Atomic spectrometry update: review of advances in atomic spectrometry and related techniques

E. Hywel Evans *a, Jorge Pisonero b, Clare M. M. Smith c and Rex N. Taylor d
aSchool of Geography, Earth, and Environmental Sciences, University of Plymouth, Drake Circus, Plymouth, UK PL4 8AA. E-mail: hevans@plymouth.ac.uk
bUniversity of Oviedo, Faculty of Science, Department of Physics, c/ Calvo Sotelo s/n, 33006 Oviedo, Spain
cSt. Ambrose High School, Blair Road, Coatbridge, Lanarkshire, UK ML5 2EW
dOcean and Earth Science, University of Southampton, NOC, Southampton, UK SO14 3ZH

Received 27th March 2018

First published on 23rd April 2018


Abstract

This review of 166 references covers developments in ‘Atomic Spectrometry’ published in the twelve months from November 2016 to November 2017 inclusive. It covers atomic emission, absorption, fluorescence and mass spectrometry, but excludes material on speciation and coupled techniques which is included in a separate review. It should be read in conjunction with the previous review and the other related reviews in the series.1–6 A critical approach to the selection of material has been adopted, with only novel developments in instrumentation, techniques and methodology being included. Novel trends are evenly spread across the techniques this year, with no single ‘bandwagon’ rolling ahead of the rest. The most significant development seems to be the advance of 2D and 3D imaging using LA-ICP-MS, and the challenges associated with how to process the large amounts of data produced. In this context, there were some useful reports of user-friendly software for imaging large datasets and methods of compressed sensing to reduce the amount of data in the first place. Single particle analysis using ICP-MS is now well established in the literature which, as well as being a useful technique for particle analysis in its own right, it also offer interesting insights into the mechanisms of atomisation and ionisation in the ICP. New methods of vapour generation using a plasma as both vapour generation and excitation/ionisation cell show some promise for the analysis of single drops of solution using a relatively simple, miniaturised instrumental set-up. Methods for IR-ICP-MS of non-radiogenic isotopes continues to grow, with more elements being added to the list in this review period.


1. Sample introduction

1.1 Liquids

1.1.1 Sample pre-treatment.
1.1.1.1 Solid phase extraction. He et al.7 reviewed (190 references) the use of advanced functional nanomaterials (AFNs) for use in SPE and determination by ICP-MS. The review focuses on AFNs with high capacity, good selectivity and fast adsorption/desorption dynamics. Several tables of data for the various types of AFNs include figures of merit and other information ordered by analyte, and another listing different types of applications. An example of using one of the AFNs described was given by Ahmad et al.,8 who synthesised magnetite graphene oxide (m-GO) for use in dispersive magnetic SPE of Au from environmental samples. Detection was by means of MIP-AES to yield an LOD of 0.005 mg L−1 and spike recoveries from fresh and seawater between 97.0 and 101%. A similar approach was developed by Zarezade et al.,9 but using magnetic ion-imprinted polymers (IIPs) for SPE of Cd. They prepared the magnetic IIP using Cd2+ as the template ion for the IIP synthesis and Fe3O4@SiO2@MPTS nanoparticles as a substrate to which the IIP was attached. The LOD using FAAS was 0.02 mg L−1 with an LDR of 0.1 to 500 mg L−1.
1.1.1.2 Elemental tagging. The use of elemental tags, or intrinsic hetero-atoms, for the quantitation of proteins, peptides and nucleic acids is an area of research which has matured into a routine procedure. A number of approaches have been adopted, usually involving ICP-MS as the analytical technique, often coupled with LC for separation. The target compound is usually determined by quantitation of an element which is either already part of the molecule (e.g. the metal moiety in an enzyme or P in a phosphoprotein), or which is incorporated by means of a chemical reaction, in single or multiple steps (e.g. a sandwich immunoassay or bio-conjugate tagging). Quantitation can then be performed either by straightforward calibration or by ID-MS. Chahrour and Malone10 reviewed both of these approaches, using ICP-MS, and discussed their implications and implementation in the field of quantitative proteomics. Liu et al.11 reviewed (245 references) the use of ICP-MS and exogenous tags for the quantitation of biomolecules. The comprehensive review separated research along the lines mentioned above, also including the use of LA for direct analysis of solids. Several useful tables, summarising the literature, were included with publications ordered by date so the historical development of the approach could be seen at a glance.

The use of an immunoassay to tag proteins has become popular in recent years. Presumably this is consequence of the ready availability of existing procedures and instrumentation, with the requirement only to make sure that one or other of the reagents contains an element that is easily detectable using ICP-MS. One approach, which has been commercialised, is to use a tagging element that can form an NP (e.g. Au) which considerably amplifies the analytical signal. Zhang et al.12 used this approach for the determination HepG2 tumour cells. They combined this approach with a DNA hybridisation reaction and magnetic bead (MB) conjugation to effect further amplification and separation respectively. They achieved an LOD of 15 HepG2 cells, a linear range of 40 to 8000 cells, and the RSD for seven replicates of 200 cells was 8.7%. This type of multi-conjugate approach can get extremely complicated. In this case the final conjugated molecule consisted of an [AuNP–streptavidin–DNAconcatamer]n–HepG2–antiEpCAM–MB ‘sandwich’. The same group used a similar approach,13 except without the DNA amplification, and using separate specific tags to simultaneously determine two types of cancer cells, MCF-7 and HepG2. In this case the tags were CdSeQD–antiASPGR for MCF-7 and AuNP–antiMUC1. This illustrates several important aspects of the approach: the specificity and multiplexing advantage of using an immunoassay approach; amplification by using NPs or nucleic acid hybridisation/PCR; and the necessity to separate the target conjugate molecule from the sample matrix prior to analysis, hence the ubiquitous use of magnetic beads. Variations on these themes have been applied by numerous other workers, sometimes using a different metal, e.g. CsNPs;14 or incorporating a multifunctional probe, e.g. a fluorescent dye in combination with Ln-containing upconversion NPs to facilitate simultaneous imaging.15,16

Tagging proteins without using an immunoassay can also be achieved by chelation or covalent bonding to one or more functional groups. The problem with chelation is that it tends not to be very specific. However, Tang et al.17 used a GaNP linked to Fe MNPs through a photocleavable o-nitrobenzyl ether to allow magnetic separation and subsequent UV-induced release of the Ga-tagged molecules. The GaNP was chosen because of the highly specific interaction between Ga and the phosphate ester on phosphorylated proteins (Kd = 3.08 × 106). This allowed the absolute quantitation of phosphotyrosine down to 30 amol using ID-ICP-MS. The method was validated using VNQIGTLSEpYIK, VNQIGTLpSEpYIK, and extracellular regulated protein kinase 1 peptide (–pTEpY–) standards, with recovery of more than 96% (n = 5). It was applied to the absolute quantitation of phosphotyrosine in human breast cancer MCF-7 cells. Of course, it is also possible to quantify phosphopeptides without tagging at all, simply by determining the P in the phosphate group. This approach, taken originally and still being developed, was exemplified by Li et al.18 who used hollow fiber (HF) supported TiO2 monolithic microextraction combined with capillary LC for separation of the matrix prior to absolute quantitation of phosphopeptides by ICP-MS. A similar outcome can be achieved by determining the intrinsic S contained in proteins. Calderon-Celis et al.19 have developed this approach using IC-ICP-QQQ-MS with post-column addition of a 34S spike for quantitation by IDMS.

Several of the advantages of immunoassays are present in methods developed for the quantitation of nucleic acids. In particular, the ability to amplify using the PCR and the inherent specificity of using hybridisation probes for tagging. Amplification was made use of by Chen et al.,20 but with a significant difference to the usual method; they used an enzyme-free strand displacement amplification procedure instead of the usual PCR. A DNA probe (P3), containing 6 T bases at both the 3′ and 5′ ends, was designed so that it formed a hairpin structure in the presence of Hg2+ (T–Hg2+–T). The hairpin loop was complementary to the target DNA so that hybridisation took place in its presence, the hairpin was broken and free Hg2+ released which was monitoring by AFS. Then, another DNA strand (P4), hybridised with target–P3 complex and free target DNA, triggering another reaction cycle leading to an increase in the AFS signal. The method was further developed for protein detection by the use of an aptamer–P2 arched structure. The P2 DNA strand was displaced by the target protein binding with the aptamer and then entered the amplification cycle previously described as the target DNA strand. This method allowed AFS detection of DNA and thrombin down to the 0.3 aM and 0.1 aM level, respectively, with high selectivity.

1.1.2 Nebulisation. Cao et al.21 developed a gas pressure-assisted sample introduction system for ICP-MS. The system was described as environmentally-friendly due to its quoted high nebulisation efficiencies, small sample volume requirements and low waste emission. The system was demonstrated to continuously nebulise 10% NaCl solution for more than an hour without clogging. The analytical performance in terms of sensitivity, LODs, precision, long-term stability, doubly charged and oxide ion levels, nebulisation efficiencies, and matrix effects of the sample introduction system were evaluated. The reported experimental results indicated that the performance of the sample introduction system, was equivalent to, or better than that obtained by conventional systems.

A flow focusing pneumatic nebuliser (FFPN) working at low liquid flow rates was evaluated for analysis of slurried samples by Ar–He MIP-OES.22 The results obtained were compared with a commercially available V-groove Babington type nebuliser. The LODs were 0.9, 0.2, 0.3, 0.2, 0.3, 0.1, 0.2, 0.4, and 0.3 ng mL−1 for Ba, Ca, Cd, Cu, Fe, Mg, Mn, Pb and Sr, respectively. RSD values from 5% to 8% were obtained for micro-slurry sampling analysis. Accuracy was demonstrated through satisfactory analysis of the CRMs NRCC DOLT-2, GBW 07302 and SRM 2710. Slurry concentration up to 3% m/v (particles <20 μm), prepared in 10% m/v HCl with ultrasonic agitation were used with standard additions calibration. The nebuliser exhibited no clogging problems throughout the study.

1.1.3 Single particle analysis. Single particle ICP-MS (spICP-MS) has evolved rapidly as a quantitative method for determining NP size and number concentration at environmentally relevant exposure levels. Central to the application of spICP-MS is a calibration approach based on the measured transport efficiency (TE) and the response of ionic standards. Liu et al.23 carried out a systematic study of the accuracy, precision and robustness of spICP-MS using the NIST RM 8017 (PVP coated nominal 75 nm silver NPs). The authors report statistically significant differences in frequency-based and size-based measures of TE and demonstrate that the size-based measure is more robust and yields more accurate results for a silver NP reference material. As the frequency-based method is more widely applied, this finding is significant. Acidified ionic standards were found to improve the measurement of ICP-MS Ag response, without degrading the accuracy of the results for silver NP suspensions in water or other diluents. Approaches for controlling silver NP dissolution were also investigated and are shown to effectively improve particle stability in dilute suspensions required for spICP-MS analysis, while minimally affecting the measured intensity and allowing for more robust analysis. The results of this study provide an important step toward the validation of spICP-MS for more general use.

There is an increasing interest in using spICP-MS to help quantify exposure to engineered NPs, and their transformation products, released into the environment. The use of this analytical technique for environmental samples can be hindered by the presence of high levels of dissolved analyte which impedes resolution of the particle signal from the dissolved. While sample dilution is often necessary to achieve the low analyte concentrations necessary for spICP-MS analysis, and to reduce the occurrence of matrix effects on the analyte signal, it can also be used to reduce the dissolved signal relative to the particulate, while maintaining a matrix chemistry that promotes particle stability. Schwertfeger et al.24 proposed a simple, systematic dilution series approach whereby the first dilution was used to quantify the dissolved analyte, the second was used to optimise the particle signal, and the third as an analytical quality control. Using suspensions of well characterised Au and Ag NPs spiked with the dissolved analyte form, as well as suspensions of complex environmental media (extracts from soils previously contaminated with engineered silver NPs), the dilution series technique was demonstrated to improve resolution of the particle signal, which in turn improved the accuracy of particle counts, quantitation of particulate mass and determination of particle size.

spICP-MS is a useful tool for characterising and quantifying suspensions of metallic NPs, but there are currently limitations on the minimum size of NPs that can be detected using this technique. Newman et al.25 demonstrated the spICP-MS capabilities of a double focusing magnetic sector ICP-MS instrument for analysis of commercially available silver NP suspensions and samples of silver NPs suspended in natural lake water, including data on particle event integration and sizing, particle counting and measurement of dissolved Ag background. Analysis of commercially available silver NP suspensions using a dwell time of 30 μs determined that the ionic background dependent size LOD was similar to 10 nm at 2.7 ng Ag per L. In samples collected from a lake in which silver NPs were added experimentally (PVP-capped, 30–50 nm size distribution), the Ag NPs were shown to have a modal diameter similar to 20 nm, with a dissolved Ag background of between 6 and 13 ng L−1. The minimum particle size detectable in the lake water samples was between 15 and 18 nm, depending on the dissolved Ag concentration. There was no evidence of matrix effects in lake water that would adversely affect the accuracy of the particle sizing. The determination of PNCs by spICP-MS was shown to be limited by particle size and losses through adsorption on sample vials. Overall, while further development of the sample preparation protocols is needed, the spICP-MS technique described in this study demonstrates improved discrimination of silver NPs from dissolved Ag and a lower particle size range relative to previously described instrumental methods. In order to facilitate determination of PNC and the size of NPs by spICP-MS without the need to correct for the particle TE, a total-consumption sample introduction system consisting of a large-bore, high-performance concentric nebuliser and a small-volume on-axis cylinder chamber was evaluated by Miyashita et al.26 The system potentially permits a particle TE of 100%, removing the need to include this term when calculating the PNC and the NP size. The TE of the particle through the sample introduction system was evaluated by comparing the frequency of sharp transient signals from the NPs in a standard of precisely known PNC, to the particle frequency for a measured NP suspension. The TE for Pt NPs with a nominal diameter of 70 nm was found to be 93% with satisfactory repeatability (RSD of 1.0% for four consecutive measurements). When the particle size was determined using a solution-standard-based calibration approach, without an NP standard, the nominal particle diameters of Pt and Ag NPs were determined to be between 30 and 100 nm, in good agreement the particle diameters determined by TEM. This was regardless of whether a correction was applied for TE.

The separate and combined effect of sampling depth and aerosol gas flow rate on signal formation in spICP-MS was studied by Kalomista et al.27 Dispersions containing Ag and Au NPs were used. The NP signal could be improved significantly by optimisation of sampling depth, which also affected the shape of the signal histograms. The effect of the aerosol dilution gas flow, which is now standard in most ICP-MS instruments, on the spICP-MS signal formation was also investigated in an effort to make spICP-MS measurements faster, by on-line dilution of the aerosol generated from nanodispersions. Results revealed that the dilution gas flow could only be used for moderate aerosol dilution if the gas flow going to the pneumatic nebuliser was proportionally reduced at the same time. However, this was found to cause a significant worsening in the operation of the sample introduction system, causing a reduction in NP signal. It was concluded that the use of the aerosol dilution gas flow, in its present form, could not be used for spICP-MS analysis.

1.2 Vapour generation

Generation of volatile species for introduction into an atom cell has been around as long as analytical atomic spectrometry. Four main methods are used to generate volatile species: chemical, electrochemical, photochemical or plasma methods. Only certain elements are usually determined in this way, including elemental Hg, hydrides of As, Bi, Pb, Te, Sb and Sn, and transition metals which form volatile organometallic species.

Chemical hydride generation, typically using NaBH4, and cold vapour generation of Hg using SnIICl2, have been around for a considerable time so scope for innovation is limited. More recently, CVG of volatile organometallic compounds has generated considerable interest, with a growing list of reactions being published. The latest addition to this catalogue was reported by Duan et al.,28 who determined Zn as a volatile chelate with detection by AFS. The optimised reaction was performed by on-line mixing of acidified (0.1 M HCl) Zn and 0.4% sodium diethyldithiocarbamate solutions, yielding an LOD of 0.33 ng mL−1, 1.33% RSD and between 33 and 85% efficiency of generation. Interferences, caused by transition metals in the concentration range from 10 to 200 μg mL−1, were noted but the method was successfully applied to the determination of Zn in spinach and rice CRMs with good agreement with the certified values.

One of the advantages of CVG is that separation of the analyte from the matrix is inherent in the process. This, together with the gaseous nature of the generated species, means that it is ideal for introduction into low energy atom cells, such as cool flames or microplasmas, in low cost and portable instruments. Matusiewicz and Slachcinski29 developed a capacitively coupled argon microwave miniplasma (μCMP) interfaced with a CCD minispectrometer and CVG sample introduction. The instrumental configuration was used for the determination of As, Hg, Sb and Se with LODs of 1.4, 3.0, 1.5 and 3.8 ng mL−1 respectively. The μCMP itself was fabricated from an air and water cooled metal block with three 1.6 mm o.d. tungsten rod electrodes placed symmetrically around the cavity with a quartz minitorch mounted over them. This block was then mounted on the upper wall of a microwave generator. Volatile species were generated using acidified NaBH4, at a flow rate of 200 μL min−1, with gas/liquid separation performed using a nebuliser and cyclonic spray chamber. A similar set-up was reported by Li et al.30 but using a point discharge (PD) as the atom cell rather than a CCP. The PD consisted of a quartz tube (3 mm i.d. × 5 mm o.d. × 5 cm long) with a PTFE cap which was pierced radially with two tungsten needle electrodes at 180° to each other. The discharge was generated in the gap between them and emission observed radially at 90° to the electrodes. The plasma was sustained in a mixture of He and H2, the latter being generated by the KBH4 reduction of As, Bi, Sb and Sn. Under optimised conditions the LODs were in the range 1 to 7 μg L−1.

The use of a plasma to both generate volatile species and act as an excitation source has been reported by Chen et al.,31 who developed a single drop solution electrode GD (SD-SEGD) for CVG of Hg without using any chemicals. The device consisted of a quartz tube (8 mm i.d. × 10 mm o.d. × 8.5 cm length), into which a tapered tungsten electrode, and a stainless steel tube (0.5 mm i.d. × 1.5 mm o.d. × 5 cm length) protruded axially, with a small gap between then. A hanging drop (the sample) was exuded from the steel tube and the plasma was ignited in the gap by a high voltage from a compact a.c. ozone generation power supply. Using carbon nanotubes as an SPE medium, they preconcentrated Hg from digests of fish tissue into 100 μL of eluent, with an LOD of 0.01 μg L−1 (0.2 pg) using AFS detection. The same group32 also developed a liquid spray DBD for CVG of Pb, and introduction into ICP-MS. The Pb was converted to volatile species in the presence of 5% (v/v) formic acid in 0.01 M HCl by direct nebulisation into a DBD generated at the nebuliser tip. The DBD was formed between two electrodes, one wrapped around the nebuliser itself and a counter electrode plate 3 mm from the nebuliser tip, with the argon nebuliser gas providing the discharge gas. Under optimised conditions, the LOD for Pb was 0.003 μg L−1. A further paper from this group33 described another variation, using a so-called solution GD electrochemical vapour generation (EVG) method with AFS detection. In this set-up, the EVG device consisted of a two-compartment electrochemical cell with approximately 5 mL reaction volumes. One compartment contained a Pt foil anode immersed in electrolyte and the other contained a microplasma as cathode. The microplasma was formed in the gap (1 mm) between the tip of a stainless steel capillary (i.d. 1 mm; o.d., 2 mm; 5 cm, length) and the surface of the sample solution by application of a d.c. voltage. Sample was continuously introduced into the cell where volatile species were formed in the plasma at the surface, then swept by an argon stream (emanating from the capillary) through a gas–liquid separator (GLS) and then to a T-tube and mixed with auxiliary H2 for detection by AFS. Volatile species of Cd and Zn, in HCl electrolyte at pH 3.2, were determined, with LODs of 0.003 and 0.3 μg L−1 respectively.

Photochemical vapour generation (PVG) has seen an upsurge of interest over the last few years. This technique is based on a free-radical reaction between the analyte and an organic acid, sometimes in the presence of a catalyst, under UV radiation, to form volatile species. Previous reviews have covered numerous reports detailing variations in the type of acid and catalyst used. The only really new development in the current review period was reported by Luo et al.34 who used NPs of an amine functionalised Ti metal organic framework (MOF) as a photocatalyst for the PVG of Se. Metal oxides with electron–hole pair properties are often used as catalysts for photochemical reactions and amine-functionalisation expands the spectral absorption range into the UV-visible. The MOF photocatalyst was mixed with sample solution in 35% formic acid and exposed to UV radiation in a flow-through PVG reactor consisting of a quartz pipe (250 mm × 3 mm i.d. × 6 mm o.d.) wrapped around a low pressure Hg vapour lamp. After gas–liquid separation Se was determined by ICP-OES with an LOD of 0.3 ng mL−1 for volatile species of SeIV and SeVI.

1.3 Solids

1.3.1 Direct methods.
1.3.1.1 Glow discharge. GD-OES and GD-MS have become well established techniques for depth profiling, offering practical insights to assist the synthesis optimisation process and the quality control of materials coated with thin or thick layers. Commercial instrumentation and depth profile quantitation methods using these two analytical tools were briefly reviewed by Lobo et al.35 Recent applications demonstrating the capabilities of GD-OES and GD-MS for fast elemental quantitative depth profiling of films from an atomic layer to >100 μm were also described alongside some illustrative applications for the characterisation of organic films. Gaiaschi et al.36 described a new in situ measurement technique using GD-OES which provided depth information during the profiling process. The setup was based on a differential interferometer, and measurement accuracy better than 5% was demonstrated for crater depths ranging from 100 nm to several tens of μm. The technique was directly applied to non-transparent coatings, and promises significant improvements in quantitation using GD-OES.

Conventional GD-OES or GD-MS requires the assumption that the surface of the sample is homogeneous. However, recent developments in GD imaging appear to offer an opportunity to obtain 3D concentration maps, in which this assumption is no longer necessary. Storey et al.37 combined experimental results, models, and a summary of earlier work to examine the sputtering behaviour of elemental and morphological heterogeneities in a sample. The theoretical model revealed gaps in current knowledge of GD sputtering of heterogeneous samples, in particular that heterogeneity in the sample leads to roughened crater bottoms and how the morphology can evolve. A 3D profiling microscope was used to characterise the effects of surface inclusions on the sputtering process in a DC-GD spectrometer in a reduced-pressure Ar environment. The results of the study provide useful information for the future of bulk analysis, depth-profiling, and elemental surface mapping with GD spectrometry. Another publication from the same group38 described a tilting-filter instrument to overcome the challenges encountered when GD is applied for surface mapping. A pulsed GD was coupled with an optical system comprising an adjustable-angle tilting filter, collimating and imaging lenses, and a gated, intensified charge-coupled device (CCD) camera, to provide surface elemental mapping of solid samples. The tilting-filter spectrometer was instrumentally simpler, produced less image distortion, and achieved higher optical throughput compared with a monochromator-based instrument, but had a more limited tunable spectral range and poorer spectral resolution. Consequentially, the tilting-filter spectrometer was limited to single-element or two-element determinations, and only when the target spectral lines fell within a defined spectral range. Spectral interferences resulting from heterogeneous impurities could be flagged and overcome by observing the spatially resolved signal response across the available tunable spectral range. The instrument was evaluated for the spatially resolved GD-OES analysis of representative samples.

A modification in the design of a GD-MS system to allow 2D elemental imaging was reported by Konarski et al.39 The authors moved the samples with respect to a fixed GD so that the image size was not limited by the size of the anode. Using an add-on ‘xyz’ high-vacuum manipulator equipped with a membrane bellow, 2D images were obtained. Pixel-by-pixel acquisition applied for xy maps (7.5 mm × 5.5 mm) consisted of 3040 and 1520 data points were obtained. The maps were acquired in 25 and 12 min respectively, with a spatial resolution of between 0.16 and 0.42 mm in the x axis. The results were obtained using a DC-GD-QMS at 1.1 kV DC voltage and 1 mA current. The technique was proposed for the determination of the homogeneity of materials, distribution of surface impurities, and 3D imaging.

An atmospheric-pressure SCGD was evaluated as an ion source for atomic, molecular, and ambient desorption/ionisation MS by Schwartz et al.40 The SCGD consisted of a DC plasma, supported in ambient air in the absence of gas flows, and sustained upon the surface of a flowing liquid cathode. Analytes were introduced into the flowing liquid as an ambient gas or as a solid held near the plasma, and vaporised and ionised by interactions within or near the discharge. Ambient gases and solids, desorbed by the plasma from a glass probe, were found to be readily ionised by the SCGD and strong analyte signals were obtained from solid samples placed at least 2 cm from the plasma. These findings indicated that the SCGD might be useful also for ambient desorption/ionisation MS.


1.3.1.2 Thermal vaporisation. Direct analysis of NPs by GFAAS was the subject of several reports over the past year. The properties of magnetic NPs (MNPs) are strongly influenced by the Fe concentration and size of the NPs. A method for the direct determination of these factors by solid sampling HR-CS-GFAAS was proposed by Alonso et al.41 A novel strategy for evaluating the area and upslope for the absorbance line of Fe at 352.614 nm was developed. This involved the optimisation of five furnace programme parameters: atomisation heating rate, atomisation temperature, pyrolysis heating rate, pyrolysis temperature and pyrolysis hold time. Satisfactory calibration curves were obtained for Fe using liquid iron standards (R ≥ 0.995), and for MNP samples with a certified size of particle (for size particle determination) (R ≥ 0.990). Determination of the size of MNPs and their Fe percentage concentration were validated by TEM and SEM, respectively. The method could be employed in the optimisation of MNP synthesis procedures.

HR-CS-GFAAS was also used to obtain information on the chemical form (ionic or as AuNPs) of Au in solutions without any additional separation step.42 Optimisation of the temperature programme, using a very slow heating ramp (150 °C s−1) during the atomisation step and a sufficiently high atomisation temperature (2200 °C) in the absence of chemical modifiers, enabled a fast and simple screening to be performed. This was possible because the signal profiles obtained for solutions and suspensions of ionic Au and AuNPs were different. In addition, when NPs were detected, it was possible to estimate the average particle size because this parameter appeared to be directly related to the time of appearance of the maximum peak height. A value of 27.7 nm ± 8.8 nm was estimated for NIST 8012 AuNPs with nominal diameter of 30 nm. The LOD was 5.5 pg (0.55 μg L−1) with a linear range up to 10 ng (1000 μg L−1), and was further validated by spiking a natural water CRM (CRM KEJIM 02). When mixtures of ionic Au and AuNPs are analysed, quantitation was more complicated because signal overlap from ionic Au and AuNPs occurred, so deconvolution of the peaks was required. GFAAS was used by Leopold et al.43 to distinguish NPs and ions in solid samples by using a similar method of data evaluation but extending the parameters measured to ‘atomisation delay’ (tad) and ‘atomisation rate’ (kat). The authors measured tad and kat in the analysis of AuNPs and solutions of AuIII. Results obtained for kat and tad values showed reproducible size correlations for both parameters in a particle size range from 2 to 100 nm. These correlations could be used for size calibration and to allow the size determination of AuNPs in monodisperse aqueous suspensions. Absorbance values and the tad and kat of 20 nm-sized AuNPs were examined over a concentration range from 1 to 50 μg L−1 revealing that tad was the most robust parameter. Mixtures of three or four different sizes of AuNPs were studied and deconvolution of the obtained sum peaks resulted in acceptable accuracy for determinations of size distribution. The potential of the proposed method for easy and fast measurement of size distribution of polydisperse AuNP suspensions was demonstrated.

1.3.2 Indirect methods.
1.3.2.1 Laser ablation. Laser ablation has become a dominant technology for direct solid sampling in analytical chemistry. Laser ablation refers to the process in which an intense burst of energy delivered by a short laser pulse is used to sample (remove a portion of) a material. The advantages of LA analysis include direct characterisation of solids, no chemical procedures for dissolution, reduced risk of contamination or sample loss, analysis of very small samples not separable for solution analysis, and determination of spatial distributions of elemental composition. In a short review article, Gonzalez44 described the most common approaches, namely LIBS and LA-ICP-MS, and provided an introduction to LAMIS. Shazzo and Karpov45 produced a review focusing on element fractionation as the main source of error in laser sampling. Elemental and isotopic fractionation occurs in the interaction between the laser and the sample surface resulting in the formation of varied sample aerosols. The review covered studies of the effect of the laser wavelength, pulse duration, pulse fluence, plasma screening, explosive boiling, and the crater geometry on elemental fractionation.

LA-ICP-MS is a technique able to obtain either quantitative elemental data or spatially resolved imaging/mapping of elements in biological tissues. This has found application in research into metallomics, nanoparticle uptake, tagging in biological systems, elemental mapping of tissues and quantitative analysis for biomedical studies. Pozebon et al.46 reviewed recent applications of LA-ICP-MS in the biological field over the last three years. Topics included: fast wash-out LA cells; novel calibration strategies such as ink printing and dried-droplets; mapping of elemental distribution in biological samples (animal, human and plant tissues); nanoparticle uptake; and protein and single cell analysis. Bussweiler et al.47 highlighted trends in high-speed, high-spatial resolution, multi-elemental imaging that have become possible due to advances including fast-washout ablation cells and fast TOF-MS analysers. The authors demonstrated this by coupling LA with ICP-TOF-MS for two applications: quantitative mapping of trace elements in a sulfide mineral (sphalerite) and imaging of the distribution of a chemotherapy drug (cisplatin) in a rat kidney. The results demonstrated that LA-ICP-TOF-MS is an effective tool to study biological and geological processes, with greater speed and in greater detail than previously possible with conventional ICP-MS instruments.

In order to improve vaporisation, atomisation, and ionisation efficiencies in LA-ICP-MS, Nakazawa et al.48 merged HCl gas with laser-ablated particles before introduction into the plasma, to convert their surface constituents from oxides to lower-melting chlorides. When particles were merged with HCl gas generated from a HCl solution at 200 °C, the measured concentrations of elements in the particles were 135% higher on average than the concentrations in particles merged with ultrapure water vapour. Particle corrosion and surface roughness were observed by SEM, and oxide conversion to chlorides was confirmed by XPS. Under the optimum conditions, the recoveries of measured elements improved by 23% and recoveries of elements with high-melting oxides (Sr, Zr, and Th) improved by a maximum of 36%.

LA-ICP-MS is a well-established technique for elemental and isotopic analyses. Ultrafast lasers offering pulse widths of tens to a few hundred fs were proposed fifteen years ago as a major development but have yet to be widely adopted. Poitrasson and d’Abzac49 listed the considerable advantages of ultrafast fs lasers. Chief among them was the significant reduction in chemical fractionation, thereby alleviating the need for matrix-matched calibration standards, and enhanced sensitivity due to a higher ablation yield compared with ns lasers. The authors suggested that the development of new measurement strategies using ns lasers has rendered fs LA-ICP-MS less of a novelty, however its versatility, accuracy, precision and improved spatial resolution have been clearly demonstrated in many applications.

The particle size distribution of dry aerosols originating from the LA of glass material was monitored by LA-ICP-MS and simultaneously using two aerosol spectrometers, FMPS and APS by Novakova et al.50 The combination of LA-ICP-MS and FMPS allowed measurement of the particle size distribution at 1 s intervals in the size range between 5.6 and 560 nm. Using APS, this range was extended to between 0.54 and 17 μm. Online monitoring of the dry aerosol was performed for two ablation modes (spot and line with a duration of 80 s) with a 193 nm excimer laser system, using the glass reference material NIST 610 as a sample. Different sizes of laser spot for spot ablation and different scan speeds for line ablation were tested. It was found that the FMPS device was capable of detecting changes in particle size distribution during the first pulses of spot LA, and was suitable for LA control. The line mode of laser ablation produced larger particles throughout the LA process, whereas spot ablation produced larger particles only at the beginning. Spot ablation also produced more primary nanoparticles (in ultrafine mode size range <100 nm) than line ablation. The authors suggested that this was caused by the reduced number of large particles released from the spot ablation crater, while larger particles scavenge ultrafine particles during line ablation.

2. Instrumentation, fundamentals and chemometrics

2.1 Instrumentation

2.1.1 Sources. A common theme of recent reviews has been miniaturisation of instrumentation. Sources for OES have been an obvious target given the ready availability of miniaturised optical spectrometers with which to couple them. The application of the DBD has been reviewed by Brandt et al.51 (95 references), including: a discussion of the theory and mechanism; sample introduction via liquid, vapour and GC; and AES, AAS, AFS and CL detection. Three main geometries of DBD were identified: the planar version, where two planar glass plates serve as dielectric layer and the electrodes are mounted on the outside; the cylindrical version where electrode is usually wrapped around a tube and the other one is located as a pin inside this tube; and the capillary version, with two ring electrodes mounted on the outer surface of the capillary. They concluded that there are now a considerable number of applications of DBDs, with LODs comparable with conventional methods provided the sample matrix is uncomplicated. Clearly, the main remaining challenge for DBD sources is to eliminate the effects of the matrix. One way of doing this was proposed by Li et al.52 who used a miniaturised ETV device fabricated from a wrapped molybdenum filament inside a cylindrical quartz tube (i.d. 10 mm). Using a cylindrical type DBD formed with argon, they achieved an LOD for Hg of 0.4 μg L−1 for a 3 μL sample and recoveries of between 86 and 101% in pre-concentrated (by SPE) spiked seawater samples. Zhang et al.53 incorporated a DBD into a LOV arrangement so that sample treatment and detection using a CCD were integrated into a compact package. Trace Cd was separated and preconcentrated by flowing through a microcolumn, eluted and derivatised with a borate buffer solution containing 7.5% (v/v) ethanol (presumably for vapour generation), then nebulised into the DBD for Cd detection by OES. An LOD of 0.06 μg L−1 was achieved, with spike recoveries of between 94 and 108% for a variety of water samples. The question remains how these systems would perform for less volatile elements and without separation of the matrix?

One novel development was reported by Liu et al.54 They used a substrate of graphene/SiO2/Si onto which the analyte was deposited. The graphene acted as a conducting electrode so that, when a voltage pulse was applied, the 10 nm SiO2 insulating layer broke down. Electrons induced at the SiO2/Si interface were emitted into void channels in the SiO2 and travelled into the graphene (anode) where the resultant electron impact caused analyte excitation and emission. However, no details were given about how this might operate with a sample.

Quite unrelated, but high in novelty, was a paper by Burger et al.55 who reported an application of AES during electrosurgery. An r.f. current was applied to a needle active electrode in contact with tissue to be excised. The small contact area resulted in a high current density, caused ohmic heating and vaporised the tissue. The gap between the electrode and tissue was filled with hot tissue vapour which primarily consisted of water vapour and tissue fragments. A strong electric field was produced in this gap by application of high voltage, which lead to ionisation and excitation of elements in the tissue. Surprisingly it was possible to distinguish between normal renal tissue and tumour tissue by monitoring the Cd atomic emission line at 228.8 nm!

2.1.2 Spectrometers. Boesl56 has published a ‘basic introduction’ to TOF-MS, aimed at helping the budding TOF-MS researcher. The tutorial was divided into sections covering basic theory, approaches to identify sources of flight time broadening and ion losses, and options for construction of a TOF-MS for those wishing to build their own instrument. The text was backed up with the relevant mathematical equations and useful diagrams. Dennis et al.57 reviewed the current status of DOF-MS, a relatively new type of MS which has been covered in the three previous ASU reviews. The review introduced the technique and compared it to other mass analysers, in particular TOF-MS with which there are many similarities. An important consideration is the choice of position-sensitive detector, so the current options were compared. The review ended by considering applications that would benefit from the technique’s ability to energy-focus ions and spatially separate m/z values without the use of magnets or cyclotron frequencies.

2.2 Fundamentals

2.2.1 Fundamental constants. A comprehensive compendium of accurately known, fundamental constants is on the wish-list of every researcher. Some constants are well known but that doesn’t mean that they cannot be improved. Revision of Avogadro’s constant, and the associated redefinition of the mole and kilogram, has been in process for several years, with the intention to base it on a crystal of silicon. Parmann et al.58 used MC-ICP-MS to determine the molar mass and isotopic composition of the silicon isotopes 28Si, 29Si, and 30Si of a new silicon crystal (Si28–23Pr11) highly enriched in the 28Si isotope. Two independent research groups analysed samples from three different axial and radial positions to establish if there was any variation in the crystal. Their results agreed within the limits of uncertainty and averaged M = 27.976942666(40) g mol−1 with a relative combined uncertainty of 1.4 × 10−9. No significant inhomogeneity in the fractions of 28Si and 29Si (within the limits of the uncertainty) was observed. There was variation in the 30Si fraction but this could not be attributed to a spatial effect.

Podesta et al.59 published a new estimate of the Boltzmann constant to replace one which they had published earlier (de Podesta et al., Metrologia, 2013, 50, 354–376). In the original paper, molar mass (M) for the Ar gas used did not correspond with the assumed value, probably due to contamination. The estimate required accurate measurement of Ar isotope ratios in a sample of gas with accurately known M, and computation involving the temperature (T), the low pressure speed of sound in Ar (c02) and NA. The revised estimate for kB was 1.38064860(97) × 10−23 J K−1 with a combined uncertainty of 0.702 × 10−6. Contributions to the combined uncertainty were: M, 28.3%; T, 26.8%; c02, 44.9%; and NA, 0.0%.

Experimental measurements of Stark widths are always useful. Manrique et al.60 measured Ti(II) lines in the range between 250 to 460 nm in a LIBS plasma using a fused glass sample. They measured 49 multiplets, including 85 widths and 72 shifts, which compared favourably with previously reported calculations for transitions between 3d24s, 4p and 5s configurations. Also, 39 lines for transitions to 3d3 were measured, but no comparative data was available. Popov et al.61 measured Stark widths of Mn(I) lines belonging to multiplets z6P° → a6S and z6D° → a6D in a ‘long’ spark induced laser plasma with an Al target. This type of plasma was deemed more homogenous than a cylindrical one. Comparison with calculations and other experimental measurements revealed considerable divergence in results.

Quinet et al.62 determined 19 radiative lifetimes for 5080 spectral lines of Co(II) using time-resolved LIF one- and two-step excitation. Seven from the high lying 3d7(4F)4d configuration in the energy range 90[thin space (1/6-em)]697 to 93[thin space (1/6-em)]738 cm−1 were new, and the other 12 from the 3d7(4F)4p configuration with energies between 45[thin space (1/6-em)]972 and 49[thin space (1/6-em)]328 cm−1 were compared with previous measurements. Experimental values agreed within the error bars of two other studies, but with a tendency to be somewhat shorter values, probably due shorter pulse length and better time resolution in the detection system. Overall agreement between theory and experiment was between 5 and 17%.

2.2.2 Diagnostics.
2.2.2.1 Plasmas. The constant battle to understand the fundamental characteristics of the ICP, coupled with both OES and MS, continues. Recent reviews have documented considerable progress in understanding formation of the ion beam, and ion sampling, in ICP-MS. Of particular note is the work of the Bogaerts group, who have published a useful tutorial review summarising their extensive research in this area.63 They start with an overview of other research in the field and then describe their own computer modelling approach. Crucially, these computer models were validated with experimental studies and provide an interesting insight into the effects of plasma operating conditions on particles of varying diameter as they travel through the central channel of the ICP, with respect to atomisation, ionisation and diffusion out of the central channel. Other parameters include the position of the ion sampling interface, i.d. of the ion sampling cone orifice, position and i.d. of the injector tube and mass flow rate. The current model only considers the introduction of particles from e.g. LA, and the formation of Cu+ ions, but work is continuing to compare the behaviour of Zn, the effects of He addition and the behaviour of aqueous droplets.

Farnsworth and Spencer64reviewed the physical processes that control ion sampling and transport in ICP-MS. In a comprehensive treatise, they cited 63 references and compared the results of theory and computerised models with experimental data. Thus, they gained insight into the distribution of ions in the plasma and their behaviour as they sequentially flow through the sampling cone, the supersonic expansion, the skimmer, the ion optics and the mass analyser. They conclude that idealised models still fall short of any accurate quantitative description because, while models for ambipolar diffusion and space charge effects work well in the absence of matrix effects and supersonic shock at the skimmer, this is alas not the case in the real world. As the authors point out, despite enormous advances in the sensitivity and robustness of ICP-MS, ion transport efficiency from the plasma to the mass analyser is typically <0.1%, so where do the rest of the ions go? As if to partly answer this question, the same group published an empirical study of the effects of the matrix on the ion beam in ICP-MS. Using planar-LIF, they observed the effects of Mg, Cs and Pb on the cross section of a Ba ion beam in the second vacuum stage and in-front of the entrance to the mass analyser. The data, also rendered as colour images captured on a CCD camera, showed that the ion beam containing only Ba was approximately 2.3 mm wide at 50% FWHM. The addition of a matrix did not change the width significantly, however, the addition of Pb did cause the beam to shift 1 mm to the right, resulting in a loss of ion transmission under compromise focusing conditions. They postulated that this lateral shift in the ion beam was caused by coulombic repulsion. Shifts in the ion beam in the collision reaction interface were also observed in the presence of He and H2 gases. These were approximately 4.8 mm to the right for a Ba beam, and 2.9 mm to the right for a Ca beam from centre to centre. This, and the large beam widths in the CRI (ca. between 2.8 and 4.7 mm) accounted for a decrease in ion transmission. In addition, the velocity distribution of Ca ions, perpendicular to the axis of the ion beam, was much narrower under normal conditions compared to CRI conditions, with hydrogen gas flow at 100 mL min−1.

Hoffmann et al. performed both an empirical study65and theoretical modelling66of the argon ICP, for two different types of torch used in ICP-OES. In the first study, a comparison was made of a Fassel torch operated at 12–20 L min−1 and a low-flow torch (SHIP torch, cooled with compressed air) operated at 0.6 L min−1. A novel imaging acousto-optical spectrometer was used to monitor emission lines of Ar, OH, Fe and Mg to determine Trot, Texc and Te, voltage and current in the load coil as well as plasma power. In addition, IR-measurements of the torch wall temperature, and RF voltages and currents were measured directly in the load coil. The results confirmed that the plasma generated in the SHIP torch had a toroidal structure in the presence of the nebuliser gas flow. Deviation from LTE was evident using both torches by the fact that Te (SHIP, 8770–10[thin space (1/6-em)]840 K, Fassel, 6130–10[thin space (1/6-em)]200 K) was much higher than Trot (SHIP, 4580–5120 K; Fassel, 3230–4250 K) and Texc (SHIP, 4360–5000 K; Fassel, 3000–5560 K). One further surprising result was that, depending on plasma conditions, the measured power differed from that indicated by the software up to a factor of two. In the second study, LTE-based and two-temperature models were developed for both torches, and results compared with those obtained empirically. The LTE-model described the reverse gas flow and formation of the toroidal plasma structure after introduction of the nebuliser gas. Unsurprisingly, since this was an LTE model, Tgas and Te were predicted to be dissimilar; a drawback admitted by the authors given that the empirical studies had demonstrated this. The two-temperature model predicted Te to higher than Tgas. However, both were lower than Tgas predicted by the LTE model and the measured value for Te, but higher that the measured values for Trot and Texc.

Hattendorf et al.67 made a study of doubly charged diatomic ions by mass spectrometry with an ICP ion source. They found that the abundance ratios were high (0.1% for BaAr2+, 0.4% for CaAr2+, 0.1% for MgAr2+ and 0.2% for SrAr2+) compared to the corresponding M2+ ions. Dissociation energies for these polyatomic ions were calculated using spectroscopic constants from the literature, and the Tgas was calculated using the equilibrium dissociation expression abundance ratios of MAr2+[thin space (1/6-em)]:[thin space (1/6-em)]M2+, for the dissociation of MAr2+ to M2+ and Ar. In increasing order, Tgas values were 2400 to 2600 K, 2600 to 2700 K, 3300 to 3700 K and 5500 to 5700 K for, SrAr2+, CaAr2+, BaAr2+ and MgAr2+ respectively. The implication of these results is that MgAr2+ ions formed predominantly in the ICP itself (i.e. higher Tgas environment) but the others formed in the ion extraction interface where the temperature was lower. The authors stressed that there is no simple relationship that explains this difference because the picture is complicated by the relative abundance of other species involved in collisions induced reactions, and internal charge transfer reactions which may occur in the polyatomic ions causing them to dissociate. They also point out that the existence of such ions are significant with respect to accurate isotope ratio measurements of Cu and Zn by MC-ICP-MS, which was their original reason for studying them.

Another empirical approach was taken by He et al.68 who undertook a study comparing different skimmer cones (‘H-’ and ‘X-’) combined with both standard and ‘Jet-’ sampling cones in ICP-MS. By studying the effects on the isotopes of B, they came to the somewhat predictable conclusion that more ions could be introduced into sample cones with larger orifices, resulting in increased sensitivity.

Fundamental studies of the ICP and other types of plasma have provided a rich seam for many research groups, with a plethora of contradictory results given that few (if any) analytical plasmas are in LTE. The GD plasma is a case in point, where ionisation and excitation mechanisms are predominantly influenced by non-LTE processes. Mushtaq et al.69 studied the effects on Cu excitation in Ar and Ne GDs with either 2% v/v added hydrogen or 2% v/v added nitrogen. The authors contended that selective excitation processes that tend to occur in the GD depend on the spin multiplicity of the initial and final states and the energy of the upper excited level, so each element must be considered individually. Their results indicated that an unexpected increased intensity of Cu(II) lines emitted from the 15.961 eV energy level (3d9(2D)4p)3Po2, in a Ne GD with added hydrogen, were most likely due to asymmetric charge transfer involving H2+ ions. A similar effect was not observed on the addition of N2 despite similarities in the vibrational levels of N2+ and H2+ and their concentrations in the plasma. They explained this by considering the potential energy curves of the two species, and showed that the N2+ ions occupy lower vibrational states compared to the H2+ ions, so therefore have insufficient energy to ionise Cu.

Matsuta70 used a CFS spectrometer to study spectra of O (I) 844.6 nm and Ar (I) 842.5 nm lines in an RF GD. The CFS spectrometer had permanent magnets set in the Faraday configuration and profiles of the aforementioned lines were obtained by scanning the wavelength using a diode laser. An RF GD discharge tube was placed between two crossed polarisers that acted as a polariser and an analyser, and blocked the probe light when there was no atomic interaction. The intermagnet spacing was 90 mm. A significant change in the CFS spectrum of the Ar (I) line was observed when the partial pressure of Ar in a He–Ar discharge was changed. This was explained by theoretical calculations of the spectra performed using Faraday functions.


2.2.2.2 Graphite furnaces. In keeping with the grand tradition of the study of furnace fundamentals, Ruiz et al.71 used XPS for the identification of the solid state species formed during the atomisation of Mo on a pyrolytic graphite surface used for ETAAS.71 As XPS is capable of probing materials at the concentrations encountered in actual ETAAS measurements, it was anticipated that it would allow obtaining more realistic results than those reported in previous works by means of techniques such as X-ray diffraction, which requires thousand-fold amounts of the analyte. The results obtained by XPS showed a distribution pattern of the oxidation states of Mo on the graphite surface as a function of the temperatures reached at each stage of the ETAAS procedure, which was correlated with the possible species formed during the drying, ashing and atomisation steps. A previously proposed atomisation mechanism for Mo was revisited and modified as a result of these findings. In particular, species such as Mo4O11 were not observed at the ashing stage nor was metallic Mo observed as a stable, intermediate phase during the Mo2C low temperature carbide formation at between 1200 and 1900 °C.
2.2.3 Interferences. Interferences generally fall into one of two categories, spectroscopic or non-spectroscopic interferences. Many interferences can be eliminated or reduced by using an appropriate sample preparation step, where the matrix is separated from the analyte. Simple methods of interference correction are preferred if possible, and manufacturers are continually searching for new ways of eliminating interferences by instrumental methods.

One instrumental method for eliminating spectroscopic interferences is ICP-QQQ-MS. Bolea-Fernandez et al.72 have reviewed this technique in a tutorial review of 110 references. They described the different modes of operation: single and double quadrupole; the role of the collision/reaction cell and reaction gases; and different scanning modes of product ion scan, precursor ion scan and neutral mass gain scan. There was a section on overcoming spectral interferences by on-mass and mass-shift approaches and a useful flow-chart to help the prospective user choose the correct approach. A table of applications was ordered by analyte ion with details of interfering ions and the reaction gas that can be used to eliminate the interference. The mass-shift and on-mass modes of operation in ICP-QQQ-MS are becoming popular. In this mode the instrument is operated such that Q1 (the first quadrupole) selects ions at the m/z of interest, allowing both analyte and interfering ions to pass to Q2 (the collision/reaction cell) which is pressurised with a reaction gas. Selective reaction with analyte ions converts them to a molecular species of different m/z, which is then selectively passed through Q3 and detected, thereby eliminating the interference. By comparison, in on-mass mode, the interfering ion is eliminated by selective reaction and only the analyte ion is passed by Q3 for detection. Virgilio et al.73 studied the effect of using O2 reaction gas on three groups of analytes: highly reactive (Ce, La, P, Sc, Ti, and Y); partially reactive (As, Ba, Mo, Si, Sr, and V); and less reactive towards oxygen (Al, Bi, Cu, Mg, Pb, and Pd). On-mass and mass-shift modes were evaluated by monitoring atomic and metal oxide ions with oxygen flow rates varying from 0.1 to 1.0 mL min−1. They concluded that the most highly reactive elements, plus As and V, had greater sensitivity when measuring their oxides, whereas Mo and Si oxides presented intermediate sensitivity. On-mass mode was generally more sensitive for less reactive elements.

Non-spectroscopic matrix effects can be reduced by mathematical correction, dilution or separation. A novel method of identifying the correct dilution factor to eliminate matrix effects, developed by the Hieftje group, has been reported previously in this review. In the most recent paper by Cheung et al.74 the method, which makes use of the transient dilution process inherent in the dispersion of the sample in an FI flowing stream, was further refined. Analyte ratios were monitored, using ICP-TOF-MS, to produce a so-called gradient–dilution curve, which was used to flag a matrix interference. The problem was that dispersion in a single FI sample was element specific, so the gradient–dilution curves were different for each analyte. Their solution was to use a gradient HPLC pump, so that variation in dispersion between elements became negligible. Hence, the optimal dilution factor to reduce the interference to an acceptable level could be found from the gradient–dilution curve as the point where the signal ratio between two elements became constant.

A mathematical method to correct for matrix interferences in ICP-MS was reported by Narukawa and Chiba.75 They proposed an ICP-index, calculated from the ionisation potentials of the analyte and matrix elements, mean free path and number of ionisation states. This reciprocally represented the ease of ionisation for each element in the plasma, such that elements with similar ICP-indexes behaved similarly, hence making it easier to choose an appropriate internal standard. The ICP-index was applied to matrix effects caused by Li (small atomic mass), K (large atomic radius), Ar (maximum matrix in ICP plasma) and Pb (large atomic mass) and C. The ICP-index of Ar was considered to be most useful for estimating the influence of the matrix and considered to be the most useful indicator for selecting an appropriate internal standard for daily operation.

An investigation into the causes of matrix effect in ICP-MS was reported by Olesik and Jiao.76 They studied the effects of high concentrations of the matrix elements Na, Cu, Y, In, Cs, Tb, Lu, and Tl on the analytes 7Li+, 11B+, 24Mg+, 45Sc+, 60Ni+, 71Ga+, 75As+, 88Sr+, 111Cd+, 138Ba+, 153Eu+, 72Yb+, 209Bi+, and 238U+ as a function of analyte mass, matrix mass, matrix concentration, lens voltage, and nebuliser gas flow rate. They found that, using their particular instrument, matrix-induced changes in analyte ion sensitivity were generally similar for all analyte ions, contrary to accepted knowledge. Hence, a single internal standard could be used to correct for effects due to 5 mM mid- or low-mass matrix elements to give recoveries between 10% and 20% for most analytes. Matrix suppression on analyte ion signal due to high-mass matrix elements (>140 amu) was greater for high-mass analyte ions compared to low-mass analyte ions.

2.3 Chemometrics

Gamez77 produced a short tutorial on compressed sensing (CS) for data acquisition in spectroscopy. The CS approach is interesting in that it is a way of reducing the amount of data while still extracting the most relevant information from a signal. This is important for 2D and 3D imaging, or very high frequency signals, because it speeds up the analysis and reduces the total amount of data acquired in the first place. Unlike other compression methods, where all the data is first acquired and then compressed by omitting the ‘information poor’ parts, CS uses group theory to randomly select the data in the first place. This was quite neatly explained using the analogy of identifying a single gold coin in a larger group, then expanded to include examples in fluorescence and mass spectrometry. A second paper by the same author78 reports on an application of CS for ICP-OES which nicely illustrated the benefits for atomic spectroscopy. Rather than using an array sensor for spatial imaging, the CS system comprised a single pixel sensor and a variable encoding mask to allow selected sections of the image (i.e. an encoded image) to simultaneously reach the detector. Hence, a series of randomly generated masks resulted in the single pixel measuring a series of encoded light intensity signals, the random nature of which ensured an equal probability of all of the sections of the image to be measured. The spatial resolution was a function of experimental and image processing parameters such as sensing matrix selection, choice of reconstruction algorithm, and sparsifying basis. Spectral images of optical emission of He, I, N2 and N2+ lines from an atmospheric plasma jet were collected. The main conclusion for the practitioner was that the system gave high quality images for an order-of-magnitude lower price compared with a conventional array detector, while avoiding the time restrictions of pixel-by-pixel rastering.

Related to the issue of how to handle large data arrays, Cumpson et al.79 addressed the problem of applying multivariate PCA to 3D imaging in XPS, OES and SIMS, arising from the large data matrices involved. Their solution was to use algorithms that improve the speed of PCA for datasets of unlimited size. Under normal circumstances, a 128 × 128 pixel, SIMS depth-profile of 120 layers, with each voxel having a 70[thin space (1/6-em)]439 channel mass spectrum associated with it (>1 Tb of data) took 27 h to process on a desktop PC. They reduced this to 10 s by using alternative algorithms and a PC with a more powerful graphical processing unit. It seems to this reviewer that this might be a perfect application for the CS approach to data sampling in the first place.

Koscielniak and Wieczorek have reviewed80 the various univariate calibration methods used for quantitative analysis in a systematic way. Variations such as matrix matching, internal standardisation, standard additions, dilution and bracketing were examined. They also investigated the effect of combining more than one calibration approach on matrix interferences. They concluded that matrix interferences could be minimised by judicious choices, but that effects caused by different forms of the analyte (the species effect) were less predictable, so more difficult to correct. Similarly, Bulska and Wagner81 reviewed calibration strategies for quantitative analysis using ICP-MS. They concluded, unsurprisingly, that semi-quantitative methods were least accurate and that ID-MS was the most accurate and traceable. Sforna and Lugli82 produced a MATLAB script, using both internal and external calibration, that allowed fast treatment and matrix correction of LA-ICP-MS data for imaging purposes. Attractive features of the script are: the user-friendly interface for the benefit of non-experts in MATLAB; the use of raw data from any LA-ICP-MS instrument; and its applicability to any type of sample. Some nice examples of 2D, elemental maps of a speleothem, a fossil tooth and a magmatic plagioclase were presented.

3. Laser-based atomic spectrometry

Key fundamental studies and instrumental developments in laser-based atomic spectrometry, published in 2017 and at the end of 2016, are highlighted in this section. Prior progress is documented in previous ASU reviews. Atomic spectrometry techniques where the laser is used as either a source of intense energy or of precise wavelength (e.g. LIBS and LAAS) are considered. Studies related to LA-ICP-MS/OES are reviewed in Section 1.3.2. The use of lasers for fundamental studies of the properties of atoms or for thin film deposition are not reviewed.

3.1 Laser induced breakdown spectroscopy (LIBS)

This section describes the latest instrumental developments and fundamental studies related to LIBS, but it does not cover detailed applications. LIBS is one of the most important widespread techniques where the laser is used as an energy source to induce an optical plasma. Recent and novel LIBS applications and developments were described in a review article by Bauer and Buckley.83 Similarly, Senesi and Senesi84 provided a historical and critical overview of the developments of LIBS for quantitative measurements of carbon in soils. Bengtson85 compared the analytical potential of LIBS with spark OES and GD-OES techniques for the analysis of metals. This comparison was carried out using one instrument with interchangeable sources, eliminating differences related to the optical system and detectors. One major conclusion was that LIBS is not likely to replace spark-OES for bulk analysis nor GD-OES for depth profile analysis, in the near future.

Several international conferences were dedicated to discuss recent progress in LIBS, in particular the 9th Euro-Mediterranean Symposium on LIBS held in Pisa, Italy on June 11–16 June, 2017 (http://www.emslibs.org/), and the 2nd Asian Symposium on Laser Induced Breakdown Spectroscopy held in Tokushima Japan, August 27–31, 2017 (http://www.me.tokushima-u.ac.jp/lplab/ASLIBS2017.html).

3.1.1 Fundamental studies. Evolution of Te and ne in the LIBS plasma has been the subject of several reports during the review period. De Giacomo and Hermann86 described how the hierarchy of the elementary mechanisms changes continuously because the electron number density and the electron temperature decrease during LIBS plasma expansion. Current literature and traditional plasma theories were critically evaluated to compare the pros and cons of emission spectroscopy for the study of LIBS plasmas in their various applications. In another critical review, experimental procedures for characterisation of ne in LIBS plasmas were summarised and evaluated by Lykovic et al.87 The proper application of the Stark shift for plasma ne measurement, and testing for self-absorption, were recommended. The authors criticised the lack of basic and relevant information (e.g. description of the deconvolution processes or calculation of the estimated uncertainty) in many publications.

Determination of Te evolution in a LIBS plasma is a challenging task. A new method was developed by Bredice et al.,88 based on decomposition in independent components of a LIBS spectra, collected at various delays after the formation of the plasma. This information was used to calculate parameters required for 3D Boltzman plots. The advantage of this method was that only knowledge of the upper energy level of the transitions was required for the determination of Te. Double-pulse LIBS spectra of a brass sample, acquired at different delay times with respect to the second laser pulse, were decomposed into independent components making use of mean field independent component analysis (MFICA) to finally obtain the temporal evolution of the plasma Te. The results agreed well with Te values obtained using a classical Saha–Boltzmann plot approach. Another approach used to investigate the fast plasma expansion present in LIBS was developed by Bukvic et al.89 They demonstrated that the characteristic oval shape of the laterally resolved spectral lines, collected in side-on LIBS experiments, were due to the Doppler effect during fast radial expansion of the plasma. Numerical procedures, without use of an inverse Abel transform, were used to evaluate the expansion velocity and temperature of heavy particles. Using this approach, the expansion of a copper LIBS plasma in ambient gas at low pressure (20 Pa) was investigated. Good agreement with existing data obtained by observing shock waves, was obtained.

Spatial confinement of a LIBS plasma has been used to improve sensitivity. Experimental and computational studies were performed by Wang et al.90 who evaluated the expansion dynamics of shockwaves in channel cavity LIBS, and also in typical flat surface cases. Computational models, in 2D and 3D, of shockwave propagation and reflection were validated against experimental results, demonstrating that reflected shockwaves compressed the plasma plume, producing a vertical stretch in the case of channel cavity LIBS. Plasma emission intensity was higher during the interaction between the plasma and the reflected shockwave compared to flat surface LIBS analysis. Furthermore, Weiss et al.91 found that plasma core position was more stable and allowed to find an optimal signal collection angle. Enhancements in Cu(I) emission intensity in cylindrical cavities were described by Wang et al.,92 when ablating a Cu sample. This was attributed to interaction of the plasma with two reflected shockwaves produced at different delay times.

Accurate and precise quantitative analysis using LIBS is challenging, requiring an optimised experimental set-up and effective data treatment. Anderson et al.93 developed a sub-model method, which required a substantial training set, for improving the accuracy of quantitative LIBS analysis of a range of diverse target materials. Several regression models, each trained on a restricted compositional range and combined using a linear weighted sum, were used to get more accurate predictions. This is applicable to any multivariate regression method, and is a useful addition to the toolbox of methods used to obtain the quantitative LIBS data. Castro et al.94 investigated several data normalisation methods for quantitative analyses of metal samples (e.g. alloys and steel), and for classification purposes. The methods were used to compensate for signal variation and sample matrix differences, and to improve the calibration response. Both multivariate and univariate models performed similarly for the determination of ten different analytes in eighty metal samples. Classification models were also proposed for rapid sample differentiation, best results being obtained with the K-nearest neighbor model. A large set of geologically-diverse samples was analysed by Dyar et al.95 as part of an investigation into the accuracy of elemental predictions using models that incorporated only the spectral regions of interest, in comparison to using the entire spectrum. It was concluded that univariate predictions based on single emission lines were the least accurate, compared with multivariate analysis (e.g. full-spectra PLS), independently of how carefully the spectral region was chosen. Similarly, Guezenoc et al.96 significantly improved the ability to determine the concentration of potassium in agricultural soils by using PLS models. Advance variable selection was employed and the PLS model only included the variables related to the edges of the selected spectral lines of K (e.g. at 766.49 nm and 769.90 nm) which exhibited unusual profiles. This work highlights the importance of correct variable selection on the prediction ability of quantitative PLS models.

The availability of CRMs for matrix-matched LIBS quantitative analysis is limited for some applications. Hence, standard-less quantitative analysis based on fitting model-generated synthetic spectra to experimental spectra is an area of interest. Monte-Carlo (MC) methods have been employed for standard-less LIBS analysis but they require long computation time. Demidov et al.97 developed an MC method, based on massively parallel computing, that reduced the analysis time from tens of minutes to several seconds per sample. Multi-element analysis with relative error within tens of % was possible, which is sufficient for industrial applications such as steel slag analysis.

3.1.2 Instrumentation. Instrumental developments in standoff-LIBS have been an area of interest over the last few years. For instance, standoff LIBS using a spatial heterodyne spectrometer, with 10 mm diffraction gratings and no moving parts, was demonstrated by Barnett et al.98 The instrument had high light throughput, such that measurements of samples located up to 20 m from the spectrometer were possible without collection optics, and with a wide field of view of up to 1°. Laser focus was optimised at standoff distances by monitoring the shock wave intensity of the LIBS plasma with a microphone and oscilloscope. Three different optical configurations with large depth of field were evaluated in order to arrive at a compromise fixed focus optical arrangement for light collection. The best collection efficiency was obtained using two spherical mirrors with the same curvature radius. Simulations agreed with experimental results obtained for the analysis of moving samples of coal.

A gas-assisted, localised, liquid discharge underwater LIBS instrument, with in situ electrosorption was developed by Jiang et al.,99 for the direct determination of CrVI in solution. This novel instrument provided an instantaneous gaseous environment for underwater LIBS measurements using flowing nitrogen to force the solution to leak out from the laser channel and plasma region. The electrosorption process was used to preconcentrate the dissolved CrVI.

Combined LIBS-Raman systems provide superior and complimentary analytical capabilities in direct solid analysis because LIBS is very sensitive to elemental composition while Raman provides information on the mineral structure and polymorphs. Such a system was developed by Choi et al.,100 using a laser beam expander to facilitate collection of the Raman signal with minimal interferences from the plasma, together with a focusing lens of small diameter to generate the strong laser induced plasma. A key part of the design was to generate the Raman scattering separately from the LIBS plasma by adjusting the position of the focusing lens to control the area of Raman scattering, allowing for sufficiently long gate width and wide area for Raman detection. An optical fibre was axially positioned in order to independently collect the Raman, LIBS, or combined Raman–LIBS signal.

A bench-top LIBS instrument was developed by Zha et al.101 using a home-made miniature laser source, for the qualitative and quantitative analyses of nickel ore. A compact double-pulse LIBS system, less than 10 kg in weight, was developed for field applications by Li et al.102

LIBS analysis of samples with uneven surfaces is challenging because of laser focusing problems. Cortez et al.103 developed a simple device, based on two commercial laser pointers, that positioned the sample surface at a reproducible distance (better than ±0.2 mm) from the laser focusing lens. This improved the precision by minimising fluctuations in laser fluence at the sample surface.

3.1.3 Novel LIBS approaches. Standoff LIBS for remote analysis (10 m) of suspended matter in a free stream of air was investigated by Álvarez-Trujillo et al.104 The major challenge for real-time monitoring of this aqueous aerosol was the variation in composition, size and density of particles, which affected the breakdown probability, line intensity and plasma emission continuum. These effects were studied using solutions containing either NaCl or Na2SO4 at different concentrations. A method to determine the average droplet size from the slope of H (I) and O (I) lines, was compared with one using the plasma continuum emission. The former method was slightly affected by the salt content in the droplets.

LIBS has great potential for the chemical analysis of submerged solids. Gavrilovic et al.105 investigated the spatial and temporal evolution of an underwater LIBS plasma formed on non-metallic targets (e.g. Al2O3), following up previous studies of conductive targets (e.g. Al). The secondary plasma, caused by backward heating and slow target evaporation into the growing vapour bubble, was studied using fast photography, shadowgraphy and Schlieren imaging. The formation of a secondary plasma, with a narrow interaction region and an almost spherical plume shape, was observed for alumina. This was in contrast to the large volume and intense secondary plasma that was observed with an Al target. The spectrum from the secondary plasma was almost free from any continuum components and showed narrow emission lines from low excited states.

The LIBS parameters affecting the spectral response of metallic targets in an oceanic environment were investigated by Lopez-Claros et al.106 As the hydrostatic pressure increased, plasma persistence and cavitation bubble size decreased. In contrast with observations at atmospheric pressure, the emission spectra contained self-absorbed atomic lines due to strong radiative recombination in the electron-rich, high pressure environment.

Direct determination of halogen elements is difficult by OES because of their low excitation efficiency in the LIBS plasma; an alternative is to observe their molecular emission. Alvarez-Llamas et al.107 developed an indirect method to determine trace amounts of F in solid samples using atmospheric-air LIBS, based on the detection of CaF molecular emission bands. The bands were observed when mixing Ca with F concentrations of between 50 and 600 μg g−1, and variable amounts of Ca. A linear relationship between CaF emission and F concentration in the sample was observed. The LOD for fluorine was improved by more than 1 order of magnitude compared with the OES method. The same authors108 developed the method further for quantitation of F in Ca-free samples by nebulisation of a Ca-containing solution onto the surface of a F containing sample, in order to obtain the desired CaF molecular emission. Nebulisation parameters were optimised to maximise the molecular emission. The LOD for F, of 50 μg g−1, was similar to that obtained for Ca-containing samples. This method opens new ways for the determination of halogens in a great variety of solid samples using LIBS.

Nanoparticle enhanced laser induced breakdown spectroscopy (NELIBS) was used by Koral et al.109 to enhance molecular band emission. They used silver NPs on the surface of Al-based alloys to enhance the AlO (B2Σ+ → χ+Σ+) system emission by more than one order of magnitude. They ascribed this to the increased production of atomic species that are the molecular precursors to the formation of AlO.

A novel method for the analysis of liquid samples via a hydrogel-based solidification technique was developed by Lin et al.110 An aqueous solution was poured directly into a hydroscopic sodium polyacrylate resin, forming a hydrogel which was amenable to LIBS analysis. The proposed method required a shorter processing time compared to other methods for liquid sample analysis. Therefore, this method might be very useful for in situ LIBS environmental analysis of liquid samples.

3.2 Laser atomic absorption spectroscopy (LAAS)

Absorption and emission spectroscopy of uranium species was investigated by Skrodzki et al.111 They used LAAS operated in a vacuum chamber between 1 and 760 Torr, with both N2 and air, to explore the mechanisms leading to U signal quenching. A significant reduction in atomic ground state transitions was observed in air compared with nitrogen, suggesting that chemical reactions with oxygen were significant. However, U emission was unaffected. Resonance absorption techniques were also used by Miyabe et al.112 to investigate the spectroscopic properties of Pu. Remote LAAS of a plutonium oxide sample was performed in a glove box. Absorption spectra for 17 transitions from ground states of Pu atoms and ions were examined to determine the absorbance, isotope shift, and hyperfine splitting by Voigt profile fitting. Transitions at 375.4 nm, 420.6 nm and 648.9 nm were selected for isotopic analysis by LAAS. Spectral interference and background from U atoms and ions was found to be negligible. The conclusion was that LAAS might be a good candidate for remote isotopic analysis of highly radioactive nuclear fuels and waste materials containing multiple actinide elements.

4. Isotope analysis

An item that is becoming prominent in isotopic analytical literature is the contrast in the terminology used to describe uncertainty between different analytical communities. This was highlighted in a paper by Pfeifer et al.113 who, possibly following encouragement by a reviewer, elaborated on this as an aside to a study redefining the Ta isotope ratio. Their paper stated their use of the guidelines set out by the Joint Committee for Guides in Metrology (Joint Committee for Guides in Metrology (JCGM), International vocabulary of metrology – basic and general concepts and associated terms (VIM), BIPM, 2012) to define measurement uncertainty. These guidelines form the standard terminology for most analytical chemists. However, they also acknowledged their own analytical geochemistry community by inserting parenthesised reference to their commonly used expressions of uncertainty. An example of this contrast is the terminology for the validity of measurements. In geochemistry this is reported as ‘external reproducibility’ which is the variability of a repeated standard expressed in standard deviations (usually 2 SD). The equivalent term used in VIM is ‘intermediate precision’ which reports repeated measurements for the same instrument over an extended time period. Likewise the term ‘internal precision’ in geochemistry refers to the variation in the mean of successive measurements that comprise the analysis of a single sample. For example in mass spectrometry each isotope ratio analysis may consist of 100 successive 5 s measurements (or time integrations). As this internal precision is effectively the confidence of the location of the mean, it is proportional to the number of measurements made. Hence this precision is defined by the term standard error (SE = SD/n−0.5; where n = number of measurements). Pfeifer et al.113 recognise that the closest equivalent in VIM terminology is repeatability, which includes replicate measurements on the same or similar objects over a short time period.

4.1 Reviews

Following a review of the usage of TIMS in nuclear science last year, Aggarwal114 have provided a review of boron isotope ratio measurement. This included a systematic assessment of the positive and negative ion TIMS techniques alongside more recent analytical developments in MC-ICP-MS and QQQ-ICP-MS. A particularly useful feature was the inclusion of a tabulated assessment of analytical parameters on the key instrument types. For example, parameters included isobaric interferences, mass fractionation, ionisation efficiency and time required for data acquisition. In addition, a figure summarised the precision of each analytical technique in relation to the amount of B needed to complete the measurement. Boron isotopes were reviewed in an issue of Elements (Elements, 13(4), 2017) as a themed collection of papers that cover a variety of Earth Science applications. These included the use of B as a paleo-pH meter to track ancient atmospheric CO2,115 tracking processes through subduction zones116 and how B behaved in the early solar system.117

Albarede118 examined the recent applications of stable Cu, S and Zn isotopes in medical cases. These isotopes have been found to vary systematically with the development of cancer in patients. Isotopic fractionation is developed by the preference of O- and N-bearing functional groups for heavy isotopes of Cu and Zn, and S-groups for light isotopes. A result of this is that Cu in liver tumours has been found to be isotopically heavy whereas Zn in breast cancer tumours is lighter than healthy tissue. Consequently the review speculated that these isotope systematics may help to evaluate homeostasis during patient treatment.

A review of the key aspects of isotope ratio determinations by LA-ICP-MS were made by Woodhead et al.119 This included a useful summary table of “when and where” in the literature particular isotope systems and tracers were first reported using this technique. The review included aspects of isotope systems and their applications, including U–Th, Sr, U–Pb as well as the stable tracers of Fe, Mg and Si. Instrumental developments such as ablation cell design, ablation protocols, detector advances and collision cells were discussed and what potential advances could be made in the future.

4.2 Isotope dilution analysis

Clases120 evaluated and devised a strategy for a novel quantitation method termed isobaric dilution analysis (IBDA). This worked in a similar fashion to ID via ICP-MS, but the spike added was a different element to the one being measured – similar to the commonly used technique of measuring the isotope ratios of an external element to estimate instrumental mass bias. This technique relied on the presence of an isobaric isotope in the spike element and the target element, as well as similar chemical properties and analytical responses. This technique has the potential to assist in the measurement of elements such as Tc.

4.3 Isotope ratio analysis

4.3.1. Radiogenic isotope ratio analysis. 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr is a radiogenic isotope system that has been used for a variety of Earth Science applications for approaching 70 years. Measurement precision of this ratio when Sr is available in microgram quantities is very high, typically around 0.002%. Bazzano121 used a desolvating nebuliser and a high-efficiency jet interface in conjunction with an MC-ICP-MS fitted with high-impedance resistor amplifier circuits to measure 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr in Antarctic snow. This was a challenging target as snow contains 5–20 pg g−1 Sr. Analysis of final solutions containing 0.4–0.8 ng mL−1 followed evaporative and chromatographic preconcentration and separation of Sr from 30 g of snow. Results indicated that precisions of ∼0.03% could be achieved using this technique, despite the analyte being around 60 times the level of the Sr blank. Results were used to demonstrate that useful discrimination could be made between different source areas of contamination in snow.

Th isotope determinations are hindered by the generally low abundance of 230Th in most natural samples. A modified Th isolation protocol was reported by Carpentier et al.122 utilising two-step extraction micro-chromatography. Reported precision was comparable between synthetic Th standards and rock reference materials, indicating no influence from sample matrix. In addition the authors reported the first 230Th[thin space (1/6-em)]:[thin space (1/6-em)]232Th for the BIR-1 and BE-N reference materials.

A combination of 204Pb–207Pb double spike and total evaporation techniques were used by Fukami123 for IR of sub-ng quantities of Pb. They monitored low-intensity TIMS ion beams to remove the larger 204Pb error from the heating integration, also boosted in abundance by the 204Pb–207Pb double spike. Final correction for fractionation was achieved by external normalisation to SRM NBS 981. Limitations of mass fractionation corrections using Tl as a fractionation proxy for Pb isotope determinations were investigated by Vaglarov et al.124 This study added typical geological major matrix elements to Pb–Tl solutions for MC-ICP-MS analysis. They found little influence on the final corrected Pb isotope ratios when Mg, Al or Fe were added. However, they found significant differences in the fractionation behaviour of the two elements when Ca was added to the analyte solution. They suggested that this effect was caused by precipitation of a Ca–Pb complex in the analytical solution. Analysis of Pb isotopes by LA-SC-ICP-MS was evaluated by Pietruszka and Neymark.125 A double-focusing SC-ICP-MS instrument with an ion-counting detector was used to measure a series of synthetic and natural glasses in order to evaluate precision. Results indicated that spot-size and matrix effects produced ∼0.1% per AMU discrepancies in fractionation corrected isotope ratios. Sensitivity and precision of this technique, was found to be compromised relative to MC-ICP-MS due to the partitioning of dwell time on each isotope of interest. Tl is also used as a mass fractionation internal standard during Hg isotope analysis. Yin et al.126 studied the effects of Hg hydride formation in MC-ICP-MS and the potential for this to disrupt the 203Tl[thin space (1/6-em)]:[thin space (1/6-em)]205Tl systematics. It was concluded that these effects can be minimised if Tl concentrations are sufficiently high in the measurement solution and Hg concentrations are matched between the samples and intervening standards.

Gonzalez-Guzman127 developed new software (IsotopeHf®) to process data from Hf–Yb–Lu isotopes measured by MC-ICP-MS through to final Lu–Hf isotope dilution concentrations and Hf isotope ratios. They developed a two-stage column separation protocol to separate Lu and hence minimise isobaric interference on 176Hf.

4.3.2 Stable isotope ratio analysis. Measurement techniques using MC-ICP-MS and negative ion mode TIMS (N-TIMS) reported in last year’s review were advanced further by Archer et al.128 who utilised 1013 Ω resistors to monitor and correct for 18O[thin space (1/6-em)]:[thin space (1/6-em)]16O isotopic variations on WO3. In addition, they reported a revised O mass fractionation line to produce more accurate oxide interference corrections. Improvements to Hg IR analysis were also reported by Berail et al.129 They combined a CVG system with a dual gold amalgamation trap to increase the purity and concentration of Hg prior to introduction to MC-ICP-MS. This allowed determination of Hg isotopic compositions in samples containing only 5 ng L−1, which is suitable for the measurement of environmental samples, such as natural waters.

MC-ICP-MS running under cold plasma conditions for Fe isotopic analysis was evaluated by Chernonozhkin et al.130 Combinations of normal or cold plasma, wet or dry plasma and standard or jet interface were investigated. The cold–wet-standard combination did not suppress ArOH+ formation, whereas cold–dry-standard conditions did, but with a concomitant increase in ArN+. Optimum conditions were cold plasma combined with a standard interface and low mass resolution.

Oxygen isotope measurement by SIMS were the subject of a study by Fabrega et al.,131 who produced a mathematical model to describe the instrumental mass fractionation of O generated by sputtering, ionisation, extraction, beam transmission and detection. Using response surface modelling, they successfully predicted the fractionation values for known and unknown samples, concluding that fractionation predictions are dominated by the electrostatic deflector parameters.

Measurement of Si isotope ratios is an expanding field and Frick et al.132 experimented with combining such measurements with simultaneous determination of major element composition. This study involved splitting a LA generated aerosol into MC-ICP-MS and Q-ICP-MS via a Y-split in the LA sample carrier. This was compared with a second set-up of simultaneous LA-MC-ICP-MS or -OES. One of the key challenges reported was maintaining stable operating conditions while partitioning the aerosol into the second instrument. Overall, split-stream laser ablation MC-ICP-MS – ICP-MS combination was favoured as this resulted in the best simultaneous major element and isotopic analysis.

Boron isotopes can be precisely measured by N-TIMS or MC-ICP-MS. Farmer et al.133 compared the two techniques and found Δ11B values in calcites were 0.5–2.7‰ lower on MC-ICP-MS. This offset was not explained by the blank, but did correlate with chemical composition of the sample. The overall conclusion was that both techniques were equally applicable to the measurement of paleo-pH reconstructions. B and Li were measured on basalt glasses, minerals and SRM61x glasses by Kimura et al.,134 to investigate the effects and origins of isotopic fractionation during LA-MC-ICP-MS. They monitored the effects of LA parameters such as crater size and laser repetition rate. One conclusion was that the introduction of mass into the ICP was the cause of isotopic fractionation, caused by ablated material lowering the plasma temperature which resulted in Rayleigh fractionation of lighter isotopes as particles broke down. The outcome was the development of a protocol to correct fractionation using the relationship between crater volume and isotopic fractionation.

Plasma source mass spectrometry, including MC-ICP-MS usually involves introduction of the analyte as an aerosol. This necessitates the sample being aspirated from either a mineral acid (e.g. HF, HCl, HNO3) or H2O. These media are known to affect the mass fractionation of B during measurement. Chen135 found that as HCl and HNO3 strength increased, isotopic mass bias reduced and B signals increased. Hydrofluoric acid produced the highest signal increases, but the mass bias did not change with acid strength. They concluded that acid strength should be optimised for signal intensity and exactly matched between samples and standards. Nitric acid was found to have a less rigorous acid match requirement compared to HCl.

Gueguen et al.136 investigated the measurement of Eu–Gd–Sm–Nd mixtures using LC-MC-ICP-MS. This was directed at measurement of isotopes of these elements for nuclear applications. In a similar instrument configuration, Gourgiotis137 examined the time lag of rare earth element isotopes entering the MC-ICP-MS from the liquid chromatograph. After correction for the electronic shift in isotope peaks, they found an earlier arrival of heavier isotopes at the detectors caused by chromatographic isotope fractionation. When this was corrected, the fractionation time lag was quantified to 0.0036 s amu−1. The authors proposed that their method of internal signal synchronisation could be used to examine fractionation in transient signals generated by other systems. Karasinski et al.138 used the LC-MC-ICP-MS coupling to determine Mg isotope ratios in natural water and rock samples. They reported precision and accuracy around 0.15‰ on the 26Mg[thin space (1/6-em)]:[thin space (1/6-em)]24Mg ratio in both wet- and dry-plasma modes, which is comparable to analysis following off-line Mg separation.

Mo isotope ratios are sensitive to redox state, so 98Mo[thin space (1/6-em)]:[thin space (1/6-em)]95Mo can be used as a proxy for reducing/sulfidic conditions in ancient oceans. Kerl et al.139 investigated the potential for isotope fractionation in thiomolybdates using LC-MC-ICP-MS. Reaction between MoO42− and variable excesses of sulfide demonstrated an Mo isotope fractionation during the formation of thiomolybdates. A novel analytical strategy for Mo isotope measurement was reported by Malinovsky et al.140 They used the technique of simultaneously measuring Sr isotopes alongside Mo to provide a monitor for instrumental mass fractionation. In their experiment the RF power of the plasma was incrementally varied to generate a well-defined relationship between the Sr and Mo systems. The results of this regression measurement promised an alternative high-precision method of calibrating Mo IR. This technique was also utilised by Oelze et al.,141 who investigated the measurement of Si isotopes in solutions doped with Mg. It was found that the addition of Mg acted as a matrix modifier, increased the sensitivity for Si and stabilised mass bias even during significant changes in the molarity of HCl.

Precision of Cr isotope determinations was improved by Li et al.142 who used an activator containing Nb2O5 powder dispersed in 0.4 M phosphoric acid to enhance ionisation from a TIMS filament. A double spike was used to correct instrumental mass fractionation. Results indicated a long-term reproducibility for 53Cr[thin space (1/6-em)]:[thin space (1/6-em)]52Cr of 0.035‰ in NIST SRM 3112a standard solution, and comparable results for rock standards such as BIR-1. Lithium IR measured on solid samples were investigated in a study by Lin et al.143 This study analysed USGS reference materials using a 193 nm excimer laser coupled to MC-ICP-MS. It was found that the 7Li[thin space (1/6-em)]:[thin space (1/6-em)]6Li ratio could be stabilised by using a high gas-flow rate (>0.8 L min−1) combined with shallower sampling of the plasma by adjusting the torch position.

Ta has two isotopes; 180Ta and 181Ta, with the latter representing ∼99.99% of its abundance. Hence it is challenging to measure minor natural variation in the ratio caused by cosmological or mass-dependent geological processes. Pfeifer et al.113 used the 1012 Ω and 1013 Ω high-impedance resistors in an MC-ICP-MS to re-evaluate the large-magnitude Ta IR. Careful correction for the 181Ta tail onto m/z 180, combined with mass bias estimations made using simultaneous measurement of Yb and Re, enabled an intermediate precision (‘external reproducibility’ in geochemistry) for the 180Ta[thin space (1/6-em)]:[thin space (1/6-em)]181Ta of ±4ε relative to a standard. The study also proposed a new absolute 180Ta[thin space (1/6-em)]:[thin space (1/6-em)]181Ta ratio of 0.00011705 and a new atomic weight for Ta.

Se isotope systems have attracted interest through the last few years. To help develop the system further, Ren et al.144 produced and calibrated a suite of Se isotopic standards with a range of isotope abundances, which are now commercially available.

Ba isotopes are other analytes of interest, with differing levels of fractionation observed in ocean water columns. Such variation promoted van Zuilen et al.145 to develop synthetic Ba isotope standards for the community, which covered a range of δ137Ba. In addition they measured δ137Ba on a range of geological reference materials in two separate mass spectrometry laboratories. The agreement between the laboratories was very close, and highlighted a significantly higher δ137Ba (∼0.1) in the Icelandic basalt standard BIR-1 (∼+0.19) relative to the other rock standards (∼+0.07). Yobregat et al.146 developed a method of column separation for Ba and Sr. Analysis used TIMS, and external reproducibility for 135Ba[thin space (1/6-em)]:[thin space (1/6-em)]136Ba was reported as ±4 ppm; around 2× better than most published data.

Ni isotopes were analysed in metal samples by LA-MC-ICP-MS by Weyrauch et al.147 Ni and Fe were investigated along a transect between taenite and kamacite phases in the Canyon Diablo meteorite. They demonstrated a significant difference in the Ni and Fe isotopes of the two phases which indicated a diffusion-driven fractionation during the replacement of taenite by kamacite.

Isotope ratio analysis of Ga has the potential to uncover specific biological processes because of fractionation on uptake by certain plants like mosses and lichens. Yuan et al.148 devised a separation procedure and mass spectrometric protocol to make a precise isotopic analysis of Ga. A two-stage column separation with AG 1-X4 and Ln-spec resin which yielded ∼99.8% recovery. Analysis using two different MC-ICP-MS instruments demonstrated precision better than 0.05‰ on δ71Ga. Measurements of a number of natural geological reservoirs revealed a significant variation in the Ga isotopic composition, being up to 1.83‰ δ71Ga.

Ir isotopes were examined using MC-ICP-MS to develop a calibrated, SI traceable, isotope ratio. Zhu et al.149 completed this using a regression model, which drew on the primary calibration points of NIST SRM 997 Tl and NIST SRM 989 Re, to correct the instrumental mass. This work also provided the source of a ‘delta zero’ reference standard (NRC Canada IRIS-1) for future Ir isotope measurements.

Isotope ratio analysis of Eu at low concentrations was reported by Arantes de Carvalho et al.,150 who separated Eu using di(2-ethylhexyl)orthophosphoric acid (HDEHP) resin, followed by measurement using MC-ICP-MS. Mass fractionation was corrected by combining internal standardisation with sample-standard bracketing. Results showed there was no discernible Eu isotopic variation between various natural waters.

A new Zn isotopic standard, AA-ETH Zn, was introduced by Archer and co-workers.151 This was developed to replace the commonly-used, but now exhausted, JMC-Lyon standard. Characterisation was by multiple techniques in multiple laboratories against all of the accepted reference materials. The volume, concentration and availability will make this standard a significant tool for Zn isotope analysis.

Double spike analysis has become the favoured method for correcting instrumental mass fractionation in the analysis of radiogenic and non-radiogenic isotope systems. To date, this has been restricted to systems with four or more isotopes. However, Coath et al.152 have devised a methodology for estimating the mass fractionation incurred by a three isotope system using a natural run and a double spike run. This relies on the double spike (mixture) run having an isotope composition where the mass bias curve and sample-double spike mixing curves are coincident. Evaluation of this technique indicated that it provided a robust method for the isotopic determination of Mg and Si isotopes.

4.3.3 Geological studies. U–Pb dating is a routine method using zircons, but Courtney-Davies et al.153 have overcome some of the issues of using hematite as a target for single mineral dating. They achieved matrix matching by mixing a U–Pb solution with a laser ablated synthetic hematite during LA-ICP-MS analysis. This enabled the validation of Pb–Pb hematite ages (1.6 Ga) from the Olympic Dam deposit in Australia and to tie the mineralisation to large igneous province magmatism in this region. Zhang et al.154 devised a method for measuring Pb isotopes in Hg-rich sulfides by LA-MC-ICP-MS. Isobaric interference by 204Hg is a significant limitation to precision of Pb isotopic measurements, but this study utilised a gas exchange device which replaced the ambient aerosol gas with Ar. This was found to almost completely remove the Hg interference and resulted in 20xPb[thin space (1/6-em)]:[thin space (1/6-em)]204Pb precision around 0.2%. Ek et al.155 described a new method of determining Pd isotope analysis. A MC-SF-ICP-MS instrument was used to accurately measure all Pd isotopes. This included 102Pd an important isotope for understanding the development of the solar system.

Isotopes of S were determined in solid sulfide ore samples using SIMS, by LaFlamme et al.156 This study measured minerals such as pyrite, chalcopyrite and pentlandite in Australian komatiite-hosted mineral deposits. Results demonstrated that all of these minerals were in isotopic equilibrium, but are different from the isotope systematics of the magmatic system. They concluded that the S isotopes were derived from the assimilation of sulfidic shale at distance from the volcanic vent rather than digesting volcanogenic sulfides in vent-proximal environments. Improvements were made to a desolvation technique, used in the measurement of S isotopes by MC-ICP-MS, by Yu et al.157 They found that by adding Na to the analytical solution, the transmission of S through the desolvator was dramatically increased. The added Na was also found to reduce the memory effect between samples during the measurement of low S pore fluids in marginal sea sediments.

Lugli et al.158 used LA-MC-ICP-MS to determine 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr in the teeth of a Pleistocene Italian Rhinoceros. They found that there were subtle changes in the 87Sr[thin space (1/6-em)]:[thin space (1/6-em)]86Sr across a tooth profile, related to the growth of the tooth. This was ascribed to seasonal migration of the beasts between a more radiogenic volcanic terrain and limestone-dominated grasslands.

Because V is redox-sensitive, its mobilisation under various conditions of pH and Eh can result in significant stable isotope fractionation up to several per mil. Schuth et al.159 examined several V-rich minerals using fs LA-MC-ICP-MS. Additional analyses of the minerals were also made using a conventional dissolution and chromatographic separation followed by solution-based MC-ICP-MS. The minerals had δ51V in the range −0.5 to +1.3‰ which greatly extends the known range of terrestrial V IR to more positive values. Potentially V isotopes may be a proxy for redox processes during the formation of V ores.

4.3.4 Nuclear forensics. Boulyga160 examined the potential for high-precision U isotope measurement in sub-ng samples. They used MC-ICP-MS equipped with a ‘jet interface’ and desolvation sample introduction system to achieve ion registration efficiencies of ∼2%. Overall uncertainties were less than 0.2% for 235U[thin space (1/6-em)]:[thin space (1/6-em)]238U ratios obtained with 0.5 ng U. Phillips et al.161 used 2D fluorescence spectroscopy to distinguish between the atomic absorption of 235U and 238U in a fs LA plume. Results indicated the potential for this technique to measure sub μg levels of U by stand-off detection under ambient conditions. RIMS was used by Savina et al.162 to measure U isotopes in solid samples. Yields of ∼24% were achieved on U metal, but were significantly lower for the oxides. Reproducibility of the 235U[thin space (1/6-em)]:[thin space (1/6-em)]238U ratio was <0.5% RSD.

The LA-ICP-MS determination of U isotopes in micrometric uranium particles is challenging, chiefly due to the transient nature of the signal. Donard et al.163 tackled this by filtering the spikes (short-term fluctuations) in the signal. This reduced the dispersion of ratios and improved accuracy of measurement. Hence, without using a multi-collector, they were able to achieve external precisions of ∼5% for 235U[thin space (1/6-em)]:[thin space (1/6-em)]238U and ∼12% for 234U[thin space (1/6-em)]:[thin space (1/6-em)]238U. An indirect method of determining U isotope ratios in U particles was proposed by Wang et al.164 They used the Pb isotope signature, measured by LA-ICP-MS, as a fingerprint of the potential origins of particles. It promised to distinguish between a mining/ore-body and anthropogenic sources.

Dunne et al.165 optimised the chemical separation of Cs for TIMS analysis, to obtain precise 135Cs[thin space (1/6-em)]:[thin space (1/6-em)]137Cs ratios. Glucose was selected as activator to yield a high ionisation efficiency for the TIMS filament source. As well as problems caused by the low abundance of 135Cs and 137Cs they encountered scatter and spectral interference using ion counting detectors. This was reduced using a Faraday cup. Their methods were evaluated using environmental samples and standards derived from regions affected by the Chernobyl and Fukushima nuclear incidents.

Low-concentration Pu isotope measurement was investigated by Inglis et al.166 The authors used a total evaporation TIMS technique and achieved 240Pu[thin space (1/6-em)]:[thin space (1/6-em)]239Pu external reproducibility of >10% RSD (2 s) on samples with less than 1 fg of 240Pu.

Conflicts of interest

There are no conflicts to declare.

5. Glossary of abbreviations

Whenever suitable, elements may be referred to by their chemical symbols and compounds by their formulae. The following abbreviations may be used without definition. Abbreviations also cover the plural form.
2DTwo dimensional
3DThree dimensional
AAAtomic absorption
AASAtomic absorption spectrometry
ADIAmbient desorption ionisation
AESAtomic emission spectrometry
AFAtomic fluorescence
AF4Asymmetric flow field-flow fractionation
AFMAtomic force microscopy
AFSAtomic fluorescence spectrometry
APDCAmmonium pyrrolidine dithiocarbamate
APGDAtmospheric pressure glow discharge
APSAerodynamic particle sizer
CCDCharge coupled detector
CCPCapacitively coupled plasma
CRMCertified reference material
CRSCavity ringdown spectroscopy
CSContinuum source
CVCold vapour
CVGChemical vapour generation
DBDDielectric barrier discharge
DCDirect current
DC-GD-QMSDirect current glow discharge quadrupole mass spectrometry
DDWDistilled deionised water
DLDiode laser
DLTVDiode laser thermal vaporisation
DMADimethylarsenic
DOF-MSDistance of flight mass spectrometry
DOTA1,4,7,10-Tetraazacyclo-dodecane N,N′,N′′,N′′′-tetra acetic acid
EDB-LIBSElectrodynamic balance trap laser induced breakdown spectroscopy
ES-MSElectrospray mass spectrometry
ETVElectrothermal vaporisation
ETV-AASElectrothermal vaporisation atomic absorption spectrometry
ETV-ICP-MSElectrothermal vaporisation inductively coupled plasma mass spectrometry
EVGElectrochemical vapour generation
FAPAFlowing atmospheric pressure afterglow
FFFlame furnace
FFPNFlow focusing pneumatic nebuliser
FIFlow injection
FMPSFast mobility particle sizer
FWHMFull width at half maximum
GDGlow discharge
GD-MSGlow discharge mass spectrometry
GD-OESGlow discharge optical emission spectrometry
GFAASGraphite furnace atomic absorption spectrometry
HG-AASHydride generation atomic absorption spectrometry
HG-AFSHydride generation atomic fluorescence spectrometry
HPLC-ICP-MSHigh performance liquid chromatography inductively coupled plasma mass spectrometry
HRHigh resolution
HR-CS-GFAASHigh resolution continuum source graphite furnace atomic absorption spectrometry
hTISISHeated torch integrated sample introduction system
HVHigh voltage
ICP-AESInductively coupled plasma atomic emission spectrometry
ICP-OESInductively coupled plasma optical emission spectrometry
ICP-DOF-MSInductively coupled plasma distance of flight mass spectrometry
ICP-MSInductively coupled plasma mass spectrometry
ICP-MS/MSTriple quadrupole inductively coupled plasma mass spectrometry
ICP-MS-MSInductively coupled plasma multi-collector mass spectrometry
ICP-OESInductively coupled plasma optical emission spectrometry
ICP-QMSInductively coupled plasma quadrupole mass spectrometry
ICP-QQQ-MSInductively coupled plasma triple quadrupole mass spectrometry
ICP-TOF-MSInductively coupled plasma time-of-flight mass spectrometry
IDIsotope dilution
IDAIsotope dilution analysis
ID-ICP-MSIsotope dilution inductively coupled plasma mass spectrometry
ID-MRM-MSIsotope dilution multiple reaction monitoring mass spectrometry
ID-MSIsotope dilution mass spectrometry
ILIonic liquid
IRIsotope ratio
IRMSIsotope ratio mass spectrometry
LALaser ablation
LA-ICP-MSLaser ablation inductively coupled plasma mass spectrometry
LA-MC-ICP-MSLaser ablation multi-collector inductively coupled plasma mass spectrometry
LAMISLaser ablation molecular isotopic spectrometry
LCLiquid chromatography
LDRLinear dynamic range
LEPLiquid electrode microplasma
LIBSLaser induced breakdown spectroscopy
LIFLaser induced fluorescence
LODLimit of detection
LOVLab-on-a-valve
LPMELiquid phase microextraction
LTPLow temperature plasma
MALDIMatrix-assisted laser desorption ionisation
MCMonte Carlo
MC-ICP-MSMulti-collector inductively coupled plasma mass spectrometry
MIPMicrowave induced plasma
MIP-OESMicrowave induced plasma optical emission spectrometry
MMAMonomethylarsenic
MNPMagnetic nanoparticle
MPTMicrowave plasma torch
MW-LIBSMicrowave assisted LIBS
Nd:YAGNeodymium doped:yttrium aluminum garnet
n e Electron number density
NELIBSNanoparticle enhanced LIBS
NISTNational institute of standards and technology
NPNanoparticles
o.d.Outer diameter
OPOOptical parametric oscillator
PCRPolymerase chain reaction
PDMSPolydimethylsiloxane
PFHPerfluorohexane
PLSPartial least squares
PNPneumatic nebuliser
PNCParticle number concentration
ppbParts per billion (10−9)
ppmParts per million (10−6)
ppqParts per quadrillion (10−15)
pptParts per trillion (10−12)
PVGPhotochemical vapour generation
PVPPolyvinylpyrrolidone
QQuadrupole
QTFQuartz tube furnace
RFRadiofrequency
RIMSResonance ionisation mass spectrometry
RSDRelative standard deviation
RVMRelevance vector machine
SALDSubstrate assisted laser desorption
SBRSignal-to-background ratio
SCGDSolution cathode glow discharge
SEMScanning electron microscopy
SNRSignal-to-noise ratio
SpSingle particle
SPMESolid phase microextraction
TDLAASTunable diode laser atomic absorption spectroscopy
T e Electron temperature
TETransport efficiency
TEMTransmission electron microscopy
T gas Gas temperature
TIMSThermal ionisation mass spectrometry
T ion Ionisation temperature
TOFTime-of-flight
TOF-MSTime-of-flight mass spectrometry
T rot Rotational temperature
UVUltraviolet
VGVapour generation
XPSX-ray photoelectron spectroscopy

6. References

  1. O. T. Butler, W. R. Cairns, J. M. Cook and C. M. Davidson, J. Anal. At. Spectrom., 2017, 32(1), 11–57 RSC .
  2. S. Carter, A. Fisher, B. Gibson, J. Marshall, B. Russell and I. Whiteside, J. Anal. At. Spectrom., 2017, 32(11), 2068–2117 RSC .
  3. R. Clough, C. F. Harrington, S. J. Hill, Y. Madrid and J. F. Tyson, J. Anal. At. Spectrom., 2017, 32(7), 1239–1282 RSC .
  4. E. H. Evans, J. Pisonero, C. M. M. Smith and R. N. Taylor, J. Anal. At. Spectrom., 2017, 32(5), 869–889 RSC .
  5. A. Taylor, N. Barlow, M. P. Day, S. Hill, M. Patriarca and M. White, J. Anal. At. Spectrom., 2017, 32, 432–476 RSC .
  6. M. West, A. T. Ellis, P. J. Potts, C. Streli, C. Vanhoof and P. Wobrauschek, J. Anal. At. Spectrom., 2016, 31(9), 1706–1755 RSC .
  7. M. He, L. Huang, B. Zhao, B. Chen and B. Hu, Anal. Chim. Acta, 2017, 973, 1–24 CrossRef CAS PubMed .
  8. H. Ahmad, A. A. Jalil and S. Triwahyono, RSC Adv., 2016, 6(91), 88110–88116 RSC .
  9. V. Zarezade, M. Behbahani, F. Omidi, H. S. Abandansari and G. Hesam, RSC Adv., 2016, 6(105), 103499–103507 RSC .
  10. O. Chahrour and J. Malone, Protein Pept. Lett., 2017, 24(3), 253–266 CrossRef CAS PubMed .
  11. Z. R. Liu, X. T. Li, G. Y. Xiao, B. B. Chen, M. He and B. Hu, TrAC, Trends Anal. Chem., 2017, 93, 78–101 CrossRef CAS .
  12. X. Zhang, B. Chen, M. He, H. Wang and B. Hu, Biosens. Bioelectron., 2016, 86, 736–740 CrossRef CAS PubMed .
  13. X. Li, B. Chen, M. He, H. Wang, G. Xiao, B. Yang and B. Hu, Biosens. Bioelectron., 2017, 90, 343–348 CrossRef CAS PubMed .
  14. J. A. Ko and H. B. Lim, Anal. Chim. Acta, 2016, 938, 1–6 CrossRef CAS PubMed .
  15. Z. Liu, B. Yang, B. Chen, M. He and B. Hu, Analyst, 2016, 142(1), 197–205 RSC .
  16. B. Yang, Y. Zhang, B. Chen, M. He, X. Yin, H. Wang, X. Li and B. Hu, Biosens. Bioelectron., 2017, 96, 77–83 CrossRef CAS PubMed .
  17. N. Tang, Z. Li, L. Yang and Q. Wang, Anal. Chem., 2016, 88(20), 9890–9896 CrossRef CAS PubMed .
  18. S. Li, B. B. Chen, M. He and B. Hu, J. Anal. At. Spectrom., 2017, 32(6), 1186–1195 RSC .
  19. F. Calderon-Celis, S. Diez-Fernandez, J. M. Costa-Fernandez, J. R. Encinar, J. J. Calvete and A. Sanz-Medel, Anal. Chem., 2016, 88(19), 9699–9706 CrossRef CAS PubMed .
  20. P. Chen, P. Wu, Y. Zhang, J. Chen, X. Jiang, C. Zheng and X. Hou, Anal. Chem., 2016, 88(24), 12386–12392 CrossRef CAS PubMed .
  21. Y. Cao, B. Deng, L. Yan and H. Huang, Talanta, 2017, 167, 520–525 CrossRef CAS PubMed .
  22. M. Slachcinski, Talanta, 2016, 161, 812–818 CrossRef CAS PubMed .
  23. J. Liu, K. E. Murphy, M. R. Winchester and V. A. Hackley, Anal. Bioanal. Chem., 2017, 409(25), 6027–6039 CrossRef CAS PubMed .
  24. D. M. Schwertfeger, J. R. Velicogna, A. H. Jesmer, R. P. Scroggins and J. I. Princz, Anal. Chem., 2016, 88(20), 9908–9914 CrossRef CAS PubMed .
  25. K. Newman, C. Metcalfe, J. Martin, H. Hintelmann, P. Shaw and A. Donard, J. Anal. At. Spectrom., 2016, 31(10), 2069–2077 RSC .
  26. S. I. Miyashita, H. Mitsuhashi, S. I. Fujii, A. Takatsu, K. Inagaki and T. Fujimoto, Anal. Bioanal. Chem., 2017, 409(6), 1531–1545 CrossRef CAS PubMed .
  27. I. Kalomista, A. Keri and G. Galbacs, Talanta, 2017, 172, 147–154 CrossRef CAS PubMed .
  28. X. C. Duan, R. Sun and J. L. Fang, Spectrochim. Acta, Part B, 2017, 128, 11–16 CrossRef CAS .
  29. H. Matusiewicz and M. Slachcinski, Microchem. J., 2017, 130, 345–352 CrossRef CAS .
  30. M. T. Li, Y. J. Deng, C. B. Zheng, X. M. Jiang and X. D. Hou, J. Anal. At. Spectrom., 2016, 31(12), 2427–2433 RSC .
  31. Q. Chen, Y. Lin, Y. Tian, L. Wu, L. Yang, X. Hou and C. Zheng, Anal. Chem., 2017, 89(3), 2093–2100 CrossRef CAS PubMed .
  32. X. Liu, Z. Zhu, H. Li, D. He, Y. Li, H. Zheng, Y. Gan, Y. Li, N. S. Belshaw and S. Hu, Anal. Chem., 2017, 89(12), 6827–6833 CrossRef CAS PubMed .
  33. X. Liu, Z. Liu, Z. Zhu, D. He, S. Yao, H. Zheng and S. Hu, Anal. Chem., 2017, 89(6), 3739–3746 CrossRef CAS PubMed .
  34. J. Luo, F. J. Xu, J. P. Tu, X. Wu and X. D. Hou, Microchem. J., 2017, 132, 245–250 CrossRef CAS .
  35. L. Lobo, B. Fernandez and R. Pereiro, J. Anal. At. Spectrom., 2017, 32(5), 920–930 RSC .
  36. S. Gaiaschi, S. Richard, P. Chapon and O. Acher, J. Anal. At. Spectrom., 2017, 32(9), 1798–1804 RSC .
  37. A. P. Storey, S. J. Ray, V. Hoffmann, M. Voronov, C. Engelhard, W. Buscher and G. M. Hieftje, Spectrochim. Acta, Part B, 2016, 126, 37–43 CrossRef CAS .
  38. A. P. Storey, S. J. Ray, V. Hoffmann, M. Voronov, C. Engelhard, W. Buscher and G. M. Hieftje, Appl. Spectrosc., 2017, 71(6), 1280–1288 CrossRef CAS PubMed .
  39. P. Konarski, M. Misnik and A. Zawada, J. Anal. At. Spectrom., 2016, 31(11), 2192–2197 RSC .
  40. A. J. Schwartz, K. L. Williams, G. M. Hieftje and J. T. Shelley, Anal. Chim. Acta, 2017, 950, 119–128 CrossRef CAS PubMed .
  41. E. V. Alonso, M. M. L. Guerrero, M. T. S. Cordero, J. M. C. Pavon and A. G. de Torres, J. Anal. At. Spectrom., 2016, 31(12), 2391–2398 RSC .
  42. M. Resano, E. Garcia-Ruiz and R. Garde, J. Anal. At. Spectrom., 2016, 31(11), 2233–2241 RSC .
  43. K. Leopold, A. Brandt and H. Tarren, J. Anal. At. Spectrom., 2017, 32(4), 723–730 RSC .
  44. J. J. Gonzalez, Spectroscopy, 2017, 32(6), 28–34 Search PubMed .
  45. Y. K. Shazzo and Y. A. Karpov, J. Anal. Chem., 2016, 71(11), 1069–1080 CrossRef CAS .
  46. D. Pozebon, G. L. Scheffler and V. L. Dressler, J. Anal. At. Spectrom., 2017, 32(5), 890–919 RSC .
  47. Y. Bussweiler, O. Borovinskaya and M. Tanner, Spectroscopy, 2017, 32(5), 14–20 Search PubMed .
  48. T. Nakazawa, S. Izumo and N. Furuta, Anal. Sci., 2016, 32(12), 1283–1289 CrossRef CAS PubMed .
  49. F. Poitrasson and F.-X. d’Abzac, J. Anal. At. Spectrom., 2017, 32(6), 1075–1091 RSC .
  50. H. Novakova, M. Hola, M. Vojtisek-Lom, J. Ondracek and V. Kanicky, Spectrochim. Acta, Part B, 2016, 125, 52–60 CrossRef CAS .
  51. S. Brandt, A. Schutz, F. D. Klute, J. Kratzer and J. Franzke, Spectrochim. Acta, Part B, 2016, 123, 6–32 CrossRef CAS .
  52. N. Li, Z. Wu, Y. Wang, J. Zhang, X. Zhang, H. Zhang, W. Wu, J. Gao and J. Jiang, Anal. Chem., 2017, 89(4), 2205–2210 CrossRef CAS PubMed .
  53. Y. J. Zhang, Y. Cai, Y. L. Yu and J. H. Wang, Anal. Chim. Acta, 2017, 976, 45–51 CrossRef CAS PubMed .
  54. S. Y. Liu, M. J. Kim and H. K. Kim, J. Vac. Sci. Technol., B: Nanotechnol. Microelectron.: Mater., Process., Meas., Phenom., 2016, 34(6), 6 Search PubMed .
  55. I. Burger, M. Scharpf, J. Hennenlotter, D. Nussle, D. Spether, A. Neugebauer, N. Bibinov, A. Stenzl, F. Fend, M. Enderle and P. Awakowicz, J. Phys. D: Appl. Phys., 2017, 50(3), 8 CrossRef .
  56. U. Boesl, Mass Spectrom. Rev., 2017, 36(1), 86–109 CrossRef CAS PubMed .
  57. E. A. Dennis, A. W. Gundlach-Graham, S. J. Ray, C. G. Enke and G. M. Hieftje, J. Am. Soc. Mass Spectrom., 2016, 27(11), 1772–1786 CrossRef CAS PubMed .
  58. A. Pramann, T. Narukawa and O. Rienitz, Metrologia, 2017, 54(5), 738–747 CrossRef CAS .
  59. M. de Podesta, D. F. Mark, R. C. Dymock, R. Underwood, T. Bacquart, G. Sutton, S. Davidson and G. Machin, Metrologia, 2017, 54(5), 683–692 CrossRef CAS .
  60. J. Manrique, J. A. Aguilera and C. Aragon, Mon. Not. R. Astron. Soc., 2016, 462(2), 1501–1507 CrossRef CAS .
  61. A. M. Popov, T. F. Akhmetzhanov, T. A. Labutin, S. M. Zaytsev, N. B. Zorov and N. V. Chekalin, Spectrochim. Acta, Part B, 2016, 125, 43–51 CrossRef CAS .
  62. P. Quinet, V. Fivet, P. Palmeri, L. Engstrom, H. Hartman, H. Lundberg and H. Nilsson, Mon. Not. R. Astron. Soc., 2016, 462(4), 3912–3917 CrossRef CAS .
  63. A. Bogaerts and M. Aghaei, J. Anal. At. Spectrom., 2017, 32(2), 233–261 RSC .
  64. P. B. Farnsworth and R. L. Spencer, Spectrochim. Acta, Part B, 2017, 134, 105–122 CrossRef CAS .
  65. M. Voronov, V. Hoffmann, C. Engelhard and W. Buscher, J. Anal. At. Spectrom., 2017, 32(1), 167–180 RSC .
  66. M. Voronov, V. Hoffmann, W. Buscher and C. Engelhard, J. Anal. At. Spectrom., 2017, 32(1), 181–192 RSC .
  67. B. Hattendorf, B. Gusmini, L. Dorta, R. S. Houk and D. Gunther, ChemPhysChem, 2016, 17(17), 2640–2644 CrossRef CAS PubMed .
  68. M. Y. He, Z. D. Jin, H. Lu, L. Deng and C. G. Luo, Int. J. Mass Spectrom., 2016, 408, 33–37 CrossRef CAS .
  69. S. Mushtaq, E. B. M. Steers, V. Hoffmann, Z. Weiss and J. C. Pickering, J. Anal. At. Spectrom., 2016, 31(11), 2175–2181 RSC .
  70. H. Matsuta, Spectrochim. Acta, Part B, 2017, 132, 37–42 CrossRef CAS .
  71. F. Ruiz, Z. Benzo, A. Garaboto, V. Leon, F. Ruette, A. Albornoz and J. L. Brito, Spectrochim. Acta, Part B, 2017, 133, 1–8 CrossRef CAS .
  72. E. Bolea-Fernandez, L. Balcaen, M. Resano and F. Vanhaecke, J. Anal. At. Spectrom., 2017, 32(9), 1660–1679 RSC .
  73. A. Virgilio, R. S. Amais, C. D. B. Amaral, L. L. Fialho, D. Schiavo and J. A. Nobrega, Spectrochim. Acta, Part B, 2016, 126, 31–36 CrossRef CAS .
  74. Y. Cheung, S. J. Ray, A. J. Schwartz and G. M. Hieftje, Appl. Spectrosc., 2016, 70(11), 1797–1805 CrossRef CAS PubMed .
  75. T. Narukawa, T. Iwai and K. Chiba, J. Anal. At. Spectrom., 2017, 32(8), 1547–1553 RSC .
  76. J. W. Olesik and S. Jiao, J. Anal. At. Spectrom., 2017, 32(5), 951–966 RSC .
  77. G. Gamez, J. Anal. At. Spectrom., 2016, 31(11), 2165–2174 RSC .
  78. J. D. Usala, A. Maag, T. Nelis and G. Gamez, J. Anal. At. Spectrom., 2016, 31(11), 2198–2206 RSC .
  79. P. J. Cumpson, I. W. Fletcher, N. Sano and A. J. Barlow, Surf. Interface Anal., 2016, 48(12), 1328–1336 CrossRef CAS .
  80. P. Koscielniak and M. Wieczorek, Anal. Chim. Acta, 2016, 944, 14–28 CrossRef CAS PubMed .
  81. E. Bulska and B. Wagner, Philos. Trans. R. Soc., A, 2016, 374(2079), 18 CrossRef PubMed .
  82. M. C. Sforna and F. Lugli, J. Anal. At. Spectrom., 2017, 32(5), 1035–1043 RSC .
  83. A. J. Bauer and S. G. Buckley, Appl. Spectrosc., 2017, 71(4), 553–566 CrossRef CAS PubMed .
  84. G. S. Senesi and N. Senesi, Anal. Chim. Acta, 2016, 938, 7–17 CrossRef CAS PubMed .
  85. A. Bengtson, Spectrochim. Acta, Part B, 2017, 134, 123–132 CrossRef CAS .
  86. A. De Giacomo and J. Hermann, J. Phys. D: Appl. Phys., 2017, 50(18), 183002 CrossRef .
  87. M. Lykovic and N. Konjevic, Spectrochim. Acta, Part B, 2017, 131, 79–92 CrossRef .
  88. F. Bredice, P. P. Martinez, R. S. Mercado, C. Sanchez-Ake, M. Villagran-Muniz, J. B. Sirven, M. El Rakwe, E. Grifoni, S. Legnaioli, G. Lorenzetti, S. Pagnotta and V. Palleschi, Spectrochim. Acta, Part B, 2017, 135, 48–53 CrossRef CAS .
  89. S. Bukvic and M. Skocic, Spectrochim. Acta, Part B, 2017, 132, 56–60 CrossRef CAS .
  90. Y. Wang, H. Yuan, Y. T. Fu and Z. Wang, Spectrochim. Acta, Part B, 2016, 126, 44–52 CrossRef CAS .
  91. J. Weiss, L. M. Cabalin and J. J. Laserna, Appl. Spectrosc., 2017, 71(1), 87–96 CrossRef CAS PubMed .
  92. Y. Wang, A. M. Chen, L. Z. Sui, S. Y. Li, X. W. Wang, Y. F. Jiang, X. R. Huang and M. X. Jin, J. Anal. At. Spectrom., 2016, 31(10), 1974–1977 RSC .
  93. R. B. Anderson, S. M. Clegg, J. Frydenvang, R. C. Wiens, S. McLennan, R. V. Morris, B. Ehlmann and M. D. Dyar, Spectrochim. Acta, Part B, 2017, 129, 49–57 CrossRef CAS .
  94. J. P. Castro and E. R. Pereira, J. Anal. At. Spectrom., 2016, 31(10), 2005–2014 RSC .
  95. M. D. Dyar, C. I. Fassett, S. Giguere, K. Lepore, S. Byrne, T. Boucher, C. J. Carey and S. Mahadevan, Spectrochim. Acta, Part B, 2016, 123, 93–104 CrossRef CAS .
  96. J. Guezenoc, L. Bassel, A. Gallet-Budynek and B. Bousquet, Spectrochim. Acta, Part B, 2017, 134, 6–10 CrossRef CAS .
  97. A. Demidov, S. Eschlbock-Fuchs, A. Y. Kazakov, I. B. Gornushkin, P. J. Kolmhofer, J. D. Pedarnig, N. Huber, J. Heitz, T. Schmid, R. Rossler and U. Panne, Spectrochim. Acta, Part B, 2016, 125, 97–102 CrossRef CAS .
  98. P. D. Barnett, N. Lamsal and S. M. Angel, Appl. Spectrosc., 2017, 71(4), 583–590 CrossRef CAS PubMed .
  99. T. J. Jiang, M. Yang, S. S. Li, M. J. Ma, N. J. Zhao, Z. Guo, J. H. Liu and X. J. Huang, Anal. Chem., 2017, 89(10), 5557–5564 CrossRef CAS PubMed .
  100. S. J. Choi, J. J. Choi and J. J. Yoh, Spectrochim. Acta, Part B, 2016, 123, 1–5 CrossRef CAS .
  101. F. F. Zha, G. H. Niu, Q. Y. Lin, S. Wang and Y. X. Duan, Instrum. Sci. Technol., 2017, 45(6), 650–658 CrossRef CAS .
  102. S. Li, L. Liu, A. Yan, S. Huang, X. Huang, R. Chen, Y. Lu and K. Chen, Rev. Sci. Instrum., 2017, 88(2), 023109 CrossRef PubMed .
  103. J. Cortez, B. B. Farias Filho, L. M. Fontes, C. Pasquini, I. M. Raimundo Jr, M. F. Pimentel and F. de Souza Lins Borba, Appl. Spectrosc., 2017, 71(4), 634–639 CrossRef CAS PubMed .
  104. L. A. Alvarez-Trujillo, V. Lazic, J. Moros and J. Javier Laserna, Appl. Opt., 2017, 56(13), 3773–3782 CrossRef CAS PubMed .
  105. M. R. Gavrilovic, V. Lazic and S. Jovicevic, J. Anal. At. Spectrom., 2017, 32(2), 345–353 RSC .
  106. M. Lopez-Claros, M. Dell’Aglio, R. Gaudiuso, A. Santagata, A. De Giacomo, F. J. Fortes and J. J. Laserna, Spectrochim. Acta, Part B, 2017, 133, 63–71 CrossRef CAS .
  107. C. Alvarez-Llamas, J. Pisonero and N. Bordel, Spectrochim. Acta, Part B, 2016, 123, 157–162 CrossRef CAS .
  108. C. Alvarez-Llamas, J. Pisonero and N. Bordel, J. Anal. At. Spectrom., 2017, 32(1), 162–166 RSC .
  109. C. Koral, A. De Giacomo, X. L. Mao, V. Zorba and R. E. Russo, Spectrochim. Acta, Part B, 2016, 125, 11–17 CrossRef CAS .
  110. Q. Y. Lin, F. Bian, Z. M. Wei, S. Wang and Y. X. Duan, J. Anal. At. Spectrom., 2017, 32(7), 1412–1419 RSC .
  111. P. J. Skrodzki, N. P. Shah, N. Taylor, K. C. Hartig, N. L. LaHaye, B. E. Brumfield, I. Jovanovic, M. C. Phillips and S. S. Harilal, Spectrochim. Acta, Part B, 2016, 125, 112–119 CrossRef CAS .
  112. M. Miyabe, M. Oba, K. Jung, H. Iimura, K. Akaoka, M. Kato, H. Otobe, A. Khumaeni and I. Wakaida, Spectrochim. Acta, Part B, 2017, 134, 42–51 CrossRef CAS .
  113. M. Pfeifer, N. S. Lloyd, S. T. M. Peters, F. Wombacher, B. M. Elfers, T. Schulz and C. Munker, J. Anal. At. Spectrom., 2017, 32(1), 130–143 RSC .
  114. S. K. Aggarwal and C. F. You, Mass Spectrom. Rev., 2017, 36(4), 499–519 CrossRef CAS PubMed .
  115. E. T. Rasbury and N. G. Hemming, Elements, 2017, 13(4), 243–248 Search PubMed .
  116. M. R. Palmer, Elements, 2017, 13(4), 237–242 Search PubMed .
  117. C. K. Shearer and S. B. Simon, Elements, 2017, 13(4), 231–236 Search PubMed .
  118. F. Albarede, P. Telouk, V. Balter, V. P. Bondanese, E. Albalat, P. Oger, P. Bonaventura, P. Miossec and T. Fujii, Metallomics, 2016, 8(10), 1056–1070 RSC .
  119. J. D. Woodhead, M. S. A. Horstwood and J. M. Cottle, Elements, 2016, 12(5), 317–322 CrossRef CAS .
  120. D. Clases, M. Birka, M. Sperling, A. Faust and U. Karst, J. Trace Elem. Med. Biol., 2017, 40, 97–103 CAS .
  121. A. Bazzano, K. Latruwe, M. Grotti and F. Vanhaecke, J. Anal. At. Spectrom., 2017, 32(5), 1004–1008 RSC .
  122. M. Carpentier, A. Gannoun, C. Pin and O. Sigmarsson, Geostand. Geoanal. Res., 2016, 40(2), 239–256 CrossRef CAS .
  123. Y. Fukami, M. Tobita, T. Yokoyama, T. Usui and R. Moriwaki, J. Anal. At. Spectrom., 2017, 32(4), 848–857 RSC .
  124. B. S. Vaglarov, J. Kimura, T. Miyazaki and Q. Chang, Geochem. J., 2016, 50(5), 423–429 CrossRef CAS .
  125. A. J. Pietruszka and L. A. Neymark, J. Anal. At. Spectrom., 2017, 32(6), 1135–1154 RSC .
  126. R. S. Yin, D. P. Krabbenhoft, B. A. Bergquist, W. Zheng, R. F. Lepak and J. P. Hurley, J. Anal. At. Spectrom., 2016, 31(10), 2060–2068 RSC .
  127. R. Gonzalez-Guzman, B. Weber, M. D. Tazzo-Rangel and L. Solari, Rev. Mex. Cienc. Geol., 2016, 33(2), 254–269 Search PubMed .
  128. G. J. Archer, A. Mundl, R. J. Walker, E. A. Worsham and K. R. Bermingham, Int. J. Mass Spectrom., 2017, 414, 80–86 CrossRef CAS .
  129. S. Berail, J. Cavalheiro, E. Tessier, J. P. G. Barre, Z. Pedrero, O. F. X. Donard and D. Amouroux, J. Anal. At. Spectrom., 2017, 32(2), 373–384 RSC .
  130. S. M. Chernonozhkin, M. Costas-Rodriguez, P. Claeys and F. Vanhaecke, J. Anal. At. Spectrom., 2017, 32(3), 538–547 RSC .
  131. C. Fabrega, D. Parcerisa, J. M. Rossell, A. Gurenko and C. Franke, J. Anal. At. Spectrom., 2017, 32(4), 731–748 RSC .
  132. D. A. Frick, J. A. Schuessler and F. von Blanckenburg, Anal. Chim. Acta, 2016, 938, 33–43 CrossRef CAS PubMed .
  133. J. R. Farmer, B. Honisch and J. Uchikawa, Chem. Geol., 2016, 447, 173–182 CrossRef CAS .
  134. J. Kimura, Q. Chang, T. Ishikawa and T. Tsujimori, J. Anal. At. Spectrom., 2016, 31(11), 2305–2320 RSC .
  135. X. F. Chen, L. Zhang, G. J. Wei and J. L. Ma, J. Anal. At. Spectrom., 2016, 31(12), 2410–2417 RSC .
  136. F. Gueguen, A. Nonell, H. Isnard, L. Vio and F. Chartier, Talanta, 2017, 162, 278–284 CrossRef CAS PubMed .
  137. A. Gourgiotis, G. Manhes, B. Martelat and H. Isnard, J. Anal. At. Spectrom., 2017, 32(7), 1428–1434 RSC .
  138. J. Karasinski, E. Bulska, M. Wojciechowski, L. Halicz and A. A. Krata, Talanta, 2017, 165, 64–68 CrossRef CAS PubMed .
  139. C. F. Kerl, R. Lohmayer, E. Bura-Nakic, D. Vance and B. Planer-Friedrich, Anal. Chem., 2017, 89(5), 3123–3129 CrossRef CAS PubMed .
  140. D. Malinovsky, P. J. H. Dunn and H. Goenaga-Infante, J. Anal. At. Spectrom., 2016, 31(10), 1978–1988 RSC .
  141. M. Oelze, J. A. Schuessler and F. von Blanckenburg, J. Anal. At. Spectrom., 2016, 31(10), 2094–2100 RSC .
  142. C. F. Li, L. J. Feng, X. C. Wang, Z. Y. Chu, J. H. Guoa and S. A. Wilde, J. Anal. At. Spectrom., 2016, 31(12), 2375–2383 RSC .
  143. J. Lin, Y. S. Liu, X. R. Tong, L. Y. Zhu, W. Zhang and Z. C. Hu, J. Anal. At. Spectrom., 2017, 32(4), 834–842 RSC .
  144. T. X. Ren, J. Wang, H. Lu, T. Zhou and Y. J. Zhou, Geostand. Geoanal. Res., 2016, 40(2), 227–238 CrossRef CAS .
  145. K. van Zuilen, T. F. Nägler and T. D. Bullen, Geostand. Geoanal. Res., 2016, 40(4), 543–558 CrossRef CAS .
  146. E. Yobregat, C. Fitoussi and B. Bourdon, J. Anal. At. Spectrom., 2017, 32(7), 1388–1399 RSC .
  147. M. Weyrauch, M. Oeser, A. Bruske and S. Weyer, J. Anal. At. Spectrom., 2017, 32(7), 1312–1319 RSC .
  148. W. Yuan, J. B. Chen, J. L. Birck, Z. Y. Yin, S. L. Yuan, H. M. Cai, Z. W. Wang, Q. Huang and Z. H. Wang, Anal. Chem., 2016, 88(19), 9606–9613 CrossRef CAS PubMed .
  149. Z. Zhu, J. Meija, A. Zheng, Z. Mester and L. Yang, Anal. Chem., 2017, 89(17), 9375–9382 CrossRef CAS PubMed .
  150. G. G. Arantes de Carvalho, P. V. Oliveira and L. Yang, J. Anal. At. Spectrom., 2017, 32(5), 987–995 RSC .
  151. C. Archer, M. B. Andersen, C. Cloquet, T. M. Conway, S. Dong, M. Ellwood, R. Moore, J. Nelson, M. Rehkamper, O. Rouxel, M. Samanta, K. C. Shin, Y. Sohrin, S. Takano and L. Wasylenki, J. Anal. At. Spectrom., 2017, 32(2), 415–419 RSC .
  152. C. D. Coath, T. Elliott and R. C. Hin, Chem. Geol., 2017, 451, 78–89 CrossRef CAS .
  153. L. Courtney-Davies, Z. Y. Zhu, C. L. Ciobanu, B. P. Wade, N. J. Cook, K. Ehrig, A. R. Cabral and A. Kennedy, Minerals, 2016, 6(3), 17 CrossRef .
  154. W. Zhang, Z. Hu, D. Gunther, Y. Liu, W. Ling, K. Zong, H. Chen and S. Gao, Anal. Chim. Acta, 2016, 948, 9–18 CrossRef CAS PubMed .
  155. M. Ek, A. C. Hunt and M. Schonbachler, J. Anal. At. Spectrom., 2017, 32(3), 647–656 RSC .
  156. C. LaFlamme, L. Martin, H. Jeon, S. M. Reddy, V. Selvaraja, S. Caruso, T. H. Bui, M. P. Roberts, F. Voute, S. Hagemann, D. Wacey, S. Littman, B. Wing, M. Fiorentini and M. R. Kilburn, Chem. Geol., 2016, 444, 1–15 CrossRef CAS .
  157. T. L. Yu, B. S. Wang, C. C. Shen, P. L. Wang, T. F. Yang, G. S. Burr and Y. G. Chen, Anal. Chim. Acta, 2017, 988, 34–40 CrossRef CAS PubMed .
  158. F. Lugli, A. Cipriani, C. Peretto, M. Mazzucchelli and D. Brunelli, Int. J. Mass Spectrom., 2017, 412, 38–48 CrossRef CAS .
  159. S. Schuth, I. Horn, A. Bruske, P. E. Wolff and S. Weyer, Ore Geol. Rev., 2017, 81, 1271–1286 CrossRef .
  160. S. F. Boulyga, A. Koepf, S. Konegger-Kappel, Z. Macsik and G. Stadelmann, J. Anal. At. Spectrom., 2016, 31(11), 2272–2284 RSC .
  161. M. C. Phillips, B. E. Brumfield, N. LaHaye, S. S. Harilal, K. C. Hartig and I. Jovanovic, Sci. Rep., 2017, 7(1), 3784 CrossRef PubMed .
  162. M. R. Savina, B. H. Isselhardt, A. Kucher, R. Trappitsch, B. V. King, D. Ruddle, R. Gopal and I. Hutcheon, Anal. Chem., 2017, 89(11), 6224–6231 CrossRef CAS PubMed .
  163. A. Donard, F. Pointurier, A. C. Pottin, A. Hubert and C. Pecheyran, J. Anal. At. Spectrom., 2017, 32(1), 96–106 RSC .
  164. W. Wang, Z. M. Li, J. Xu, G. Q. Zhou, L. H. Zhai, M. Li, Y. Y. Su and G. Y. Wei, Chin. J. Anal. Chem., 2016, 44(7), 1053–1058 CAS .
  165. J. A. Dunne, D. A. Richards and H. W. Chen, Talanta, 2017, 174, 347–356 CrossRef CAS PubMed .
  166. J. D. Inglis, J. Maassen, A. Kara, R. E. Steiner, W. S. Kinman and D. Lopez, J. Radioanal. Nucl. Chem., 2017, 312(3), 663–673 CrossRef CAS .

This journal is © The Royal Society of Chemistry 2018