Jeffrey R. Bacon*a, Jeffrey S. Crainb, Luc Van Vaeckc and John G. Williamsd
aThe Macaulay Institute, Craigiebuckler, Aberdeen, UK AB15 8QH
bDow Chemical Company, Technical Center, PO Box 8361, South Charleston, USA WV25303
cMicro- and Trace Analysis Centre, Department of Chemistry, University of Antwerp, Universiteitsplein 1, Wilrijk, Belgium B-2610
dNu Instruments Ltd., Unit 74, Clywedog Road South, Wrexham Industrial Estate, Wrexham, UK LL13 9XS
First published on 11th July 2002
This Update follows the same format as last year's.1 Although an attempt is made to consider all relevant refereed papers, conference abstracts, reports, book chapters and patents for inclusion in this review, the content of the review is highly selective. The selection of papers is based on criteria applied to focus sharply on the most significant developments in instrumentation and methodology or improved understanding of the fundamental phenomena involved in the MS process. With the increasing importance of speciation and the blurring of boundaries between atomic and molecular MS, a high degree of judgement is required in considering papers for inclusion. The main ruling criterion for all speciation papers is that the work should involve or be intended for the study of natural systems. For example, the study of synthetic metal clusters is generally not included whereas the determination of organometallic compounds in environmental samples is.
Applications of atomic MS are not covered in this Update and readers are referred to the Updates on Industrial analysis: metals, chemicals and advanced materials,2 Environmental analysis3 and Clinical and biological materials, food and beverages.4 There have been few general reviews of note. That of Walczyk5 gave a good introduction to the use of inorganic MS in human nutrition research.
The trends noted over the last few years have continued. In particular, the growing interest in speciation studies has been reflected by the number of papers on the subject. As the number of species identified increases then the danger of relying solely on chromatographic retention times for identification has become a recurring warning in papers. For this reason, the importance of GDMS, ESMS and S-SIMS for species identification has been widely recognised. In all these studies, sample preparation and introduction have generally received most attention, with the aim of ever improved analysis. The increasing use of multicollector ICP-MS instruments for isotope geology studies has resulted in common ground with traditional TIMS methods, in particular in sample preparation.
Throughout this review, the term molecular ion will be restricted to denote only the positive or negative radical ion formed by removal or capture, respectively, of an electron. In contrast, addition of a proton or cation to a neutral molecule gives molecular adduct ions. Deprotonated molecules are considered as fragments.
The development and application of compact radiocarbon instruments for use in biomedical and pharmaceutical research is the main growth area of AMS. A number of informative articles7–10 from researchers with direct experience were of particular value. Not only were the attractions of the technique highlighted but also current limitations and the need for improvements. Major limitations of the technique were the inability to provide chemical or structural information and, in pharmaceutical applications, the need to incorporate the radioisotope label within the molecule of interest. The rate-limiting step in the analysis of samples remained the sample preparation. In general, increased use of AMS in biomedical applications will depend on the availability of small and cheap instrumentation and development of new methods for the processing of samples. On-line chromatography was considered to be a solution for increasing sample throughput and structural information. A limitation with no obvious solution was a shortage of human subjects prepared to receive doses of radiocarbon-labelled drugs even at the very low level made possible by AMS.
Friedrich et al.12 gave preliminary details of an instrument with an air-insulated 100 kV tandem accelerator dedicated to the depth profiling of samples with high 3H concentrations. A dedicated instrument was essential in order to avoid contamination of an accelerator used for other applications and to safeguard workers from exposure to radioactivity. Initial tests with a carbon foil stripper (40 µg cm−2) gave beam losses of up to 98% and unstable operation conditions. Use of a nitrogen gas stripper improved transmission to about 10% with high stability and reproducibility.
Injection systems with 90° magnets have been seen as the key to improved performance of low energy (3 MeV) facilities. The Lund system has been developed for the measurement of 26Al13 and heavy ions.14 A serious limitation of the original injector design when ion masses heavier than 14C were measured was the use of only a single focusing injection magnet with a bending angle of just 15°. The new injector design, incorporating both an electrostatic and a 90° magnetic analyser, improved mass and energy separation by up to a factor of ten. The use of this injector in the measurement of 26Al reduced the interference from Mg by two orders of magnitude and a new ion source increased the beam current threefold. These modifications improved the LOD for 26Al from 10−14 to about 10−16 g. A first-order analytical technique was applied to ion-optical calculations used in the design of the new injector.14 Facilities for AMS are now available in India following an upgrade to a 3 MV accelerator facility to include an AMS beam line for 10Be and 14C measurements.15 The entire injection system was replaced by a commercial, state-of-the-art, fast-bouncing injector consisting of a 45° spherical electrostatic analyser and a 90° injection magnet.
Improvements in sample preparation techniques have been crucial for the use of compound-specific radiocarbon analysis (CSRA), in particular for the analysis of individual components in atmospheric and marine samples. Matsumoto et al.17 analysed individual fatty acids in an aerosol sample following separation by preparative capillary GC. A limitation of this approach was the need to perform 60 consecutive GC runs in order to isolate μg quantities of the individual fatty acids. The C16 to C22 fatty acids had modern 14C ages, suggesting that the acids were emitted from living plants or marine organisms. The C24 to C26 fatty acids, on the other hand, had much older 14C ages, suggesting that they had been stored in geochemical reservoirs for thousands of years prior to being emitted to the atmosphere. The recent development of radiocarbon AMS for small samples (200 µg–1 mg C) allowed Megens et al.18 to use CSRA for the characterization of particulate organic matter in coastal waters. A sequence of chemical treatments separated polysaccharides, proteins, lipids and compounds resistant to hydrolysis. Temporal variations in 14C∶12C ratios were observed for the first time and used to identify the origin of the particulate matter as being phytoplankton with young 14C ages or resuspended surface sediments with much older (up to 7300 years) 14C ages.
Obtaining a reliable estimate of the biogenic sources of volatile organic compounds (VOC) is an essential step towards framing legislation for control of particulate matter emissions. In a study of samples taken at four European sites, Larsen et al.19 used radiocarbon analysis of bulk airborne carbonyl compounds after collection on 2,4-dinitrophenylhydrazine-coated silica gel cartridges and chromatographic isolation of the hydrazones formed. This analysis of the samples by HPLC-chemical ionization MS provided their species composition. At all sites the carbonyl compounds were of mixed biogenic and anthropogenic origin but it was concluded that the background levels were predominantly (50–80%) of a biogenic origin. An assumption of the approach was that anthropogenic VOC in the atmosphere was of fossil origin with old 14C ages.
Protein sequencing at the attomole level using Edman degradation has been achieved through the use of AMS as a detection system.20 Fractions collected from an automated sequencer were graphitized using standard procedures. Typical AMS measurement times were 3 minutes per sample with a counting precision of 1.4–2.0% and a RSD of 1–3% (n = 3–7). Currently the method can only be used for 14C-labelled proteins but methods were being developed for extrinsic labelling of proteins to provide labelled amino acids. Although the method was considered promising for protein sequencing at the attomole level, improved sample throughput and wider availability of low cost instruments were seen as necessary for widespread application.
The use of freeze-drying instead of dissolution in chloroform for the disruption of microbial cells allowed Rumpel et al.21 to avoid the possible contamination of the sample with C. They demonstrated the possibility of measuring the 14C activity in soil microbial biomass at natural 14C abundance following direct extraction of freeze-dried soil. This method provided a tool to determine the dead carbon content in soil microbial biomass and thereby to follow the degradation of fossil material (e.g., lignite) in soil.
The half-life of 79Se, a long-lived fission product, has been recalculated as being about one third lower than its previous value.23 The use of projectile X-ray detection in the AMS analysis allowed effective separation of 75Se and 79Se from isobaric 75As and 79Br, respectively. The background ratio of <10−10 for both 79Se∶Se and 75Se∶Se was negligible in comparison to the measured ratios in samples of 10−7–10−8. The half-life of 79Se was measured relative to that of 75Se, which is known with high precision.
Determination of U and the transuranic elements in marine samples by AMS is proving to be a powerful tool for monitoring emissions from nuclear reprocessing plants and fallout from nuclear weapon testing. Marsden et al.24 developed a method for the determination in environmental samples of 236U, which is produced only by neutron irradiation of U and is a potential marker for anthropogenic U in the environment. Sample preparation was based on the radiochemical separation of U from sediment samples using TRU.spec resins. Analysis of blank samples revealed a sensitivity of 10−6 for the 236U∶238U ratio which is adequate for analysis of contaminated samples but not for samples taken further away from reprocessing plants. In order to reduce the number of 238U ions reaching the detector and being counted as 236U ions, a velocity filter was to be installed after the final magnetic analysis. Samples taken from close to the Sellafield fuel reprocessing plant revealed levels of 236U substantially above natural concentrations, with the implication that the U is preserved to some extent in the sediments and may reflect discharge history.
The study of Np in the environment is of increasing interest because of its relatively high mobility, its incorporation into the environment and the time scales involved. An AMS method for the determination of 237Np in water samples extracted from marine sediments was found to have the required sensitivity to analyse samples of limited volume (700 ml) and low activity (0.06–0.79 mBq l−1, 2.30–30.3 pg l−1).25 Neptunium was separated from matrix elements and other transuranic elements using an ion-exchange method and 242Pu added as an internal standard. Targets, prepared by mixing samples with iron and aluminium, were sputtered with a Cs+ beam to produce NpO− and PuO− ions. Although the method was considered to be capable of measuring amounts as low as 0.4 µBq (3.9 × 107 atoms), this was in practice not possible because of a systematic low level 237Np contamination in each sample derived from the 239Np yield monitor used in the separation procedure. Elimination of this contamination would give a method for the determination of very low concentrations in samples of limited volume.
A comparison of several procedures for the determination of Pu isotopes in the marine environment showed that the two MS techniques (AMS and ICP-MS) were highly sensitive and useful tools for identifying the origin of Pu.26 The sample preparation up to the final step was the same for both procedures and was based on several co-precipitation and chromatography steps. Sample targets for AMS analysis were prepared by mixing with iron and aluminium. In comparison to ICP-MS, the advantages of AMS were the low LOD (<106 atoms of 239Pu), which minimized sample size requirement, and low matrix and interference effects, which removed the need for complete sample clean-up. On the other hand, AMS analysis was more complex and expensive. Analysis of IAEA RM allowed the origin of Pu in the samples to be identified.
Song et al.28 designed a see-through type hollow cathode GD cell for analysis of gaseous and solid samples in a portable MS instrument for field applications. The hollow cathode was preferred over a Grimm type cell because of the stability of its plasma. A skimmer was used to interface the cell to an ion trap made in-house. Use of helium as buffer gas improved the Penning ionization process. Feasibility tests yielded spectra similar to those obtained by conventional electron ionization in the gas phase.
The simple, small dc GD ion source for direct analysis of planar solids described by Pisonero et al.29 was designed to be interchangeable with the ICP source in a commercial on-axis TOFMS instrument. The internal GD volume was about 100 mm3 and the distance between the cathode and sampler cone was 4 mm. Enlarging the orifice diameters of the commercial skimmer and sampler cones improved the transmission. The mass resolution was 1520 (full width at half maximum, FWHM) at m/z 120. The quasi-simultaneous extraction of the ion bunch into the TOFMS instrument eliminated the variations in isotope ratio measurements arising from changes in the plasma conditions. A precision of 0.8–1.8% and an accuracy of 1–5% were obtained for Sn isotope ratios. Ion currents from a certified brass RM (31X-B7-H) varied less than 15% over 1 h. Depth profiling over a distance of 25 µm at a sputter rate of 3.1 µm min−1 yielded craters with perpendicular walls and flat bottoms.
Hodoroaba31 investigated the addition of H2 to Ar and Ne in dc GDMS for the analysis of metallic samples. The sputter rate decreased more upon the addition of H2 to Ne than to Ar, despite the slight increase in the discharge current. Mixing H2 with Ar or Ne increased the curvature of the crater bottom but reduced its roughness. Comparative experiments on copper and titanium targets showed that 2% v/v H2 in Ar quenched the Ar+ peak but increased the Cu+ and Ti+ signals by a factor of 5 and 1.5, respectively. The disappearance of the Ar+ signal was explained by a drop in the Ar metastable population and by the formation of ArH+ ions. The Ne+ signal increased upon addition of H2 when copper was analysed but stayed constant in the case of titanium targets. The atomic ions increased by a factor 1.5–3 and 7, respectively. The data pointed to the quenching of the Ar metastables by Penning excitation of hydrogen molecules. The difference in energy levels explained why this did not occur in Ne.
Budtz-Jorgensen et al.32 investigated the chemical and physical sputtering in an Ar–H2 plasma. Pulsed-mode measurements allowed the kinetic energy (KE) distributions of the bombarding Ar+, Ar2+, ArH+, H2+ and H3+ to be measured during the sputtering of Al2O3 by Ar containing 0–80% H2. Without H2, Ar+ KE distributions showed low mean ion energies as a result of the symmetric charge transfer between Ar+ and Ar whereas Ar2+ KE distributions showed higher mean energies. Addition of H2 increased the discharge current density and thereby the mean energies of all ion species. Also the Ar2+ contribution to the plasma increased significantly. The KE distributions of hydrogen-containing ions characterized the attenuation of H2+ by the increased symmetric charge transfer process whereas a large amount of non-colliding H3+ existed. Sputter rates were calculated from the KE distributions making the assumption that physical sputtering prevailed. The maximal sputter for Al was calculated to occur with 20% H2 whereas the experimental maximum was found for 80% H2. In contrast, calculations and experiments agreed for the sputter rate of Au, where physical sputtering prevailed. This clearly suggested that hydrogen-enhanced chemical sputtering took place in the analysis of Al2O3.
The concentration dependence of relative sensitivity factors (RSF) was studied by Saka et al.34 for the dc GDMS determination of Al in metals. The RSF of Al, referenced to the matrix element, increased as a function of the Al concentration. On the other hand, the RSF of Cu, Fe, Ni and Ti decreased with concentration. Identical trends were observed for pin and disk samples. The data indicated the existence of a non-linear process affecting the ionization of elements.
Shimizu et al.35 found that the resolution of rf GDMS depth profiling was mainly limited by the curvature of the crater bottom. Anodic alumina films of different thickness and containing δ-layers of about 2 nm at various depths were studied. In the middle of the crater, the sputter rate was constant within 1% under the optimum Ar pressure. The depth resolution degraded linearly from 1.3 nm at a depth of 25 nm to 7 nm at a depth of 350 nm. Absence of significant roughening in the crater bottom, as shown by atomic force microscopy (AFM), and the minimal atomic mixing explained why the depth resolution in the first 25 nm approached the sputter depth profiling limit of 0.7–1 nm.
In a study of high resolution dc GDMS depth profiling and quantitative trace element analysis, Spitsberg and Putyera36 found that the optimized conditions for the profiling of a 30–40 µm thick Pt–Al coating on Ni-based superalloy yielded a sputter rate between 0.1 and 0.2 µm min−1 over a distance of 50–100 µm. Selected isotopes were measured at 0.5–1 µm depth intervals during a 5–7 h analysis and 58Ni was used as internal standard. An additional spacer placed in front of the sample reduced back-deposition on the insulator ring. A mass resolution of 4000 proved to be sufficient to avoid interferences. Atomic ions were quantified using RSF determined from the analysis of Ni-based alloys. The instrumental precision of depth profile measurements was about 20%. Depth scale was calibrated by weighing the loss of the sample after analysis and by metallurgical microscopy. The true depth was between 1.2 and 1.7 times higher than that calculated from weight loss because of redeposition. At a depth of 120 µm, roughness was observed with 20 µm deviations but the influence of the crater wall effects on the signal was negligible relative to the contribution from the crater surface.
Aldave de las Heras et al.37 investigated the depth profiling of trace elements in ZrO2 layers using dc GDMS. The Zr alloy was corroded to ZrO2 in an autoclave by exposure to solutions containing B and Li. Profilometry was used to determine the crater depth and sputter rate. The concentrations of B and Li were 0.002–0.06% and 0.2–7.5%, respectively. Using an erosion rate of 2.87 nm s−1, the depth resolution was 0.1–0.3 µm at a depth of 13 µm.
The compositional characterization and depth profiling of CrN films on stainless steel was studied by Kumar et al.38 using dc GDMS. Proton backscattering spectrometry and RBS with α-particles were complementarily used to estimate the surface composition in the first few µm of the surface without using RM. Depth profiling of Cr, Fe and N by GDMS showed absence of Cr in the stainless steel. No mixed interlayer of more than a few monolayers occurred. The erosion rate of 1 nm s−1 for Cr allowed the film thickness to be estimated as 1.25 µm whereas RBS yielded 1.2 µm.
In a study of the double pulse method for enhancement of analytical signals, Yang et al.39 measured the effect of changing the voltage, current, pressure and time between the pulses. The first pulse created an initial bunch of sputtered atoms, which diffused slowly into the discharge gas. The second pulse allowed these neutrals to be excited and ionized efficiently. The method featured a production of analyte enhanced by a factor of 10 in comparison to the single pulse mode and allowed the discharge gas contribution to be discriminated from that of the analyte. The method was foreseen to have a great potential for GD TOFMS.
Guzowski and Hieftje41 used a hexapole collision cell fitted to an ICP-TOFMS instrument. Signal levels were influenced by the pressure in the cell, the operating frequency of the ion guide and the rf voltage applied to the rods. Peak height precisions were slightly improved (from 7.3 to 5.4% RSD) by the addition of the collision cell but isotope ratio precisions were limited by counting statistics (0.04% RSD for Ag). Using H2 as a collision gas reduced Ar+ signals by 4 orders of magnitude. Mass resolution was also improved from 1500 with static ion optics to 2900 using the hexapole ion-guide assembly. Leach and Hieftje42 used an octapole collision cell to determine elements such as Ca, Fe and K without isobaric overlaps. Increased resolution (up to 80%) due to collisional cooling reduced the problems caused by hydrocarbon background ions formed within the cell. The LOD for heavier elements were similar to those obtained using conventional ion optics. Lighter ions showed a smaller gain and consequently LOD were degraded.
Longerich and Diegor43 evaluated the effects of a water-cooled spray chamber set at a range of temperatures (4–30 °C) used in conjunction with an ICP-TOFMS instrument. At lower temperatures, less water vapour was transported to the ICP. However increasing the nebulizer gas flow allowed the maximum sensitivity to be maintained at all temperatures. Oxide and doubly charged ion formation, background levels and LOD all remained constant at all temperatures.
The performance of different mass analysers coupled to an ICP-MS instrument has received a considerable amount of attention. Encinar et al.44 made a comparison of three instruments, one quadrupole and two magnetic sector (single and multiple collector (MC)) for the determination of Pb concentrations (by ID) and isotope ratios in biological matrices. Accurate concentration data were obtained using all instruments. Uncertainty calculations for isotope ratios showed that the main source of error for the quadrupole instrument was found in the measurements, whereas for the MC instrument the uncertainty was dominated by the spike concentration. The single collector instrument demonstrated an intermediate performance, whilst showing the greatest influence of detector dead time on ratio measurements. These results highlighted the differences between the detectors used in each instrument. Boulyga et al.45 evaluated a single collector magnetic sector ICP-MS instrument and two quadrupole instruments (one conventional and one with a hexapole collision cell) for the analysis of thin barium strontium titanate perovskite layers. The figures of merit (LOD, precision and sensitivity) were best for the magnetic sector instrument and poorest for the conventional quadrupole system.
Solyom et al.46 tested a custom-designed Mattauch–Herzog mass spectrometer coupled to an ICPsource. Sensitivities of 108–109 cps ppm−1 isotope−1 and LOD of 1–500 pg l−1 were achieved using an ultrasonic nebulizer. Isotope ratio accuracies of 1% standard error and precisions of 1–2% RSD were obtained. Spectral peak shapes (mass resolution = 400) were reported to be disappointing.
Helium plasmas have been used in an attempt to reduce Ar-based polyatomic interferences. Duan et al.47 used a He-plasma to determine Ca isotope ratios. Backgrounds were suppressed without compromising analyte sensitivity by using “off-cone” sampling. Using a microwave plasma TOFMS system, precisions of between 1.2 and 2.5% RSD could be obtained for major Ca isotope ratios. Absolute measurement errors were found to be ≤0.8% for the commonly used Ca isotopes.
The use ofMIP continues to receive attention. The ion KE in He-MIP was studied by Su et al.48 using an orthogonal acceleration TOFMS system in an “off-cone” mode. Theoretical calculations and experimental results showed that lighter ions had a higher velocity and lower KE (e.g., 7Li+ = 14.9 eV) than heavier ions (KE of 208Pb+ = 16.8 eV). Shirasaki et al.49 used a N2-MIP to determine As, Cr, Mo, Se and V (LOD 8.3, 0.2, 26, 4 and 0.3 ng l−1, respectively) in freshwater samples. Interference of 14N223Na+ ions on 51V was minimized by increasing the microwave power to 1.3 kW. Results for SRM were in good agreement with certified values.
Using a cold plasma (600 W rf power) with a “conventional” quadrupole ICP-MS instrument, Huang and Lin50 determined Fe using the 54Fe and 56Fe isotopes. Reduced interference from 40Ar16O gave a LOD of 16 ng l−1 for 56Fe, 60 times lower than that obtained using normal plasma conditions (1200 W).
Houk and Zhai51 investigated the temperature and electron density just outside the sampling orifice of an ICP-MS using MS and optical techniques. Using a matrix-free solution, ion density was typically found to be 1.6 × 1015 ions cm−3, rotational temperature was 3340 K and ionization temperatures, measured by both MS and AES, were approximately 7000 K. Ionization temperatures for Cd were found to be 300–400 K higher than for Zn. Although addition of 1000 ppm Na (as NaNO3) had little effect on electron density, 2000 ppm Na caused the electron density to rise to 2.1 × 1015 ions cm−3. As a result the signal levels fell, but no significant change in the ionization temperature was observed.
Factors influencing analyte transport through the sampling orifice of an ICP-MS were studied by Macedone et al.52 using laser-excited ionic fluorescence. Using the potential recorded by a floating probe positioned 1 mm behind the sampling orifice, ion transport efficiency was seen to change with sample composition, rf power, nebulizer flow and torch shield configuration.
The formation of a plasma boundary by interaction of the plasma behind the skimmer cone with the electric field of the extraction lens was discussed by Sakata et al.53 An ion trajectory simulation study suggested that the location and shape of the plasma boundary affected the focusing and transmission of the extraction lens.
The effects of auxiliary quadrupolar excitation applied to the first stability region of a quadrupole mass filter were investigated by Konenkov et al.54 Improved peak shapes were demonstrated by the measurement of m/z 39 in the presence of 40Ar. Use of auxiliary excitation increased sensitivity by over 1000 times, with only a weak peak broadening with increased translational ion energy (up to 20 eV). The 69Ga+ peak was separated from the interfering 137Ba2+ peak.
Campbell and Burns55 presented evidence for the non-linearity of detector response when measuring U isotope ratios using analog detection and pulse counting methods. Despite a good linear correlation for the dual calibration, the slopes of the calibration curve for pulse counting and analog detection were found to be different, leading to bias when the concentrations in calibration standards were not matched to those in the samples.
Al-Ammar et al.56 studied memory effects for I and Th in ICP-MS. Thorium was found to be volatilized from the spray chamber walls, whereas I was adsorbed on to the walls of the nebulizer tubing, in addition to being volatilised as HI and I2 from the spray chamber. Addition of ammonia, either as a solution or as a gas introduced directly into the spray chamber, was found to eliminate the memory effects after a 2 min washout.
The reassessment of various calibration strategies has been undertaken by a number of authors. Abbyad et al.57 investigated the use of standard addition as an appropriate calibration for automated ICP-MS. Bracketing the spiked sample with unspiked aliquots gave results improved over regular standard addition, particularly where significant drift was seen. Increasing spike concentrations, from 7 to 50 times that of the analyte, had little effect on precision. This allowed the addition of a large spike without accurate knowledge of the elemental concentration in the unknown sample.
Moreda-Pineiro et al.58,59 studied systematic errors in ICP-MS. The variables considered were the number of digestions, the number of replicates and calibration. Errors were studied using chemometric approaches (experimental design and PCA) in combination with analysis of variance and trilinear parallel factor analysis. The calculated systematic errors varied using the different models. In the analysis of a Chinese tea CRM, parallel factor analysis was found to be robust and easy to use.
Fundamental physical properties of LA in analysis have been investigated by several authors. Yoo et al.63 studied the phenomenon of phase explosion in silicon and its dependence on both the laser beam spot size and wavelength. Laser irradiances of >1011 W cm−2, above the threshold for phase explosion (>1010 W cm−2), were found to improve entrainment and transport efficiency into the ICP-MS, leading to increased sensitivity. In a study of transport processes in a laser-induced aerosol, Bleiner and Gunther64 identified the cell volume as the overriding influence on overall transport efficiency. A mathematical model was constructed to describe the structure of the signal in relation to the parameters of the ablation set-up. A reduction in the sample cell volume from 63 cm3 to 0.25 cm3 was found to increase the aerosol density by a factor of 6 and to decrease the aerosol dispersion by a factor of 7. Smaller cell volumes were less efficient due to increasing aerosol–wall interactions. The length and internal diameter of the transfer tube to the ICP-MS also affected aerosol dispersion.
The influence of different wavelengths on fractionation and matrix effects was investigated by Motelica-Heino et al.65 Whereas UV (266 nm) lasers produced sub-microscopic particles with similar composition to the original matrix, IR (1064 nm) lasers gave rise to larger (1–10 µm) particles, enriched in refractory elements (Al, Ca). Fractionation effects were limited to the volatile elements using the UV laser and were related to the oxide melting point of the elements. Matrix effects were limited by the use of La as an internal standard. Gunther et al.66 undertook experiments using a Nd:YAG (299 nm) and an ArF excimer (193 nm) laser coupled to an ICP-TOFMS instrument to study the wavelength-dependent ablation behavior of different materials. The investigations indicated that sample removal is both wavelength- and material-dependent. The reduction of interferences was also investigated using a DRC. Becker and Tenzler67 used a range of wavelengths to determine the trace-element composition of quartz glasses, obtaining LOD in the low ng g−1 to pg g−1 range.
Quartz glasses were studied by Tibi and Heumann68 using a LINA-Spark atomizer (an IR-LA system). The laser was focused 15 mm behind the sample surface. Aerosol moistening and the use of SiAr+ as an internal standard allowed precisions of better than 5% RSD to be obtained. The LOD for bulk analysis were in the ng g−1 range.
The accuracy of compositions determined using single pulse LA was assessed by Liu et al.69 using the NIST 610 glass SRM. Elemental fractionation showed significant variability between subsequent laser pulses. The Pb∶U ratios were higher for the first ablation, with an average of two ablations giving the most representative ratio. It was also noted that the mass of material removed per ablation was greater for the second pulse. Particle-induced spikes were dependent on both irradiance and matrix composition. In the semi-quantitative analysis of metal samples using single shot LA-ICP-TOFMS, elemental signals normalized to the total ion current gave the relative percentage of analytes.70 A dynamic range of up to four orders of magnitude was observed and the technique was found to be largely free of matrix effects.
Depth profiling using LA-ICP-MS has been the subject of several studies. Horn et al.71 have quantified the ablation rates of selected metals, the NIST 600 series glass SRM and natural calcium fluoride, using both excimer (193 nm) and Nd∶YAG (266 nm) lasers at a fluence of between 3.5 and 35 J cm−2. The results may be used to aid the selection of optimum laser parameters for future depth profile analyses. Mason and Mank72 investigated the potential of lasers of 193 and 266 nm wavelengths for depth analysis in layered glasses and metals. Optimization of the laser power density, crater geometry and gas medium allowed the determination of layer composition at up to 200 µm depth to be made. The resolution was found to be limited by beam-induced mixing of distinct layers. Thickness determinations by LA-ICP-MS were compared by Plotnikov et al.73 with those obtained using the classical calotte grinding technique and GD-OES. An ablation rate of ≤100 nm per shot using 1.5 mJ beam energy, 120 µm spot size and 5 Hz pulse rate gave a depth resolution of ≤2.5 µm with 5% RSD.
Calibration strategies for LA-ICP-MS remain a subject for discussion. Synthetic, matrix-matched standards have been developed for obsidian rock samples and archaeological artefacts74 and for a range of Ti-bearing minerals.75 Solution calibration for high-purity platinum analysis has been investigated by Becker et al.76 Multi-element standard solutions were aspirated with an ultrasonic nebulizer during ablation of a high-purity platinum target. Results for the NIST 681 Platinum SRM were in good agreement with certified values with RSD between 2 and 10%.
The coupling of LA systems to magnetic sector ICP-MS instruments has allowed the precise in situ measurement of isotope ratios to be achieved. Hirata and Ohno77 utilized the reduction in polyatomic interferences achieved when using a dry plasma to study 54Fe∶56Fe and 57Fe∶56Fe ratios. Following an “on-peak” baseline subtraction, precisions over a four month period were 0.5–1%. These variations were attributed to instability of the high voltage power supply which caused fluctuations in the ion energy of analyte ions. Bracketing samples with analyses of NIST 665 resulted in internal precisions of better than 0.1%, suggesting that this approach successfully corrected for both instrumental drift and background contributions. Hirata78 determined the isotopic composition of Zr in terrestrial and meteoritic zircon and baddeleyite samples. Precisions (2σ) were 0.01–0.02%, 0.02–0.03% and 0.03–0.04% for 92Zr∶90Zr, 94Zr∶90Zr and 96Zr∶90Zr, respectively. Hafnium and Pb isotope ratios were determined for ancient zircons by Knudsen et al.79 Li et al.80 investigated methods of improving the precise, in situ dating of zircons using 206Pb∶238U ratios. Fractionation of Pb from U was minimized by using ablation along a linear scan, rather than a depth profile. This allowed ratios to be determined with a precision of between 0.8 and 5% RSD in relatively large (>100 µm), homogeneous zircons. Davidson et al.81 found LA-ICP-MS to be a rapid method for the study of the variation of Sr isotope ratios in single feldspar crystals. Abrupt changes in the 87Sr∶86Sr ratio could be linked to changing conditions during the formation of the crystal.
Other geological applications of note included the use of a DRC with conventional LA-ICP-MS to reduce polyatomic interferences and lower the LOD for the most abundant isotopes of Ca (by 2.5 orders of magnitude) and Fe (by a factor of 20) in single melt and fluid inclusions.82 Hydrogen combined with a buffer gas to enhance thermalization of the ions was found to be particularly efficient at removing Ar-based interferences. The fingerprinting of sapphires to distinguish their origin requires the use of a minimally destructive technique. Guillong and Gunter83 developed a method for the determination of 40 isotopes in a 2 s analysis. Optimum conditions were found to be a laser fluence of 6 J cm−2, crater diameter 120 µm and repetition rate of 10 Hz, resulting in a total sample mass consumption of 60 ng per 2 s analysis. Calibration was carried out using the NIST 612 glass SRM with Al as internal standard.
The analysis of platinum group elements (PGE) has been undertaken in road sediments by Motelica-Heino et al.84 using both UV (266 nm) and IR (1064 nm) Nd:YAG lasers. Data were compared with results from solution high resolution (HR) ICP-MS. Good agreement between techniques was obtained for Pt and Rh: however, interferences on Pd remained a problem. The LOD were estimated to be in the lower ng g−1 range. Rauch et al.85 continued this work by studying the variations in PGE concentrations in airborne particles. High concentrations and Pt∶Rh ratios of up to 12.3∶1 were found in a few particles on PM10 filters and were attributed to catalyst ageing.
Other novel environmental applications of LA-ICP-MS included the determination of annual variations in trace elements concentrations in ice cores by Reinhardt et al.86 A specially constructed sample chamber allowed a spatial resolution of 300–1000 µm to be obtained, much better than that for conventional solution techniques. Borisov et al.87 investigated the analysis of pressed powder pellets used in Pu disposal. The effects of different laser conditions, methods of sample preparation and the matrix composition were studied using pellets of CeO2, Bi2O3 and PtO2.
In a related paper, Venable et al.91 described a multiplexing ETV system capable of performing three determinations per minute. A series of 300 W tungsten filaments served as atomizers, and the temperature program for each filament was staggered to achieve the desired improvement in vaporization cycle time. Other figures of merit for the system were similar to those of a traditional ETV. However, the use of tungsten was found to interfere with the determination of several elements, most notably Hg and Mo. The authors suggested that other filament materials, e.g., rhenium or tantalum, could be used in place of tungsten, thereby eliminating the interference problems.
Having observed that ETV peak broadening occurred at a rate much higher than that expected from longitudinal diffusion alone, Venable and Holcombe92 subsequently studied parameters that influenced extra-diffusional broadening. Diffusion coefficient, tube diameter, transport tube length, and gas flow rate were built into a Monte Carlo simulation of the ETV system. Using typical operating parameters, the dispersion was found to be primarily a consequence of laminar flow conditions in the carrier gas stream. Dispersion decreased when the carrier gas flow was outside the laminar flow regime. The simulation was also used to calculate average particle size for different masses of sample. Particle size was found to be largely independent of sample size over a mass range of 0.01 to 10 ng, indicating that mass transport occurred as a result of increased particle number rather than increased particle size. The authors speculated that their conclusions could be generalized to understand better the behaviour of other transient sample introduction systems.
Friese et al.93 studied the ETV transport efficiencies of nine elements under “wet” plasma conditions. Two distinctly different atomizers and interfaces were evaluated by comparing peak integrals for ETV introduction with those obtained by pneumatic nebulization. The efficiency of the nebulizer was calibrated by determining the mass of aerosol captured on a filter. Over a range of atomization conditions, a “tube and nozzle” atomizer design was found to be the most efficient. Transport efficiency was 15–50% with the modified atomizer, whereas the standard “tube” atomizer was only 10–35% efficient. When 0.5% (v/v) trifluoromethane was added to the carrier gas, efficiencies were 20–70% and 70–100% for the standard and modified atomizers, respectively.
The suitability of ETV sample introduction for multielement determinations was re-examined by Resano et al.94 Three elements (Cd, Co and Ti) with significantly different vaporization properties were studied, and the effects of dwell time and data processing methods examined. Precision, sensitivity and LOD were not affected by changes in dwell time, as long as the transient signal was well-defined by 3 or 4 signal measurements. The authors suggested that as many as 20 isotopes could be determined simultaneously in a typical 1.5–2 s analysis, a significant increase on the number of isotopes typically determined.
Yu et al.95 used a water-free aerosol to decrease the amount of O2 present in the ID-ETV-ICP-MS determination of S in fossil fuels. Addition of N2 to the plasma further reduced the interfering 16O2 peak. This allowed the measurement of 32S∶34S to be made with a precision of 0.3% in spiked samples and 0.7% in unspiked samples. The results were within 0.3% of those determined using TIMS. The LOD were 4 ng l−1. The work of Okamoto96 was noteworthy in that it allowed determination of F, which is not generally ionized in an ICP. A LOD of 0.29 µg was achieved for aqueous F by removing water which would otherwise have suppressed ionization. Precision for a F concentration of 5 µg was 3.6% RSD (n = 10). Tetramethylammonium hydroxide was added into a tungsten boat furnace to prevent loss of F. Interference from 18O1H+ was reduced by increasing the plasma/interface sampling distance.
Slurry sampling ETV with ID was used by Maia et al.97 to determine Tl in sediments. Analysis of NIST 2704 Buffalo River sediment SRM and other freshwater sediments indicated that 973 K was a suitable pyrolysis temperature for this analysis. However, tests with marine sediments gave concentrations much higher than expected. This was attributed to an excess of labile chlorine (presumably chloride) in the marine sediment samples that readily reacted with the spike isotope. This lead to the premature vaporization of TlCl at temperatures in excess of 673 K. The problem was eliminated by operating at a pyrolysis temperature of 673 K for marine sediments. The LOD of this method was estimated at 3 ng g−1 from analysis of HISS-1 sediment RM. The precision of Tl determinations at the 0.5–1.0 µg g−1 level was found to be 1.3–4.0% RSD (n = 5).
A headspace extraction technique for the determination of tributyltin in aqueous solutions was described by Mester et al.98 The analyte was volatilized by reaction with chloride present in sea-water and in aqueous solutions of HCl and sodium chloride. The tributyltin chloride vapour was concentrated on a solid phase microextraction (SPME) fibre coated with polydimethylsiloxane–divinylbenzene and, upon thermal desorption, the Sn was detected by ICP-TOFMS. The LOD for Sn in aqueous solution was 5.8 ng l−1 and results from the determination of tributyltin in the PACS-2 (sediment) CRM agreed well with certified values.
Christopher et al.99 developed an ID CV procedure for the determination of Hg in several NIST SRM. The authors noted that CV Hg introduction significantly reduced the memory effect and long washout times characteristic of Hg determination with solution nebulization. Instrument sensitivity for CV introduction of 201Hg was 2 × 106 ions s−1 per ng g−1. At optimum dwell time, the precision of the 201Hg∶202Hg isotope ratio in NIST 3133 (mercury spectrometric solution) SRM was 0.34% RSD (n = 3) with a bias of only 0.13% relative to the natural isotope ratio. Determinations of Hg in NIST 1641d (1.6 mg l−1 Hg in water) SRM by CVAAS and ID-CV-ICP-MS agreed to within 0.3%.
Myojo et al.100 described a sample introduction system for the analysis of aerosol particulates separated according to size. Particles in undifferentiated aerosols were separated using a differential mobility analyser (DMA) and then re-suspended in argon to form an “argonsol”, which was injected into an ICP-MS for analysis. Using this system, the concentration of lead nitrate in particles ranging in size from 30 to 140 nm could be measured. The absolute sensitivity of the DMA argonsol system for Pb was reported to be in the fg range. It was speculated that the system would be a suitable replacement for existing sample introduction systems in ICP-MS and ICP-AES.
In the high-efficiency cross-flow micronebulizer (HECFMN) described by Li et al.,101 the diameter of the gas nozzle was only 75 µm. A 50 µm id fused silica capillary was used for sample uptake. At 1.1 kW ICP forward power, the optimum sample uptake rate for the HECFMN was between 5 and 120 µl min−1 for gas flows between 0.8 and 1.0 l min−1. In direct comparison with a standard cross-flow nebulizer, aerosol particles produced by the HECFMN were found to be smaller on average, and the particle size distribution was narrower. Sensitivity, precision and LOD were equal to or better than those of the standard nebulizer, despite the fact that the high efficiency device consumed 20 times less sample. The authors speculated that the HECFMN has excellent potential for service as an interface between ICP-MS, capillary electrophoresis and microbore HPLC.
Minnich et al.102 used optical patterning for the spatial characterization of aerosols produced by the direct-injection high-efficiency nebulizer (DIHEN). Aerosol cone diameter and cone angle were reported at selected cumulative volume per cent. of the total aerosol, as a function of nebulizer gas flow, and at three axial locations downstream from the DIHEN tip. The primary aerosol was more confined as gas flow increased and “satellite droplets” were more prevalent as the axial distance was increased and the gas flow was decreased. The implications of these observations with respect to noise and plasma–sample interactions were discussed.
A heated knotted reactor FI system was described by Chen and Beauchemin106,107 for the isolation and concentration of trace metals in saline water. A 2% solution of sodium diethyldithiocarbamate in 1 M ammonium acetate was used to precipitate Co, Cr, Cu, Fe, Mn and Ni from the SLEW series of estuarine water RM. The precipitate was collected in the knotted reactor, where it was washed with high purity water and dissolved with 50 µl of 4 M HNO3 containing 10 µg l−1 Ga as internal standard. With five-fold concentration of the original sample and using external calibration, a complete analysis took only 3.5 min and the measured concentrations were in good agreement with certified values.
Benkhedda et al.108 also used a knotted reactor procedure in the determination of REE in natural waters at sub ng l−1 concentrations using ICP-TOFMS. The REE were selectively chelated with 1-phenyl-3-methyl-4-benzylpyrazol-5-one at pH 3.7–4.6 and the complexes were concentrated by sorption on the walls of the PTFE knotted reactor. The immobilized analytes were eluted using 2% HNO3 containing 0.5 µg l−1 each of In and Rh and the eluate introduced into the plasma via an ultrasonic nebulizer. Concentration factors of 15–22 were attained and LOD ranged from 3 to 40 pg l−1. For 100 ng l−1 of analyte, precision was better than 5% RSD (n = 9).
A novel system for the indirect determination of thyroid stimulating hormone (TSH) in human serum was developed by Zhang et al.109 Anti-TSH monoclonal anitibodies were immobilized on a solid support and used to capture TSH present in the serum sample. A mixture of biotinylated anti-TSH antibodies and Eu-labelled streptavidin were then passed over the stationary phase to form a captured TSH–antibody–Eu–streptavidin complex. The bound Eu was extracted by 1% HNO3 and detected by ICP-MS. The calibration was linear to 170 mIU l−1 of TSH, and the LOD for TSH was 0.5 mIU l−1. Within- and between-run RSDs (n = 6) were better than 10%, and assay results were strongly correlated with radioimmunoassay data (r2 = 0.9639).
Sequential injection analysis (SIA) is a flow technique in which the sample volume is regulated by reverse aspiration through a multiport selection valve. Unlike conventional FI, the sample volume is not defined by the volume of an injection loop. Egorov et al.110 described an SIA manifold for the separation and determination of actinides in solutions of vitrified nuclear waste. An extraction chromatography medium was employed to eliminate spectral interferences due to direct overlap of Am, Np and Pu isotopes as well as to protonated species (e.g., 238UH+). The extraction procedure also helped to overcome what would have otherwise been inadequate abundance sensitivity. The authors found that uranium matrix removal was particularly efficient when the sample was treated with ferrous sulfamate prior to separation.
Wang and Hansen111 developed a SIA scheme using a renewable microcolumn for preconcentration of Bi and Ni in the determination by DIHEN-ICP-MS. The microcolumn was created by aspirating 15 µl of a suspension containing Sephadex C-25 cation-exchange resin and was then loaded with sample solution. After removal of matrix components using a dilute (ca. 1%) HNO3 carrier, the analytes were eluted using ca. 8% HNO3. The resin was discarded after each elution and a new column aspirated for the next operation. Data acquisition for one sample was performed in parallel with the preconcentration process for the following sample. Precision was 1.7 and 2.9% RSD for 0.8 µg l−1 solutions of B and Ni, respectively. Sampling frequency was 12 per hour and LOD (3 σ) for Bi and Ni were 4 and 15 ng l−1, respectively.
In the ID determination of Cd using the programmable microflow system described by Packer et al.,112 the software controlled the sample∶spike (112Cd) ratio by adjusting the relative contributions of sample and spike solutions to the uptake rate of 65 µl min−1. Samples were introduced into the plasma using a microconcentric nebulizer. Programmed and observed114Cd∶112Cd ratios were strongly correlated (r = 0.997) for isotope ratios from 6 to 25. A comparison of determined and expected Cd concentrations between 2 and 20 µg l−1 showed good agreement at the 95% confidence interval. It proved possible to perform 30 ID analyses per hour with consumption of only 33 to 168 pg112Cd per analysis.
Lambertsson et al.117 used a double-spike ID protocol with GC-ICP-MS for the determination of methylmercury and the extent of methylation and de-methylation in brackish water sediments. Divalent 199Hg was added to sediment extracts to monitor changes in mercury methylation during sample handling, whilst 200Hg-enriched methylmercury was added as a species-specific ID spike. The extracts were treated with sodium tetraethylborate to form ethylated derivatives that were captured in a Tenax trap prior to GC analysis. The LOD was 3 ng g−1 for Hg and the precision 2–3% RSD for methylmercury concentrations of 2–5 ng g−1. The LOD for methylation and demethylation reactions were 0.1 and 0.2 ng g−1, respectively. Demuth and Heumann118 also used ID to validate determinations of methylmercury in aquatic systems. Changes in mercury speciation were noted during the ethylation process and the authors noted that halides promoted demethylation. Interestingly, the authors found that propylation had no effect upon either mercury speciation or the accuracy of the methylmercury determination.
Encinar et al.119 used ID-GC-ICP-MS for the determination of butyltin compounds in sediments. A 119Sn-enriched spike was prepared by direct butylation of Sn metal. The species composition of the spike was determined by GC-ICP-MS after derivatization with sodium tetraethylborate. The species concentrations were determined using reverse ID with natural abundance standards containing mono-, di- and tributyltin. The reverse ID experiment also indicated that no species conversion took place during the ethylation process. Results for the determination of butyltin compounds in the PACS-2 RM using the prepared spike agreed well with the known species concentrations.
Haas et al.120 used cryotrapping GC-ICP-TOFMS for the species-specific isotopic analysis of volatile Sb and Sn compounds. Various compounds (hydrides and alkylmetal species) were collected in Tedlar bags and analysed using Cd and Rh, added post-column, as internal standards for mass bias and drift corrections, respectively. Isotope ratio precision was limited by counting statistics to 0.3–0.5% RSD in these determinations in both analog detection and pulse counting modes. Accuracy was about 3%. The authors felt that this level of isotope ratio precision might not be adequate for the study of isotopic fractionation due to biovolatilization processes. Despite these concerns, the same group applied a similar approach to the determination of volatile compounds of As, Sb and Sn in CO2-rich atmospheres.121 Although recoveries of As and Sn compounds were of the order of 80–90%; sample pretreatment with NaOH (to eliminate water and CO2 from the sample) reduced Sb species recovery by as much as two-fold.
Krachler and Emons122 used USN with membrane desolvation as an interface for the HPLC determination of Sb species. The authors found that the optimal desolvation temperature for this determination was less than that considered normal for USN. Under such conditions, LOD ranged from 9 ng l−1 for trimethylantimony to 14 ng l−1 for SbIII. The authors attempted to use this system to determine Sb species in urine but problems with co-elution forced them to use a HG interface in place of USN.123 With the HG interface, LOD were 8 and 20 ng l−1 for SbIII and SbV, respectively.
The combination of HPLC and DIHEN was used by Acon et al.124 in the determination of cobalamin species (with reversed-phase HPLC) and organolead and organomercury species (with ion-pairing HPLC). For the latter analysis, peak area precision for successive 10 µl injections was better than 5.1% RSD (n = 5). Absolute LOD were ca. 1 pg. Up to 20% organic content in the mobile phase had no adverse effect upon the DIHEN.
Elimination of background species from the reversed-phase HPLC determination of inorganic Cr species has been achieved through use of a DRC.125 Interferences due to ArC+ and ClO+ were reduced by three orders of magnitude when 0.65 ml min−1 NH3 was added to the DRC. Peak area reproducibility at m/z 52 was better than 2% RSD and the LOD for CrIII and CrVI were both 0.06 µg l−1. Chromium recovery from local water samples was 90–110%.
Latkoczy et al.126 determined Sr isotope ratios in skeletal remains by HPLC and magnetic sector ICP-MS. The chromatographic separation allowed accurate determination of 87Sr abundance to be achieved by removing the 87Rb interference. Use of a shielded torch and USN increased Sr sensitivity 100-fold. At 5 ng g−1 Sr, the 87Sr∶86Sr ratio in NIST 987 SRM was measured with a precision of 0.068% RSD (n = 5) and a bias of 0.064%.
The combination of IC with ICP-MS for speciation analysis of aqueous samples has been reported for a diverse range of applications. Wallschlager and Roehl127 found that a membrane suppressor significantly improved instrument performance, presumably by removal of NaOH from the column effluent, in the determination of Se speciation in estuarine water. Using a 1 ml sample, LOD for Se were in the range from 1.6 to 2.6 ng l−1. Reproducibility was 7–11% RSD at 50 ng l−1 Se and 0.4–8% RSD at the 500 ng l−1 level. Gurleyuk and Wallschlager128 used suppressed IC for the determination of CrIII and CrVI species in aqueous media. The CrIII species was complexed with EDTA to separate the anionic complex from the anionic CrVI species. Using 1 ml samples, the LOD for CrIII and CrVI were 5 and 12 ng l−1, respectively, and the reproducibility (n = 5) was better than 5% RSD at 50 ng l−1 Cr. Similar systems were used for the determination of anionic halogen, phosphorus and sulfur species in natural and waste waters.129,130 These studies focused strongly on bromate (a regulated species), for which recovery from 0.1 to 0.5 ml samples was 101% with a LOD of 1 µg l−1.
Work aimed at improving the electrophoresis-ICP-MS interface continued apace during the period covered by this Update. Polec et al.131 described a CZE-ICP-MS system with a self-aspirating (6 µl min−1) total consumption micronebulizer for the determination of metallothioneins. The absence of nebulizer suction improved separation efficiency and the low uptake obviated the need for a post-column make-up flow of liquid carrier, thereby improving sensitivity. The LOD for bound Cd of 10 µg l−1 (as the metallothionein–Cd complex) was obtained within 10 min at near baseline separation. Li et al.132 described a high-efficiency cross-flow interface for CE-ICP-MS in which uptake was as little as 5 µl min−1 and dead volume only 65 nl. These authors also noted improved separation efficiency as a result of the low nebulizer suction, and although a make-up carrier flow was employed, LOD for 10 REE were 0.05–1 µg l−1. Peak widths were 9–18 s, and peak area precision between 6 and 12% RSD (n = 5). Da Rocha et al.133 determined Hg species by CE-ICP-MS using a vapour generation interface. Inorganic, methyl-, and ethylmercury species were separated and volatilized by post-capillary reaction with sodium borohydride. Mercury vapour was extracted in a gas–liquid separator and detected by both quadrupole and magnetic sector ICP-MS. Under optimized operating conditions, Hg LOD in the low µg l−1 range were achieved.
Various aspects of interferences in ICP-MS have been studied by Nonose and Kubota.135,136 In the first paper, physical properties of a plasma in HR-ICP-MS were measured using an optical fibre. Behind the sampling cone, band emissions due to water predominated and the excitation temperature for Ar (Ar Tex) was 3230 K. Behind the skimmer, band emissions for N2 were strong and Ar Tex was 8120 K. Changing the accelerating voltage to the interface did not affect experimentally measured temperatures behind the sampler. However, increasing the voltage led to a reduction in Ar Tex and the vibrational temperature for N2 behind the skimmer. Theoretical dissociation equilibrium temperatures for ArX+ also decreased with increasing accelerating voltage. In the second paper, other effects of changing the accelerating voltage were investigated in an attempt to explain the signal enhancement in HR-ICP-MS. No change in the response curve was found at different voltages but the ratio of analyte signal in the presence of a matrix to the analyte signal in the absence of a matrix increased with increasing voltage. The amount of signal enhancement was greater for matrix ions with low ionization potential and higher mass. The authors concluded that the higher ionization and/or transportation efficiencies were achieved behind the sampling cone and/or the skimmer by using higher acceleration voltages.
Houk and Praphairaksit137 investigated the dissociation of polyatomic ions in a plasma. The origin of polyatomic ions was identified using a dissociation hypothesis and measured analyte ratios, mass bias corrections and estimates of the density of the neutral component to give the kinetic temperature of the gas. Plasma temperatures were estimated using the pressure difference between Ar at plasma and room temperatures. Depending on the plasma potential, species such as ArN+ and ArO+ were either dissociated or formed between the sampling cone and skimmer. Levels of refractory oxide ions were consistent with a gas kinetic temperature of 5300 K, but an excess of ClO+ was observed and higher densities than expected of some other polyatomics were present in the plasma.
The elimination of interfering species using DRC-ICP-MS was discussed by Latino et al.138 Highly reactive gases could be used to reduce interferences, with product ions being expelled from the reaction cell. Simpson et al.139 used a DRC with O2 reactant gas to form multiple oxide species (MOn+). This allowed the accurate measurement of noble metals (Hf, Nb, Ta and Zr) in the absence of interfering oxides (MO+) in a soil RM to be made. The reaction mechanism was traced by monitoring oxide species. Isotope ratio precisions and LOD were significantly improved in comparison with results for standard ICP-MS configurations. A DRC with CH3F as the reaction gas was used by Moens et al.140 to overcome the isobaric overlap of 87Rb on 87Sr. By forming and determining SrF+, 87Sr∶86Sr ratios could be measured as 87Sr19F∶86Sr19F directly in solutions without the need for chemical separations. Ratio precisions were close to those predicted by counting statistics. Hattendorf et al.141 investigated the problem of CrAr+ overlap with isotopes of Nb and Zr. Hydrogen reactant gas was used in the DRC to suppress metal argide formation. Signal-to-background ratios were increased by adjustment of the bandpass quadrupole. The LOD for Nb and Zr in pure chromium metal were 2 and 5 ng g−1, respectively. Chang and Jiang142 used NH3 as the reactant gas in a DRC to facilitate the determination of Cr in water and urine. Interferences from species such as 40Ar12C, 40Ar12C1H, 37Cl16O, 35Cl16O1H and 34S18O were reduced by 2–3 orders of magnitude, giving LOD of 0.015 and 0.024 ng ml−1 for 52Cr and 53Cr, respectively. Sample replicate precisions were better than 20%.
A review of ion–molecule interactions in an rf-driven collision/reaction cell was presented by Bandura et al.143 At nearly thermal ion energies (<0.1 eV), the number of collisions increased, suggesting that chemical reactions were more efficient at improving signal-to-background ratios than collisional fragmentation. High chemical resolution was achieved by oxidation, hydroxylation or chlorination of analyte ions. The optimization of gas type, pressure and ion energies for the removal of interferences was discussed.
A hexapole collision cell filled with H and He was used by Boulyga and co-workers144,145 to reduce Ar-based interferences in ICP-MS analysis. Figures of merit included: an increase in sensitivity for heavy elements (m/z > 100) of up to 10 times; the reduction of signals for 40Ar, 40Ar16O and 40Ar2 by 4, 2 and 5 orders of magnitude, respectively; a decrease of up to 10 times in metal oxide formation and an increase of 5 orders of magnitude in the S/N for 80Se∶40Ar2. Ratios of 44Ca∶40Ca, 56Fe∶57Fe and 78Se∶80Se were determined using a Meinhard (Ca and Fe) or an ultrasonic (Se) nebulizer with precisions of 0.26, 0.19 and 0.12% and accuracies of 0.13, 0.25 and 0.92%, respectively.144 Using mixtures of H2 and He, Ingle et al.146 measured the signal for Ar and water-based species to identify the major chemistries affecting collision cell performance. Ratios of interfering species were used to distinguish between different cell conditions. The LOD for 56Fe, 80Se and 28Si were improved at the expense of higher oxide levels and poorer sensitivity.
In a study of mixed Ar-N plasmas, Xiao and Beauchemin147 found the addition of 10% N2 to provide robust conditions. Matrix suppression effects caused by sodium ions were converted into slight signal enhancements in Ar–N2 and mass discrimination was reduced relative to that seen for a 100% Ar plasma. The LOD and sensitivity across the mass range were generally poorer with N2 than without, but improvements were seen for elements suffering from Ar-based interferences, such as Fe and Se.
Zhou et al.148 used a range of calibration strategies to overcome the interference of copper and nickel argides on PGE. Standard addition allowed appropriate corrections to be made using both internal and external calibration procedures. Tromp et al.149 used “total interference levels” (TIL) for pairs of interferants (chosen from Al, Ba, Cs, K and Na) to indicate whether a simple calibration strategy could be used for analysis. The results showed that calibration errors could be successfully predicted using the TIL model for ICP-MS.
Moor et al.150 demonstrated the potential of HR-ICP-MS in the field of ultra-trace element detection. Studies of pure waters found that the limiting factor for elemental detection was the blank levels present. Sensitivity was high enough for the majority of applications. Preliminary analyses were deemed necessary for sediments and soils in order to identify possible interferences and establish the levels of total dissolved solids.
Henry et al.151 presented a review of the use of quadrupole ICP-MS for routine measurements in the nuclear industry. They suggested that the high number of studies in this area was mainly due to the sensitivity that is now available for both quadrupole and magnetic sector ICP-MS, as well as rapid measurement. Inn et al.152 compared ICP-MS, TIMS and fission track analyses for the measurement of 239Pu at concentrations of between 3 and 56 µBq in synthetic urine samples. All techniques were found to be capable of making the measurements with a LOD of 1 µBq. McLean et al.153 used a double focusing instrument equipped with a shielded torch and a DIHEN (5 µl min−1) to achieve sensitivities of between 1300 counts fg−1 for 226Ra and 1700 counts fg−1 for 238U. Ultratrace element and isotope ratio measurements were successfully undertaken for a range of long-lived radionuclides (241Am, 237Np, 226Ra, 230Th, 232Th, 233U, 235U and 238U) in standards and environmental radioactive waste solutions.
The determination of actinides following on-line extraction with a TRU.spec resin has been investigated by two research groups. Truscott et al.,154 using a sector-field ICP-MS, achieved absolute LOD of 0.6, 0.65, 0.7 and 0.85 fg for 241Am, 243Am, 137Np and 239Pu, respectively. Analyses were performed on a variety of environmental RM including human liver, human lung, spiked cabbage and Rocky Flats soil. In order to eliminate interference from 238U1H+ on 239Pu, a sequential separation of Pu from U was required. This used an on-column reduction with TiCl3, followed by elution of Pu with 4 M HCl.
An improved separation procedure using tri-n-octylamine columns allowed Ji et al.155 to achieve instrumental LOD of 0.46 pg ml−1 (1.2 × 10−5 Bq ml−1) in the determination of 237Np. As a result ICP-MS can be considered as a rapid and cost efficient alternative to traditional techniques.
A review of MC-ICP-MS by Rehkamper et al.156 covered both the instrumentation and its applications. Performance characteristics and strategies employed to correct for mass discrimination, spectral interferences and matrix effects were discussed. Doucelance and Manhès157 effectively illustrated the changing status of MC-ICP-MS by using it as the benchmark against which the quality of Pb isotope ratios determined by TIMS were judged.
Blichert-Toft158 reviewed the various techniques (TIMS, SIMS, MC-ICP-MS) used to determine Hf isotopes in geological materials. Poor ionization is a major limitation of TIMS analysis and the advent of MC-ICP-MS has enabled the routine determination of Hf isotopes for the first time. High precisions and good reproducibility were easily obtained using MC-ICP-MS after separation of Hf from the matrix.
The determination of the isotopic composition of Ag using MC-ICP-MS was undertaken by Carlson and Hauri.159 Samples were spiked with Pd to allow fractionation of Ag to be corrected. The 107Ag∶109Ag ratios were measured with an external precision of 1.3 ε, sufficient to distinguish between low Pd∶Ag objects including including pallasites, group IA iron meteorites, chondrites and terrestrial samples.
Galy et al.160 used MC-ICP-MS to determine Mg isotope ratios in terrestrial materials. Following minimization of interferences caused by C2, CN and MgH, the long-term repeatability of 26Mg∶24Mg ratios relative to the NIST 980 (Mg) SRM was 0.12‰ and that of 25Mg∶24Mg was 0.06‰ (95% confidence). The 26Mg∶24Mg ratio was increased by 0.2‰ by the addition of a solution containing Al, Ca and Na. This mass-dependent bias was observed to be greater in the presence of Ca than Na. Interference from 48Ca2+ was only significant when Ca∶Mg was greater than 0.5.
The fractionation of Mo isotopes has been quantified using MC-ICP-MS. Siebert et al.161 used a Mo double spike to correct for laboratory and instrumental mass fractionation. Using a minimum sample size of 1 µg Mo, highly precise results (±0.06‰, 2σ, for 98Mo∶95Mo) were obtained for hydrothermal molybdenites and fine grained sediments. Anbar et al.162 used a Zr or Ru spike to correct for instrumental Mo isotope fractionation. Ratios of 95Mo∶97Mo were determined with precisions of 0.2‰ (2σ). Ion-exchange chromatography caused a fractionation of 1.5‰ per amu and a bias of −0.3‰ was observed between a laboratory standard and natural MoS2. Barling et al.163 applied similar methodology to the analysis of 97Mo∶95Mo in anoxic sediments and ferromanganese nodules. The precision of the measurements (0.25‰) was sufficient to detect natural Mo fractionation resulting from changing redox conditions.
Inaccuracies in isotope ratio measurements made using MC-ICP-MS were noted by Thirlwall.164 This problem was exacerbated by instruments with poor abundance sensitivity. Corrections made using conventional TIMS techniques were found to be inappropriate. Instead, a tail correction using a mono-isotopic ion beam was applied. Using this method, inaccuracies in published Pb isotope ratios were explained and corrected.
Al-Ammar and co-workers developed a common-analyte internal-standardization chemometric-algorithm165 to improve the precision of boron isotope ratiomeasurements using a quadrupole ICP-MS instrument.166 Precisions of 0.05% RSD were obtained for 11B∶10B ratios after correction for short-term noise, instrumental drift over the 20 min counting time, matrix effects and mass bias. The method had been validated using both synthetic solutions and sea-water samples analysed with conventional techniques.
Applications utilising the high resolution capacity of sector field ICP-MS include the quantification of Mg isotope ratios in fish intestines.167 A mass resolution of 300 was found to be sufficient to separate interferences such as C2+, C2H+, CN+, NaH+ and 48Ca2+. Cool plasma conditions were also evaluated, but interferences remained a problem. The 26Mg∶24Mg ratios were measured with a precision of 0.2–0.5% RSD.
Trace concentrations of Si in biological and clinical samples were determined using ID-HR-ICP-MS by Klemens and Heumann.168 Samples were digested in either HNO3 or HNO3–HF using a microwave procedure and spiked with 30Si. The LOD were dependent on blank levels which were minimised by the use of a PFA spray chamber, sapphire injection tube and a silicon nitride torch. For samples with a concentration range of 1–600 µg g−1, precision was between 2 and 4% RSD. The LOD were 0.15 and 0.2 µg g−1 for the HNO3 and HNO3–HF digestions, respectively.
Makishima and Nakamura169 reported the determination of total S in silicate RM and the Allende meteorite by ID-HR-ICP-MS. The S was oxidized to sulfate by the addition of bromine. Total blanks of 46 ng could be reduced to 8 ng through the use of a clean air system during evaporation. The LOD for 32S+ were 1 ng ml−1 for an ideal solution and 0.07 µg g−1 for silicate samples. Those for 34S+ were 6 ng ml−1 and 0.3 µg g−1, respectively. Reproducibility was better than 9%. Evans et al.170 developed a method for the determination of S in fuels. Natural and spiked 32S∶34S ratios were determined at a mass resolution of 3000 to avoid spectral interference from O2+ molecules. Mass bias was corrected using the natural isotopic ratio of Si. Uncertainty on S concentrations determined at ∼50 µg g−1 (the EC limit on S content in fuels as from 2003) was 1–2%, similar to that obtained using routine methods.
Concentrations of Cd, Cu, Ni, Pb and Zn were determined in an estuary water RM using ID by Evans and Fairman.171 Polyatomic interferences, such as 23Na40Ar on 63Cu, were resolved using the precise mass calibration. Combined uncertainty calculations indicated that the majority of the 2–3% systematic uncertainty was derived from the isotope ratio measurement.
The high sensitivity of magnetic sector ICP-MS has been utilized by Choi et al.172 for the determination of 231Pa and 230Th in sea-water by ID. Following pre-concentration, samples were aspirated using a desolvating micro-nebulizer. Internal precision was ∼2% and reproducibility ∼5% at the 95% confidence level. The LOD for 231Pa and 230Th were 0.02 and 0.15 fg kg−1, respectively.
Boulyga and co-workers173–175 used a range of techniques to investigate the isotopic composition of Pu and U in contaminated soils. A quadrupole ICP-MS instrument with a hexapole collision cell and shielded torch was used to determine 236U concentrations and U isotope ratios.173 A LOD of 3 pg l−1 was achieved for 236U, with the formation of UH+ ions restricted to 2 × 10−6 (UH+∶U+) by the use of ultrasonic nebulization (USN) with membrane desolvator. Ratio precisions were 0.45 and 0.88% for 235U∶238U and 236U∶238U, respectively, when using the USN device and 0.13 and 0.33%, respectively, using a Meinhard nebulizer. Measured ratios, however, were found to be consistently lower than those calculated for synthetic mixtures. A similar experiment was undertaken using a magnetic sector ICP-MS instrument, also with a shielded torch.174 Using a MicroMist™ nebulizer, a detection efficiency of 4 × 104 counts atom−1, a UH+∶U+ ratio of 1.2 × 10−4 and precisions of 0.11 and 1.4% for 235U∶238U and 236U∶238U, respectively, were achieved. Using a DIHEN, the detection efficiency was 10−3 counts atom−1, the UH+∶U+ ratio 1.4 × 10−4 and ratio precisions 0.25 and 1.9%, respectively. Isotopic ratios for Pu and U were determined using a magnetic sector ICP-MS instrument coupled to a USN with membrane desolvation.175 This configuration resulted in LOQ for both 239Pu and 236U of 0.6 pg l−1 in solution and 0.13 pg g−1 in soil. Sensitivity at 238U was up to 4 × 109 counts s−1 ppm−1, and UH+∶U+ ratios were 3 × 10−5.
Taylor et al.176 measured Pu isotope ratios using a MC-ICP-MS instrument. Solutions were doubly spiked with 233U and 236U prior to analysis to allow corrections for instrumental drift and mass bias to be calculated. This technique allowed 240Pu∶239Pu ratios to be determined to within 1.4% (2σ) at 100 fg ml−1 and <0.3% at >3pg ml−1. Ratio precisions for 242Pu∶239Pu were better than 2% and 10% in solutions containing 30 and 5 fg 242Pu, respectively. Precisions of less than 1% are required to distinguish between different sources of anthropogenic Pu so ICP-MS can be considered a practical technique for the analysis of environmental samples.
Isotope dilution HR-ICP-MS was used by Agarande et al.177 to determine the concentration of 241Am in sedimentsat pg kg−1 levels. A number of steps were required to separate out the major and other transuranic elements. Resulting solutions were purified by anion exchange and extraction chromatography.
Laser-inducedchemical damage is a fundamental issue for depth profiling applications. Therefore, Kasaai et al.179 studied the chemical composition of the ejected material and the material remaining after multi-shot irradiation of fused silica with 150 fs pulses at a wavelength of 800 nm and a repetition rate of 1 kHz. The gas phase composition was monitored by LIMS. The composition of the irradiated sample was measured by SIMS and SEM-EDXRF. During irradiation, a marked increase of the H2O content in the vacuum was observed but SIMS revealed an identical elemental composition for irradiated and fresh surfaces. The original silicon composition was shown by SEM-EDXRF to be preserved in the molten material splashed back on the sample, whereas X-ray diffraction showed the development of crystalline phases in the redeposited material. These observations pointed to the predominance of thermal processes. Indeed, if electronic excitation prevailed, dissociation of sample species and their subsequent reaction with residual vacuum components, such as CO, CO2 and H2O, would lead to oxidation or carbonization of the surface.
The determination of 235U trace concentrations in as-received samples is important in studies of the U-fuel cycle. In a study of U isotope ratio measurements in metallic uranium without sample pretreatment, Vors et al.181 ablated the metal with 10 ns pulses at an incidence angle of 0°. The laser-produced jet, collimated by a 2 mm orifice, was post-ionized at a distance of 75 mm from the sample by the third harmonic of a Nd∶YAG laser (4 ns pulse). Analysis was performed using a 10 Hz linear TOFMS instrument, made in-house. The signals were averaged over 30 shots. The instrument was optimized using natural samples with a relative abundance of 235U∶(238U + 235U) of 0.72%. Up to 10 235U ions could be detected from a single laser shot. A mass resolution of 400 was achieved when the ionizing laser was focused to achieve an overlapping area with the atomic beam of 0.09 cm2. The isotope ratio measured in the non-selective ionization mode was 0.71 ± 0.05%. Calculations yielded a good agreement with the experimental saturation intensity of 20 J cm−2 on the condition that a virtually unpopulated U metastable level was assumed in the photoionization zone.
Stentz et al.182 studied the speciation capabilities of TOF LIMS for sodium-doped and undoped lead borosilicate glass components. A nitrogen laser providing 150 mJ per 10 ns pulse at a repetition rate of 10 Hz was used because the wavelength of 337 nm matched the absorption of lead. Mass spectra were accumulated over 200 shots and the mass accuracy was better than 0.2 u. The negative ion mass spectra showed high m/z ions from the borate and silicate meso-structures. The positive ion mass spectra of lead borosilicates gave evidence for the formation of boroxol rings with one lead cation and of a limited lead oxide network at high contents of boron and lead, respectively. Sodium-doped glass data showed a preference of the sodium for boron units and sodium-containing silicate units with 10–24 mole% Na2O. The authors provided good reasoning for why they considered the detected ions to be structural units instead of species re-assembled in the laser plume. Specifically, the mass spectra showed no dependence on the laser power density, as would be expected for ion formation in the laser plume, the required laser power density was low, and the good mass resolution excluded occurrence of space charging, inevitable in real plasmas.
Yorozu et al.183 investigated the depth profiling of one-step desorption-ionization under resonant LMMS conditions for D-implanted graphite samples. Use of a laser wavelength of 243 nm, corresponding to a 2 + 1 resonant ionization scheme, increased the yield of H and D by a factor of 1000 in comparison with the use of a non-resonant wavelength. The etching rate was 0.2 µm per shot at a laser pulse energy of 1.8 mJ and the crater depth depended linearly on the number of shots up to 2000. The LOD was 1016 atoms cm−3. The experimental depth profile showed a shift towards larger depth and a broader tail than predicted by Transport of Ions in Matter (TRIM) calculations. Assuming a nearly Gaussian etching profile, a formula to account for the increasing crater diameter by consecutive shots was elaborated. In this way, the deviation between the calculated and experimental position of the distribution was improved by a factor of 2 and the widths of the distributions agreed within 1–4 µm. Therefore, LIMS was considered valuable for fast depth profiling over distances of up to 400 µm.
The ion yield in a TOF LMMS may vary between individual particles depending on the particle size, the morphology, the absorption characteristics and the ablation threshold as a function of the wavelength. Kane et al.184 used laser wavelengths of 193 and 266 nm to study the detection of elemental ions from 30–100 nm particles with different matrices using aTOF LMMS. An aerodynamic lens focused a wide range of particle sizes. Signals were detected from almost all particles larger than 100 nm but the probability of detection became poorer for smaller particles, in particular for those with organic matrices. The dependence of the ion yield on laser wavelength and energy was consistent with the requirement of a threshold power density, which was a function of the size and chemical composition of the particles.
Kane and Johnston185 described a method for enhancement of the ablation efficiency of sulfate particles in aTOF LMMS. Specifically, the aerosol was exposed to a supersaturated vapour of 1-naphthyl acetate in a heated mixing chamber at the entrance of the MS instrument. In this way, particles were coated with a strong UV absorber. The procedure preserved the original size distribution except for its shift over a distance of twice the coating thickness of about 6 nm. Whereas the coating increased the probability of detecting ions from 136 nm ammonium sulfate aerosols by a factor of 20, detection of oleic acid and ammonium nitrate particles was not improved. Application to ambient aerosols allowed about 20% of the particles, undetected without coating, to be measured.
A fast adaptive resonance algorithm (ART-2a) for the automated classification of the large data sets in aTOF LMMS was developed by Phares et al.186 Seven types of representative particles, generated from solution using a nebulizer system, were used to evaluate the algorithm. Matrices included inorganic salts and organic compounds. The vigilance parameter had to be adjusted in order to be large enough to distinguish between different classes but small enough to overcome experimental fluctuations. The learning rate had to be optimized to avoid modification of the vector of a class upon the addition of each spectrum. Problems existed with the recognition of ammonium sulfate as a result of its poor ionization yield and with the creation of classes due to the fragmentation of organic compounds on smaller particles.
Watanabe et al.188 investigated the use of RIMS for the detection of Kr or Xe tracer gas leaked into the cover gas of a fast nuclear reactor. A Nd:YAG pumped dye-laser was focused to a spot with diameter of 5 µm to perform a 2 + 1 ionization of the analyte. Measured isotopic abundances of Kr and Xe in argon agreed with the reference values within the experimental uncertainty. The precision was typically better than 2% except for the abundances of <0.1%, for which the RSD increased to 25%. From the extrapolation of the results to the threshold of 1 detected ion, the LOD was estimated to be 6.7 and 0.18 µg g−1 for Kr and Xe, respectively.
Müller et al.189 studied the ultra-trace determination of 41Ca by double- or triple-resonance excitation with narrow-band extended-cavity diode lasers and subsequent non-resonant photoionization of the collimated atomic beam. The quadrupole analyser was optimized for background reduction and neighbouring mass suppression. An overall detection efficiency of 5 × 10−5 was achieved with a selectivity of >108 for isobars. The precision and accuracy for the measurement of 41Ca∶40Ca isotope ratios in synthetically enriched samples were 1 and 5%, respectively. The accuracy was primarily determined by the uncertainty on the reference sample composition. Calibration was linear over 3 decades down to abundances of 10−9. The 41Ca abundance in meteorite samples could be determined at levels as low as 10−9 to 10−11 with a precision of about 10%. Experimental and theoretical investigation of single-, double- and triple-resonance excitation schemes allowed Bushaw et al.190 to push the limits further down to isotopic abundances of 2.5 × 10−13 and a LOD of 2 × 105 atoms using a triple-resonance scheme. The method was ultimately limited by the background from neutral particles and/or Rydberg atoms ionized within the MS. Use of off-axis atomization was expected to give a significant improvement in LOD.
In a study of the processes of laser ablation and non-selective photoionization to be used for the measurement of U-isotope ratios in metallic samples, Semerok et al.191 ablated natural samples directly by using 10 ns pulses from the second harmonic of a Nd∶YAG laser (fluence, 1 J cm−2; spot diameter, 500 µm). For isotopically enriched samples, a heated furnace could be also used to generate the neutral beam. The neutrals were collimated in an atomic jet and post-ionized by 4 ns pulses of the third harmonic of the Nd:YAG laser at a repetition rate of 10 Hz. Fluence was between 10−2 and 7 × 103 J cm−2. Ions were analysed using a linear TOFMS instrument. Comparison of the experimental data and simulations showed strong suppression of the ionization from the metastable level of the U atom. The low population of this level was related to the expansion and resulting population cooling.
Pibida et al.192 determined 135Cs∶137Cs isotope ratios in samples of low-level radioactivity for dating of radioactive waste by using single-resonance excitation of Cs with an extended cavity diode laser (852 nm) and photoionization by an argon ion laser (488 nm). An optical selectivity of 1000 between 135Cs or 137Cs and 133Cs was achieved. The overall selectivity, including the MS detection, was 2 × 109, although 1010 was theoretically possible. The overall efficiency was typically 1–3 × 10−6. A ratio of 138Ba to 135,137Cs of up to 108 could be tolerated by this technique, which is in contrast to TIMS analysis, for which Ba interference is a major limitation. Comparison of RIMS and TIMS data showed an agreement within the experimental uncertainty of 1.5%.
The SIMS literature of last year reflects the increasing use of modelling to push the limits in ultrashallow depth profiling. Improved understanding of beam-induced artefacts is needed to refine the depth resolution function (DRF), which is used to convert the experimental depth profile into the “original” distribution. On the other hand, the use of polyatomic primary ions for depth profiling has emerged as an efficient alternative to low energy monoatomic ions under optimized incidence angle. The incidence and emission angles, mentioned throughout this section, are expressed as the angle between the ion beam and the normal to the sample surface. Applications of S-SIMS for the study of organic compounds are increasing in number but these fall outside the scope of this review.
A primary ion source for polyatomic anions was developed by Gillen et al.194 for use in stigmatic imaging and depth profiling. Selected targets were bombarded with Cs+ ions to produce polyatomic ions such as Aln−, CsCn− (n up to 8) and Cn− (n up to 10) ions. The beam stability was a few per cent. over 20 min and the lifetime of the source was typically one to several weeks. The current measured on the sample was over 1 µA for 10 keV C2− ions and still tens of nA for Cn− (n = 4–10) or CsCn− (n = 2–8) ions. The diameter of the spot on the sample was a few µm. Bombardment of organic films with a static dose of C10− primary ions allowed the yield of structural ions to be increased by a factor of up to 800 in comparison with the use of C1− primary ions. In the depth profiling of a 1 keV As implant in silicon, the decay length was 1.5 nm with a 3 keV CsC6− beam at an incidence angle of 52° but 9 nm with a 14.5 keV Cs+ beam at an incidence angle of 25°.
Beam deflection in current magnetic sector instruments limits the reduction of primary ion energies to about 1.5 keV. Napolitani et al.195 developed a tilted sample holder to be used with 0.65 keV O2+ primary ions in magnetic sector SIMS. In the depth profiling of a 0.25 keV B implant in silicon, use of 0.65 keV O2+ primary ions improved the decay length from 1.5 to 1.0 nm in comparison with the use of 1.5 keV O2+ primary ions. Sensitivity was the same for both beams. The erosion rate could be as high as 16.5 nm min−1 but varied up to 15% during depth profiling with 0.65 keV O2+ primary ions, whereas it stayed constant within 2% with 1.5 keV O2+ ions. Under 0.65 keV O2+ bombardment, significant roughening occurred from a depth of 20 nm onwards.
In a dedicated instrument for MCs+ detection developed by Wirtz et al.,196 a third column (for Cs neutrals) was added to the Cs+ and Ga+ guns. In this way, the Cs concentration at the surface could be finely tuned. The design of the sample stage and ion optics allowed the sputter yield to be optimized by varying the impact angle at constant incident energy. The useful yields of Ag, B, In, Mg, S (detected as MCs+) and F (detected as FCs2+) were between 7 and 80 times higher in the dedicated instrument than in the commonly used commercial one.
Nagashima et al.197 developed a stigmatic image detector based on stacked CMOS-type active pixel sensors (SCAPS). The device had an area of 7.2 × 6.8 mm2 and was comprised of 512 × 490 pixels (14 × 14 µm2 area with an active area of 12 × 12 µm2). Although the detector was robust and allowed direct, two-dimensional, detection of charged particles to be achieved, linearity was poor. Calibration of the SCAPS with a Faraday cup allowed a correction formula for the non-linearity of the transistors in a pixel to be derived. The precision at low count rates was 0.23%, twice the statistical error, and the accuracy of the counting was 1%. The applicability to the detection of up to 105 ions per pixel matched the requirements of imaging. Kunihiro et al.198 determined the noise characteristics of SCAPS in stigmatic imaging with O− primary ions. The source of dark current was found to change from thermal to electron leakage with decreasing device temperature. Hence, shot- and signal-noise at low count rates was reduced significantly by cooling the SCAPS to 77 K. The signal frame fixed-pattern noise due to inter-pixel variations in sensitivity became dominant at high count rates. Use of a non-destructive readout operation based on reset-frame-subtraction reduced the noise to the level of statistical fluctuation of the incident ions. The dynamic range was 80 dB.
Herec et al.200 measured the secondary ion yield of Cs+ and Cl−from silicon as a function of the implanted Cs+ or Cl− primary ions, generated by a thermionic source. Both the emission of ions and the sputtering of neutrals were measured as a function of the implanted ion dose. The production of Cs+ and Cl− secondary ions peaked sooner than the neutrals in the transient regime. However, whereas the emission of Cs+ ions and Cs neutrals reached the same level in the saturation regime, the Cl− ion production kept decreasing exponentially. Hence, different RSF were needed for quantification in either the transient or the saturation regime.
In a study of the sputtering and secondary ion emission from Ni(100) covered with up to 2 monolayers of copper Karolewski and Cavell201 found that the probability of ionization for atoms and dimers sputtered from the Cu/Ni(100) sample was attenuated by a factor of 2–3 in comparison to the sputtering from clean Ni(100). Simulations were performed using molecular dynamics (MD) to predict sputter yields. Comparison of two models to describe the ionization showed that the electron tunnelling model yielded a closer agreement with the experimental data than the thermalization model. The production of Cu+ ions relative to Ni+ ions for 1 monolayer Cu on Ni(100) pointed to an information depth of 0.39 ± 0.15 monolayer. Between 69 and 81% of the atoms in the Cu2+, CuNi+ and Ni2+ ions originated from the surface layer.
Vogan and Champion202 investigated the effects of adsorbed or chemisorbed oxygen on Si(100) on secondary anion and electron emission under bombardment with 50–500 eV Na+ primary ions. The adsorbate coverage ranged from none to over half a monolayer. Adsorbed oxygen enhanced the emission of secondary anions but reduced that of the secondary electrons. Whereas the former observation was also found for metallic substrates, the latter was not. Simulations showed the importance of electronic excitation of SixO− for ion emission. Electrons would be emitted through a process similar to Penning ionization, in which a vacancy due to the excitation of SixO−, could be filled by a valence band electron.
Serrano et al.203 simulated the Si+ and O+ ion emission in the transient regime under bombardment of silicon by oxygen. The implantation, sputtering, replacement/relocation and diffusion (ISRD) code was used to calculate the time-dependent oxygen concentration at the surface and the sputtering yields of silicon and oxygen. The data were fed into a second model, which factorized the ionization probabilities from the bond breaking and the electron tunnelling models. The discrepancy between experimental and simulated depth profiles was tentatively explained by the charge build-up in insulating materials, which was, however, not accounted for in the modelling.
An MD code, which accounted for covalent bonds and van der Waals interactions, was used by Krantzman et al.204 to simulate the sputtering of polyatomic species from a molecular solid. It was shown that molecules hit by 0.3 keV particles initially broke into fragments, which were driven together by their attraction potential and reacted. The resulting reaction energy created more fragments and caused intact molecules to move and eventually be ejected. Computing power limited the simulations to 3–5 ps after impact and a distance of 1 nm in the solid.
Dynamic Monte Carlo simulations consider the slowing down processes as well as thermal processes such as diffusion and segregation. Ishida et al.205 performed Monte Carlo simulations of depth profiling through multilayers of 18 nm Ta2O5 separated by 0.5 nm SiO2 δ-layers. The binary-collision-based code accounted for the generation of interstitial atoms and vacancies, annihilation of vacancies, diffusion of interstitial atoms or primary ions and the relaxation of target materials. The simulations predicted well the experimental shift to the surface of the δ-layer position over 1–3 nm under 3–8 keV O2+ ion bombardment. The shift resulted from the expansion due to the defects behind the δ-layer as collision mixing only broadened the profile. The use of low ion energy or a large incidence angle should reduce the peak shift.
Aoki et al.206 used MD and Monte Carlo simulations for the sputtering and mixing of δ-layers in Si under bombardment with Ar+ ions of less than 0.5 keV. The Monte Carlo calculations were performed with an improved TRIM code which allowed recoil atoms, primary and secondary knock-on atoms to be followed. The MD calculations used model potentials for the Si–Si and Ar–Si interactions. The MD and Monte Carlo approaches yielded similar results with respect to the sputtering yield, depth distribution of the projectiles and mixing of substrate as a function of the incident ion energy. The mixing depth of 4 nm for 0.5 keV Ar+ primary ions was derived from the primary ion implant range. A marker layer within that implant range extended itself isotropically whereas markers at greater depth underwent an apparent shift to surface.
Lenaerts et al.207 stressed the importance of the ion extraction field in the modelling of depth profiles from sub-µm cubic particles with core-shell structure. A TRIM code was used to describe the sputter yield from each point of the crystal and the resulting change in morphology during analysis. An ion trajectory simulation code known as SIMION was applied to characterize the secondary ion extraction. Only 10% of the secondary ions generated from the side surface reached the detector whereas most of those from the top face were extracted. This explained the similarity of depth profiles from cubic microcrystals and from a macroscopic flat surface.
In an investigation into the mixing-roughness-information depth (MRI)-model as a valid basis for DRF, Hofmann208 introduced assumptions to account for the effects from preferential sputtering or segregation. The predictions agreed within 0.2 monolayers with the experimental depth profiles of GaAs/AlAs δ-layers and thick layers. Segregation and compound formation in the mixing zone were the main sources of errors in the MRI applications. The MRI-based reconstruction of the depth composition from SIMS and Auger electron spectroscopy data, was shown for samples with 1–36 monolayers of AlAs separated by 44 monolayers of GaAs.209 Depth profiling was achieved with 3 keV Ar+ primary ions at an incidence angle of 52° and 58° in SIMS and Auger electron spectroscopy, respectively. The fitting of the MRI parameters yielded a mixing length of 3.0 ± 0.3 nm, an information depth of 0.4 nm and a roughness parameter of between 1.4 (SIMS) and 2 nm (Auger electron spectroscopy). The latter discrepancy was due to different experimental conditions with respect to the incidence angle and use of sample rotation in Auger electron spectroscopy. The calculated and experimental intensities scaled linearly in spite of the large variations in local concentrations.
The broadening of the dopant depth distribution during the annealing of ion-implanted materials has become an important issue because of the use of these materials as RM. Mannino et al.210 performed calculations on the transient-enhanced diffusion in ultra-low-energy-B-implanted silicon. Two models were compared. One was based on the kick-out mechanism and the other on the breaking-up of initially implanted boron clusters. The average migration length proved to be 13.4 nm for 1 keV B implanted in silicon at a dose of 1014 cm−2 and annealed at 750 K. The experimental profile broadening was well simulated. The values of the fitting parameters pointed to the formation of defect B-containing clusters, which immobilized a large fraction of the implanted B.
A modified UT–Marlowe Monte Carlo code was used by Wang et al.211 to model oxygen recoil during ion implantation in thin silicon dioxide layers on crystalline (instead of amorphous) silicon. The stoichiometry of the silicon layer was dynamically changed as oxygen atoms were knocked out of the layers. Reduction techniques were introduced to alleviate the computing power requirements. Simulations of the As and BF2 implantation through silicon dioxide layers of different thickness agreed well with the experimental depth profiles measured on HF-etched samples by using standard-derived RSF and depth calibration by stylus profilometry.
Shima et al.212 investigated the combination of front-side and backside depth profiling to localize a buried interface and to reveal the segregation and pile-up of B at the silicon dioxide–silicon interface of ultra-shallow p-junctions. The thickness of a 200 nm layer of α-silicon on the surface was reduced to 100 nm by chemical and mechanical polishing in order to use backside SIMS with a 1 keV Cs+ primary ion beam. The front-side SIMS depth profile was deconvoluted with a DRF based on Gaussian-like broadening of the interface. The atomic mixing length and surface roughness parameter in backside experiments were 1.5 nm and 2.0 nm, respectively. This approach constituted an interesting way of obtaining adequate experimental results on dopant profiles for simulations.
The oxygen-beam-induced oxidation of Si in the transient regime was simulated using codes developed by De Witte et al.213 Dynamic incorporation into the ISRD calculations of the changes in surface composition predicted by a TRIM Monte Carlo code allowed the sputter yields and the erosion rate to be predicted for the altered layer. Equations were introduced to account for the oxidation of silicon to silicon dioxide and for the out-diffusion of excess oxygen in the sample. Simulations were refined with experimental data for the sputter yields as a function of the energy, the incidence angle, the ion dose and the surface receding due to oxygen incorporation and sputtering. The code predicted well the initial and final sputter yields, critical angle for full oxide formation, the oxide thickness and the surface recession rate.
Hernández-Mangas et al.214 developed a code with only one fitting parameter for calculating the depth distribution of low implantation doses in semiconductors (Si, GaAs and SiC). The simulations were based on the binary collision approximation but additionally accounted for specific interatomic potentials and recent improvements in physical models for inelastic stopping. A periodic, ab initio, full-bond electron density function for the target was included and a modified model from the literature introduced to describe the damage accumulation. Also, statistical noise reduction algorithms were incorporated. The code yielded an excellent agreement with experimental data for different targets, even under channelling conditions.
The scaling-down of semiconductor devices requires ion implantation profiles to be simulated in topographically complex structures. Li et al.215 developed a three-dimensional Monte Carlo code for simulation of the ion implantation into multi-material systems with intricate topography. A three-dimensional trajectory-replication algorithm allowed the calculation time to be reduced by two orders of magnitude. Although the code yielded an excellent agreement with experimental data for implantation into an amorphous network, chanelling effects caused discrepancies for implantation into wafers containing silicon with (110) orientation.
The use of polyatomic primary ions instead of monoatomic projectiles is generally expected to improve the depth resolution in SIMS and to increase the yield of structural ions in S-SIMS. The latter feature is particularly important for speciation analysis with the specificity of molecular information. Furthermore, the currently existing incompatibility between molecular information and depth profiling might be no longer valid according to the study of Fuocco et al.216 on the sputter and ionization yield by SF5+ and Ar+ primary ions. The study was conducted on polymethylmethacrylate films, which were more suitable than inorganic materials for identifying beam-induced chemical damage. In comparison with the use of monoatomic Ar+ primary ions of 3–5.5 keV, the use of polyatomic SF5+ primary ions increased the sputter yield, determined using a quartz microcrystal balance and stylus profilometry, by a factor of 2–3 and secondary ion yield by a factor of 10. Additionally, X-ray photon electron spectroscopy (XPS) showed that the chemical composition of the surface stayed intact with 3 keV SF5+ primary ions whereas 3 keV Ar+ ion bombardment caused formation of a carbon rich layer. The surface modification when using 0.7 keV SF5+ primary ions proved to be more important than when 3 keV SF5+ ions were used. It was concluded that 3–5.5 keV polyatomic ions sputtered a zone equal in size or larger than the one in which chemical damage occurred. Hence, depth profiling of structural speciation ions would be feasible.
Yamazaki218 investigated the use of energy filtering for improving the LOD of Ar in silicon. The detection of Ar+ secondary ions suffered from interferences by CSi+ and C2O+ secondary ions when O2− ion bombardment was applied. The MCs+ method presented similar problems. It was found that the KE of Ar+ ions ranged from −70 to +50 eV in contrast to the 0–50 eV range for the KE of C2O+, CSi+ and CsAr+ ions. Hence, energy filtering allowed the interfering ions to be discriminated against. A LOD for Ar of 1017 atoms cm−3 was obtained at an erosion rate of 1 nm s−1. The KE data also indicated that Ar+ secondary ions were primarily formed in the selvedge at a distance of 100 µm, whereas C2O+ secondary ions were generated from the surface.
Two studies identified the projectile penetration depth and the energy density deposited in the subsurface as important factors influencing secondary ion yield. A Be target was bombarded with 4 keV Ar+, SF5+ or Xe+ primary ions at a flux of 1 µA cm−2 in an investigation into the enhancement of secondary ion yield through the use of polyatomic primary ions.219 The Be+ secondary ion intensity per ion dose increased when going from Ar+ to Xe+ to SF5+ primary ions. Polyatomic projectiles additionally produced Be3+ and Be6+ cluster ions, the intensity of which increased when the impact energy was reduced from 8 to 2 keV. Belykh et al.220 investigated the ion emission when silicon was bombarded with Aum− (m = 1–3) or Alm− (m = 1, 2) primary ions of 6–18 keV energy. The emission of large cluster ions Sin+ (n = 4–17) increased non-linearly with the number of atoms in the projectile and its mass. The generation of small cluster ions Sin+ (n < 5) was optimized by lowering the impact energy.
Konarski et al.221 investigated the use of sample rotation for improving depth profiling of 5–25 µm particles with core-shell structure. Specifically, illite particles with a TiO2 shell between one monolayer and 500 nm were studied. Simulations of the erosion were performed by means of a finite element Monte Carlo code, using TRIM for sputter yield data. Calculations showed equal recession of the side surfaces for a hemispherical object whereas the top surface exhibited a tendency to become flattened. Sample rotation limited the reduction of the original radius in all directions to 23%, whereas without sample rotation, the radius was reduced by up 47% in some directions. Improved resolution between the shell and core signals was obtained with sample rotation. The benefits of sample rotation were also demonstrated by the depth-resolved analysis of aerosols in the size range 0.3–15 µm, collected in steel works and glass plants.222 The occurrence of Cl, F and Pb in shells thinner than 50 nm on particles with iron and manganese cores was demonstrated. Depth profiling could be performed over a distance of 200 nm in aerosol particles of 0.6–1 µm.
The availability of adequate RM is essential to the development of quantitative analysis or depth profiling. Laursen et al.223 prepared standards of Si1 − xGex (0 < x < 0.35) implanted with 5.0 ± 0.1 × 1015 C atoms cm−2. The film thickness and implantation range were 300 ± 3 and 105 ± 5 nm, respectively. The Ge concentrations were verified within 0.5% by RBS. The SIMS RSF of C decreased by 50% and 20% (referenced to Ge and Si, respectively) when the Ge content was increased from Si0.91Ge0.09 to Si0.66Ge0.34. In general, the RSF had values within 10% of the mean for the typical application range of 0.5–10% Ge. Rocholl et al.224 verified the micro-analytical use of NIST glass SRM 610, 611 and 616 doped with sixty trace elements at concentrations between 20 ng g−1 and 500 µg g−1. As the materials were designed for bulk analysis, the homogeneous distribution of several elements on a microscopic scale was not assured but proved to be adequate. Specifically, the within-sample RSD for all elements except Li were 0.5 and 2.5% at lateral resolutions of 50 and 200 µm, respectively. The RSD for Li was about 3%. Ecker et al.225 developed a 400 keV Sb-implanted silicon wafer as thin layer RM. High purity silicon covered by 100 nm amorphous silicon dioxide was implanted to a surface density of 5 × 1016 ions cm−2. The largest deviation from the mean density was 1.3%. Wätjen et al.226 developed a micro-structured RM, allowing the beam spot size and scanning characteristics of microprobes to be evaluated. The RM consisted of permalloy (81% Ni, 19% Fe) strips (2–100 µm width) arranged in patterned modules of 650 × 650 µm2 on a silicon chip of 5 × 5 mm2. The height of the strips was 0.5 µm and the slopes were within about 0.2 µm. The uncertainty on the distances between patterns and widths of the broad lines was less than 1 µm. Sets of metal squares separated by gaps of different width were provided to check the beam resolution in the raster scanning mode.
Matrix effects in the determination of H in silicates due to the presence of water are hard to correct because of the lack of reliable standards. Ottolini and Hawthorne228 studied the matrix dependence of the RSF for H in a set of kornerupines (Mg- and Al-rich silicates with 38–45% Al2O3, 29–32% SiO2, 15–20% MgO, <11% FeO and 2–4% B2O3). The RSF of H referenced to Si varied by up to 30% but increased linearly with the atomic concentration of Mg (r2 = 0.73) and Al (r2 = 0.90) whilst it decreased linearly with the summed atomic concentrations of Fe and Mn (r2 = 0.80). Hence, the H2O content could be quantified with an accuracy of 2–10% by means of only 2 reference samples with different concentrations of Fe. The overall analytical precision was 3%.
Parks229 determined the relative etch rates, sputter and useful ion yields (number of detected secondary ions per sputtered atom) of 16 elements in different matrices such as silicon, silicides, metals, inorganic dielectrics and organic polymers. The data were normalized to the signals of Cl− and K+ secondary ions in the case of Cs+ and O2+ primary ions, respectively. As expected from the yield saturation hypothesis, the K+ secondary ion yield under O2+ ion bombardment was the same for all matrices and quantification could be readily performed by multiplying the RSF in a given matrix with the relative useful yield and etch rate. The yield of Cl− secondary ions under Cs+ ion bombardment approached a saturation limit for a titanium matrix but decreased for matrices composed of elements with high sputter yields such as copper. The negative ion yields for elements with low yield in different matrices showed significant variations, e.g., by up to a factor of 50 for C− ions generated from copper or titanium targets.
The use of polyatomic secondary ions for reducing matrix effects during depth profiling of matrix and dopant elements was demonstrated by Dong et al.230 in a study of Si1−xGex (x < 0.5) layers on silicon using O2+ primary ion bombardment. For x < 0.25, the intensity ratio of Ge+∶Si+ ions varied almost linearly with x and the same calibration could be used for different primary ion energies. However, for x > 0.25, the SiGe+∶Ge+ ion intensity ratio had to be used with a primary ion energy of 1 keV to extend the linear dependence on x up to x < 0.5. The stoichiometry obtained agreed within 10% with reference values from double-crystal X-ray diffraction.
The use of M+ or MCs+ secondary ion detection for quantitative analysis is still a matter of debate as the latter method features fewer matrix effects but poorer LOD. Yamazaki231 compared detection of Ar+ and ArCs+ secondary ions in quantitative depth profiling of Ar in tungsten films using O2+ and Cs+ primary ions, respectively. The interference from C2O+ on Ar+ secondary ions compromized the LOD using O2+ ion bombardment. In contrast, the yield of ArCs+ secondary ions was better than that of Ar+ ions and only negligible interference from C2OCs+ ions occurred. As a result, the LOD of 4 × 1018 atoms cm−3 obtained when using Ar+ secondary ions could be improved to 3 × 1016 atoms cm−3 when using ArCs+ ions at a primary ion current density of 0.16 mA cm−2. The depth resolution also improved because Cs+ primary ions developed less topographical variation than O2+ projectiles. Gard et al.232 combined M+ and MCs+ secondary ion detection for depth profiling across ZnSe/GaAs interfaces. Mass interferences encouraged the use of As+ and CsGa+ instead of AsCs+ and Ga+ secondary ions, respectively. The useful ion yields of Se and Zn in GaAs and of As and Ga in ZnSe were determined and the LOD of As and Ga was 2 × 1019 and 1 × 1017 and atoms cm−3, respectively.
Jiang et al.233 developed the combination of depth profiling data at incidence angles of both 45° and 60° to minimize the variations of RSF as well as sputtering rates across the silicon dioxide–silicon interface at a depth of 100 nm. Bombardment with 1 keV O2+ primary ions at an incidence angle of 45° yielded identical sputtering rates for silicon dioxide and silicon. Use of AFM revealed that the crater bottom surface remained smooth in silicon dioxide and at the interface with silicon but significantly roughened afterwards. The roughness could be minimized through the use of an incidence angle of 60°. Therefore, each sample was analysed using two incidence angles. An angle of 45° was used to obtain information in the range 0–15 nm and an angle of 60° for the deeper regions. The depth scale was established by applying the sputtering rates for silicon dioxide and silicon before and after the interface, the position of which was derived from a 50% reduction of the 30Si+ signal. The error caused by the intermediate oxidation states was estimated to be 20%.
Quantitative depth profiling of sub-2 nm nitrided SiO2 films was investigated by Novak et al.234 The ion mixing depth was estimated from the intercept in the plot of the film thickness as a function of the sputtering time needed for a 50% decay of the oxygen signal. The incidence angle of between 60 and 80° affected the mixing depth more than the primary ion energy of between 0.3 and 1 keV. A mixing depth of only 0.3 nm was achieved with 0.3 keV Cs+ primary ions at an incidence angle of 75° but sputter yield was poor. A 0.5 keV ion beam at the same angle provided a more practical compromise. Determinations of N surface densities in films of 1.7–2.4 nm by SIMS and XPS agreed within 5–10%. Because the ion mixing depth was significantly less than the film thickness, SIMS could characterize both the distribution and amount of N within the coating.
The 238Pu∶234U, 239Pu∶235U and 240U∶236U ratios in individual micro-particles on polished carbon were determined by Wallenius et al.235 by rastering 15 keV O2+ primary ions across a surface area of 250 × 250 µm2 and applying energy filtering of the secondary ions. Particles were first localized by stigmatic imaging and subsequently analyzed with a focused beam. Using TIMS as a reference method, the RSF of Pu referenced to U was found to be 2.41 and 2.34 in metallic oxide samples and in the plutonium sulfate SRM 946/947, respectively.
Layne and Sims236 found a mass resolution of 2000 to be adequate for the measurement of 232Th∶230Th in volcanic rocks using a medium-resolution high-transmission ion microprobe. An O− primary ion beam with diameter of 50–75 µm was rastered over an area of 250 µm2. Instrumental fractionation was <0.2% per u and the RSD on the 232Th∶230Th data was 0.5%–1.0%. The ionization efficiency of 2.4% allowed analysis of just 10 ng of sample to be achieved which compared well with the 300 ng of sample needed for TIMS analysis. The abundance sensitivity was better than 4 ppb. The measurement of 232Th∶230Th in a test sample by SIMS and TIMS gave (1.710 ± 0.012) × 105 and (1.706 ± 0.014) × 105, respectively.
The precision and accuracy of isotope ratio determinations in TOF SIMS can be hampered by the typically low erosion rate. Fahey and Messenger237 analysed sub-μm particles of SiC and MgAl2O4 using a focused Ga+ LMIG. Software developed in-house carried out automatically (1) dead time correction, based on reference isotope ratio measurements, (2) peak location and identification, (3) correction of interferences by peak fitting and subtraction and (4) peak integration. The total uncertainty on the isotope ratios was typically twice that due to the counting statistics so the precision of TOF SIMS approached that typical of magnetic instruments.
The use of the maximum and of the centroid of the depth profile was compared by Wittmaack et al.239 for localization of δ-layers in silicon under bombardment with 0.25–1 keV O2+ primary ions. It was shown that markers grown by molecular beam epitaxy often contained doping artefacts which caused the centroid to deviate significantly from the expected value. Therefore, the peak positions were derived by third order polynomial fitting of the profile in the 20–100% intensity range. The peak positions showed a more pronounced, non-linear energy-dependent shift than the centroids. The dramatic effect of shape distortions on the centroids could lead to experimental deviations from results obtained with the peak maximum method of up to 0.8 nm. Therefore, the latter procedure was highly recommended for determination of apparent marker locations.
Oxygen flooding in ultrashallow depth profiling was found by Ng et al.240 to be essential when using low energy primary ions. Both sample rotation and oxygen flooding were studied using O2+ primary ions of 0.5 keV at an incidence angle of 56°. Samples were comprised of 16 B δ-layers in silicon with interlayer distances of between 5 nm and 15 nm. When no sample rotation or O2 flooding was used, δ-profiles broadened and became asymmetrical. Use of sample rotation narrowed the profile but oxygen flooding was additionally needed to optimize both width and symmetry. Crater bottom characterization by AFM at a depth of 90 nm yielded root-mean-square (rms) roughness parameters of 1 and 0.2 nm without and with sample rotation, respectively. According to simulations with a modified MRI model, roughness should be similar when using either rotation or flooding but the mixing parameter would improve from 2 nm when using rotation to 1.2 nm when using oxygen flooding without rotation. Another paper from the same group considered the effect of oxygen flooding on the composition and roughness of the crater bottom.241 Targets of silicon implanted with 2 keV B were bombarded with 1 keV O2+ primary ions at an incidence angle of 56°. The oxidation state and the thickness of the oxide layer at the crater bottom was determined by XPS. Specifically, silicon and silicon dioxide were found to be the major components and only small contributions of silicon suboxides occurred. The calculated equilibrium oxide thickness increased with the oxygen partial pressure whereas the sputter-induced roughening decreased. The onset of surface roughening occurred when the crater bottom composition became inhomogeneous.
Group Ia elements are known to migrate under the influence of an electrical field. An incidence angle of at least 27° was found to be necessary in a study by Deenapanray and Petravic242 of the angular dependence of the profile broadening of F-, Li- and Na-implanted silicon bombarded with 10 keV N2+ and O2+ primary ions. Using O2+ ions, the decay length of Li and Na at the silicon dioxide/silicon interface decreased exponentially with the incidence angle. If N2+ ion bombardment was used, the decay length was minimum for incidence angles >15°. Field-induced segregation of Li at the interface caused more broadening in the low resistivity n-type silicon than in the high resistivity p-type silicon when bombardment with N2+ ions at an incidence angle of 35° was used. Under these conditions a stoichiometric nitride layer was formed. The broadening of the F profile in O2+ SIMS became maximum at an incidence angle of 33°. Bombardment with O2+ ions at an incidence angle below the critical angle for oxide formation resulted in sharper profiles due to the uniform distribution of F in the silicon dioxide layer. The decay length of F decreased exponentially with the surface charging. In contrast, use of N2+ primary ions resulted in a gradual decrease of the decay length as a function of the incidence angle with a plateau between 20 and 30°. This can be explained by the reduced mobility of F in the nitride layers resulting in segregation, while collisional mixing increased at low incidence angles.
Alkemade and Jiang243 studied the surface roughening of silicon as a function of the O2+ beam energy (0.5–2 keV) and incidence angle (45°–80°). Samples were Si(100) wafers grown by atmospheric pressure chemical vapour deposition with removal of the native oxide layer before growth. The thickness of the deposited layers was 200 nm. The presence of buried Ge δ-layers at 15 or 11 nm depth, used to estimate the erosion depth, had no influence on the surface roughening. Under all conditions, topographical features developed in the form of regular ripples, irregular bumps and/or triangular depressions or elevations with or without superimposed ripples. The surface roughening started earlier with lower energy primary ions and the ripple wavelength varied with the impact energy E according to E0.63. The period of the ripple systematically increased until saturation occurred. A small change in incidence angle of 0.7–1 keV primary ion beams had a large effect on the surface roughening. At incidence angles between 65–75°, ripple formation occurred with the shape of the ripples being dependent on the incidence angle. At incidence angles above 80°, the surface remained smooth, most likely because surface protrusions were shaved off.
Lau et al.244 investigated the development of topographical features in Si(001) and Si(111) surfaces under 1 keV O2+ ion bombardment as a function of the sputtered depth and the incidence angle between 44 and 56°. The highest incidence angle caused roughening to start at depths of 1 and 1.5 µm in Si(001) and Si(111), respectively. The transition roughening depth varied “exponentially” with the incidence angle. The rms parameter steeply increased from 200 nm onwards with an incidence angle of 44°. The roughening rate was higher for Si(001) than for Si(111), in particular at high incidence angle. The formation of (111) facets on Si(001) and of (111) facets on Si(001) caused development of prominent elongated triangular and rectangular features, respectively.
The need for vigilence in the storage of samples was highlighted by the study of Alkemade and Liu245 in which anomalously long transient regimes of up to 40 nm were found in the depth profiling of B-doped silicon. In-diffusion of B was excluded as an explanation because correct depths and the expected variation of the apparent depth with the incidence angle were obtained. Sample rotation eliminated the possibility of local segregation of B. The transient duration could be reduced to the normal level of about 6 nm by removal of the native oxide by HF. Hence, contamination by an organic compound from the storage container was pinpointed as the cause for the long transient.
Nakamura et al.246 exploited isotopic labelling for easy and accurate depth profiling. Specifically, metallorganic chemical vapour deposition in an 18O2-enriched atmosphere was used to deposit strontium and titanium oxide films. Dual beam depth profiling in a TOF SIMS instrument yielded an agreement within 10% between the signal intensity ratios M18O+∶M16O+ (M = Sr, Ti) and 18O−∶16O−. The latter signals gave better LOD than the oxide cations. Isotope studies facilitated the quantitative depth profiling because the beam-induced artefacts could be neglected.
A wide range of applications in biological studies has been reported. Chandra and Lorey249 studied the calcium storage and intracellular element distribution in cancerous and normal cells. Sample preparation involved freeze-fracturing and freeze-drying. Coating the specimen with gold reduced charge build-up. Isotopic images were obtained with a lateral resolution of 0.5 µm and digitized directly from the microchannel plate using a slow charge-coupled-device camera. Quantification was performed with RSF using 12C as internal standard. Chehade et al.250 investigated the uptake mechanism of iodobenzamides using stigmatic imaging of the 127I in skin sections under 10 keV O2+ primary ion bombardment. The lateral resolution was 1 µm. Images of CN− secondary ions were used to recognise the general morphology of the sample and the PO2− ions served to localize the cell nuclei. Hindié et al.251 imaged 129I in thyroid samples of rats to elucidate the increased occurrence of thyroid cancer upon exposure to nuclear radiation. A mass resolution of 3000 avoided interferences by polyatomic ions. Accumulation of the 129I in subcellular structures could be observed in 150 × 150 µm2 frames. Florent et al.252 used EPMA and SIMS to image several group IIIa-elements, lanthanides and actinides in rat enterocytes after oral administration of soluble salts. Atomic ions were imaged over an area of 250 × 250 µm2 with a resolution of 0.5 µm using O2+ primary ions. The LOD were 10−19–10−20 g at an erosion rate of 0.3 nm s−1. The time required for imaging elements with low ionization potentials was 1–900 s depending on the concentration.
Oba et al.253 evaluated the use of SIMS with LMIG and cold sample stage as an alternative to SEM EDXRF to study the element distribution in goblet cells of rat conjunctiva. Attention was paid to preserving the element distribution during sample preparation. The tissue was quenched in liquid propane and nitrogen. The semi-thin sections were freeze-dried, carbon-coated and analysed using the cold stage (−40 °C) in the SIMS instrument. Using the O2+ primary ion beam, images were made over an area of 250 × 250 µm2 with a beam spot diameter of 20 µm. The Ga+ LMIG allowed an area of 100 × 100 µm2 to be rastered with a spot diameter of 55 nm. Analysis time was 180 s. The resolution of the negative ion image was better than 1 µm (total size of the feature to be observed). The sensitivity of SIMS was superior to that of EDXRF. Comparison of the morphological features in images from SIMS and SEM EDXRF revealed that the small raster size in SIMS resulted in better images than obtained with SEM EDXRF.
Detection of both atomic and structural information adds value to imaging applications of SIMS in material sciences. In a study of the effects of magnetic disc corrosion, Choa254 demonstrated a direct link between the hydrogenated carbon overcoat and cobalt migration. Micro-scale recovery of the deposits and IC were used to verify the TOF SIMS results. There was a good agreement between the two methods for the atomic ion concentrations.
Hirokawa et al.256 investigated the use of the valence of positive and negative ions and electronegativity to explain the ion formation and yield from 30 inorganic compounds. It was claimed that the mass spectra could be predicted qualitatively by using the developed rules. A more quantitative prediction of ion intensities would have required detailed knowledge of the effects on the mass spectra of the primary ion and its energy, energy transfer of atoms and ions, their sputtering and ionization yields, the transmission probability and the detector efficiency. Hence, use of RSF remained mandatory.
The combination of detailed speciation and imaging was demonstrated by O'Dea et al.257 in their study on the interaction between xanthate dithiocarbonates and freshly cleaved or oxidized galena at neutral or alkaline pH. Atomic ions such as O− and S−, as well as structural ions such as the OCS2−, CH2OCS2− and C2H5OCS2−, were imaged from (sub)monolayers on galena. Atomic and structural ion imaging had lateral resolutions of 5 and 10 µm, respectively, and minimum detectable features in the images of about 10 and 20 µm, respectively. The relative intensity of the ion images correlated well with the independently determined amount adsorbed.
Godfrey et al.258 studied the F distribution in non-porous, mesoporous or microporous polystyrene beads to follow plasma-chemical fluorination. Additionally, XPS and FT-IR spectroscopy were used to calibrate the composition with an information depth of 2 nm and 1–5 µm, respectively. Analysis of cross sections by S-SIMS allowed surface analysis with a fine beam spot to be performed. In this way, the local concentration of the analyte at a large distance from the original surface could be characterized without the beam-induced artefacts found in conventional depth profiling. Analysis of microporous particles showed that F was present only within 5 µm of the surface whereas mesoporous particles were fluorinated throughout the bulk due to their larger pore and particle sizes.
The importance of obtaining full, organic and inorganic, structural information was demonstrated by Takatsuji259 in the investigation into low level impurity contamination of thin film transistor liquid crystal displays. Mapping was performed over an area of 12 × 12 mm2 with a mass resolution of 4000 (FWHM) at m/z 29. Recording the complete mass spectra allowed detailed data processing to be made after analysis. Specifically, the atomic ion images could be complemented with the corresponding images for polyatomic speciation ions or organic fragments. The tremendous amount of information accessed would be hard to obtain with a sequential mass spectrometer. Specifically, the presence of In+, Sn+ and In2OH+ ions in the mass spectrum revealed contamination by indium tin oxide splash during manufacture of the pixel electrodes. The presence or absence of polyimide could be linked to the non-uniform alignment of the liquid crystal molecules.
Peled et al.260 demonstrated imaging with sub μm resolution for the different species present in electrodes of Li-ion batteries. Specifically, atomic ions such as F, Li and O were imaged together with C2H−, PO3− and structural ions like C2H3O2− and C2H3O−. In this way, the elemental composition could be linked to the presence of hydrocarbons or polymers and specific reaction products.
Hodoroaba et al.262 used high-frequency-mode plasma SNMS for the depth profiling of insulating multilayers, specifically, sandwiches of SiO2/TiO2 and Si3N4/SiO2 on glass. A depth resolution of 5 nm was achieved for the interface at a depth of 100 nm in Ni/Cr layer stack samples. A low rf voltage was needed to obtain flat crater bottoms for a multilayer of 133 nm SiO2 and 76 nm TiO2 on BK7 glass. The depth resolution ranged from 16 to 70 nm, depending on the interface depth. Each type of sample required a given rf voltage. A 100 nm SiO2/100 nm TiO2 layer stack on glass was used to compare SNMS with TOF SIMS imaging of cross sectioned samples. The latter method was more sensitive but the sample preparation was much more difficult than that for SNMS.
Chevalier et al.263 exploited oxygen isotope labelling to investigate the effect of Nd2O3, Pr2O3, Sm2O3, Y2O3 and Yb2O3 coatings on the oxidation of Fe–Cr alloys. Two-stage exposure to air containing 16O2 and 18O2 was applied and the 18O tracer distribution determined by SNMS and SIMS. The latter method provided sensitive information by means of the 18O+∶16O+ signal intensity ratio while SNMS was preferred for depth profiling because of the absence of matrix effects. The chromium oxide growth mechanism in uncoated samples proved to be controlled by chromium cation diffusion, whereas external diffusion of chromium ions was inhibited by the reactive element coating. This type of refined information was hard to get other than through the use of isotope labelling and isotope-specific methods.
Although the coupling of SIRMS to HPLC has been used previously, there have been few reports of its application, which should allow isotopic analysis of non-volatile analytes to be achieved. Abramson et al.267 built a system in which various HPLC columns were connected to a SIRMS instrument through a microwave-powered chemical reaction interface (CRI), and applied it to the analysis of a range of involatile compounds. The reactant gas in the CRI was O2 in order to produce CO2 for analysis. Reversed-phase LC was used for the analysis of intact proteins, normal-phase LC for highly polar species such as sugars, ion-exchange chromatography for the hydrolysis products of nucleotides and SEC for a synthetic conjugate vaccine and its component parts. It was considered that the availability of precise isotopic measurements for macromolecular and non-volatile species should simplify greatly isotopic tracer experiments by lowering the amount of tracer needed, removing the requirement for derivatization and hydrolysis of biopolymers.
An on-line system for the preconcentration of atmospheric methane, described by Rice et al.,268 consisted of several stages designed primarily to remove contaminants. After removal of water by an initial liquid-nitrogen trap, methane was adsorbed quantitatively on a Hayesep D precolumn to purge more than 99.99% of the N2 and O2 entrained in the sample stream. Limiting the amount of N2 and O2 in the system dramatically lowered the SIRMS background signal and improved the overall precision of measurement. Methane was separated from residual gas components (for example, Ar, CO, CO2, N2, O2 and hydrocarbons) by GC on a PoraPLOT Q column. Whereas isothermal analysis was possible for hydrogen analysis, temperature programming was required to avoid interference from an unidentified contaminant peak in carbon analysis. The methane eluted from the separation column was either oxidized to CO2 and H2O, using a mixture of CuO, NiO and Pt catalysts, for carbon isotope analysis or pyrolysed to H2 and O2 for hydrogen isotope analysis. The precision of measurement (1σ) for δ13C and δ2H in methane was 0.05 and 1.5‰, respectively, using 120 ml of ambient air. In comparison to off-line combustion and conventional dual-inlet analysis, the analysis time was shortened from 1–6 h to 30–40 min and sample volume reduced from 80–1500 l to 120 ml without a degradation in precision and accuracy. It was demonstrated that the precision was adequate for determining spatial and temporal trends in atmospheric methane.
Gaschnitz et al.269 developed an on-line pyrolysis GC-combustion-SIRMS system in a study of the natural kinetic isotope fractionation that controls the isotopic signature of gases thermally generated from coal. Samples of carrier gas and pyrolysis products were injected into the GC unit by means of a 2 ml sampling loop. A feature of the procedure was the injection of a gas sample immediately after elution of the previous sample from the GC column rather than wait for the previous sample to clear the complete instrument. This reduced the sampling interval to 12 min but required stable retention times and accurate switching of columns. Reproducibility (better than 0.5‰) and sensitivity (4.65 ng carbon equivalent per compound) were good. Results indicated a substantial range in isotopic composition for each of the seven pyrolysis products.
The ratios of stable isotopes of C, Cl, H and, possibly, Br in atmospheric halomethanes could provide valuable information on global sources and sinks. Any stable isotope fractionation associated with production or degradation processes influences the isotopic composition of atmospheric CH3Cl and CH3Br. Kalin et al.271 have developed a CSIA technique for measurement of δ13C values in halomethanes from both natural and anthropogenic sources. Optimization of GC parameters allowed CH3Br, CH3Cl, CH3I, CH3SH, CO2 and O2 from a single sample to be identified and quantified. Although the temperature programme fully resolved CO2 from N2, thereby avoiding N2O interference in isotope measurement, it did not separate CO2 from O2. The concentrations of CO2 and O2 were obtained by integration of the peaks at m/z 44 and 32, respectively, and normalisation to the N2 peak at m/z 28. The precision of δ13C measurements decreased with increasing mass (±0.5, ±0.3 and ±1.3 for CH3Br, CH3Cl and CH3I, respectively). Experimental results indicated strong isotope discrimination in halomethanes with variations in δ13C in atmospheric methane of up to 20‰.
An understanding of palaeoclimates is critical for interpreting the history of the earth and the biotic response to global change. The measurement of δ18O in fossilized rodent teeth is attractive as a potential proxy in palaeoclimate reconstruction in that the oxygen is derived directly from body water. Lindars et al.272 used a direct laser fluorination method in which samples of as little as 1–2 mg were heated in the presence of excess BrF5 using a CO2 laser (power, 25 W; wavelength, 10.66 µm) to release 100% of the phosphate oxygen and leave a pure CaF2 residue. A key feature of the procedure was that each run of samples consisted of only one biogenic apatite and three standard samples. The reason for this was that any additional apatites present in the source reacted prematurely when laser heating the first. The first of two in-house quartz standards was sacrificed to allow the sample chamber to be heated. The second was used for calibration. The third standard sample was NIST SRM 120c (phosphate-bearing sediment). To measure only the phosphate oxygen component of biogenic apatite, a pretreatment was required to remove non-phosphate oxygen (carbonate, hydroxyl, water and organic matter). Although heating at 1000 °C removed all the non-phosphate oxygen, some samples were susceptible to oxygen exchange with atmospheric water. Samples were instead heated to 400 °C to remove organic matter and fused by laser within the sample chamber under high vacuum to remove remaining water and carbonate. Results from the analysis of teeth from modern rodents showed that single teeth could not be used because of large variations within both a single specimen and a population. Hence, a mean of the analysis of at least five post-weaning teeth was used for palaeoclimate reconstruction.
A major thrust in SIRMS analysis in recent years has been the development of continuous flow (CF) methods as more rapid and convenient alternatives to the traditional dual-inlet techniques. A technique for the CF determination of both δD and δ18O in water and hydrous minerals involved reaction with glassy carbon at 1450 °C in a He carrier gas.273 Water samples (as little as 0.1 µl) were injected into the He stream whereas solid samples were wrapped in silver foil and introduced into the furnace using an autosampler. Attempts to load water samples in silver capsules were unsuccessful due to excessive evaporation prior to entrainment in the He stream. The authors presented detailed discussions on the introduction of reference gases and the calculation of the H3+ factor. Precisions (1σ) of ±2 and ±0.2‰ for δD and δ18O, respectively, were as good as those achieved by conventional analysis for water samples. In contrast to the previous study, Kelly et al.274 appear to have had no problems in enclosing water samples (0.7 µl) in tin capsules and introducing them into a furnace from an autosampler. Water and organic samples were reduced over chromium metal at 1000 or 1260 °C, respectively, in a stream of He. The memory effect in the CF-SIRMS system was negligible when analysing water at natural abundance. The precision (1σ) of the measurement of δD in water samples was ±2.3‰ and accuracy, from the analysis of RM, 2.2‰. The use of Cr reduction had the benefit over the glassy carbon pyrolysis reactor of being able to measure directly δ13C in calcium formate, which retains site-specific isotope information from organic precursors and could be used for PSIA.
A potential limitation of both CF methods was that they required instrumentation designed specifically to overcome the presence of a 4He tail at m/z 3. In addition, a correction for the presence of 1H3+ ions at m/z 3 must be applied, using a so-called H3 factor. In a study of two proposed but fundamentally different approaches, “pointwise” and “peakwise”, for this correction, Sessions et al.275 found the latter to be valid only for peaks with the same shape and only when background signals were constant. Such conditions are seldom encountered in SIRMS and the potential for significant error is great. The “pointwise” correction was more versatile and could be used when peak shapes and sizes and background signals varied significantly. Limitations remained, however, and accuracy was restricted by signal-broadening effects of electronic time constants, the analog-to-digital conversion frequency and the highest frequency of the sample signal. Analog-to-digital conversion rates of at least 4 Hz and electronic time constants of <500 ms were recommended to minimize errors under typical GC conditions. Using the pointwise algorithm, a series of 14 homologous n-alkanes with concentrations over a five-fold range could be analysed with a mean precision of 2.3‰ and no systematic error. The correction depended crucially on the accuracy and stability of the measured H3 factor. In a critical evaluation of four methods for determining the H3 factor, the same authors found that static measurement over a range of H2 signals, the approach commonly used in dual-inlet analysis, was highly precise but could lead to systematic errors in δD when applied to CF systems.276 In comparison, peak-based measurements consistently produced more accurate δD values but less precise correction factors. The value of the H3 factor remained stable throughout GC analyses despite small changes in carrier gas composition. It was concluded that a single value of H3 factor could be used to correct accurately a wide range of peak sizes across a chromatogram.
A CF method for the rapid determination of δ34S in sulfide and sulfate minerals had significant advantages over the traditional extraction method in terms of the reduced sample quantity (1 mg) and rapid analysis time (450 s).277 In addition, complete separation of sulfides from sulfur-free minerals was not necessary as the SO2 produced upon combustion was separated from other gases by GC. The method was, however, sensitive to a number of operational parameters, including sample weight and the O2 saturation of the Cu reduction reactor. An optimum sample size (ranging from 1 mg for pyrite to 3.5 mg for galena) was adopted in order to avoid saturation of the ion peak and to minimize variations in the quality of the SO2 gas produced. A number of RMs were used to establish the calibration, which was observed to drift slightly and had to be monitored through analysis of an in-house RM. The drift was attributed to the increased presence of exchangeable oxygen in the oxidized CuO and could be minimized by flushing the system with He between sessions. Measured S contents were within 1–1.5% of expected values and the reproducibility (1σ) of δ34S values was ±0.1‰.
Lipid biomarkers are biochemicals derived from a restricted range of organisms and thus provide a highly selective means of isolating material from a specific source. Most of the hydrogen in lipids is bound to C and is non-exchangeable. Sauer et al.281 investigated the use of certain algal sterols in freshwater systems as aquatic biomarkers for the reconstruction of lakewater δD. Sterols were solvent-extracted from sediment samples and derivatized for analysis by GC-combustion-SIRMS. Of a number of derivatization methods investigated, acetylation was considered the most attractive. It avoided the introduction of fluorine and introduced only three H atoms. During acetylation, no bonds to H retained in the product were made or broken so there was no opportunity for large isotopic fractionation. A secondary isotope effect was observed, however, consistent with acetate H being depleted by 60‰ relative to H in acetic anhydride. No correction was attempted for this effect which resulted in an inaccuracy of 4.4 ± 4.2‰. It was concluded that the hydrogen isotope analysis of algal sterols provided a less ambiguous reconstruction of δD in environmental waters than methods based on analyses of kerogen or other operationally defined organic matter fractions. Lakewater δD could be reconstructed to within 10‰.
Li et al.282 measured δD values in n-alkanes and acyclic isoprenoid alkanes in crude oils to evaluate their use as parameters for oil-source correlation and for paleoenvironmental reconstructions. Alkanes fractionated from crude oil samples were analysed by GC-combustion-SIRMS using co-injected 5β-androstane as internal standard. The value of δD in individual alkanes varied by up to 130‰. This was high compared to analytical precision of 3‰ so the technique gave much better resolution than carbon isotope analysis, for which δ13C typically falls into a range of only 2–6‰.
Wanek et al.283 developed a simple and reliable preparation procedure for the CF-SIRMS measurement of δ13C in lipids, soluble sugars, starch and cellulose from 100 mg of dried plant material. The procedure was based on methanol–chloroform–water extraction of dried and ground leaf material (100 mg), recovery of the chloroform (lipids) and methanol–water phases (soluble carbohydrates) and enzymatic hydrolysis (starch) and solvolysis of the residue (cellulose). Of three starch preparation methods evaluated, only the enzymatic hydrolysis by α-amylase was free of isotopic fractionation. It also gave better recovery, accuracy and precision.
The objective of the work described by Ogawa et al.284 was to develop a simple and rapid pre-treatment method for the determination of δ15N of nitrate in groundwater. A powdered super-absorbent with hygroscopic properties was used to absorb and retain the nitrate. Samples, which had been checked for the absence of ammonium nitrogen and for Cl− concentrations below 100 mg l−1, were evaporated (40 °C) to produce a nitrate-nitrogen concentration of >100 mg l−1. The concentrated solution (5 ml) was absorbed on the polymer resin (50 mg), dried (80 °C, 2 h) and introduced in a tin cup into an elemental analyser. Isotope fractionation was not observed during either sample evaporation or resin drying. Although this CF method had the advantage of fast analysis in comparison with conventional methods, it required >60 µg N and was only suited to the analysis of contaminated waters.
As in any GC procedure, derivatization of organic compounds plays a key role in successful SIRMS analysis so there is a continuous search for improved methods of derivatization. Zaideh et al.285 developed a procedure for non-polar amino acids (2 g) based on NaBH4–I reduction to amino alcohols in order to avoid the introduction of derivative C, which occurs in most commonly used derivatization procedures. A 5 µm thick stationary phase was used for GC separation as it gave much sharper peaks than the conventional 0.25 µm thickness. Although the average precision of 0.19‰ for analysis of purified standards confirmed the feasibility of the approach, analyses of real samples remained to be demonstrated and further work was required to overcome the involatility of some of the amino alcohols. The use of trimethylsilyl derivatives, prepared under optimized conditions, for the GC-combustion-SIRMS of sterols was found by Prevost et al.286 to provide well-resolved chromatographic peaks, no peak tailing, reproducible (0.32‰) δ13C measurements, good S/N and slow degradation of CuO in the combustion furnace. The alternative use of acetylated steroids was, on the other hand, found to give incomplete derivatization and poor chromatography. The method has been applied as a new approach to controlling the misuse of androgens in breeding animals. The δ13C and δ15N enrichments in glutamine in rat plasma samples were measured by GC-combustion-SIRMS following derivatization as the N(O,S)-ethoxycarbonyl ether esters.287 The derivative was stable on storage and gave accurate (3.2‰) analysis.
The dead-time of a pulse-counting system has been measured by Rameback et al.290 using both a ratio measurement approach and an electronic method. A large and as yet unresolved discrepancy between the dead-times measured by the two approaches (38.5 and 43.7 ns, respectively) revealed a serious problem within the detection system. It was considered that dead-times should be determined using pulse timing measurements in order to gain a better understanding of the detection system.
Kawai et al.291 found a temperature dependent change occurred also in the ionization parameters of alkaline iodides on double filament assemblies. At low evaporation filament temperatures, the sample vapour arrived at the ionization filament in the chemical form in which it was loaded. Therefore, the sample vapour gained the sum of the dissociation and ionization energies at the ionization filament. At higher temperatures, however, the vapour could dissociate before reaching the ionization filament, thereby changing the apparent ionization parameter. Three different processes were involved in ionization (MI2 → M+, MI → M+, M → M+) with the dominant process shifting from the first to the third at higher temperatures.
The Re–Os isotopic system is currently limited as a chronometer because of the lack of accurate gravimetric standards for Os spike calibration and uncertainty of the 187Re half-life, which itself is also dependent on the accuracy of spike calibration. To calibrate an Os spike precisely and accurately, an Os compound with an accurately known composition is required. Although (NH4)2OsCl6 has been widely adopted as primary gravimetric standard, Yin et al.293 recommended that it should be abandoned because it was hygroscopic and decomposed during heating. Use of high-purity, stoichiometric and anhydrous K2OsCl6 as an alternative standard improved the current level of accuracy from 1–2% to ±0.2%.
An alternative application of uranium analysis is the determination of oxygen isotopic composition, as reported by Pajo and colleagues in two papers.296,297 This application, for which TIMS has rarely if ever been used, was seen as a possible tool for identifying the origin of seized samples of uranium oxides in nuclear forensic science. Whereas uranium oxides of different geographical origin have significantly different 18O∶16O ratios, different samples of the same origin have constant ratios. Experimental conditions for TIMS analysis of solid uranium oxide were investigated and optimized. Benzene suspensions of uranium oxide (few µg) were loaded on to double rhenium filaments in small amounts and the benzene evaporated completely between each addition. In order to avoid contamination of the specimens with ambient oxygen, the ion source was flushed twice with dry N2 before measurements were taken. The optimum ionization yield for MO+ occurred at an ionization filament current of 4.2A, corresponding to a temperature of 2300 K. The 238U16O+ and 238U18O+ ions were measured simultaneously on two Faraday detectors under total evaporation conditions in which the whole sample was evaporated from the filament and measured, thereby minimizing the effect of isotopic fractionation. In a comparison with analysis by GDMS and SIMS, TIMS analysis was considered the only technique that provided data relatively quickly and sufficiently precisely (0.04%) for the application.
Improvements to B isotope analysis continue to be sought with particular emphasis on sample preparation and filament loading for positive ion analysis. Deyhle298 gave a very good summary of the historical development of B isotope analysis and of the problems associated with each approach. The author analysed NIST boric acid SRM 951, using static multicollection of CsBO2+ ions, with the objectives of improving precision and accuracy and of establishing the critical amount of B required. Some of the key improvements to analysis were made to filament loading and heating procedures. Prior to sample loading, tantalum filaments were degassed and allowed to oxidize for three days. Whereas in previous studies samples were loaded on to filaments already coated with graphite, in this study the loading order was reversed. Loading the sample followed by the graphite resulted in a more stable beam (1 × 10−11 A at m/z 309) with less fractionation. Loaded filaments were kept warm to retain complete dryness until placed in the mass spectrometer. The reproducibility (2σmean) of measured 11B∶10B in SRM 951 (100 ng B) was ±0.006%, equivalent to an external uncertainty (2σ) of 0.024%. That small amounts of natural samples with low B contents could be measured reliably was demonstrated by the external reproducibility for natural samples of 0.011–0.016%. Although the method was suitable for analysis of as little as 50 ng B, 100 ng B was recommended as the optimum amount for static, positive ionization analysis. One of the problems associated with B analysis is the risk of B loss by volatilization during sample preparation. Whereas Deyhle followed the conventional procedure of adding mannitol to reduce B volatilization, Wang et al.299 developed a new procedure to avoid the need to add mannitol. Separation and purification of B was based on chromatography first on a boron-specific resin, from which B was eluted in HCl, and then on a mixed ion-exchange resin. The latter adsorbed all anions and cations except B, thereby removing HCl and ions eluted from the first column. In this way losses due to removal of HCl by evaporation could be avoided. Samples of 1–2 µg were loaded on to single tantalum filaments (already coated with graphite) and the 11B∶10B ratio in CsBO2+ ions measured using a peak jumping procedure. The value obtained for 11B∶10B in NIST SRM 951 (11–40 µg B) was 4.052 ± 0.002 (1σ) and compared well with the value of 4.0522 ± 0.0009 (2σ) obtained by Deyhle for 1 µg B.
Modified loading conditions and the use of MC analysis allowed Numata et al.300 to achieve better precision than previously reported for chlorine isotope analysis. The recovery of Cl from both organic and inorganic matrices was based on lengthy chemical separation in which Cl was recovered as AgCl, converted to CsCl and purified using ion-exchange chromatography. In order to get small, homogeneous spots on the tantalum filaments, samples (2 µg Cl in aqueous solution) were mixed with graphite (50 µg) prior to loading. Use of wide filaments (1.5 mm) gave slightly higher ion currents and easier control of the filament current than when standard filaments (0.76 mm wide) were used. A static, multi-collector technique was used to measure the Cs2Cl− ions at m/z 301 and 303. Replicate analysis of RM gave an external precision (1 σ) of 0.1–0.2‰ for 2 µg Cl. This was considered to be good enough to identify differences in the Cl isotopic composition of chlorine compounds but still requiring improved sample preparation and loading procedures. Whereas the range of δ37Cl in inorganic samples (−2.5 to +0.9‰) was similar to that of most Cl in surface and subsurface water, that in organic compounds (−5.0 to +2.9‰) was almost twice as large.
A TIMS procedure for Zn isotope analysis was found suitable for the determination of amounts of Zn produced by neutron transmutation doping in highly pure samples of copper.301 Use of a MC configuration with nine Faraday cups enabled simultaneous measurement of five Zn ion currents. Loading 150 ng samples of Zn on a single Re filament using the silica gel technique gave an ion current of 1 × 10−11 A for 64Zn. The RSD of measured ratios were 0.05%.
Although Cd mass fractionation has been recorded for meteorites, none has until now been observed in lunar samples. Positive, mass-dependent fractionation has now been determined experimentally for Cd in three lunar soil samples in which the light isotopes were depleted with respect to the heavier isotopes.302 An extraction efficiency of 96% was achieved for separation of Cd from the lunar soil, thereby making separation-induced fractionation unlikely. Analytical blanks (10–90 pg) were at most 0.1% of the total Cd in the analysed samples. Cadmium extracted from the soils was loaded in silica gel and phosphoric acid on to single Re filaments to give stable ion currents of 1 × 10−11 to 2 × 10−13 A, which lasted 1–2 h. Potential isobaric interferences from In, Pd and Sn were shown to be negligible with the exception of 116Sn, which had to be taken into account when measuring 116Cd.
Problems associated with purification of stripping agents are a serious drawback in the use of the extraction material Pb.spec for the isolation of Pb from geological methods. A new scheme described by Deniel and Pin304 for the simultaneous separation of Pb and Sr was based on the selectivity and high capacity of the alternative extraction chromatographic material, Sr.spec. Samples were loaded on to a column of the material in 2 M HNO3 and, after removal of matrix elements, Ba, Sr and Pb were eluted in sequence with 7 M HNO3, 0.05 M HNO3 and 6 or 8 M HCl, respectively. The procedure gave straightforward separation of Pb and Sr on 150 µl columns with high Pb yield (>96%), good purity and satisfactory blank levels. In order to avoid poor ionization of Pb on the filaments, it was found to be essential to pretreat the Sr.spec material before use to remove traces of organic components released during chromatography. The small volumes of mineral acids required and significant gain in time were advantages over conventional ion-exchange methods. Even though Pb was eluted last in the procedure, it could be separated with blanks (150 pg) as low as those obtained with the conventional methods. Although the Sr.spec material could not be recycled for further Pb separations, it could be decontaminated sufficiently for further use in routine Sr separations.
In spite of its potential, the NdO+ technique has not been widely applied because of a combination of problems such as the uncertainty in oxygen isotope ratios, long acquisition times needed on single collector instruments and the need for good chromatographic separation of Nd. A simple and highly efficient chromatographic procedure developed by Griselin et al.305 gave a high-yield separation of Nd from other light REEs (LREEs) and offered the opportunity to measure Nd isotope ratios on very low abundance and/or small size samples (1–5 ng Nd loaded on the filament). After separation of the LREE fraction from the bulk matrix using the specific chromatographic resin TRU.spec, Nd was separated from other LREEs using HPLC with anion-exchange resin and methanol–acetic acid–HNO3 eluent. The low blanks (10–16 pg) significantly reduced interferences during MS measurements so that 140CeO+∶144NdO+ and 141PrO+∶144NdO+ ratios were below recommended values. The external precision (±0.004%, 2σ) was as good as that obtained by conventional analyses on larger (several tens of ng of Nd) samples. A compilation of oxygen isotope ratios published in the literature revealed that variations in ratios were mainly due to different loading techniques used. The source of oxygen, the nature of the chemical emitter and the type of ionization filament all influenced the measured ratios. It was recommended that the same loading technique be used for both NdO+ and oxygen isotope measurements and that the temperature range, over which the NdO+ isotope composition was measured, be well controlled.
There is a continuing need to improve the relatively poor sensitivity of ESMS, in particular when matrix components are present. An increasing number of papers has reported improved sample preparation methods and analytical procedures, in particular the so-called multidimensional separation procedures which use two or more separation methods in series. McSheehy et al.306 investigated in some depth the isolation of individual As species from algae extracts in a solution free of sample matrix. They paid particular attention to the chromatographic purity of peaks and the optimization of buffers for their removal prior to ESMS analysis. For example, the Tris-buffer commonly used in chromatography was replaced with ammonium carbonate. An initial separation using SEC was required to remove high molecular weight matrix components, e.g., polysaccharides and proteins, which could interfere with subsequent separation steps or suppress ionization. The As fraction was then subject to anion-exchange chromatography followed, if MS-MS analysis were to be used, by reversed phase HPLC. These procedures could be used to identify unambiguously eight of the 13 As species, including one new species (5-dimethylarsinoylribofuranose), found in an oyster sample.307 However, there were still limitations in that the five unidentified species either co-eluted with matrix components or were suppressed by the presence of buffer.
The application of ESMS to Sb speciation has been rare. This is partly because, in comparison with As, concentrations in the environment are low and partly due to the lack of suitable standard compounds. The analytical methodology needed to determine non-volatile Sb species in the environment generally remains to be developed. Zheng et al.308 investigated the ESMS of Sb compounds most often used as chromatographic standards with the objective of providing fundamental data for the chromatographic separation and ESMS identification of Sb compounds found in the environment. The sample cone voltage was investigated as an important factor controlling the degree of ion fragmentation in the ion source. A voltage of 15 V was optimum in terms of sensitivity of intact molecular ions. Operation of an ES-TOFMS instrument in the positive ion mode and with a high sample cone voltage (100 V) was suitable for identification of the two commonly encountered organic Sb compounds studied, trimethylantimony dichloride and dihydroxide. These compounds cannot be differentiated by HPLC-ICP-MS because of their similar chromatographic behaviour but have different characteristic ESMS spectra. Negative ion ESMS was found to be more suitable for identification of inorganic compounds (SbIII and SbV) than operation in the positive ion mode. Peaks at m/z 183 and 185 were used to identify trimethylantimony dichloride following separation by SEC. The method used by Craig et al.309 for a study into the oxidation in air of trimethylstibine did not require clean-up of water and methanolic extracts of the sample. A complex series of products was identified, including a range of cyclic and linear oligomers.
Two groups have reported Se speciation in biological matrices following several separation steps and using MS-MS. McSheehy et al.310 identified six Se compounds in yeast extracts through the use of two-dimensional SEC and reversed phase HPLC for separation of the selenospecies. The separated fractions were analysed separately by ICP-MS for monitoring the eluted compounds and by ES-MS-MS for identification of the eluted species. The addition of the SEC step improved the purity of the reversed phase HPLC peaks and reduced the noise level in ESMS. The authors considered molecular mass and fragmentation pattern alone to be insufficient to give complete identification of the eluted compounds, especially those not previously reported. Isolation of the compounds followed by chemical reduction or enzymatic cleavage was considered necessary for unambiguous identification of complex structures. Cao et al.311 used a single-step reversed-phase HPLC separation procedure and ESMS-MS on a triple quadrupole instrument to identify two of the six organoselenium species found in human urine. These species, selenomethionine and selenocystamine (11 and 40 µg l−1, respectively), were identified for the first time in human urine. The effluent from the HPLC column was split, with 1% going to an ICP-MS instrument and 99% to a fraction collector. Fraction collection commenced when a Se signal was recorded by ICP-MS. The two seleno-species were monitored using multiple reaction monitoring (MRM), in which one or more transitions of precursor to product ions were monitored in tandem. The MRM scan only gave a signal if the precursor and product ions were transmitted through their two respective quadrupoles. This mode was more sensitive than the full-scan mode and had a ten-fold lower sample volume requirement (30–50 µl).
A particular problem in As speciation by ESMS is that As is monoisotopic. This means that a peak in a MS spectrum cannot be assigned to an arsenic compound without some form of fragmentation. The group at Pau306,307 has generally used collision induced dissociation of molecular species and tandem MS-MS for unambiguous species identification, but other studies by Francesconi and colleagues312–314 have reported the successful use of a single quadrupole analyser, variable source fragmentor voltages and selected ion monitoring. The latter system was used to study the microbial demethylation of arsenobetaine in sea-water,314 arsenic biotransformation by algae313 and the human metabolism of a pure arsenosugar.312 The systems used, in particular the chromatography and a reliance on retention data, still require refinement to improve sensitivity. Only three of the 12 As metabolites detected in the last study could be positively identified, especially as there was some evidence for transformation of As species during storage of urine samples.
The most commonly used technique for the determination of organotin compounds is GC. This technique has a high separation power and can be connected to very sensitive detectors but it requires derivatization and clean-up procedures prior to analysis. An attractive alternative to GC is HPLC coupled to MS, in particular ESMS, which can provide direct species identification. A simple and sensitive method (linear calibration 0.5–200 ng ml−1; LOD 0.05 ng ml−1) for the determination of tributyltin in aqueous and environmental samples involved automated in-tube solid-phase microextraction (SPME) coupled to HPLC and a quadrupole ESMS instrument.315 A major advantage in the use of SPME for ESMS analysis was, in comparison to direct injection, removal of matrix ions prior to HPLC separation. This resulted in a 20-fold improvement in sensitivity. The commercially available Supel-Q PLOT (Porous Layer Open Tubular) capillary column (60 cm long, 30 µl internal volume) was placed between the sample injection loop and the injection needle of the HPLC autosampler. Desorption of analytes from the SPME capillary was achieved with the HPLC mobile phase. The MS parameters optimized were fragmentor voltage (variable), capillary voltage (2500 V), nebulizer gas pressure (N2, 55 psi) and drying gas flow rate and temperature (N2, 9 l min−1, 350 °C). The optimum fragmentor voltage, which controls ion fragmentation and transmission, was different for each ion of interest so fast switching of fragmentor voltage was required to monitor simultaneously all the selected ions. Extraction efficiency was insensitive to all sample matrices and pHs tested with the exception of saturated NaCl. The use of microcapillary LC coupled to an ES ion trap instrument allowed Jones-Lepp et al.316 to identify unequivocally the organotin species leaching from PVC pipes. It was concluded that dibutyltin can leach from PVC pipes to give, under ambient conditions, concentrations of 1 µg l−1 in water.
Measurement of metal species in soil solution is important for assessing the availability and mobility of contaminants. Svete et al.317 used synthetic solutions of hydrated Zn2+ species and Zn complexes with citrate, oxalate and EDTA to develop a method for Zn based on convective interactive media (CIM) fast monolithic chromatography, FAAS detection and ESMS structure identification. It was found that Zn interacts with various buffers and that careful adjustment of pH with dilute KOH solutions was required. The base peak in the mass spectra of the Zn species studied was due to the deprotonated ligand (M–H)−, which was selected as the precursor ion for further collision-induced fragmentation. The LOD for the separated Zn species was 10 µg l−1 with good reproducibility (2–4%) at 2 mg l−1. Although high salt loadings eluted during the chromatographic separation were tolerated by the ion source used, it should be noted that the polluted samples tested had atypically high Zn concentrations and that not all species eluted could be identified.
The use of ESMS for isotope ratio determination is new and has potential in some applications in that analysis time can be much reduced in comparison with other techniques. Moraes et al.318 found that spectra of tetrafluoroborate solutions obtained under mild ES conditions were free of interference with baseline resolution at m/z 86 and 87, corresponding to 10BF4− and 11BF4−, respectively. The only other peaks were due to Na(BF4)2− at m/z 195, 196 and 197. A drawback in the use of BF4− is its significant memory effect, which was minimized by replacing the original pepper pot counter electrode with a cross-flow counter electrode and by applying a cleaning process based on mannitol injection. Corona discharges occurred only rarely in the source but changed the spectrum so dramatically that robust statistics, based on the median and the median of absolute deviations, were needed. The origin of mass discrimination and drift effects could not be identified and so the effects had to be corrected through successive injections of NIST SRM 951 (boric acid), converted to NaBF4. An isotope ratio precision of 0.4‰ was obtained for 5 min injections of 100 µM boron solution at 10 µl min−1. The conversion of B in real samples into BF4− was not covered in this paper and further studies were required to establish the true potential of the technique. Isotope distributions in phosphate are useful measurements in geochemical systems as they give information on their origin and environmental oxygen exchange processes. The ESMS determination of 18O in orthophosphate solutions has been shown by Alvarez et al.319 to be rapid with only mg sample requirements. The absolutely essential requirement of lack of oxygen exchange under the experimental conditions used was demonstrated by injecting a mixture of unlabelled phosphate and isotopically enriched water into the mass spectrometer. Although various aspects of the analysis were discussed, no analytical figures of merit were presented so the true potential of the method cannot be assessed.
One anticipated difficulty in coupling an ES ion source to TOF analysers is the inherent incompatibility of a continuous ionization source with a pulsed detection system. The ion flux in ESMS is very small (pA) so it is important to capture as many ions as possible. Heated capillary tube introduction is a simple interface design not requiring pneumatic assistance, curtain gas and rf multiple cooling. It can be coupled easily to any mass analyzer. Hang and Majidi320 have reported the detailed investigation of ion transport processes in aheated capillary tube interface for ES-TOFMS. Sample solution was introduced into the capillary needle through a fused silica capillary (360 µm od, 40 cm long). The spray needle was mounted on a translation stage to allow it to be adjusted in three dimensions with respect to the interface tubing. A series of experiments was designed to evaluate the critical parameters such as spray needle position, spray potential, tubing temperature, skimmer potential, ion energy distribution and ion fragment formation. Adjusting the potential placed on the capillary tube altered the collision energy in the system so that both atomic and polyatomic ions could be generated. The interface was considered a promising approach to reducing the KE of ions generated from an ES source. Its simplicity eliminated the need for a countercurrent gas flow and a rf-only multipole.
The large increase in recent years in GC-EIMS developments driven by the need for speciation analysis has not continued into the period covered by this Update. Although the technique is seen to have several valuable advantages, relatively poor sensitivity appears to be limiting its usefulness. A critical comparison of three GC detectors for the speciation of lead in rainwater found GC-EIMS to be intrinsically the most specific option, allowing analytes to be identified on the basis of not only their retention times but also their mass spectra.321 In addition, the interface between the GC column and the detector was simplest for GC-EIMS. However, the levels of analytes in rainwater samples (5 ml) fell below the LOD of 4–12 pg ml−1, thereby making GC-EIMS unsuitable. In comparison the LOD for GC-ICP-TOFMS was 0.4–0.6 pg ml−1. The precision of analysis (3–5%) was controlled by the FI sample introduction system and was therefore similar for all the detectors evaluated.
Gómez-Ariza et al.322 built a pervaporation unit for the GC-EIMS analysis of Se in sediments but found that the levels of Se species in samples studied were below the LOD of the method (0.3–0.6 ng in 5 g sample). The pervaporation module consisted of two compartments separated by a hydrophobic membrane. The sample was placed in the lower compartment, which was maintained at 70 °C, and the volatile analytes collected in the upper compartment in a carrier gas. The analytes were carried in a flow of N2 to a preconcentration trap (0 °C) from which they were thermally desorbed (150 °C) and introduced into the GC column in a flow of He. The major advantages over headspace methods were the ease of automation with minimum investment in equipment, better recoveries and repeatability and shorter analysis time as equilibrium conditions were not required. An advantage over the purge-and-trap method is that removal of water vapour was not necessary, as water did not pass through the hydrophobic membrane.
The use of SPE techniques for GC-EIMS has continued to receive attention, in particular for the determination of organotin compounds. In the method of Nogueira et al.,323 tributyltin, extracted from environmental samples with HCl (5%, 30 ml) and trapped on the SPE tube (1 ml, 100 mg, C18), was eluted with ethyl acetate and derivatized with CH3MgBr in a diethyl ether solution. This extraction procedure was found to be easy and fast in comparison with other methods, in particular liquid–liquid extraction. Operation in the selected ion monitoring mode (m/z 193) gave a linear calibration (0.01–10 ng) with high selectivity. Repeatability was, however, quite poor, as shown by analysis of an in-house standard (50 ng) for which a value of 44 ± 21 (n = 15) was obtained. Cardellicchio et al.324 used SPME in the determination of organotin compounds in marine sediments. The procedure was based on the extraction of the organotins with HCl–methanol, in-situ derivatization with sodium tetraethylborate (NaBEt4) and headspace SPME extraction using a fibre coated with poly(dimethylsiloxane). The attraction of headspace rather than direct SPME sampling was reduced extraction time resulting from the much higher diffusion of analytes in the vapour phase than in the aqueous phase. Derivatization with NaBEt4 can be performed in the aqueous phase and is therefore simpler and quicker than the use of Grignard reagents. The method was optimized with respect to the derivatization and extraction conditions and gave linear calibration over the range 30–1000 ng l−1 (as Sn) and LOD from 0.7 to 1 ng g−1 (as Sn).
The problems posed by high chloride matrix and acidity in the chromatographic determination of low concentrations (µg l−l) of bromide in sea-water were avoided by Mishra et al.325 by oxidation of Br− using 2-iodosobenzoic acid in the presence of substituted phenols, 2,6-dimethylphenol or 2-tert-butyl-4-methylphenol, to form 4-bromo-2,6-dimethylphenol or 6-bromo-2-tert-butyl-4-methylphenol, respectively. Before derivatization, the organic matter in the sample was removed by SPE using C18-bonded silica or a polymeric sorbent based on styrene-divinylbenzene copolymer. The bromophenol derivative was extracted, pre-concentrated and washed to remove ionic impurities using SPE or solvent microextraction. Good calibration linearity (0.02–100 µg−l), LOD (5 ng l−1 for 250 ml samples) and precision (4%) were obtained.
Lott327 has given details of a new cryogenic trap design, which employed polished stainless steel as the active trapping surface. In comparison with the charcoal trap systems commonly used, the new trap offered significant advantages of lower release temperatures and improved separation. These resulted in shorter processing times, reduced blanks and memory effects and improved purity of the prepared gas.
The use of RMs is a fundamental feature of any analytical measurement, especially when highly precise and accurate analysis is required. A large batch of high-purity krypton separated from the atmosphere has been prepared as a primary isotopic gas standard.328 This new isotopic RM has certified values with small uncertainties for isotope ratios, isotopic composition and molar mass of Kr.
Baba and Waki331 have successfully mass analysed just 30 ions of a Xe isotope by laser-cooled fluorescence MS using laser-cooled barium. The four parts of the instrument were an ion source, a linear ion trap, an ion guide and an electron multiplier. The principle of the technique is based on energy transfer from molecular ions to the laser-cooled ions through Coulomb collisions when the secular motion of the sample ions is excited. Sample and barium ions were created in the ion source section by electron ionization at 150 eV. The secular frequency of sample ions, proportional to the specific charge of the ion, was measured by detecting laser-induced fluorescence emitted by the laser-cooled ions as sample ions oscillated in the ac electric field. The absolute amount of the sample ions that contributed to the spectrum was measured using the electron multiplier, which was calibrated using crystallized laser-cooled barium ions as a standard amount of trapped ions. Separation of the ion source from the ion trap section prevented coating of the surface of the ion trap with barium metal. High transfer efficiency was achieved using the ion guide placed between the ion trap and the electron multiplier. The absolute LOD of the method was estimated to be 3 ions.
AFM: atomic force microscopy
AMS: accelerator mass spectrometry
aTOF: aerosol time-of-flight
BCR: Bureau Communautaire de Reference
CE: capillary electrophoresis
CF: continuous flow
CIM: convective interactive medium
CRI: chemical reaction interface
CRM: certified reference material
CSIA: compound-specific isotope analysis
CSRA: compound-specific radiocarbon analysis
CV: cold vapour
CZE: capillary zone electrophoresis
dc: direct current
DIHEN: direct injection high efficiency nebulizer
DMA: differential mobility analyser
DRC: dynamic reaction cell
DRF: depth resolution function
EDTA: ethylenediaminetetraacetic acid
EDXRF: energy dispersive X-ray fluorescence
EI: electron ionization
EPMA: electron probe microanalysis
ESMS: electrospray mass spectrometry
ETV: electrothermal vaporization
FI: flow injection
FTIR: Fourier transform infrared
FWHM: full width at half maximum
GC: gas chromatography
GC-EIMS: gas chromatography electron ionization mass spectrometry
GDMS: glow discharge mass spectrometry
HECFMN: high efficiency cross-flow nebulizer
HG: hydride generation
HPLC: high-performance liquid chromatography
HR: high resolution
IC: ion chromatography
ICP-MS: inductively coupled plasma mass spectrometry
id: internal diameter
ID: isotope dilution
IR: infrared
ISRD: implantation, sputtering, replacement/relocation and diffusion
KE: kinetic energy
LA: laser ablation
LC: liquid chromatography
LEAFS: laser-excited atomic fluorescence spectrometry
LIMS: laser ionization mass spectrometry
LIPS: laser-induced plasma spectrometry
LMIG: liquid metal ion gun
LMMS: laser-microprobe mass spectrometry
LOD: limit of detection
LOQ: limit of quantification
LREE: light rare earth element
MC: multiple collector
MD: molecular dynamics
MIC: multiple ion counting
MIP: microwave-induced plasma
MRI: mixing-roughness-information depth
MRM: multiple reaction monitoring
MS: mass spectrometry
NIST: National Institute of Standards and Technology
PB: particle beam
PCA: principal components analysis
PGE: platinum group element
ppb: parts per billion
ppm: parts per million
PSIA: position-specific isotope analysis
PTFE: poly(tetrafluoroethylene)
RBS: Rutherford backscattering spectrometry
REE: rare earth element
rf: radiofrequency
RIMS: resonance ionization mass spectrometry
RM: reference material
RPQ: retarding potential quadrupole
RSD: relative standard deviation
RSF: relative sensitivity factor
SCAPS: stacked CMOS-type active pixel sensors
SEC: size-exclusion chromatography
SIA: sequential injection analysis
SIFT: selected-ion flow tube
SIMS: secondary ion mass spectrometry
SIRMS: stable isotope ratio mass spectrometry
SNMS: sputtered neutral mass spectrometry
SPME: solid phase microextraction
SRM: standard reference material
S-SIMS: static secondary ion mass spectrometry
SSMS: spark source mass spectrometry
TEM: transmission electron microscopy
TIC: thermal ionization cavity
TIL: total interference level
TIMS: thermal ionization mass spectrometry
TOF: time-of-flight
TOFMS: time-of-flight mass spectrometry
TRIMz: transport of ions in matter
TSH: thyroid stimulating hormone
USN: ultrasonic nebulization
UV: ultraviolet
VOC: volatile organic compound
WARP: wide aperture retardation potential
XPS: X-ray photoelectron spectroscopy
XRF: X-ray fluorescence
YAG: yttrium aluminium garnet
This journal is © The Royal Society of Chemistry 2002 |