 0.3 × 109 L−1 with a higher mean diameter 21.9 ± 0.8 nm | 320 |
| Ag–Au, Au and SiO2 NPs | sNP-ICP-MS | Determination of the porosity of nano- and sub-micron particles. Accuracy and precision were comparable to those of the reference methods, small angle X-ray scattering (SAXS), gas adsorption or TEM imaging | 321 | |
| Pt NPs | sNP-ICP-MS, TEM | Characterization of Pt NPs for fuel cell applications. Results for particles with diameters of 30, 50 and 70 nm were compared with those from TEM. Technique then used for the characterization of 10 nm Pt NP | 322 | |
| Ni NPs | sNP-ICP-MS | Characterization of Ni NPs which included a flow injection approach | 323 | |
| Au NPs and Au–DNA conjugates | sNP-ICP-MS | DNA assay via Au NPs based on a hybridisation-chain-reaction-mediated spherical nucleic acid assembly. The assembly process generated large gold nanoparticle aggregates, and the number of aggregates could be counted by SP-ICPMS, which was closely correlated to the concentration of the target DNA. This simple homogeneous assay could analyse DNA within the range of 5 fM to 10 pM with excellent selectivity | 324 | |
| Se NPs | sNP-ICP-MS, TEM | Size determination of Se NPs. Data obtained using SP-ICP-MS were consistent with those from TEM for the core diameter. Consecutive size analysis by SP-ICP-MS for the fractions collected from flow FFF was proposed for sizing of SeNP mixtures | 289 | |
| Various NPs | Waste water | sNP-ICP-TOF-MS | A holistic study of particle behaviours during wastewater treatment. A new data analysis approach was reported. Waste water treatment plants had average of 90% and 94% removal efficiencies of single-metal and multi-metal NPs, respectively | 325 |
| Fe3O4 NPs | sNP-ICP-MS, TEM | Detection of Fe3O4 NPs in petroleum hydrocarbon materials. Helium used as a collision gas to control spectral interferences from ArO and CaO on Fe at m/z 56. Different cell gas flow rates (3, 3.5, and 4 mL min−1) were tested and 4 mL min−1 was optimal | 326 | |
| Ag NPs | Toothpaste | sNP-ICP-MS | Ultrasonic extraction combined with SP-ICP-MS was developed to detect the size distribution, number concentrations and mass concentrations of silver nanoparticles. After optimization, detection limits of particle size, number concentration and mass concentration were 23 nm, 3.9 × 108 particles per kg and 26 ng kg−1, respectively were obtained | 327 |
| TiO2 NPs | sNP-ICP-MS, TEM | Characterization of 15 batches of E171 food additive. After optimization of the sample preparation with method validation, a significant variation in the particle size and shape distributions, the crystallographic structure (rutile versus anatase), and the physicochemical form (pearlescent pigments versus anatase and rutile), was identified | 328 | |
| Graphene oxide–Au NPs composite | sNP-ICP-MS | Bio-marker assay using graphene oxide–Au NPs with thrombin as the target analyte. AuNPs modified with thrombin aptamers were first non-selectively adsorbed onto the surface of graphene oxide (GO) to form GO/AuNPs composites. In the presence of thrombin, the AuNPs desorbed from the GO/AuNPs composites. The desorbed AuNPs were proportional to the concentration of thrombin | 329 | |
| CeO2 | sNP-ICP-MS | Study of CeO2 number and size distribution from different chemical mechanical planarization processes. Particle number concentrations were measured for the 21 to 559 nm size range at 1 nm size resolution | 330 | |
| Pd NPs | ICP-MS, XRD, HR-TEM, EDX, SEM | Pd NPs synthesised and characterized to produce a risk assessment of particulate pollution on the murine macrophage cell lines | 331 | |
| TiO2 NPs | Personal care products | CE-sNP-ICP-MS | Screening for TiO2 NPs in personal care products. Sensitivity was enhanced by cysteine which acts as a nanoparticles stabiliser. Method applied to lip balm and toothpaste samples | 332 |
| Halloysite-doped ZnO NPs | Sunscreen products | ICP-OES, TEM XRD | Synthesis and characterization of Halloysite-doped ZnO NPs for use in sunscreen | 333 |
| CuO, NiO, TiO2 NPs | Cell culture medium and water | ICP-OES | Dissolution behaviour of metal oxide NPs in water and Dulbecco’s modified Eagle’s medium. Solubility of the metal oxides decreased in the order CuO ≥ NiO > TiO2 in both media, solubility trends displayed by nano-TiO2 were the opposite of those displayed by nano-CuO and nano-NiO | 334 |
| Ag NPs | TOF-SIMS | Proof-of-concept study. Different shapes (spherical, triangular and rod) of silver nanoparticles embedded within a hydrogel matrix of polyacrylamide. Anti-bacterial and mechanical properties studied | 335 | |
| Au nanorods | Solution cathode GD-OES | Evaluation of solution-cathode GD-OES for the analysis of NP suspensions. Metallic nanoparticles with diameters from 5 nm to 150 nm were directly analysed and found to exhibit lower, and size-dependent, elemental sensitivity when compared with dissolved free-ion standard solutions. Correlation of NP boiling point with difference in sensitivity between free-ion and NP solutions supported the conclusion that delayed vaporization of nanoparticles is the source of the morphological matrix effect | 336 | |
| CuxS NPs | SEC-ICP-MS, TEM | Study on the effect of natural organic matter (fulvic acid) on copper sulfide NP growth, stability and dissolution over a period of four weeks. At low Cu and S concentrations, fulvic acid restricted particle growth by up to 25%. The CuxS nanoparticles exhibited great chemical stability against oxidative dissolution and were only dissolved when both fulvic acid and oxygen were present | 337 | |
| Ag and Au ions and NPs | HDC-ICP-MS | Evaluation of HDC-ICP-MS for the speciation of dissolved and NP Ag and Au. Optimization of the mobile phase identified that the addition of 0.05 mM penicillamine allowed the quantitative recovery of ionic gold and gold nanoparticles up to 50 nm, whereas 1 mM penicillamine was necessary for quantitative recovery of ionic silver and silver nanoparticles up to 40 nm. Best-case mass concentration detection limits for gold and silver species were 0.05 and 0.75 μg L−1, respectively | 338 | |
| Au–Ag NPs | ICP-MS, TEM, SAXS | Determination of the morphology and concentration of core–shell Au–Ag NPs. The shell thickness, overall size polydispersity and number particle concentration obtained by the various methods were in good agreement | 339 | |
| Au NPs | TOF-SIMS | Inkjet printing of Au NPs onto porcine gelatin coated Si wafers to create quantitative test materials. Spatially resolved quantification using TOF-SIMS was achieved for 30, 80, 100, and 150 nm particles deposited onto gelatin with loadings ranging from 34 fg up to 67000 fg per spot |
340 | |
| Au NPs | TOF-SIMS | Study on the detection of Au cysteine thiolate complexes on Au NPs. The presence of NaCl or a 2-(N-morpholino)ethanesulfonic acid buffer disabled the detection of Au NPs on the Al foil. In their absence, TOF-SIMS was demonstrated to be a good tool for the study | 341 | |
| Al NPs | Copper thin film | TOF-SIMS, scanning TEM EDX | Elemental characterization of Al NPs buried under a copper thin film. The capability of TOF-SIMS to spatially resolve individual tens of nanometer large nanoparticles under ultrahigh vacuum as well as high vacuum conditions was demonstrated | 342 |
| Bronze NPs | LIBS | Optical trapping LIBS characterization of NPs generated by laser ablation of bulk targets in air. The main advantages of the technique were the circumvention of possible material losses owing to transference into the inspection instrument while providing the high absolute sensitivity of single-particle LIBS analysis | 343 | |
| Au NPs | TOF-SIMS | Numerical evaluation of polyethylene glycol ligand conjugation to Au NPs. Statistical correlation values calculated from the signals of PEG and Au measured using TOF-SIMS imaging on the sample spots made by a micro-liquid inkjet printing system had better reproducibility and improved correlation values compared with the pipette spotting | 344 | |
| Pt NPs | sNP-ICP-MS | Study on the effects of NP size and natural organic matter composition on aggregation of polyvinylpyrrolidone coated Pt NPs. The effects of nanoparticle size (20–95 nm) and different sources/compositions of natural organic matter were tested; with the latter being the key factor | 345 | |
| Pt–Ti NPs | GI-XRF, NEXAFS | Study on the interaction of nanoparticle properties and X-ray-based analytical techniques. NEXAFS measurements were performed to investigate the binding state of titanium in the core–shell nanoparticles. This was amorphous TiO2. The nanoparticles were characterized using reference-free GIXRF employing radiometrically calibrated instrumentation | 346 | |
| Ag, Au, Cd, Eu NPs | LA-ICP-MS, ICP-MS | A study of different metal NPs for immune-analysis. Commercial kits of gold nanoparticles were the most suitable for the preparation of conjugates model monoclonal antibody (DO-1, recognizing p53 protein). The LA-ICP-MS detection gave a long linear range (0.1–14 ng) and had a LOD of 1.3 pg of P53 protein | 347 | |
| CdTe quantum dots | CV-AFS, ICP-MS | Study on the use of nanomaterials for use as fluorescence label-free multimode bioassays | 348 | |
| Ag2S, ZnS NPs | Environmental waters | LC-ICP-MS | Speciation of Ag2S and ZnS NPs in environmental waters by cloud point extraction. Precision was <4.9% RSD. Detection limits were 8 ng L−1 for Ag2S-NPs and 15 ng L−1 for ZnS-NPs. Recoveries of 81.3–96.6% for Ag2S-NPs and 83.9–93.5% for ZnS-NPs were achieved when spiked into three environmental water samples | 349 |
5 Cultural heritage
The analysis of cultural heritage samples continues to be a very popular topic. As with most years, the majority of the papers do not discuss the analytical experiments in any great detail, concentrating instead on the archaeological or historical aspects. These papers are therefore not discussed in this section. Instead, those papers that do go into greater detail will be discussed and others, that have some novelty, e.g. the use of chemometrics to discuss the provenance, manufacturing process, etc., or those that show a good level of method validation, will be discussed in brief in tables. It is usual to want to preserve most of these samples. Therefore, techniques such as LIBS, LA and XRF that are either non-destructive or minimally destructive, proliferate.5.1 Metallic artefacts of cultural heritage
A lengthy review containing 160 references of the use of non-traditional heavy stable isotopes (e.g. Ag, Cu, Fe, Hg, Sb and Sn) in archaeological research was made by Stephens et al.350 Although the stable isotopes of these analytes do not have any radioactive parents, their ratios may undergo limited fractionation from various causes. The review was split into numerous sections, including sections for each of the analytes. The paper reviewed the recent work (since 2010) for archaeological purposes and investigated whether or not they may be used for determining the geological provenance of materials such as bronze and glass. The conclusion from the literature was that, in general, they could not. The Cu and, to a lesser extent thus far, Ag was useful for determining the type of ore (supergene or hypogene) the material had been made from though. Applications as well as future developments were discussed.Dechlorination is a method of stabilizing archaeological iron artefacts. Reguer et al. investigated the use of XANES as a technique for the analysis of Cl K-edge during the dechlorination treatment of wrought iron bars raised from Roman shipwrecks.351 The paper presented new results but also gave a summary of the work undertaken using X-ray absorption spectroscopy over the last 10 years. The historical work has aimed to decipher the corrosion patterns of Cl ions in iron artefacts and also to understand the dechlorination process. Samples complete with corrosion products were subject to dechlorination in either aerated solution (five samples) or de-aerated solution (two samples). In both cases, the solution was 0.5 M NaOH. The dechlorination process was stopped for some samples so that the dechlorination front could be located. Samples were dried at 50 °C for 10 min using a heat gun prior to analysis. As well as XANES, XRF and EXAFS were also performed. The XANES enabled the chemical form of the Cl, its distribution and its role in each phase’s stability to be determined. The Cl ions maybe trapped inside the structure of the iron oxyhydroxide as akageneite (beta FeO1−x(OH)(1+x)Clx) and beta or gamma-ferrous hydroxychlorides, e.g. beta-Fe2(OH)3Cl. Some may also be adsorbed to the surface of the grains of the various phases present. It was concluded that XANES was very suitable for the task.
Other applications of the analysis of metal artefacts of cultural heritage are summarised in Table 2.
| Analyte | Matrix | Technique | Comments | Reference |
|---|---|---|---|---|
| C | Archaeological ferrous materials | LIBS | A large laser spot enabled the bulk C concentration. However, it was not capable of analysing different steel phases. Therefore, micro-LIBS was developed that enabled C to be determined in the different phases and even for non-equilibrium structures | 352 |
| Pb and Sn | Copper alloys from the bronze age/iron age transition | Hand-held LIBS, p-XRF | Hand-held LIBS was compared with pXRF for the analysis. Hand-held LIBS was affordable, rapid and micro-destructive and was thought to have the potential for significant usage. However, results from pXRF were deemed to be better, despite its limitations in penetrating corrosion | 353 |
| Pb isotopes | Twelve bronze coins and 10 lead objects from 2nd to 1st century BCE | TIMS | Lead isotopes used to elucidate provenance of materials, All of the coins have Pb isotopes indicating that it originated from the mining district of Cartagena-Mazarron. The lead objects were more diverse and were split into three clusters. All of the objects from the bath complex also originated from the same place. However, workshop and household objects were unresolved; potentially because of re-smelting | 354 |
| Various | Armour from 13th to late 16th century from Italy, Germany and France | LA-ICP-MS | Microscopic silicon slag inclusions in the armour were analysed. Two LA systems used: one operating at 193 nm and the other at 257 nm. Analytical data interrogated using PCA and Linear Discriminant Analysis (LDA). Different provenance groups identified | 355 |
| Various (Ag, Au, Bi, Cu, Fe, Pb and Zn) | Incuse coins (31) from four cities from Southern Italy from 550 BCE to 470/440 BCE | XRF | Both PCA and HCA used to interrogate XRF data. All coins contain Ag (>95%) and Cu. All coins had a similar trace metal signature, indicating the ore originated from the same mine. Some coins appeared to have been overstruck | 356 |
| Various and Pb isotopes analysis | Warring states period bronze bracelets from Huili county, China | pXRF, MC-ICP-MS | Small spot on bracelets polished for pXRF analysis and then the pieces removed during polishing acid digested and analysed using ICP-MS. All bracelets contain approx. 85% Cu and 15% Sn. The Pb data indicate two main groups: one that used local copper without adding Pb and the other added Pb. The Pb isotope ratios indicate that this may originate from the middle reaches of the Yangtse region | 357 |
| Various | Early iron age ferrous remains from Arabia | XRF, ICP-MS, LA-ICP-MS | Slag inclusions analysed using LA-ICP-MS. Rocks and ores from different origins fused using lithium metaborate and then analysed using XRF for the major elements and ICP-MS for the minor/trace ones. Analytical data treated using PCA and HCA using Ward’s method to enable provenance determination. Long distance trade with Iran, UAE and Oman potentially identified | 358 |
| Various (6) | Seal dies made from copper alloy | EDXRF | Samples (95) from 13th to 20th century underwent EDXRF analysis for Ag, Cu, Pb, Sb, Sn and Zn. Data interrogated using PCA. The elemental composition was correlated with the historical data and the location of manufacture | 359 |
| Various (8) | Mixed iron–copper chunks and iron objects from Jordan | LA-ICP-MS, TIMS | Two stage analysis: LA-ICP-MS to detect Au, Ir, Os, Pd, Pt, Re, Rh and Ru and then TIMS analysis for Os isotopes. No connection between high analyte concentrations and Os isotopes. The conclusion was that the objects were not locally produced. In addition, it provided evidence that an exchange network existed in the iron age Eastern Mediterranean | 360 |
| Various | Early medieval silver jewellery | LA-ICP-MS, SEM-EDX | Metallic components determined using SEM-EDX. Then, to obtain provenance, LA-ICP-MS used to determine Pb isotope ratios. Analytical data treated using LDA. Results indicated that all artefacts were made using re-melted metal from multiple sources. The most likely origin of the Ag was from Uzbekistan, Afghanistan or Germany | 361 |
| Various | Ancient silver coins | μ-XRF, LA-ICP-MS | μ-XRF used to detect Ag and Cu enrichment/depletion on coin surfaces. Achieved using Ag Kα/Ag Lα, Cu Kα/Ag Kα, and Cu Lα/Ag Lα for the coins and comparing with Ag–Cu standards of a similar composition. Those coins demonstrating Ag enrichment were analysed using LA-ICP-MS to measure trace elements in the core compared with the surface. Surfaces found to be enriched in Au, but depleted in As, Co, Ni and Pt | 362 |
| Various | Silver coins | LA-ICP-MS | Chronological discrimination of 266 coins based on inter-element ratios. NIST 610–617 series glasses used for calibration. Elemental composition and visual inspection helped identify many coins. Those that remained unknown had their data inserted to LDA. This enabled the sovereign and hence approximate age of the coin to be identified | 363 |
| Various | Silver coins from colonial Brazil | p-XRF, μ-XRF | Samples (17) of 960 Reis coins were analysed. Analytical data input to robust PCA so that they could be classified according to their elemental composition. The surface enrichment of Ag is a known phenomenon and this could potentially interfere with the XRF measurements which only measure the surface few microns. Initial experiments were conducted to ensure that this was not the case | 364 |
5.2 Cultural heritage samples of organic origin
Several reviews or overviews of the analysis or restoration of artworks have been produced during this review period. Included in this number is one by Borg et al. who reviewed (with 131 references) the application of state of the art technologies to support artwork conservation.365 Some of the technologies discussed were: XRF, holographic interferometry, Raman spectroscopy, thermal quasi-reflectography, FT-IR and terahertz imaging. The techniques allow analysis of the paintings, statues and murals, providing information about the pigments, restoration history and media constituents. The review was conveniently split into sections covering each of the techniques, so any reader could simply read the section applicable to them. Another paper, by Dudeja, containing 18 references, gave an overview of laser technology for analysis, cleaning and restoration of artworks.366 Although by no means a comprehensive review, it did highlight the use of techniques such as LIBS, micro-Raman spectroscopy, optical coherence tomography, LIF, etc. and would, perhaps, be a useful read for anybody interested in the area.Macro-XRF is a very useful technique for the analysis of artworks and other historical artefacts and is employed almost routinely by many workers. This is because it can give valuable information on pigments but also, because of the penetrative nature of X-rays, it can give information on layers beneath the surface. It can therefore potentially identify pictures that have been painted over. The development and application of a highly mobile and versatile macro-XRF scanner for the in situ analysis of painted works of art was reported by Pouyet et al.367 The system enabled the real-time elemental mapping of large painted surfaces. This was facilitated by the system having an XRF head that was capable of moving 84 × 84 × 20 cm in the X, Y and Z directions, respectively. The technical details of the scanner were described in detail in the paper. Briefly, the X-ray beam size was adaptable and the detection system comprised a 50 mm2 active area coupled to a CUBE preamplifier and DANTE digital pulse processor. The system was controlled using a custom software including a graphical user interface that had been programmed in Python. Detection limits for K-lines of elements with an atomic number of <50 were typically at the 10s of mg kg−1, but this increased rapidly at higher z. For these higher z elements, L-lines were required. The authors applied it to the successful analysis of paintings.
Another paper to have utilised macro-XRF was presented by Watteeuw et al. who combined it with multi-light reflectance imaging to study medieval illuminated manuscripts.368 The combination of the complementary techniques enabled the analysis of the pigments to be made in an advantageous way compared with normal analysis. Comparing and combining the macro-XRF data and the reference datasets for the multi-light reflectance imaging, enabled the ink and paint layer compositions to be characterised. The genuine 13th century materials and the areas that had been retouched in the 19th century could clearly be distinguished. This is just one example of the numerous papers that have employed complementary techniques for the analysis of materials in an attempt to glean as much information as possible. This use of complementary techniques is probably used most in the analysis of artworks.
A third paper to report the use of macro-XRF for the analysis of pigments was presented by Orsilli et al.369 These authors combined macro-XRF with Statistically Tailored Elemental Angle Mapper for the analysis of a painting by Giotto. The paper is partially an overview (with 54 references) that discussed the relative merits and limitations of using macro-XRF. The data handling aspect is based on the spectral angle mapper algorithm and on global spectrum analysis. It first maps the elemental distribution and then, in synergy with XRF data at set points, it exploits the information gathered from the mapped area(s) as a database to extend the comprehension to data outside the scanned region(s). The methodology was explained fully in the text of the paper.
A paper by Lazic et al. discussed the pitfalls associated with LIBS depth-profiling of painted layers.370 Simply taking into account the line intensities or applying a single normalization approach based on the analyte line and a matrix reference line, e.g. Ca, led to misleading results. This was attributed to the crater formation and different overlaid materials affecting the plasma dynamics and hence, sensitivity. The authors set about picking Cu and Ca atom lines that were less affected by variations of the plasma temperature and that had a similar energy gap between ground and excited state. Each analyte therefore had its own reference Ca line. A total of 20 analytes were determined per laser pulse. In this way it was possible to detect thin over-painted layers of similar colour and to recognise the painting technique used. In some instances, it was even possible to establish geographical origin of the materials used.
Other applications of the analysis of organic samples of cultural heritage are described in Table 3, below.
| Analytes | Matrix | Technique | Comments | Reference |
|---|---|---|---|---|
| Various | Wall paintings in Rome | XRF, time-gated LIF, reflectance spectroscopy, Raman, FTIR | A multi-technique approach where the experimental data was interrogated using PCA (for LIF and spectral angle mapper) and partial least squares discriminant analysis (PLSDA) for reflectance spectrometry. Raman and XRF used to characterise the pigments. It was also possible to identify areas that had been touched up using acrylic-based compounds | 371 |
| Various | Ancient murals | LIBS | Green painted layers of different pigment sizes were analysed using LIBS. Preliminary experiments optimising the LIBS parameters and minimizing signal fluctuations were undertaken. Then, LIBS analysis on different pigment size was performed on simulated samples. When real murals were analysed, the data were input to PCA so that different pigment sizes were classified | 372 |
| Various | Pictorial layers of a harpsichord | XRF, LIBS, FTIR | Numerous restorations identified. Vermillion was used for the red areas, lead white, copper-based pigments for the blue and violets (as well as Prussian blue), chromium oxides for the green tones and iron oxides for the red, yellow and ochre colours. Examination of samples cross-section indicated the presence of up to 10 layers | 373 |
| Various | Leonardo da Vinci drawings | XRF, macro-XRF, Raman | The combination of techniques allowed an improved description of the drawing technique to be made. In addition, it identified an unusual copper-based metalpoint and a peculiar bone preparation | 374 |
| Various | Historical paintings | Macro-XRF | Curvature of a painting surface can affect the measurement process of a macro-XRF scanner, hence changing sensitivity of the analytes. This work used NIST 610 at different distances from the instrument to set up an algorithm that could correct this. Once derived, it was applied to the analysis of some Rembrandt paintings | 375 |
| Various | Historical inks | LA-ICP-MS | Iron gall inks can damage the paper they are written on. This study used standards of historic iron gall inks and then used test papers containing bathophenanthroline to extract the metals. The Fe could be detected colorimetrically (a red colour forms). Other elements detected using LA-ICP-MS. This enabled an approximation of the trace element concentration on historical documents | 376 |
| Various | Cross-sections of paintings | TOF-SIMS | Dual beam TOF-SIMS using both a bismuth cluster liquid metal ion gun and an argon gas cluster beam was used to determine whether the binding medium of each layer of the painting was oil, egg yolk (oil and protein) or glue (proteinaceous). Unexpectedly, it turned out to be oil-based. Pigments also analysed and several metal-based components identified. Sub-μm resolution obtained | 377 |
| Various | Surface and interface treatments for wooden artefacts | XRF, FTIR, Raman, optical coherence tomography | Multi-analytical, non-invasive approach adopted to study the potential and limitations for analysis of surface treatments. The abilities and limitations for each technique were nicely summarised in tabular form. Mock-up pieces of wood were treated and then exposed to different lengths of artificial aging before analysis. Analytical data from FTIR and XRF interrogated using PCA. Valuable insights into aging and light effects obtained | 378 |
| Various | Sixteenth century illuminated printed book | pXRF | Portable XRF combined with Monte Carlo simulations for depth assessment of painted objects. Spectra obtained from painted and non-painted areas. Data compared with simulations. Results provided quantitative compositional and stratigraphic data. Although limitations exist, these preliminary results encourage future work | 379 |
| Various | Ceiling paintings | pXRF, XRD, FTIR, optical coherence tomography | Multi-technique, non-invasive approach that was performed in situ with portable instruments. Pigments identified using XRF and XRD included hydrocerussite, vermillion, smalt, copper-based blue or greens; lead-tin yellow and iron oxide. The FTIR and optical coherence tomography enabled the condition of the varnish layers to be established | 380 |
5.3 Ceramic objects of cultural heritage
The analysis of ceramic objects is always a popular area of research in cultural heritage studies. A paper by Niedzielski et al. evaluated the effects of sample processing on the XRF results from archaeological pottery.381 A total of 41 samples of pottery from the late bronze age/early iron age were used in the study. Of these, some were hand made and others had been wheel-thrown pottery. All had been collected from western Andalusia, Spain. The results from intact samples had significantly higher levels of Al, Ca, Co, Fe, K, Nb, S, Sb, Si, Sn and Sr compared with pulverised ones. Cluster analysis and PCA were used with the datasets. Cluster analysis failed to identify any clusters. However, PCA identified two outlying samples in the intact (i.e. non-destructively prepared) samples; indicating that they had a different chemical composition. Analysis of the same samples that had been pulverised did not make that distinction. The clear conclusion was that grinding the sample loses vital information (as well as destroying part of the sample).The use of LIBS has become popular in cultural heritage studies because it is quick, cheap and quasi-non-destructive. It does, however, suffer from problems associated with calibration and the production of huge amounts of information that can require powerful computing capacity. Some researchers therefore use statistical techniques to help remove redundant or insignificant data, thereby improving the efficiency of data interpretation. Such a paper has been produced by Ruan et al. who developed a novel hybrid filter/wrapper method for feature selection for the LIBS analysis of archaeological ceramics.382 The method was based on the mutual information algorithm-based filter method (to remove redundant or insignificant data) and the bi-directional selection algorithm-based wrapper method (to obtain the optimal feature subset with good productive performance). The method was implemented through a random forest-based non-linear multivariate classification method. A total of 35 archaeological ceramics were analysed in the study. Prior to the use of the method, the LIBS spectra were treated using wavelet transform de-noising. The overall method was compared with other methodologies and was found to have superior performance in terms of classification ability.
A study presented by Odelli et al. used portable XRF followed by data analysis using an assortment of chemometric methods to classify the provenance of 73 transport amphorae.383 These authors used PCA, cluster analysis and a method called graph clustering for the classification. The analysis took place on a fresh cut surface. Classification using the new method was by no means universally successful. A success rate of approximately 75% was achieved. However, the authors did note that the majority of failures were for samples that had coarse grained clay pastes, implying that the method is especially useful for fine-grained materials. Despite this obvious flaw, the method was rapid, easy and relatively non-destructive.
Vannoorenberghe et al. used LA-ICP-MS for the analysis of the clay fraction of archaeological pottery samples with a view to elucidate provenance.384 Assorted different calibration strategies were tested on the reference material NIST 679 brick clay before being applied to the archaeological samples. These strategies were: external calibration with no internal standard, external calibration with Si as an internal standard, sum normalization approach with estimation of the Si concentration as internal standard and another sum normalization method. The calibration materials tested were a matrix matched clay and five glass reference materials. When no internal standard was used, the glass materials provided incorrect data. The matrix matched material did, however, produce reliable data. When the normalizations or the internal standard were used, either the glass materials or the matrix matched sample could be used for calibration. Once the method had been developed, it was applied to a series of Mesolithic and early Neolithic ceramic materials from the Scheldt valley. Data for 44 analytes from these analyses were treated using linear discriminant analysis (LDA). This was capable of classifying the different sediments used.
Other applications of the analysis of ceramic cultural heritage objects are shown in Table 4.
| Analyte | Matrix | Technique | Comments | Reference |
|---|---|---|---|---|
| Pb isotopes | Roman lead glazed archaeological artefacts | SIMS | 206Pb/204Pb, 207Pb/204Pb, 207Pb/206Pb, 208Pb/204Pb, 208Pb/206Pb were measured on the glazed coatings of each sample. No mass bias correction was necessary. Treatment of data using PCA and comparing with the OXALID database indicated that since the 2nd century, the Pb in the glaze most probably originated in Britain. The technique of SIMS has several advantages over other techniques such as easy sample preparation, quick acquisition times, minimal destruction of samples and the possibility to choose the location of each analysis point | 385 |
| Various | Thirty four archaeological ceramics from 17th century shipwreck | ICP-MS, XRD, SEM | Sample pulverized and fused using lithium borate and then the melt was acid digested using 1N nitric acid containing a few drops of hydrofluoric acid prior to analysis, reference materials JB-3, JG-2, JG-1a and JA-2 used for method validation. The ICP-MS data were treated using PCA. Samples identified as coming mainly from Seville but also from northern Portugal | 386 |
| Various (10 major oxides and 12 minor analytes) | Ancient pottery from Hachijojima Island, Japan | WDXRF | Samples cleaned, pulverized, fused into a glass bead and analysed. Analytical data obtained input to PCA and cluster analysis (using Ward’s method). The two chemometric techniques produced similar results. Of 47 samples analysed, 36 originated in Honshu. It was concluded that cultural exchange was happening 14000 years ago between the island and Honshu, 190 km away |
387 |
| Various | Byzantine pottery from Romania | PIXE | Shards (45) of pottery from Romania were milled and pressed into a pellet. Samples analysed using PIXE with Basalt BHVO-2 as a CRM. Data input to PCA. Two groups identified. One made from kaolinitic clays (high Al) and the other from alluvial sediments | 388 |
| Various (37) | Roman amphorae fragments | pXRF, ICP-MS, FTIR, Voltammetry | Sample (1 g) was pulverized into a powder for XRF determination of Al, Ca, Fe, K, Si and Ti. For ICP-MS analysis, a total acid dissolution employing hydrofluoric acid was used. Validation for both methods was achieved using GBW07408 soil CRM. Analytical data input to PCA enabled classification of amphorae types | 389 |
| Various | 20 ceramic artefacts from eastern Siberia | TXRF, GIXRF, ICP-MS, WDXRF | Several sample preparation methods compared for TXRF analysis: suspension after dry grinding, suspension after wet grinding, acid leaching and acid decomposition. All preparation methods ended by injecting an aliquot onto a quartz carrier and air drying prior to analysis. The TXRF data were validated through analysis of GeoPT33 ball clay CRM. Both suspension methods and acid decomposition provided good data. Data from TXRF compared with those from other techniques. Data from TXRF analysis input to PCA for provenance classification | 390 |
| Various (15) | 142 ceramic artefacts and local clays from Greece | pXRF | Non-destructive analysis of ceramic artefacts and eight local clay sources and their mixtures using pXRF. Clays made into briquettes and fired between 750 and 1050 °C prior to analysis. Results input to hierarchical cluster analysis, biplots, boxplots, PCA and Euclidean and Mahalanobis distances. Results indicated the ceramics were made from local clays and their mixtures | 391 |
| Various | Chinese blue and white pottery sherds | WDXRF, pXRF | Comparison of two techniques for the analysis of 12 sherds of pottery. Chemical composition of the body, blue pigment and glaze determined. Provenance classification possible using both major (Al, Ca, K, Mg and Si) and minor (Rb, Sr and Zr) analytes | 392 |
| Various (42) | On-shore and offshore Roman ceramics from Croatia | ICP-MS, FTIR, SEM-EDS | Samples (20) were fused and then diluted with nitric acid prior to analysis. Of the 20 samples, five came from the excavation site at Crikvenica and 15 came from the coastline. Those from the sea also underwent a desalination process prior to analysis. Analytical data input to PCA and box and whisker plots. Although the provenance of all the samples was known to be the same, the samples recovered from the sea formed a very different group using PCA than those from the land | 393 |
| Various | Black glazed wares from two archaeological excavations in Sicily | μ-XRF, μ-XANES | Possible elemental variations in the glaze and materials evaluated. Speciation of Fe investigated using XANES. Cross-sectional analysis indicated a range of compositions of the black glaze, including some containing Zn. Samples from the attic had glaze that is mainly composed of magnetite (Fe3O4) and hercynite (FeAl2O4). Colonial samples had mainly magnetite with minor amounts of hematite (Fe2O3) | 394 |
5.4 Glass samples of cultural heritage
This has been a relatively quiet area of research in this review period with only two papers of note to be discussed. One, by Comite et al., used LA-ICP-MS, SEM-EDS, FTIR and XPS to identify the degradation products that show up as iridescent black patinas or incrustations on Byzantine glasses from Northern Tunisia.395 All of the analytical techniques were either non-destructive or, at worst, minimally destructive. The LA-ICP-MS was used to identify the chemical composition of the glass and the degradation products whereas the XPS was used to identify the elements present at the surface and their chemical state. The iridescent areas were found to be depleted in alkali and alkaline earth elements. The incrustations were a result of deposits of calcium carbonate or oxyhydroxides.A paper that critically compared the performance of four non-invasive techniques for the differentiation of glass types in historical leaded windows was presented by Cagno et al.396 The techniques were: macro-XRF, UV–Vis-IR, Raman and IR thermography. All of the techniques could be used in situ but the macro-XRF was least portable. The non-quantitative nature of macro-XRF and its inability to measure low z elements such as Mg and Na meant that it was not ideal. However, it was capable of distinguishing two glass types based on their K content. It was also the technique that provided the most chemical information. The UV–Vis-IR broadly matched the macro-XRF findings. The Raman spectroscopy has had a good previous track record of historical glass analysis. However, in this case where colourless glass was analysed, there was very little fluorescence detected.
6 Abbreviations
| 2D | two dimensional |
| 3D | three dimensional |
| AAS | atomic absorption spectrometry |
| ABS | acrylonitrile butadiene |
| AES | Auger electron spectrometry |
| AFFF | asymmetric field flow fractionation |
| AF4 | asymmetric flow-field flow fractionation |
| AFS | atomic fluorescence spectrometry |
| AFM | atomic force microscopy |
| AMS | accelerator mass spectrometry |
| ANOVA | analysis of variants |
| ARECV | average relative error of cross-validation |
| ASTM | American Society for Testing of Materials |
| ATR | attenuated total reflection |
| BCR | Community Bureau of Reference |
| CCD | charge coupled device |
| CE | capillary electrophoresis |
| CFFF | centrifugal field flow fractionation |
| CIGS | copper indium gallium selenide |
| CRM | certified reference material |
| CS | continuum source |
| CT | computerised tomography |
| CV | cold vapour |
| DA | discriminant analysis |
| DLS | dynamic light scattering |
| DLTV | diode laser thermal vaporisation |
| DP-RLIBS | double pulse resonance laser induced breakdown spectrometry |
| DMSO | dimethylsulfoxide |
| DRC | dynamic reaction cell |
| EAST | experimental advanced super conducting tokamak |
| EDS | energy dispersive spectrometry |
| EDXRD | energy dispersive X-ray diffraction |
| EDXRF | energy dispersive X-ray fluorescence |
| ELM | extreme learning machine |
| EPMA | electron probe microanalysis |
| ESI-MS | electrospray ionisation mass spectrometry |
| ETAAS | electrothermal atomic absorption spectrometry |
| ETV | electrothermal vaporisation |
| EXAFS | extended X-ray absorption fine structure |
| FAAS | flame atomic absorption spectrometry |
| FFFF | flow field flow fractionation |
| FI | flow injection |
| FIB | focused ion beam |
| FI-CVG | flow injection chemical vapour generation |
| FTIR | fourier transform infrared |
| FWHM | full width at half maximum |
| GC | gas chromatography |
| GD-MS | glow discharge mass spectrometry |
| GD-OES | glow discharge optical emission spectrometry |
| GF-AAS | graphite furnace atomic absorption spectrometry |
| GI-SAXS | grazing incidence small angle X-ray scattering |
| GIXRD | grazing incidence X-ray diffraction |
| GIXRF | grazing incidence X-ray fluorescence |
| HDC | hydrodynamic chromatography |
| HG | hydride generation |
| HPLC | high performance liquid chromatography |
| HR-CSAAS | high resolution continuum source atomic absorption spectrometry |
| IAEA | International Atomic Energy Agency |
| IBA | ion beam analysis |
| ICA | independent component analysis |
| ICP | inductively coupled plasma |
| ICP-MS | inductively coupled plasma mass spectrometry |
| ICP-OES | inductively coupled plasma optical emission spectrometry |
| ICP-QMS | inductively coupled plasma quadrupole mass spectrometry |
| ICP-TOFMS | inductively coupled plasma time-of-flight mass spectrometry |
| ID | isotope dilution |
| IL-DLLME | ionic liquid-dispersive liquid–liquid microextraction |
| IP | Institute of Petroleum |
| IRMS | isotope ratio mass spectrometry |
| ISO | International Organisation for Standardisation |
| ITER | international thermonuclear experimental reactor |
| JET | Joint European Torus |
| K-SVM-RFE | k-fold support vector machine recursive feature elimination |
| LA | laser ablation |
| LASIL | laser ablation of sample in liquid |
| LC | liquid chromatography |
| LDA | linear discriminant analysis |
| LIAS | laser induced ablation spectrometry |
| LIBS | laser induced breakdown spectrometry |
| LIBS-LAMS | laser induced breakdown spectrometry-laser ablation mass spectrometry |
| LIF | laser induced fluorescence |
| LIPS | laser induced plasma spectroscopy |
| LOD | limit of detection |
| LOQ | limit of quantification |
| MALDI-TOF | matrix-assisted laser desorption ionisation time-of-flight |
| MALS | multi-angle light scattering |
| MC | multicollector |
| MIBK | methylisobutylketone |
| MIP | microwave induced plasma |
| MIP-AES | microwave plasma atomic emission spectrometry |
| MS | mass spectrometry |
| MSFA | mono-segmented flow analysis |
| MWCNT | multi-wall carbon nanotubes |
| NAA | neutron activation analysis |
| NAAR | neutron activation autoradiography |
| ND | neutron diffraction |
| NDI-MS | near-field desorption ionization mass spectrometry |
| Nd:YAG | neodymium doped-yttrium aluminium garnet |
| NEXAFS | near edge X-ray fine structure |
| NIST | National Institute of Standards and Technology |
| NMR | nuclear magnetic resonance |
| NRA | nuclear reaction analysis |
| OES | optical emission spectrometry |
| PARAFAC | parallel factor analysis |
| PBS | phosphate buffered saline |
| PCA | principal component analysis |
| PCR | principal component regression |
| PET | polyethylene terephthalate |
| PFC | plasma facing components |
| PGAA | prompt gamma neutron activation analysis |
| PGM | platinum group metals |
| PIGE | particle induced gamma ray emission |
| PIXE | particle-induced X-ray emission |
| PLAL | pulsed laser ablation in liquids |
| PLS | partial least squares |
| PLS-DA | partial least squares discriminant analysis |
| PLSR | partial least squares regression |
| ppb | parts per billion |
| ppm | parts per million |
| RAFM | reduced activation ferritic/martensitic |
| RBS | Rutherford backscattering spectrometry |
| RDA | regularised discriminant analysis |
| REE | rare earth elements |
| rf | radiofrequency |
| RIMS | resonance ionisation mass spectrometry |
| RMSECV | root mean square error of cross validation |
| RMSEP | root mean squared error of prediction |
| RoHS | restriction of hazardous substances |
| RSD | relative standard deviation |
| RSF | relative sensitivity factor |
| SDS–PAGE | sodium dodecyl sulfate–polyacrylamide gel electrophoresis |
| SEC | size exclusion chromatography |
| SEM | scanning electron microscopy |
| SEM-EDS | scanning electron microscopy-energy dispersive spectrometry |
| SF | sector field |
| SIA | sequential injection analysis |
| SIBS | spark induced breakdown spectrometry |
| SIMCA | soft independent modelling of class analogy |
| SIMS | secondary ion mass spectrometry |
| SNR | signal to noise ratio |
| SP | single particle |
| SPION | superparamagnetic iron oxide nanoparticles |
| SR | synchrotron radiation |
| SRM | standard reference material |
| SXRF | synchrotron X-ray fluorescence |
| SVR | support vector regression |
| STXM | scanning transmission X-ray microscopy |
| TEM | transmission electron microscopy |
| TDS | thermal desorption mass spectrometry |
| TGA | thermogravimetric analysis |
| TIMS | thermal ionisation mass spectrometry |
| TOF | time of flight |
| T-PGAA | time-resolved prompt gamma activation analysis |
| TPR | temperature programmed reduction |
| TXRF | total reflection X-ray fluorescence |
| UAE | ultrasound assisted extraction |
| UV–Vis | ultra-violet–visible |
| WDXRF | wavelength dispersive X-ray fluorescence |
| WEEE | waste electrical and electronic equipment |
| XAFS | X-ray absorption fine structure spectrometry |
| XANES | X-ray absorption near edge structure |
| XAS | X-ray absorption spectroscopy |
| XPS | X-ray photoelectron spectroscopy |
| XRD | X-ray diffraction |
| XRF | X-ray fluorescence |
| XRPD | X-ray powder diffraction |
| XRR | X-ray reflectometry |
7 Conflicts of interest
There are no conflicts to declare.8 References
- S. Carter, R. Clough, A. Fisher, B. Gibson, B. Russell and I. Whiteside, J. Anal. At. Spectrom., 2020, 35(11), 2410–2474 RSC.
- C. Vanhoof, J. R. Bacon, U. E. A. Fittschen and L. Vincze, J. Anal. At. Spectrom., 2020, 35(9), 1704–1719 RSC.
- M. Patriarca, N. Barlow, A. Cross, S. Hill, A. Robson, A. Taylor and J. Tyson, J. Anal. At. Spectrom., 2021, 36(3), 452–511 RSC.
- J. R. Bacon, O. T. Butler, W. R. L. Cairns, O. Cavoura, J. M. Cook, C. M. Davidson and R. Mertz-Kraus, J. Anal. At. Spectrom., 2021, 36(1), 10–55 RSC.
- R. Clough, C. F. Harrington, S. J. Hill, Y. Madrid and J. F. Tyson, J. Anal. At. Spectrom., 2020, 35(7), 1236–1278 RSC.
- E. H. Evans, J. Pisonero, C. M. Smith and R. N. Taylor, J. Anal. At. Spectrom., 2020, 35(5), 830–851 RSC.
- V. C. Costa, D. V. Babos, J. P. Castro, D. F. Andrade, R. R. Gamela, R. C. Machado, M. A. Speranca, A. S. Araujo, J. A. Garcia and E. R. Pereira, J. Braz. Chem. Soc., 2020, 31(12), 2439–2451 CAS.
- D. X. Zhang, H. Zhang, Y. Zhao, Y. L. Chen, C. Ke, T. Xu and Y. X. He, Appl. Spectrosc. Rev., 2020 DOI:10.1080/05704928.2020.1843175.
- L. B. Guo, D. Zhang, L. X. Sun, S. C. Yao, L. Zhang, Z. Z. Wang, Q. Q. Wang, H. B. Ding, Y. Lu, Z. Y. Hou and Z. Wang, Front. Phys., 2021, 16(2), 22500, DOI:10.1007/s11467-020-1007-z.
- G. S. Senesi, R. S. Harmon and R. R. Hark, Spectrochim. Acta, Part B, 2021, 175, 106013 CrossRef CAS.
- S. Legnaioli, B. Campanella, F. Poggialini, S. Pagnotta, M. A. Harith, Z. A. Abdel-Salam and V. Palleschi, Anal. Methods, 2020, 12(8), 1014–1029 RSC.
- W. T. Li, X. Y. Li, X. Li, Z. Q. Hao, Y. F. Lu and X. Y. Zeng, Appl. Spectrosc. Rev., 2020, 55(1), 1–25 CrossRef CAS.
- S. K. H. Shah, J. Iqbal, P. Ahmad, M. U. Khandaker, S. Haq and M. Naeem, Radiat. Phys. Chem., 2020, 170 DOI:10.1016/j.radphyschem.2019.108666.
- A. Limbeck, L. Brunnbauer, H. Lohninger, P. Porizka, P. Modlitbova, J. Kaiser, P. Janovszky, A. Keri and G. Galbacs, Anal. Chim. Acta, 2021, 1147, 72–98 CrossRef CAS PubMed.
- W. D. Zhou, Y. H. Guo and R. R. Zhang, Front. Phys., 2020, 15(5) DOI:10.1007/s11467-020-0969-1.
- P. S. Rahimabadi, M. Khodaei and K. R. Koswattage, X-Ray Spectrom., 2020, 49(3), 348–373 CrossRef.
- Z. Z. Wang, Y. Deguchi, F. J. Shiou, S. Tanaka, M. C. Cui, K. Rong and J. J. Yan, ISIJ Int., 2020, 60(5), 971–978 CrossRef CAS.
- Q. D. Zeng, G. H. Chen, X. G. Chen, B. Y. Wang, B. Y. Wan, M. T. Yuan, Y. Liu, H. Q. Yu, L. B. Guo and X. Y. Li, Plasma Sci. Technol., 2020, 22(7) DOI:10.1088/2058-6272/ab8a0b.
- V. N. Lednev, P. A. Sdvizhenskii, A. Y. Stavertiy, M. Y. Grishin, R. S. Tretyakov, R. D. Asyutin and S. M. Pershin, Spectrochim. Acta, Part B, 2021, 175, 106032 CrossRef CAS.
- U. A. Taparli, T. Kannengiesser, K. Cieslik, D. Mory and A. Griesche, Spectrochim. Acta, Part B, 2020, 167, 105826 CrossRef CAS.
- J. Petersson, M. Gilbert-Gatty and A. Bengtson, J. Anal. At. Spectrom., 2020, 35(9), 1848–1858 RSC.
- F. Chang, H. L. Lu, H. Sun and J. H. Yang, J. Anal. At. Spectrom., 2020, 35(11), 2639–2648 RSC.
- V. N. Lednev, P. A. Sdvizhenskii, R. D. Asyutin and A. V. Perestoronin, J. Anal. At. Spectrom., 2021, 36(3), 540–547 RSC.
- Y. Q. Zhang, C. Sun, Z. Q. Yue, S. Shabbir, W. J. Xu, M. T. Wu, L. Zou, Y. Q. Tan, F. Y. Chen and J. Yu, Opt. Express, 2020, 28(21), 32019–32032 CrossRef CAS PubMed.
- F. Poggialini, B. Campanella, S. Legnaioli, S. Pagnotta, S. Raneri and V. Palleschi, Rev. Sci. Instrum., 2020, 91(7) DOI:10.1063/5.0012669.
- S. Kashiwakura and K. Wagatsuma, ISIJ Int., 2020, 60(6), 1245–1253 CrossRef CAS.
- M. C. Cui, Y. Deguchi, C. F. Yao, Z. Z. Wang, S. Tanaka and D. H. Zhang, Spectrochim. Acta, Part B, 2020, 167 DOI:10.1016/j.sab.2020.105839.
- R. W. Liu, K. Rong, Z. Z. Wang, M. C. Cui, Y. Deguchi, S. Tanaka, J. J. Yan and J. P. Liu, ISIJ Int., 2020, 60(8), 1724–1731 CrossRef CAS.
- J. B. Cai, M. R. Dong, Y. S. Zhang, Y. X. Chen, H. J. Chen, Y. C. Liang, W. J. Li and J. D. Lu, At. Spectrosc., 2021, 42(2), 43–50 CrossRef CAS.
- W. Zhang, R. Zhou, K. Liu, Q. Z. Li, Z. Y. Tang, C. W. Zhu, X. Y. Li, X. Y. Zeng and C. He, J. Anal. At. Spectrom., 2021, 36(2), 375–379 RSC.
- H. Wang, Y. H. Jia, Y. L. Li, L. Zhao, C. Yang and D. W. Cheng, Spectrochim. Acta, Part B, 2020, 171 DOI:10.1016/j.sab.2020.105927.
- W. Wang, L. X. Sun, P. Zhang, L. F. Qi, L. M. Zheng and W. Dong, Microchem. J., 2020, 158 DOI:10.1016/j.microc.2020.105267.
- G. S. Senesi, O. De Pascale, A. Bove and B. S. Marangoni, Appl. Sci., 2020, 10(23) DOI:10.3390/app10238461.
- F. F. Hu, C. H. Wang, J. D. Li, P. Y. Liu, H. Liu and L. Zhang, At. Spectrosc., 2021, 42(1), 25–31 CrossRef CAS.
- A. V. Kuptsov, A. V. Volzhenin, V. A. Labusov and A. I. Saprykin, Spectrochim. Acta, Part B, 2021, 177 DOI:10.1016/j.sab.2020.106047.
- A. V. Kuptsov, A. V. Volzhenin, V. A. Labusov and A. I. Saprykin, J. Anal. At. Spectrom., 2020, 35(11), 2600–2605 RSC.
- S. Klaes, T. Lostak, E. Pappert, K. Kerpen and U. Telgheder, Talanta, 2021, 224, 121783 CrossRef CAS PubMed.
- J. F. Shao, J. Guo, Q. Y. Wang, A. M. Chen and M. X. Jin, Plasma Sci. Technol., 2020, 22(7) DOI:10.1088/2058-6272/ab7901.
- A. V. Kuptsov, A. V. Volzhenin, V. A. Labusov and A. I. Saprykin, J. Anal. At. Spectrom., 2021, 36, 829–835, 10.1039/d0ja00429d.
- J. Y. Zhang, T. Zhou, Y. J. Cui, Y. C. Tang and B. Wu, J. Anal. At. Spectrom., 2020, 35(11), 2712–2722 RSC.
- M. C. Cui, Y. Deguchi, Z. Z. Wang, S. Tanaka, B. W. Xue, C. F. Yao and D. H. Zhang, Spectrochim. Acta, Part B, 2020, 168 DOI:10.1016/j.sab.2020.105873.
- M. C. Cui, Y. Deguchi, Z. Z. Wang, C. F. Yao, L. Tan and D. H. Zhang, Front. Phys., 2020, 8 DOI:10.3389/fphy.2020.00237.
- M. Bonin, D. Lariviere and P. P. Povinec, Anal. Methods, 2020, 12(17), 2272–2278 RSC.
- Y. R. Wang, R. H. Li and Y. Q. Chen, Spectrochim. Acta, Part B, 2021, 177 DOI:10.1016/j.sab.2021.106077.
- Y. R. Wang, Y. Q. Chen, R. H. Li, J. Kang and J. K. Gao, J. Anal. At. Spectrom., 2021, 36(2), 314–321 RSC.
- Z. Q. Yue, C. Sun, L. Gao, Y. Q. Zhang, S. Shabbir, W. J. Xu, M. T. Wu, L. Zou, Y. Q. Tan, F. Y. Chen and J. Yu, Opt. Express, 2020, 28(10), 14345–14356 CrossRef CAS PubMed.
- Y. J. Dai, S. Y. Zhao, C. Song and X. Gao, Microw. Opt. Technol. Lett., 2021, 63, 1629–1634, DOI:10.1002/mop.32810.
- G. D. Wang, L. X. Sun, W. Wang, T. Chen, M. T. Guo and P. Zhang, Plasma Sci. Technol., 2020, 22(7) DOI:10.1088/2058-6272/ab76b4.
- Y. Fugane, S. Kashiwakura and K. Wagatsuma, Anal. Sci., 2020, 36(11), 1415–1421 CrossRef CAS PubMed.
- Y. Fugane, S. Kashiwakura and K. Wagatsuma, Surf. Interfaces, 2020, 20 DOI:10.1016/j.surfin.2020.100602.
- S. Kugler, A. Nagy, J. Osan, L. Peter, V. Groma, S. Pollastri and A. Czitrovszky, Spectrochim. Acta, Part B, 2021, 177 DOI:10.1016/j.sab.2021.106110.
- X. Li, T. Z. Zhao, Q. X. Zhong, S. Z. Nie, H. Xiao, S. H. Zhao, W. D. Huang and Z. W. Fan, J. Anal. At. Spectrom., 2020, 35(10), 2171–2178, DOI:10.1016/j.sab.2021.106110.
- V. N. Lednev, P. A. Sdvizhenskii, M. Y. Grishin, A. Y. Stavertiy, R. S. Tretyakov, R. D. Asyutin, A. N. Fedorov and S. M. Pershin, Appl. Opt., 2021, 60(5), 1144–1149 CrossRef CAS PubMed.
- D. L. Li, Z. X. Liu, L. Zhao, X. J. Shen and H. Z. Wang, Spectrochim. Acta, Part B, 2020, 169 DOI:10.1016/j.sab.2020.105896.
- N. S. Medvedev, A. V. Volzhenin and A. I. Saprykin, Microchem. J., 2020, 157 DOI:10.1016/j.microc.2020.104970.
- N. S. Medvedev, A. V. Volzhenin and A. I. Saprykin, At. Spectrosc., 2021, 42(2), 71–78 CAS.
- O. V. Lundovskaya, N. S. Medvedev, A. R. Tsygankova, A. V. Volzhenin and A. I. Saprykin, Spectrochim. Acta, Part B, 2021, 177 DOI:10.1016/j.sab.2020.106049.
- J. Sieber, A. Marlow, R. Paul, C. Barber, L. Wood, L. Yu, A. Rieke, R. Carl, A. Kutnerian, J. McCandless and C. Wallace, X-Ray Spectrom., 2021, 50, 210–223, DOI:10.1002/xrs.3216.
- K. Harouaka, E. W. Hoppe and I. J. Arnquist, J. Anal. At. Spectrom., 2020, 35(12), 2859–2866 RSC.
- M. Habibpour, P. Parvin and R. Amrollahi, Appl. Opt., 2021, 60(5), 1099–1109 CrossRef PubMed.
- J. Brocchieri, E. Scialla and C. Sabbarese, Nucl. Instrum. Methods Phys. Res., Sect. B, 2021, 486, 73–84 CrossRef CAS.
- J. K. Gao, J. Kang, R. H. Li and Y. Q. Chen, Appl. Opt., 2020, 59(13), 4091–4096 CrossRef CAS PubMed.
- K. L. LeBlanc, P. M. Le, J. Meija, J. F. Ding, J. E. Melanson and Z. Mester, J. Anal. At. Spectrom., 2021, 36(2), 416–428 RSC.
- M. Lagerstrom and E. Ytreberg, Talanta, 2021, 223 DOI:10.1016/j.talanta.2020.121820.
- S. Kummel, A. Horst, F. Gelman, H. Strauss, H. H. Richnow and M. Gehre, Anal. Chem., 2020, 92(21), 14685–14692 CrossRef PubMed.
- A. Gubal, V. Chuchina, N. Ivanenko, R. Qian, N. Solovyev and A. Ganeev, Spectrochim. Acta, Part B, 2020, 173, 105986, DOI:10.1016/j.sab.2020.105986.
- Z. E. Ahmed, M. Abdelhamid, Z. A. Abdel-Salam, V. Palleschi and M. Abdel-Harith, J. Chemom., 2021, 35 DOI:10.1002/cem.3334.
- F. F. Hilario, M. L. de Mello and E. R. Pereira, Anal. Methods, 2021, 13(2), 232–241 RSC.
- Z. B. Amorim, M. P. Machado, M. K. Moro, L. O. Reboucas, B. M. Dalfiior, W. Romao, P. R. Filgueiras, M. Carneiro and G. P. Brandao, J. Braz. Chem. Soc., 2021, 32(1), 47–58 CAS.
- M. Horiguchi and S. Kunimura, Anal. Sci., 2020, 36(11), 1311–1315 CrossRef CAS PubMed.
- J. Yu, Y. J. Kang, Q. F. Lu, H. Luo, Z. H. Lu, L. J. Cui and J. L. Li, Microchem. J., 2020, 159 DOI:10.1016/j.microc.2020.105507.
- H. J. F. Fabrino, C. P. Demicheli, F. J. G. Frezard and L. M. Costa, J. Braz. Chem. Soc., 2021, 32(1), 11–19 CAS.
- M. Elcan, H. Erdogan and O. Acar, Spectrosc. Lett., 2021, 54, 140–150, DOI:10.1080/00387010.2021.1877157.
- J. C. Garcia-Mesa, P. Montoro-Leal, A. Rodriguez-Moreno, M. M. L. Guerrero and E. I. V. Alonso, Talanta, 2021, 223, 121795 CrossRef CAS PubMed.
- V. S. de Souza, L. S. G. Teixeira, J. Lima, U. Cerqueira, O. M. C. de Oliveira, A. F. D. Queiroz and M. A. Bezerra, Appl. Spectrosc. Rev., 2020, 55(2), 128–157 CrossRef.
- C. Andreu, M. F. Gazulla, M. Rodrigo, M. Orduna and M. J. Ventura, Anal. Lett., 2020, 54, 2081–2095, DOI:10.1080/00032719.2020.1839478.
- V. D. Ferreira, J. S. Almeida, V. A. Lemos, O. M. C. de Oliveira, K. S. Garcia and L. S. G. Teixeira, Talanta, 2021, 222, 121514 CrossRef CAS PubMed.
- M. A. Aguirre, A. Canals, I. Lopez-Garcia and M. Hernandez-Cordoba, Talanta, 2020, 220 DOI:10.1016/j.talanta.2020.121395.
- R. Sanchez, F. Chainet, V. Souchon, S. Carbonneaux, C. P. Lienemann and J. L. Todol, J. Anal. At. Spectrom., 2020, 35(10), 2387–2394 RSC.
- X. L. Li, L. Zhang, Z. H. Tian, Y. Bai, S. Q. Wang, J. H. Han, G. F. Xia, W. G. Ma, L. Dong, W. B. Yin, L. T. Xiao and S. T. Jia, J. Anal. At. Spectrom., 2020, 35(12), 2928–2934 RSC.
- J. Xie, X. D. Niu, K. Q. He, M. D. Shi, S. J. Yu, C. G. Yuan and J. F. Liu, Fuel, 2021, 292 DOI:10.1016/j.fuel.2021.120228.
- W. H. Zhang, Z. Zhuo, P. Lu, T. F. Sun, W. L. Sun and J. Q. Lu, Spectrochim. Acta, Part B, 2021, 177 DOI:10.1016/j.sab.2021.106076.
- M. R. Dong, L. P. Wei, J. J. Gonzalez, D. Oropeza, J. Chirinos, X. L. Mao, J. D. Lu and R. E. Russo, Anal. Chem., 2020, 92(10), 7003–7010 CrossRef CAS PubMed.
- F. Deng, Y. Ding, Y. J. Chen and S. N. Zhu, Plasma Sci. Technol., 2020, 22(7) DOI:10.1088/2058-6272/ab77d5.
- N. Mketo and P. N. Nomngongo, Sci. Rep., 2020, 10(1) DOI:10.1038/s41598-020-71963-2.
- N. S. Mdluli, P. N. Nomngongo and N. Mketo, Crit. Rev. Anal. Chem. DOI:10.1080/10408347.2020.1781591.
- K. Li, X. Wu, Z. M. Chen, J. Luo, X. D. Hou and X. M. Jiang, J. Anal. At. Spectrom., 2020, 35(11), 2656–2662 RSC.
- M. Vinic, E. Aruffo, F. Andreoli, M. Ivkovic and V. Lazic, Spectrochim. Acta, Part B, 2020, 164 DOI:10.1016/j.sab.2020.105765.
- K. Hasme and A. Wassmuth, Spectroscopy, 2020, 35(3), 18 CAS.
- F. R. Adolfo, P. C. do Nascimento, D. Bohrer, C. Viana, L. M. de Carvalho, M. C. C. Cravo and L. Nascimento, Fuel, 2020, 277 DOI:10.1016/j.fuel.2020.118098.
- G. Chauhan and A. de Klerk, Energy Fuels, 2020, 34(3), 2870–2879 CrossRef CAS.
- G. Gonzalez, S. Acevedo, J. Castillo, O. Villegas, M. A. Ranaudo, K. Guzman, M. Orea and B. Bouyssiere, Energy Fuels, 2020, 34(10), 12535–12544 CrossRef CAS.
- V. Garcia-Montoto, S. Verdier, Z. Maroun, R. Egeberg, J. L. Tiedje, S. Sandersen, P. Zeuthen and B. Bouyssiere, Fuel Process. Technol., 2020, 201 DOI:10.1016/j.fuproc.2020.10634.
- R. Arthur and P. A. Scherer, X-Ray Spectrom., 2020, 49, 560–571, DOI:10.1002/xrs.3151.
- H. El Hadri, J. Gigault, S. Mounicou, B. Grassl and S. Reynaud, Mar. Pollut. Bull., 2020, 160 DOI:10.1016/j.marpolbul.2020.111716.
- J. Jimenez-Lamana, L. Marigliano, J. Allouche, B. Grassl, J. Szpunar and S. Reynaud, Anal. Chem., 2020, 92(17), 11664–11672 CrossRef CAS PubMed.
- D. Chen, T. Wang, Y. W. Ma, G. W. Wang, Q. W. Kong, P. F. Zhang and R. L. Li, Sci. Total Environ., 2020, 743 DOI:10.1016/j.scitotenv.2020.140850.
- L. A. Holmes, R. C. Thompson and A. Turner, Environ. Pollut., 2020, 261 DOI:10.1016/j.envpol.2020.114107.
- L. Hildebrandt, M. von der Au, T. Zimmermann, A. Reese, J. Ludwig and D. Profrock, PLoS One, 2020, 15(7) DOI:10.1371/journal.pone.0236120.
- R. Junjuri and M. K. Gundawar, Waste Manage., 2020, 117, 48–57 CrossRef CAS PubMed.
- L. Brunnbauer, S. Larisegger, H. Lohninger, M. Nelhiebel and A. Limbeck, Talanta, 2020, 209 DOI:10.1016/j.talanta.2019.120572.
- A. P. M. Michel, A. E. Morrison, V. L. Preston, C. T. Marx, B. C. Colson and H. K. White, Environ. Sci. Technol., 2020, 54(17), 10630–10637 CrossRef CAS PubMed.
- K. Liu, D. Tian, X. Yue and G. Yang, J. Instrum., 2020, 15(4) DOI:10.1088/1748-0221/15/04/t04002.
- V. C. Costa, M. L. de Mello, D. V. Babos, J. P. Castro and E. R. Pereira-Filho, Microchem. J., 2020, 159 DOI:10.1016/j.microc.2020.105558.
- Y. Nishiwaki, S. Honda, T. Yamato, R. Kondo, A. Kaneda and S. Hayakawa, J. Forensic Sci., 2020, 65, 1474–1479 CrossRef CAS PubMed.
- S. Aidene, V. Semenov, D. Kirsanov, D. Kirsanov and V. Panchuk, Measurement, 2021, 172 DOI:10.1016/j.measurement.2020.108888.
- F. Kuczelinis, J. H. Petersen, P. Weis and N. H. Bings, J. Anal. At. Spectrom., 2020, 35(9), 1922–1931 RSC.
- L. F. Rodrigues, R. F. Santos, R. C. Bolzan, F. A. Duarte, J. C. P. Mattos and E. M. M. Flores, Talanta, 2020, 218 DOI:10.1016/j.talanta.2020.121129.
- E. de Almeida, F. L. Melquiades, J. P. R. Marques, E. Margui and H. W. P. de Carvalho, Spectrochim. Acta, Part B, 2020, 167 DOI:10.1016/j.sab.2020.105818.
- K. Ehelebe, J. Knoppel, M. Bierling, B. Mayerhofer, T. Bohm, N. Kulyk, S. Thiele, K. J. J. Mayrhofer and S. Cherevko, Angew. Chem., Int. Ed., 2021, 60, 8882–8888, DOI:10.1002/anie.202014711.
- F. D. Speck, M. T. Y. Paul, F. Ruiz-Zepeda, M. Gatalo, H. Kim, H. C. Kwon, K. J. J. Mayrhofer, M. Choi, C. H. Choi, N. Hodnik and S. Cherevko, J. Am. Chem. Soc., 2020, 142(36), 15496–15504 CrossRef CAS PubMed.
- A. Kormanyos, F. D. Speck, K. J. J. Mayrhofer and S. Cherevko, ACS Catal., 2020, 10(19), 10858–10870 CrossRef CAS.
- I. Khalakhan, M. Bogar, M. Vorokhta, P. Kus, Y. Yakovlev, M. Dopita, D. J. S. Sandbeck, S. Cherevko, I. Matolinova and H. Amenitsch, ACS Appl. Mater. Interfaces, 2020, 12(15), 17602–17610 CrossRef CAS PubMed.
- D. J. S. Sandbeck, N. M. Secher, M. Inaba, J. Quinson, J. E. Sorensen, J. Kibsgaard, A. Zana, F. Bizzotto, F. D. Speck, M. T. Y. Paul, A. Dworzak, C. Dosche, M. Oezaslan, I. Chorkendorff, M. Arenz and S. Cherevko, J. Electrochem. Soc., 2020, 167(16) DOI:10.1149/1945-7111/abc767.
- D. J. S. Sandbeck, N. M. Secher, F. D. Speck, J. E. Sorensen, J. Kibsgaard, I. Chorkendorff and S. Cherevko, ACS Catal., 2020, 10(11), 6281–6290 CrossRef CAS.
- C. Van Pham, M. Buhler, J. Knoppel, M. Bierling, D. Seeberger, D. Escalera-Lopez, K. J. J. Mayrhofer, S. Cherevko and S. Thiele, Appl. Catal., B, 2020, 269 DOI:10.1016/j.apcatb.2020.118762.
- M. Senila, O. Cadar, L. Senila, S. Boringer, K. Seaudeau-Pirouley, A. Ruiu and P. Lacroix-Desmazes, Materials, 2020, 13(22) DOI:10.3390/ma13225136.
- F. Y. Zhang, G. A. Zhang, J. Xu, Z. Zhao and L. Xu, Chin. J. Anal. Chem., 2020, 48(11), 1590–1596 Search PubMed.
- A. Anthemidis and A. Tzili, Int. J. Environ. Anal. Chem., 2021 DOI:10.1080/03067319.2021.1906424.
- X. W. Zhu, Q. He, F. J. Xu, X. X. Ma, Z. Xing, S. C. Zhang and X. R. Zhang, J. Anal. At. Spectrom., 2020, 35(4), 804–809 RSC.
- X. W. Zhu, F. J. Xu, Z. Xing, S. C. Zhang and X. R. Zhang, Microchem. J., 2020, 156 DOI:10.1016/j.microc.2020.104926.
- X. W. Zhu, F. J. Xu, Q. He, Z. Xing, S. C. Zhang and X. R. Zhang, Microchem. J., 2021, 161 DOI:10.1016/j.microc.2020.105762.
- F. Horak, A. Nagl, K. Fottinger and A. Limbeck, Microchim. Acta, 2020, 187(12) DOI:10.1007/s00604-020-04609-9.
- F. B. Madeira, V. G. P. Saide, M. Castro, C. M. Barra, S. Rocha, V. G. K. Almeida, A. V. S. Canuto, G. G. Shimamoto, M. Tubino and J. G. Rocha, J. Braz. Chem. Soc., 2020, 31(12), 2470–2478 CAS.
- I. M. D. Ferreira, B. F. Braz, L. da Silva, A. S. Luna and R. E. Santelli, Spectrochim. Acta, Part B, 2021, 177 DOI:10.1016/j.sab.2021.106087.
- M. Aliste, S. Arranz, A. Sánchez-Ortega, M. C. Sampedro, N. Unceta, A. Gómez-Caballero, A. Vallejo, M. A. Goicolea and R. J. Barrio, J. Forensic Sci., 2020, 65(4), 1094–1101 CrossRef CAS PubMed.
- A. Akmeemana, P. Weis, R. Corzo, D. Ramos, P. Zoon, T. Trejos, T. Ernst, C. Pollock, E. Bakowska, C. Neumann and J. Almirall, J. Chemom., 2021, 35 DOI:10.1002/cem.3267.
- P. Becker, C. Neff, S. Hess, P. Weis and D. Gunther, J. Anal. At. Spectrom., 2020, 35(10), 2248–2254 RSC.
- R. Corzo, T. Hoffman, T. Ernst, T. Trejos, T. Berman, S. Coulson, P. Weis, A. Stryjnik, H. Dorn, E. Pollock, M. S. Workman, P. Jones, B. Nytes, T. Scholz, H. F. Xie, K. Igowsky, R. Nelson, K. Gates, J. Gonzalez, L. M. Voss and J. Almirall, Forensic Chem., 2021, 22 DOI:10.1016/j.forc.2021.100307.
- J. Almirall, A. Akmeemana, K. Lambert, P. Jiang, E. Bakowska, R. Corzo, C. M. Lopez, E. C. Pollock, K. Prasch, T. Trejos, P. Weis, W. Wiarda, H. F. Xie and P. Zoon, Spectrochim. Acta, Part B, 2021, 179 DOI:10.1016/j.sab.2021.106119.
- R. Corzo and E. Steel, X-Ray Spectrom. DOI:10.1002/xrs.3179.
- V. Sharma, R. Acharya, H. K. Bagla and P. K. Pujari, J. Anal. At. Spectrom., 2021, 36(3), 630–643 RSC.
- A. S. Bryukhova, A. A. Kuznetsov, I. V. Seliverstova, A. M. Popov, T. A. Labutin and N. B. Zorov, J. Appl. Spectrosc., 2020, 87, 800–804 CrossRef CAS.
- T. Volker, S. Millar, C. Strangfeld and G. Wilsch, Constr. Build. Mater., 2020, 258 DOI:10.1016/j.conbuildmat.2020.120345.
- Y. Cruz-Hernandez, M. Chrysochoou and K. Wille, Spectrochim. Acta, Part B, 2020, 172 DOI:10.1016/j.sab.2020.105949.
- N. N. Mijatovic, A. M. Terzic, L. L. Pezo, L. R. Milicic and D. Z. Zivojinovic, J. Serb. Chem. Soc., 2020, 85(12), 1605–1619 CrossRef CAS.
- L. Zou, C. Sun, M. T. Wu, Y. Q. Zhang, Z. Q. Yue, W. J. Xu, S. Shabbir, F. Y. Chen, B. Liu, W. H. Liu and J. Yu, J. Anal. At. Spectrom., 2021, 36(2), 303–313 RSC.
- R. Patidar, B. Rebary, G. R. Bhadu and G. Patel, Int. J. Mass Spectrom., 2020, 454 DOI:10.1016/j.ijms.2020.116356.
- A. Muller, D. Pozebon and A. S. Ribeiro, Anal. Methods, 2020, 12(20), 2638–2644 RSC.
- T. Y. Gusel'nikova, A. R. Tsygankova and A. I. Saprykin, J. Anal. Chem., 2020, 75(5), 582–586 CrossRef.
- Y. X. Zhang, S. S. Ge, Z. H. Yang and C. Dong, Environ. Sci. Pollut. Res., 2020, 37887–37893, DOI:10.1007/s11356-020-09884-w.
- V. H. Nguyen, M. Ramzan, D. Kifle and G. Wibetoe, J. Anal. At. Spectrom., 2020, 35(11), 2594–2599 RSC.
- J. T. Sloop, S. C. Allen, J. A. Carter, G. L. Donati and B. T. Jones, Microchem. J., 2021, 160 DOI:10.1016/j.microc.2020.105740.
- S. Dhara, J. Anal. At. Spectrom., 2021, 36(2), 352–360 RSC.
- G. Paudel and M. Di Sabatino, J. Anal. At. Spectrom., 2020, 35(11), 2748–2757 RSC.
- S. Maji, S. Kumar and K. Sundararajan, Optik, 2020, 207 DOI:10.1016/j.ijleo.2019.163801.
- S. Kumar, V. Ngo, J. Park, K. S. Ham, V. K. Singh, S. H. Nam and Y. Lee, Bull. Korean Chem. Soc. DOI:10.1002/bkcs.12259.
- S. U. Choi, S. C. Han, J. Y. Lee and J. I. Yun, J. Anal. At. Spectrom., 2020, 35(10), 2378–2386 RSC.
- R. Hedwig, I. Tanra, I. Karnadi, M. Pardede, A. M. Marpaung, Z. S. Lie, K. H. Kurniawan, M. M. Suliyanti, T. J. Lie and K. Kagawa, Opt. Express, 2020, 28(7), 9259–9268 CrossRef PubMed.
- Z. P. Yang, S. E. Jackson, L. J. Cabri, P. Wee, H. P. Longerich and M. Pawlak, J. Anal. At. Spectrom., 2020, 35(3), 534–547 RSC.
- L. Q. Guo, Q. Li, Y. R. Chen, G. X. Zhang, Y. S. Xu and Z. Wang, J. Anal. At. Spectrom., 2020, 35(7), 1441–1449 RSC.
- K. Yahiaoui, S. M. Aberkane and A. Naitbouda, Mater. Chem. Phys., 2021, 259 DOI:10.1016/j.matchemphys.2020.124045.
- L. Reclaru, L. C. Ardelean, C. A. Miu and A. F. Grecu, Materials, 2020, 13(7) DOI:10.3390/ma13071697.
- O. Prymak, L. E. Vagiaki, A. Buyakov, S. Kulkov, M. Epple and M. Chatzinikolaidou, Materials, 2021, 14(4) DOI:10.3390/ma14041049.
- J. Chen, G. Mallmann, I. Zhukova and H. O'Neill, J. Sustain. Met., 2020, 6, 691–699, DOI:10.1007/s40831-020-00308-0.
- P. Devangad, V. K. Unnikrishnan, M. Yogesha, S. D. Kulkarni and S. Chidangil, J. Chemom., 2020, 34(11) DOI:10.1002/cem.3310.
- B. Stefanuk, C. Sanchez-Ake, J. L. Benitez, O. Depablos-Rivera, T. Garcia-Fernandez, S. Negrete-Aragon and M. Villagran-Muniz, Spectrochim. Acta, Part B, 2020, 167 DOI:10.1016/j.sab.2020.105823.
- C. Gerhard, A. Taleb, F. Pelascini and J. Hermann, Appl. Surf. Sci., 2021, 537 DOI:10.1016/j.apsusc.2020.147984.
- G. S. Maurya, A. Marin-Roldan, P. Veis, A. K. Pathak and P. Sen, J. Nucl. Mater., 2020, 541 DOI:10.1016/j.jnucmat.2020.152417.
- Z. Cao, Y. T. An, F. J. Gou, Z. Zhang, L. Deng, L. Guo, Z. Zheng and C. A. Chen, Fusion Eng. Des., 2020, 155 DOI:10.1016/j.fusengdes.2020.111580.
- C. R. Feng, R. Z. Yang, Q. Li, X. Q. Ye, J. L. Wu, C. G. Chen, X. F. Wang and X. H. Chen, Appl. Opt., 2020, 59(9), 2866–2873 CrossRef CAS PubMed.
- J. Oelmann, E. Wust, S. Brezinsek, C. Li, D. Y. Zhao, M. Rasinski, C. P. Dhard, M. Mayer, D. Naujoks, Y. Gao and W. X. Team, Nucl. Mater. Energy, 2021, 26 DOI:10.1016/j.nme.2021.100943.
- P. Gasior, Acta Phys. Pol., A, 2020, 138(4), 601–607 CrossRef CAS.
- D. Y. Zhao, D. Wu, J. Oelmann, S. Brezinsek, Q. M. Xiao, R. X. Yi, L. Z. Cai and H. B. Ding, J. Anal. At. Spectrom., 2020, 35(12), 2867–2879 RSC.
- P. Veis, A. Marin-Roldan, V. Dwivedi, J. Karhunen, P. Paris, I. Jogi, C. Porosnicu, C. P. Lungu, V. Nemanic and A. Hakola, Phys. Scr., 2020, T171(1) DOI:10.1088/1402-4896/ab7ebd.
- I. Jogi, P. Paris, M. Laan, J. Kozlova, H. Mandar, M. Passoni, D. Dellasega, A. Hakola and H. J. van der Meiden, J. Nucl. Mater., 2021, 544 DOI:10.1016/j.jnucmat.2020.152660.
- H. Sattar, J. L. Shi, H. Ran, M. Imran, W. Ding, P. Das Gupta and H. B. Ding, J. Nucl. Mater., 2020, 540 DOI:10.1016/j.jnucmat.2020.152389.
- Y. H. Jin, R. He, X. X. Zhang, L. Qiao and P. Wang, Rare Met. Mater. Eng., 2020, 49(10), 3498–3504 Search PubMed.
- N. Catarino, A. Widdowson, A. Baron-Wiechec, J. P. Coad, K. Heinola, M. Rubel, N. P. Barradas, E. Alves and J. E. T. Contributors, Phys. Scr., 2020, T171(1) DOI:10.1088/1402-4896/ab4df7.
- S. Moon, E. Bang, P. Petersson, D. Matveev, S. H. Hong, A. Kirschner and M. Rubel, Phys. Scr., 2020, T171(1) DOI:10.1088/1402-4896/ab51f4.
- E. Lang, C. N. Taylor and J. P. Allain, J. Nucl. Mater., 2020, 532 DOI:10.1016/j.jnucmat.2020.152047.
- A. Pisarev, I. Arkhipov, Y. Babich, M. Berdnikova, Y. Gasparyan, K. Gutorov, S. Grashin, V. Efimov, M. Isaenkova, S. Krat, O. Krymskaya, V. Kurnaev, T. Stepanova, E. Vovchenko, I. Vizgalov and M. Zibrov, Fusion Eng. Des., 2021, 162 DOI:10.1016/j.fusengdes.2020.112105.
- J. Miskovicova, M. Angus, H. van der Meiden and P. Veis, Fusion Eng. Des., 2020, 153 DOI:10.1016/j.fusengdes.2020.111488.
- D. Wu, X. L. Mao, G. C. Y. Chan, R. E. Russo, V. Zorba and H. B. Ding, J. Anal. At. Spectrom., 2020, 35(4), 767–775 RSC.
- D. Y. Zhao, N. Gierse, J. Oelmann, S. Brezinsek, M. Rasinski, Y. F. Liang, C. Linsmeier and H. B. Ding, Fusion Eng. Des., 2020, 151 DOI:10.1016/j.fusengdes.2019.111379.
- P. Veis, S. Atikkuke, A. M. Roldan, V. Dwivedi, M. Veis, P. Barton, M. Jerab and R. Dejarnac, Nucl. Mater. Energy, 2020, 25 DOI:10.1016/j.nme.2020.100809.
- R. Zaplotnik, V. Nemanic, M. Zumer, J. Kovac and M. Mozetic, Nucl. Fusion, 2021, 61(2) DOI:10.1088/1741-4326/abc934.
- A. de Castro, C. Moynihan, S. Stemmley, M. Szott, D. Andruczyk and D. N. Ruzic, Nucl. Mater. Energy, 2020, 25 DOI:10.1016/j.nme.2020.100829.
- Z. Cao, Y. T. An, X. L. Wang, C. A. Chen and Y. Li, Materials, 2020, 13(1) DOI:10.3390/ma13010240.
- Z. Cao, Y. T. An, Z. J. Wang, L. Guo, C. A. Chen, F. J. Gou and Y. Li, Nucl. Mater. Energy, 2020, 24 DOI:10.1016/j.nme.2020.100786.
- L. Wang, X. Q. Ye, Y. J. Feng, X. F. Luo, Z. H. Hong, J. Yan, B. P. Gong, H. B. Liao, X. Y. Wang, C. D. Zhu, J. J. Yang and M. Gong, Fusion Eng. Des., 2020, 153 DOI:10.1016/j.fusengdes.2020.111490.
- Z. B. Ye, W. Y. Yang, L. Shu, Z. J. Wang, Q. C. Liu, Q. Yan, J. J. Wei, K. Zhang and F. J. Gou, Fusion Sci. Technol., 2020, 76(2), 157–162 CrossRef.
- T. Hernandez, F. J. Sanchez, A. Morono, M. Aristu and J. F. Marco, J. Nucl. Mater., 2021, 545 DOI:10.1016/j.jnucmat.2020.152614.
- A. Saha, K. Kumari, S. B. Deb and M. K. Saxena, J. Anal. At. Spectrom., 2021, 36(3), 561–569 RSC.
- H. X. Cong, C. X. Liu, R. F. Li, Y. X. Liu, Q. Dou, H. Y. Fu, L. Zhang, W. Zhou, Q. N. Li and W. X. Li, J. Radioanal. Nucl. Chem., 2019, 322(3), 2025–2032 CrossRef CAS.
- A. Sengupta, B. Rajeswari and R. M. Kadam, J. Anal. At. Spectrom., 2020, 35(1), 169–177 RSC.
- A. Pandey, F. A. Khan, A. Kelkar, P. Purohit, P. Kumar, D. B. Sathe, R. B. Bhatt and P. G. Behere, J. Radioanal. Nucl. Chem., 2020, 324(2), 731–736 CrossRef CAS.
- J. Namitha, U. K. Maity, T. Selvalakshmi, P. Manoravi, G. Kumar, M. Joseph and N. Sivaraman, J. Anal. At. Spectrom., 2020, 35(7), 1412–1422 RSC.
- S. Diez-Fernandez, H. Isnard, A. Nonell, C. Bresson and F. Chartier, J. Anal. At. Spectrom., 2020, 35(12), 2793–2819 RSC.
- K. A. Adams, S. S. Morrison, D. E. Cherkasov and N. A. Wall, J. Radioanal. Nucl. Chem. DOI:10.1007/s10967-020-07240-3.
- S. Xing, M. Y. Luo, N. Yuan, D. Q. Liu, Y. Yang, X. X. Dai, W. C. Zhang and N. Chen, At. Spectrosc., 2021, 42(2), 62–70 CrossRef CAS.
- B. Russell, H. Mohamud, M. G. Miranda, P. Ivanov, H. Thompkins, J. Scott, P. Keen and S. Goddard, J. Anal. At. Spectrom., 2021, 36, 845–855 RSC.
- V. K. Do, T. Furuse, E. Murakami, R. Aita, Y. Ohta and S. Sato, J. Radioanal. Nucl. Chem., 2021, 327, 542–553, DOI:10.1007/s10967-020-07503-z.
- L. C. Zhu, X. L. Hou and J. X. Qiao, Talanta, 2021, 221 DOI:10.1016/j.talanta.2020.121637.
- M. P. Dion, K. W. E. Springer, R. I. Sumner, M. L. P. Thomas and G. C. Eiden, Int. J. Mass Spectrom., 2020, 449 DOI:10.1016/j.ijms.2019.116273.
- Y. Shibahara, S. Nakamura, A. Uehara, T. Fujii, S. Fukutani, A. Kimura and O. Iwamoto, J. Radioanal. Nucl. Chem., 2020, 325, 155–165, DOI:10.1007/s10967-020-07198-2.
- S. Asai, M. Ohata, Y. Hanzawa, T. Horita, T. Yomogida and Y. Kitatsuji, Anal. Chem., 2020, 92(4), 3276–3284 CrossRef CAS PubMed.
- K. Sasa, M. Honda, S. Hosoya, T. Takahashi, K. Takano, Y. Ochiai, A. Sakaguchi, S. Kurita, Y. Satou and K. Sueki, J. Nucl. Sci. Technol., 2020, 72–79, DOI:10.1080/00223131.2020.1801530.
- M. Schiffer, L. Bussmann, S. Herb, R. Spanier, A. Stolz, C. Mueller-Gatermann, G. Hackenberg, S. Heinze, E. Strub and A. Dewald, Nucl. Instrum. Methods Phys. Res., Sect. B, 2020, 464, 19–22 CrossRef CAS.
- J. Gong, M. He, Q. Z. Zhao, P. Hu, Y. J. Pang, Y. W. Bao, Q. B. You, K. N. Li, S. Y. Su, Y. M. Hu and S. Y. Wu, Nucl. Instrum. Methods Phys. Res., Sect. B, 2020, 465, 31–36 CrossRef CAS.
- D. G. Pacesila, A. R. Petre, E. C. Calvo, A. I. Rotaru, A. N. State, V. D. Mosu, M. Virgolici and G. Cata-Danil, UPB Sci. Bull. A: Appl. Math. Phys., 2020, 82(2), 241–250 Search PubMed.
- K. McHugh and E. Krogstad, Int. J. Mass Spectrom., 2021, 459 DOI:10.1016/j.ijms.2020.116471.
- Z. T. Wang, Y. Xie, J. X. Lin, Z. J. Li and Z. Y. Tan, J. Radioanal. Nucl. Chem., 2021, 328, 137–147, DOI:10.1007/s10967-021-07627-w.
- K. Kolacinska, T. A. Devol, A. F. Seliman, J. Dudek and M. Trojanowicz, Microchem. J., 2020, 152 DOI:10.1016/j.microc.2019.104426.
- N. N. Wanna, A. Dobney, K. Van Hoecke, M. Vasile and F. Vanhaecke, Talanta, 2021, 221 DOI:10.1016/j.talanta.2020.121592.
- B. T. Manard, D. A. Bostick, S. C. Metzger, B. W. Ticknor, N. A. Zirakparvar, K. T. Rogers and C. R. Hexel, Spectrochim. Acta, Part B, 2021, 179 DOI:10.1016/j.sab.2021.106097.
- K. Liu, Y. L. Liu, Z. F. Chai and W. Q. Shi, Sep. Purif. Technol., 2021, 265 DOI:10.1016/j.seppur.2021.118524.
- W. I. Mortada, A. A. Awad, M. M. El-defrawy and M. E. Khalifa, Microchem. J., 2021, 161 DOI:10.1016/j.microc.2020.105778.
- E. Phillip, K. S. Khoo, M. A. W. Yusof and R. O. A. Rahman, J. Environ. Manage., 2021, 280 DOI:10.1016/j.jenvman.2020.111703.
- R. Nakanishi, M. Saeki, I. Wakaida and H. Ohba, Appl. Sci., 2020, 10(24) DOI:10.3390/app10248985.
- Y. Ikeda, J. A. Ofosu and I. Wakaida, Spectrochim. Acta, Part B, 2020, 171 DOI:10.1016/j.sab.2020.105933.
- H. Miura, Y. Kurihara, M. Yamamoto, A. Sakaguchi, N. Yamaguchi, O. Sekizawa, K. Nitta, S. Higaki, D. Tsumune, T. Itai and Y. Takahashi, Sci. Rep., 2020, 10(1) DOI:10.1038/s41598-020-68318-2.
- Y. Kurihara, N. Takahata, T. D. Yokoyama, H. Miura, Y. Kon, T. Takagi, S. Higaki, N. Yamaguchi, Y. Sano and Y. Takahashi, Sci. Rep., 2020, 10(1) DOI:10.1038/s41598-020-59933-0.
- S. V. Jovanovic, P. K. Weber, A. J. Pidduck, A. M. Gaffney, P. Girard, F. Pointurier, M. Hedberg, A. J. Simons, V. Stebelkov, T. Kell, K. Knight, T. Parsons-Davis, M. Kristo, R. W. Williams, K. C. Treinen, N. J. Montgomery, J. King, A. Wickenden, D. Knight, A. L. Faure, A. Hubert, N. Albert, M. C. Vincent, M. Wallenius, I. A. Elantyev, K. D. Zhizhin, J. M. Schwantes, O. Marsden and F. Taylor, J. Radioanal. Nucl. Chem., 2020, 326, 1853–1866, DOI:10.1007/s10967-020-07470-5.
- Z. Varga, M. Wallenius, A. Nicholl and K. Mayer, Spectrochim. Acta, Part B, 2020, 171 DOI:10.1016/j.sab.2020.105920.
- Z. Varga, M. Wallenius, A. Nicholl, K. Mayer, I. Balan and V. Benea, J. Radioanal. Nucl. Chem., 2019, 322(3), 1585–1591 CrossRef CAS.
- M. Krachler, M. Wallenius, A. Nicholl and K. Mayer, RSC Adv., 2020, 10(28), 16629–16636 RSC.
- K. Yamamoto, H. Asanuma, H. Takahashi and T. Hirata, J. Anal. At. Spectrom., 2021, 36(3), 668–675 RSC.
- G. Craig, M. S. A. Horstwood, H. J. Reid and B. L. Sharp, J. Anal. At. Spectrom., 2020, 35(5), 1011–1021 RSC.
- R. K. Bhatia, V. K. Yadav, M. M. Gulhane, R. Datta, K. D. Joshi, A. M. Kasbekar, S. Das, K. Sreeramulu, T. K. Saha, E. Ravisankar and V. Nataraju, Rapid Commun. Mass Spectrom., 2021, 35(3) DOI:10.1002/rcm.8963.
- C. G. Lee, R. Park, J. Park and S. H. Lim, Nucl. Eng. Technol., 2020, 52(7), 1532–1536 CrossRef CAS.
- C. Rinaldi, M. Pozzi, N. Boggio and J. Vorob, Spectrochim. Acta, Part B, 2020, 167 DOI:10.1016/j.sab.2020.105841.
- E. Rollin, O. Musset, D. Cardona and J. B. Sirven, Spectrochim. Acta, Part B, 2020, 166 DOI:10.1016/j.sab.2020.105796.
- X. Y. Yu, J. Yao, E. C. Buck and Z. H. Zhu, Surf. Interface Anal., 2020, 52(7), 454–459 CrossRef CAS.
- H. W. Paing, B. T. Manard, B. W. Ticknor, J. R. Bills, K. A. Hall, D. A. Bostick, P. Cable-Dunlap and R. K. Marcus, Anal. Chem., 2020, 92(12), 8591–8598 CrossRef CAS PubMed.
- E. D. Hoegg, B. T. Manard, E. M. Wylie, K. J. Mathew, C. F. Ottenfeld and R. K. Marcus, J. Am. Soc. Mass Spectrom., 2019, 30(2), 278–288 CrossRef CAS PubMed.
- V. A. Azov, L. Mueller and A. A. Makarov, Mass Spectrom. Rev., 2020 DOI:10.1002/mas.21669.
- W. Giurlani, E. Berretti, M. Innocenti and A. Lavacchi, Coatings, 2020, 10(12) DOI:10.3390/coatings10121211.
- M. Czyzycki, M. Kokkoris and A. G. Karydas, J. Anal. At. Spectrom., 2020, 35(12), 2964–2973 RSC.
- M. D. Huang, N. Esser, K. Hinrichs, Q. Tan, J. Rappich, N. H. Nickel and T. Dittrich, Spectrochim. Acta, Part B, 2021, 177 DOI:10.1016/j.sab.2020.106050.
- S. Torrengo, D. Eichert, Y. Mazel, M. Bernard, Y. Menesguen, M. C. Lepy and E. Nolot, Spectrochim. Acta, Part B, 2020, 171 DOI:10.1016/j.sab.2020.105926.
- J. S. Wu, Q. Gao, G. X. Wei, J. S. Xiu, Z. Li and H. Q. Liu, Thin Solid Films, 2021, 722 DOI:10.1016/j.tsf.2021.138572.
- S. M. Liu, Q. Gao, J. S. Wu, Y. M. Xin, J. S. Xiu, Z. Li and H. Q. Liu, Appl. Opt., 2020, 59(29), 9152–9158 CrossRef PubMed.
- C. Herzig, J. Frank, A. K. Opitz, J. Fleig and A. Limbeck, Talanta, 2020, 217 DOI:10.1016/j.talanta.2020.121012.
- L. Lobo, B. Fernandez, M. Aranaz, A. F. Lorenzo, J. I. Martin-Carbajo and R. Pereiro, Spectrochim. Acta, Part B, 2020, 168 DOI:10.1016/j.sab.2020.105865.
- V. S. Smentkowski and S. Goswami, J. Vac. Sci. Technol., A, 2021, 39(3) DOI:10.1116/6.0000905.
- T. Miyayama and S. Iida, J. Vac. Sci. Technol., B: Nanotechnol. Microelectron.: Mater., Process., Meas., Phenom., 2020, 38(3) DOI:10.1116/6.0000096.
- A. Priebe, T. L. Xie, G. Burki, L. Petho and J. Michler, J. Anal. At. Spectrom., 2020, 35(6), 1156–1166 RSC.
- V. Spampinato, M. Dialameh, A. Franquet, C. Fleischmann, T. Conard, P. van der Heide and W. Vandervorst, Anal. Chem., 2020, 92(16), 11413–11419 CrossRef CAS PubMed.
- K. Higgins, M. Lorenz, M. Ziatdinov, R. K. Vasudevan, A. V. Ievlev, E. D. Lukosi, O. S. Ovchinnikova, S. V. Kalinin and M. Ahmadi, Adv. Funct. Mater., 2020, 30 DOI:10.1002/adfm.202001995.
- Z. J. Sha, E. Cali, G. Kerherve and S. J. Skinner, J. Mater. Chem. A, 2020, 8(40), 21273–21288 RSC.
- R. X. Yi, D. Y. Zhao, J. Oelmann, S. Brezinsek, M. Rasinski, M. Mayer, C. P. Dhard, D. Naujoks, L. W. Liu, J. L. Qu and W. X. Team, Appl. Surf. Sci., 2020, 532 DOI:10.1016/j.apsusc.2020.147185.
- Y. T. Liu, M. Lorenz, A. V. Ievlev and O. S. Ovchinnikova, Adv. Funct. Mater. DOI:10.1002/adfm.202002201.
- H. Shakeel, S. U. Haq, V. Contreras, Q. Abbas and A. Nadeem, Optik, 2020, 211 DOI:10.1016/j.ijleo.2020.164627.
- H. L. Hu, M. C. Qin, P. W. K. Fong, Z. W. Ren, X. J. Wan, M. Singh, C. J. Su, U. S. Jeng, L. Li, J. J. Zhu, M. J. Yuan, X. H. Lu, C. W. Chu and G. Li, Adv. Mater., 2021, 33 DOI:10.1002/adma.202006238.
- D. Remeteiova, S. Ruzickova, V. Mickova, M. Laubertova and R. Slezakova, Metals, 2020, 10(11) DOI:10.3390/met10111511.
- D. V. Babos, A. Cruz-Conesa, E. R. Pereira-Filho and J. M. Anzano, J. Hazard. Mater., 2020, 399 DOI:10.1016/j.jhazmat.2020.122831.
- J. A. Garcia, J. R. A. da Silva and E. R. Pereira, Microchem. J., 2021, 164 DOI:10.1016/j.microc.2021.106007.
- J. P. Castro, E. R. Pereira and R. Bro, J. Anal. At. Spectrom., 2020, 35(6), 1115–1124 RSC.
- K. R. Adair, M. N. Banis, Y. Zhao, T. Bond, R. Y. Li and X. L. Sun, Adv. Mater., 2020, 32 DOI:10.1002/adma.202002550.
- S. Smetaczek, M. Bonta, A. Wachter-Welzl, S. Taibl, R. Wagner, D. Rettenwander, J. Fleig and A. Limbeck, J. Anal. At. Spectrom., 2020, 35(5), 972–983 RSC.
- Y. Yamagishi, H. Morita, Y. Nomura and E. Igaki, ACS Appl. Mater. Interfaces, 2021, 13(1), 580–586 CrossRef CAS PubMed.
- W. D. Cheng, H. L. He, X. X. Liu, Y. F. Liu, Z. F. Zhang, S. H. Li, R. Zhang, X. X. Wang, Z. H. Wu and Z. J. Wu, Mater. Chem. Phys., 2021, 260 DOI:10.1016/j.matchemphys.2020.124157.
- A. K. Haridas, Q. A. Nguyen, T. Terlier, R. Blaser and S. L. Biswal, ACS Appl. Mater. Interfaces, 2021, 13(2), 2662–2673 CrossRef CAS PubMed.
- J. I. Robledo, J. J. Leani, S. Chauque, O. Camara, F. Y. Oliva and H. J. Sanchez, J. Anal. At. Spectrom., 2020, 35(12), 2948–2955 RSC.
- D. R. Wu, L. M. Housel, S. J. Kim, N. Sadique, C. D. Quilty, L. J. Wu, R. Tappero, S. L. Nicholas, S. Ehrlich, Y. M. Zhu, A. C. Marschilok, E. S. Takeuchi, D. C. Bock and K. J. Takeuchi, Energy Environ. Sci., 2020, 13(11), 4322–4333 RSC.
- L. Miner and D. Beauchemin, J. Anal. At. Spectrom., 2020, 35(7), 1295–1299 RSC.
- A. Wojcik, W. Kolkowski, I. Pasternak, W. Strupinski, S. Kozdra and P. P. Michalowski, J. Anal. At. Spectrom., 2021, 36(1), 178–184 RSC.
- W. J. Duan, ACS Appl. Electron. Mater., 2020, 2(11), 3711–3716 CrossRef CAS.
- S. Iida, T. Murakami, Y. Kurosawa, Y. Suzuri, G. L. Fisher and T. Miyayama, Rapid Commun. Mass Spectrom., 2020, 34(7) DOI:10.1002/rcm.8640.
- U. Makanda, A. Voinot, R. Kandel, Y. Wu, M. Leybourne and P. Wang, J. Anal. At. Spectrom., 2020, 35(11), 2672–2678 RSC.
- S. Lu, X. W. Wang, T. Z. Wang, X. R. Qin, X. L. Wang and Z. D. Jia, Appl. Sci., 2020, 10(7) DOI:10.3390/app10072617.
- F. Laborda, A. C. Gimenez-Ingalaturre, E. Bolea and J. R. Castillo, Spectrochim. Acta, Part B, 2020, 169 DOI:10.1016/j.sab.2020.105883.
- D. Mozhayeva and C. Engelhard, J. Anal. At. Spectrom., 2020, 35(9), 1740–1783 RSC.
- R. M. Galazzi, K. Chacón-Madrid, D. C. Freitas, L. F. da Costa and M. A. Z. Arruda, Rapid Commun. Mass Spectrom., 2020, 34 DOI:10.1002/rcm.8726.
- J. Wojcieszek, J. Jimenez-Lamana, L. Ruzik, J. Szpunar and M. Jarosz, Nanomaterials, 2020, 10(8) DOI:10.3390/nano10081480.
- D. Semenova and Y. E. Silina, Nanomaterials, 2019, 9(12) DOI:10.3390/nano9121673.
- D. Bleiner, L. Juha and D. Qu, J. Anal. At. Spectrom., 2020, 35(6), 1051–1070 RSC.
- J. Baumann, Y. Kayser and B. Kanngiesser, Phys. Status Solidi B, 2020, 258 DOI:10.1002/pssb.202000471.
- F. Bennet, A. Muller, J. Radnik, Y. Hachenberger, H. Jungnickel, P. Laux, A. Luch and J. Tentschert, J. Visualized Exp., 2020,(163) DOI:10.3791/61758.
- S. Heilgeist, R. Sekine, O. Sahin and R. A. Stewart, Water, 2021, 13(5) DOI:10.3390/w13050734.
- O. Pryshchepa, P. Pomastowski and B. Buszewski, Adv. Colloid Interface Sci., 2020, 284 DOI:10.1016/j.cis.2020.102246.
- S. Meyer, R. G. de Vega, X. X. Xu, Z. Q. Du, P. A. Doble and D. Clases, Anal. Chem., 2020, 92(22), 15007–15016 CrossRef CAS PubMed.
- S. Alavi, X. M. Guo, S. M. Javid, A. Ebrahimi and J. Mostaghimi, Anal. Chem., 2020, 92(17), 11786–11794 CrossRef CAS PubMed.
- A. Gundlach-Graham and K. Mehrabi, J. Anal. At. Spectrom., 2020, 35(9), 1727–1739 RSC.
- V. Kinnunen, S. Peramaki and R. Matilainen, Spectrochim. Acta, Part B, 2021, 177 DOI:10.1016/j.sab.2021.106104.
- I. Strenge and C. Engelhard, J. Anal. At. Spectrom., 2020, 35(1), 84–99 RSC.
- A. Williams, A. Al Hejami and D. Beauchemin, J. Anal. At. Spectrom., 2020, 35(10), 2165–2170 RSC.
- A. Williams and D. Beauchemin, Anal. Chem., 2020, 92(19), 12778–12782 CrossRef CAS PubMed.
- M. Tharaud, P. Louvat and M. F. Benedetti, Anal. Bioanal. Chem., 2021, 413, 923–933, DOI:10.1007/s00216-020-03048-y.
- S. Yamashita, A. Miyake and T. Hirata, J. Anal. At. Spectrom., 2020, 35(12), 2834–2839 RSC.
- A. Rua-Ibarz, E. Bolea-Fernandez, G. Pozo, X. Dominguez-Benetton, F. Vanhaecke and K. Tirez, J. Anal. At. Spectrom., 2020, 35(9), 2023–2032 RSC.
- S. Cuello-Nunez, I. Abad-Alvaro, D. Bartczak, M. E. D. Busto, D. A. Ramsay, F. Pellegrino and H. Goenaga-Infante, J. Anal. At. Spectrom., 2020, 35(9), 1832–1839 RSC.
- D. Rosenkranz, F. L. Kriegel, E. Mavrakis, S. A. Pergantis, P. Reichardt, J. Tentschert, N. Jakubowski, P. Laux, U. Panne and A. Luch, Anal. Chim. Acta, 2020, 1099, 16–25 CrossRef CAS PubMed.
- Y. U. Hachenberger, D. Rosenkranz, F. L. Kriegel, B. Krause, R. Matschass, P. Reichardt, J. Tentschert, P. Laux, N. Jakubowski, U. Panne and A. Luch, Materials, 2020, 13(6) DOI:10.3390/ma13061447.
- O. Mestek, M. Loula, A. Kana and M. Vosmanska, Talanta, 2020, 210 DOI:10.1016/j.talanta.2019.120665.
- F. Laborda, C. Trujillo and R. Lobinski, Talanta, 2021, 221 DOI:10.1016/j.talanta.2020.121486.
- A. Barber, S. Kly, M. G. Moffitt, L. Rand and J. F. Ranville, Environ. Sci.: Nano, 2020, 7(2), 514–524 RSC.
- L. Maknun, J. Sumranjit and A. Siripinyanond, RSC Adv., 2020, 10(11), 6423–6435 RSC.
- S. Yamashita, M. Ishida, T. Suzuki, M. Nakazato and T. Hirata, Spectrochim. Acta, Part B, 2020, 169 DOI:10.1016/j.sab.2020.105881.
- M. Velimirovic, S. Wagner, F. A. Monikh, T. Uusimaki, R. Kaegi, T. Hofmann and F. von der Kammer, Talanta, 2020, 215 DOI:10.1016/j.talanta.2020.120921.
- M. Velimirovic, S. Wagner, R. Koeber, T. Hofmann and F. von der Kammer, Nanoimpact, 2020, 19 DOI:10.1016/j.impact.2020.100233.
- D. Ojeda, M. V. Taboada-Lopez, E. Bolea, J. Perez-Arantegui, P. Bernejo-Barrera, A. Moreda-Pineiro and F. Laborda, Anal. Chim. Acta, 2020, 1122, 20–30 CrossRef CAS PubMed.
- B. Li, S. L. Chua, A. L. Ch'ng, D. Y. Yu, S. P. Koh, H. Phang and P. K. T. Chiew, Anal. Bioanal. Chem., 2020, 412, 5499–5512, DOI:10.1007/s00216-020-02770-x.
- F. Loosli, J. J. Wang, M. Sikder, K. Afshinnia and M. Baalousha, Sci. Total Environ., 2020, 715 DOI:10.1016/j.scitotenv.2020.136927.
- A. Sanchez-Cachero, S. Lopez-Sanz, N. R. Farinas, A. Rios and R. Martin-Doimeadios, Talanta, 2021, 222 DOI:10.1016/j.talanta.2020.121513.
- P. Tan, J. J. Yang and V. Nischwitz, J. Anal. At. Spectrom., 2020, 35(3), 548–559 RSC.
- M. A. Alghuthaymi, K. A. Abd-Elsalam, A. Shami, E. Said-Galive, E. V. Shtykova and A. V. Naumkin, J. Fungi, 2020, 6(2) DOI:10.3390/jof6020051.
- M. J. Cao, B. Li, M. Y. Guo, Y. Liu, L. L. Zhang, Y. L. Wang, B. Hu, J. Y. Li, D. S. Sutherland, L. M. Wang and C. Y. Chen, Nanotoxicology, 2021, 15, 131–144, DOI:10.1080/17435390.2020.1860264.
- P. Honicke, A. Andrle, Y. Kayser, K. V. Nikolaev, J. Probst, F. Scholze, V. Soltwisch, T. Weimann and B. Beckhoff, Nanotechnology, 2020, 31(50) DOI:10.1088/1361-6528/abb557.
- K. V. Nikolaev, V. Soltwisch, P. Honicke, F. Scholze, J. de la Rie, S. N. Yakunin, I. A. Makhotkin, R. W. E. van de Kruijs and F. Bijkerk, J. Synchrotron Radiat., 2020, 27, 386–395 CrossRef CAS PubMed.
- S. Jung, J. Lee, H. Cho, T. Kim and S. J. Ye, IEEE Trans. Nucl. Sci., 2020, 67(11), 2311–2320 CAS.
- S. Reich, J. Gottlicher, A. Ziefuss, R. Streubel, A. Letzel, A. Menzel, O. Mathon, S. Pascarelli, T. Baumbach, M. Zuber, B. Gokce, S. Barcikowski and A. Plech, Nanoscale, 2020, 12(26), 14011–14020 RSC.
- A. Brandt, B. Gomez-Nieto, J. Friedland, R. Guttel and K. Leopold, Spectrochim. Acta, Part B, 2020, 173 DOI:10.1016/j.sab.2020.105976.
- A. Brandt, K. Kees and K. Leopold, J. Anal. At. Spectrom., 2020, 35(11), 2536–2544 RSC.
- X. L. Cheng, T. T. Wang, Z. B. Yin and W. Hang, J. Anal. At. Spectrom., 2020, 35(5), 927–932 RSC.
- J. Kruszewska, J. Sikorski, J. Samsonowicz-Gorski and M. Matczuk, Anal. Bioanal. Chem., 2020, 412, 8145–8153, DOI:10.1007/s00216-020-02948-3.
- A. Wroblewska and M. Matczuk, Electrophoresis, 2020, 41(5–6), 394–398 CrossRef CAS PubMed.
- R. Yang, Y. Lin, B. Y. Liu, Y. B. Su, Y. F. Tian, X. D. Hou and C. B. Zheng, Anal. Chem., 2020, 92(5), 3528–3534 CrossRef CAS PubMed.
- X. Gao, H. J. Pan, Y. C. Han, L. X. Feng, J. P. Xiong, S. Z. Luo and H. M. Li, Anal. Chim. Acta, 2021, 1148 DOI:10.1016/j.aca.2020.12.072.
- D. Metarapi and J. T. van Elteren, J. Anal. At. Spectrom., 2020, 35(4), 784–793 RSC.
- Q. L. Chan, M. Entezarian, J. S. Zhou, R. Osterloh, Q. Huang, M. Ellefson, B. Mader, Y. Liu and M. Swierczek, J. Membr. Sci., 2020, 599 DOI:10.1016/j.memsci.2020.117822.
- N. D. Donahue, E. R. Francek, E. Kiyotake, E. E. Thomas, W. Yang, L. Wang, M. S. Detamore and S. Wilhelm, Anal. Bioanal. Chem., 2020, 412, 5205–5216, DOI:10.1007/s00216-020-02783-6.
- S. Fernandez-Trujillo, M. Jimenez-Moreno, A. Rios and R. D. R. Martin-Doimeadios, J. Anal. At. Spectrom., 2021, 36(3), 528–534 RSC.
- Y. Gao, X. J. Ge, L. Chen, R. X. Liu, Y. T. Guo and G. L. Ge, Chin. J. Anal. Chem., 2021, 49(2), 271–281 Search PubMed.
- O. Geiss, I. Bianchi, C. Senaldi, G. Bucher, E. Verleysen, N. Waegeneers, F. Brassinne, J. Mast, K. Loeschner, J. Vidmar, F. Aureli, F. Cubadda, A. Raggi, F. Iacoponi, R. Peters, A. Undas, A. Muller, A. K. Meinhardt, E. Walz, V. Graf and J. Barrero-Moreno, Food Control, 2021, 120 DOI:10.1016/j.foodcont.2020.107550.
- B. Gomez-Gomez, M. Corte-Rodriguez, M. T. Perez-Corona, J. Bettmer, M. Montes-Bayon and Y. Madrid, Anal. Chim. Acta, 2020, 1128, 116–128 CrossRef CAS PubMed.
- B. Gomez-Gomez, M. T. Perez-Corona and Y. Madrid, Anal. Chim. Acta, 2020, 1100, 12–21 CrossRef CAS PubMed.
- X. L. He, H. T. Zhang, H. L. Shi, W. Y. Liu and E. Sahle-Demessie, J. Am. Soc. Mass Spectrom., 2020, 31(10), 2180–2190 CrossRef CAS PubMed.
- I. Jreije, A. Azimzada, M. Hadioui and K. J. Wilkinson, Molecules, 2020, 25(23) DOI:10.3390/molecules25235516.
- A. Keri, A. Sapi, D. Ungor, D. Sebok, E. Csapo, Z. Konya and G. Galbacs, J. Anal. At. Spectrom., 2020, 35(6), 1139–1147 RSC.
- R. P. Lamsal, A. Hineman, C. Stephan, S. Tahmasebi, S. Baranton, C. Coutanceau, G. Jerkiewicz and D. Beauchemin, Anal. Chim. Acta, 2020, 1139, 36–41 CrossRef CAS PubMed.
- R. P. Lamsal, M. S. E. Houache, A. Williams, E. Baranova, G. Jerkiewicz and D. Beauchemin, Anal. Chim. Acta, 2020, 1120, 67–74 CrossRef CAS PubMed.
- B. R. Li, H. Tang, R. Q. Yu and J. H. Jiang, Anal. Chem., 2020, 92(3), 2379–2382 CrossRef CAS PubMed.
- K. Mehrabi, R. Kaegi, D. Gunther and A. Gundlach-Graham, Environ. Sci.: Nano, 2021, 8, 1211–1225, 10.1039/D0EN01066A.
- J. Nelson, A. Saunders, L. Poirier, E. Rogel, C. Ovalles, T. Rea and F. Lopez-Linares, J. Nanopart. Res., 2020, 22(10) DOI:10.1007/s11051-020-05033-z.
- R. X. Shi, Y. P. Zhao, P. Guan, W. X. Liang and H. Xun, Chin. J. Anal. Chem., 2020, 48(4), 523–529 Search PubMed.
- E. Verleysen, N. Waegeneers, F. Brassinne, S. De Vos, I. O. Jimenez, S. Mathioudaki and J. Mast, Nanomaterials, 2020, 10(3) DOI:10.3390/nano10030592.
- Y. Q. Xing, J. Han, X. Wu, D. T. Pierce and J. X. J. Zhao, Analyst, 2020, 145(24), 7932–7940 RSC.
- L. Zazzera, Q. L. Chan, J. Stomberg, A. Simpson, C. Loesch, D. LeHuu, D. Muradian, U. R. K. Lagudu and B. Mader, ECS J. Solid State Sci. Technol., 2021, 10(3), 034009 CrossRef CAS.
- Aarzoo, S. Naqvi, N. B. Agarwal, M. P. Singh and M. Samim, RSC Adv., 2021, 11(3), 1850–1861 RSC.
- C. Adelantado, A. Rios and M. Zougagh, Talanta, 2020, 219 DOI:10.1016/j.talanta.2020.121385.
- C. Aguzzi, A. Donnadio, G. Quaglia, L. Latterini, C. Viseras and V. Ambrogi, ACS Appl. Nano Mater., 2019, 2(10), 6575–6584 CrossRef CAS.
- M. L. Avramescu, M. Chenier, S. Palaniyandi and P. E. Rasmussen, J. Nanopart. Res., 2020, 22(8) DOI:10.1007/s11051-020-04949-w.
- C. Ferrag, S. P. Li, K. Jeon, N. M. Andoy, R. M. A. Sullan, S. Mikhaylichenko and K. Kerman, Colloids Surf., B, 2021, 197 DOI:10.1016/j.colsurfb.2020.111397.
- N. Hazel, J. Orejas and S. J. Ray, Spectrochim. Acta, Part B, 2021, 176 DOI:10.1016/j.sab.2020.106040.
- K. Hoffmann, S. Bouchet, I. Christl, R. Kaegi and R. Kretzschmar, Environ. Sci.: Nano, 2020, 7(4), 1163–1178 RSC.
- M. S. Jimenez, M. Bakir, D. Isabal, M. T. Gomez, J. Perez-Arantegui, J. R. Castillo and F. Laborda, Anal. Bioanal. Chem., 2021, 413, 1689–1699, DOI:10.1007/s00216-020-03132-3.
- J. L. Lyu, V. Geertsen, C. Hamon and D. Constantin, Nanoscale Adv., 2020, 2(10), 4522–4528 RSC.
- S. Muramoto and J. Bennett, Surf. Interface Anal., 2020, 52, 717–724, DOI:10.1002/sia.6860.
- H. Y. Nie, E. Romanovskaia, V. Romanovski, J. Hedberg and Y. S. Hedberg, Biointerphases, 2021, 16(2) DOI:10.1116/6.0000910.
- A. Priebe, J. P. Barnes, T. E. J. Edwards, E. Huszar, L. Petho and J. Michler, Anal. Chem., 2020, 92(18), 12518–12527 CrossRef CAS PubMed.
- P. Purohit, F. J. Fortes and J. J. Laserna, Anal. Chem., 2021, 93(4), 2635–2643 CrossRef CAS PubMed.
- H. K. Shon, J. G. Son, S. Joh, J. H. Moon and T. G. Lee, Biointerphases, 2020, 15(3) DOI:10.1116/6.0000106.
- M. Sikder, J. J. Wang, B. A. Poulin, M. M. Tfaily and M. Baalousha, Environ. Sci.: Nano, 2020, 7(11), 3318–3332 RSC.
- R. Unterumsberger, P. Honicke, Y. Kayser, B. Pollakowski-Herrmann, S. Gholhaki, Q. M. Guo, R. E. Palmer and B. Beckhoff, J. Anal. At. Spectrom., 2020, 35(5), 1022–1033 RSC.
- M. Vlcnovska, A. Stossova, M. Kuchynka, V. Dillingerova, H. Polanska, M. Masarik, R. Hrstka, V. Adam, V. Kanicky, T. Vaculovic and M. Vaculovicova, Molecules, 2021, 26(3) DOI:10.3390/molecules26030630.
- X. Wang, X. Chen, R. H. Zhou, P. Y. Hu, K. Huang and P. P. Chen, Anal. Chem., 2021, 93(8), 3889–3897 CrossRef CAS PubMed.
- X. X. Zhou, L. W. Jiang, D. J. Wang, S. He, C. J. Li and B. Yan, Anal. Chem., 2020, 92(7), 4765–4770 CrossRef CAS PubMed.
- J. A. Stephens, M. N. Ducea, D. J. Killick and J. Ruiz, J. Archaeol. Sci., 2021, 127 DOI:10.1016/j.jas.2021.105334.
- S. Reguer, F. Kergourlay, E. Foy, D. Neff, D. Vantelon, M. Cotte, F. Mirambet and P. Dillmann, J. Anal. At. Spectrom., 2020, 35(10), 2358–2368 RSC.
- X. S. Bai, H. Allegre, M. Gosselin, P. Dillmann, M. Lopez, F. Tereygeol and V. Detalle, Spectrochim. Acta, Part B, 2020, 172 DOI:10.1016/j.sab.2020.105964.
- S. Wallace, N. Smith and N. Nerantzis, Archaeometry, 2020, 63, 343–371, DOI:10.1111/arcm.12595.
- A. G. Sinner, M. Ferrante, S. Nisi and P. R. Trincherini, Archaeol. Anthropol. Sci., 2020, 12(7) DOI:10.1007/s12520-020-01073-7.
- E. Berard, C. Pecheyran, P. Dillmann, S. Leroy, E. Vega, A. Williams, C. Verna and V. Toureille, J. Anal. At. Spectrom., 2020, 35(11), 2582–2593 RSC.
- J. Brocchieri, R. Vitale and C. Sabbarese, Nucl. Instrum. Methods Phys. Res., Sect. B, 2020, 479, 93–101 CrossRef CAS.
- D. Chen, Y. D. Yang, J. Du, X. Tang and W. G. Luo, Heritage Sci., 2020, 8(1) DOI:10.1186/s40494-020-00413-z.
- I. S. Stepanov, L. Weeks, K. A. Franke, B. Overlaet, O. Alard, C. M. Cable, Y. Y. Al Aali, M. Boraik, H. Zein and P. Grave, J. Archaeol. Sci., 2020, 120 DOI:10.1016/j.jas.2020.105192.
- M. Necemer, P. Kump and T. Lazar, Archaeometry, 2020, 62, 965–973, DOI:10.1111/arcm.12579.
- B. Liss, T. E. Levy and J. M. D. Day, J. Archaeol. Sci., 2020, 122 DOI:10.1016/j.jas.2020.105227.
- E. Mista-Jakubowska, R. C. Blonska, W. Duczko, A. M. Gojska, P. Kalbarczyk, G. Zabinski and K. Trela, Archaeol. Anthropol. Sci., 2019, 11(12), 6705–6723 CrossRef.
- M. Hrnjic, G. A. Hagen-Peter, T. Birch, G. H. Barfod, S. M. Sindbaek and C. E. Lesher, Nucl. Instrum. Methods Phys. Res., Sect. B, 2020, 478, 11–20 CrossRef CAS.
- L. Gentelli, Archaeometry, 2021, 63, 156–172, DOI:10.1111/arcm.12628.
- V. S. Felix, M. O. Pereira, R. P. Freitas, P. J. M. Aranha, P. C. S. Heringer, M. J. Anjos and R. T. Lopes, Appl. Radiat. Isot., 2020, 166 DOI:10.1016/j.apradiso.2020.109409.
- B. Borg, M. Dunn, A. Ang and C. Villis, J. Cult. Herit., 2020, 44, 239–259 CrossRef.
- J. P. Dudeja, Lasers Eng., 2021, 48(1–3), 1–15 CAS.
- E. Pouyet, N. Barbi, H. Chopp, O. Healy, A. Katsaggelos, S. Moak, R. Mott, M. Vermeulen and M. Walton, X-Ray Spectrom., 2020, 50, 243–412, DOI:10.1002/xrs.3173.
- L. Watteeuw, M. Van Bos, T. Gersten, B. Vandermeulen and H. Hameeuw, Microchem. J., 2020, 155 DOI:10.1016/j.microc.2019.104582.
- J. Orsilli, A. Galli, L. Bonizzoni and M. Caccia, Appl. Sci., 2021, 11(4) DOI:10.3390/app11041446.
- V. Lazic, R. Fantoni, S. Falzone, C. Gioia and E. M. Loreti, Spectrochim. Acta, Part B, 2020, 168 DOI:10.1016/j.sab.2020.105853.
- M. Romani, G. Capobianco, L. Pronti, F. Colao, C. Seccaroni, A. Puiu, A. C. Felici, G. Verona-Rinati, M. Cestelli-Guidi, A. Tognacci, M. Vendittelli, M. Mangano, A. Acconci, G. Bonifazi, S. Serranti, M. Marinelli and R. Fantoni, Microchem. J., 2020, 156 DOI:10.1016/j.microc.2020.104920.
- Y. P. Yin, Z. R. Yu, D. X. Sun, M. G. Su, Z. Wang, Z. W. Shan, W. W. Han, B. M. Su and C. Z. Dong, Anal. Methods, 2021, 13(11), 1381–1391 RSC.
- I. M. Cortea, L. Ghervase, L. Ratoiu, M. Dinu and R. Radvan, Heritage Sci., 2020, 8(1) DOI:10.1186/s40494-020-00401-3.
- M. Bicchieri, P. Biocca, C. Caliri and F. P. Romano, Microchem. J., 2020, 157 DOI:10.1016/j.microc.2020.104844.
- M. Alfeld, V. Gonzalez and A. van Loon, X-Ray Spectrom., 2021, 50, 351–357, DOI:10.1002/xrs.3198.
- B. Wagner and A. Czajka, Talanta, 2021, 222 DOI:10.1016/j.talanta.2020.121520.
- C. Bouvier, H. Glanville, L. de Viguerie, C. Merucci, P. Walter and A. Brunelle, Anal. Chem., 2021, 93(10), 4463–4471 CrossRef CAS PubMed.
- C. Invernizzi, G. Fiocco, M. Iwanicka, P. Targowski, A. Piccirillo, M. Vagnini, M. Licchelli, M. Malagodi and D. Bersani, Coatings, 2021, 11(1) DOI:10.3390/coatings11010029.
- S. Bottura-Scardina, A. Brunetti, C. Bottaini and C. Miguel, Eur. Phys. J. Plus, 2021, 136(3) DOI:10.1140/epjp/s13360-021-01326-x.
- C. Vlachou-Mogire, P. Moretti, L. Monico, A. Chieli, M. Iwanicka, P. Targowski, V. Detalle, E. Bourguignon, K. Laclavetine, F. Mirambet, T. Tong and S. Pinchin, Microchem. J., 2020, 156 DOI:10.1016/j.microc.2020.104797.
- P. Niedzielski, M. Krueger and D. Brandherm, Mater. Manuf. Processes, 2020, 35(13), 1455–1460 CrossRef CAS.
- F. Q. Ruan, L. Hou, T. L. Zhang and H. Li, Analyst, 2021, 146(3), 1023–1031 RSC.
- E. Odelli, V. Palleschi, S. Legnaioli, F. Cantini and S. Raneri, Spectrochim. Acta, Part B, 2020, 172 DOI:10.1016/j.sab.2020.105966.
- M. Vannoorenberghe, T. Van Acker, J. Belza, D. Teetaert, P. Crombe and F. Vanhaecke, J. Anal. At. Spectrom., 2020, 35(11), 2686–2696 RSC.
- L. Medeghini, M. Fayek, S. Mignardi, F. Coletti, A. Contino and C. De Vito, Microchem. J., 2020, 154 DOI:10.1016/j.microc.2019.104519.
- J. G. Inanez, J. Bettencourt, I. P. Coelho, A. Teixeira, G. Arana, K. Castro and U. Sanchez-Garmendia, Archaeol. Anthropol. Sci., 2020, 12(8) DOI:10.1007/s12520-020-01109-y.
- S. Ichikawa, T. Matsumoto, T. Nakamura and T. Kurisaki, X-Ray Spectrom., 2020, 49(4), 515–532 CrossRef CAS.
- R. Bugoi, C. Talmatchi, C. Haita and D. Ceccato, Nucl. Instrum. Methods Phys. Res., Sect. B, 2020, 477, 80–86 CrossRef CAS.
- M. Ramacciotti, G. Gallello, D. Navarro-Martos, A. Domenech-Carbo, C. Roldan, E. Hernandez, S. Garrigues and A. Pastor, Appl. Clay Sci., 2020, 198 DOI:10.1016/j.clay.2020.105857.
- A. S. Maltsev, G. V. Pashkova, R. Fernandez-Ruiz, E. I. Demonterova, A. N. Shuliumova, N. N. Umarova, D. L. Shergin, M. M. Mukhamedova, V. M. Chubarov and E. A. Mikheeva, Spectrochim. Acta, Part B, 2021, 175 DOI:10.1016/j.sab.2020.106012.
- I. Liritzis, V. Xanthopoulou, E. Palamara, I. Papageorgiou, I. Iliopoulos, N. Zacharias, A. Vafiadou and A. G. Karydas, J. Cult. Herit., 2020, 46, 61–81 CrossRef.
- G. S. Franci, Appl. Spectrosc., 2020, 74(3), 314–322 CrossRef PubMed.
- T. Z. Ferri, S. Roncevic, G. L. Vrkljan and A. Konestra, J. Cult. Herit., 2020, 43, 12–25 CrossRef.
- A. Gianoncelli, S. Raneri, S. Schoeder, T. Okbinoglu, G. Barone, A. Santostefano and P. Mazzoleni, Microchem. J., 2020, 154 DOI:10.1016/j.microc.2020.104629.
- V. Comite, M. Andreoli, D. Atzei, D. Barca, M. Fantauzzi, M. F. La Russa, A. Rossi, V. Guglielmi and P. Fermo, Appl. Sci., 2020, 10(21) DOI:10.3390/app10217523.
- S. Cagno, G. van der Snickt, S. Legrand, J. Caen, M. Patin, W. Meulebroeck, Y. Dirkx, M. Hillen, G. Steenackers, A. Rousaki, P. Vandenabeele and K. Janssens, X-Ray Spectrom., 2021, 50, 293–309, DOI:10.1002/xrs.3185.
| This journal is © The Royal Society of Chemistry 2021 |