J.J.A.M. Vrakking

Detection limit including selectivity as a criterion for line selection in trace analysis using inductively coupled plasma-atomic emission spectrometry (ICP-AES)—a tutorial treatment of a fundamental problem of AES

Abstract This paper proposes the “true detection limit” as a quantitative criterion for line selection in inductively coupled plasma-atomic emission spectrometry. The “true detection limit” is denned as the sum of (1) a selectivity term that accounts for the additional noise resulting from line overlap and (2) the “conventional detection limit”, which covers the common noise sources for the interference-free situation and, additionally, the effect of line wings. The theoretical part discusses the various arguments as well as the links with the previous work from which the present approach evolved. The theoretical discussion is concluded with the formulation of the mathematical expressions to be subsequently used in the experimental part, where the approach is applied to a case study dealing with the selection of the best analysis line(s) for the determination of traces of In in binary mixtures of W and Mo, the composition of which is assumed to vary from pure W to pure Mo. Both high and medium spectral resolution are covered, as well as the effects of changing the ICP operating parameters or modifying the transport rate of the metal sample to the plasma, as achieved by varying the concentration in the solution or using an ultrasonic instead of a pneumatic nebulizer. It is shown that the approach permits straightforward and unambiguous decisions on line choice. It is argued that the approach should be applicable to multi-component samples of whatever complexity and that it should prove useful not only for a priori line selection, as elaborated in this work, but also for a posteriori line selection in methods employing “multiple line analysis”. Although the experimental data were directly derived from digitized wavelength scans for pure analytes and interferents recorded with a particular spectroscopic apparatus, it is pointed out that the approach has every prospect of being applied in software systems for line selection using physically resolved spectral data in combination with a variable spectral instrumental function.

The widths and shapes of about 350 prominent lines of 65 elements emitted by an inductively coupled plasma

Abstract This work describes the measurement of the widths and shapes of about 350 prominent lines of 65 elements emitted by an inductively coupled plasma (ICP). The experimental procedure is an improved version of an earlier described approach using a 1.5-m echelle monochromator with predisperser. For most measurements the practical spectral bandwidth was smaller than 1.5 times the physical line width. Results are reported for both simple lines with chiefly Doppler broadening and complex structures with unresolved or partly resolved hyperfine structure (HFS). An atlas with the spectral scans of about 90 interesting line profiles is included and wavelengths of HFS components are tabulated. The latter were accurately determined in the case of well resolved structures and roughly estimated for poorly resolved or unresolved structures. The data were collected chiefly with a view to spectrochemical analysis, with a threefold aim: (a) to provide the basic data needed for comparing detection limits obtained with spectroscopic apparatus of different bandwidths; (b) to identify HFS components which, under high resolution conditions, may be useful as separate prominent lines in order to circumvent spectral interference and (c) to establish a basis for models that can be used in software for line selection such that the number of data needed in view of differences among line shapes and widths can be reduced to a minimum. It appears from the results that these targets are closely approached. Further work is required, however, for the full implementation of the results.

Mutual spectral interferences of rare earth elements in inductively coupled plasma atomic emission spectrometry. III: Pseudo physically resolved spectral data: complete results and evaluation

Abstract This paper directly links up with Part II [ Spectrochim. Acta 43B , 1365 (1988)], which treats the complete theory and contains representative examples. The present work produces the “complete” results and their evaluation. The results essentially consist of pseudo physically resolved spectral data for 14 rare earth elements (REE) in 80-pm spectral windows of 26 prominent analysis lines of Ce, La, Nd, Pr, and Sm, viz. , peak wavelengths, hyperfine structure widths, and Q -values of lines and background normalized to a spectral bandwidth of 1 pm. These data, along with an application program, permit the simulation of spectral scans for any specified bandwidth. The paper provides an extensive evaluation and assessment of the vast amount of new spectroscopic data. Internal consistency is checked by comparing simulated and experimental scans for two bandwidths of the echelle spectrometer with which the primary data were measured in a 50-MHz ICP. In this assessment, both the overall likelihood and—quantitatively—the equality of the experimental and calculated Q -values at the wavelengths λ a of the analysis lines ( Q Ij a ) are used as criteria. External consistency is tested by comparing the present results with those of recent measurements in a 27-MHz ICP using a common grating spectrometer. Peak wavelengths, peak Q -values and Q Ij a -values (and the corresponding Q Wj a ) are used in this comparison. Finally, the paper also compares the present results with the data listed in the MIT and NBS Tables. The number of REE lines identified in the 26 spectral windows of 80 pm width was 1075, which compares with 358 in the compilation for the other ICP, 260 in the MIT Tables, and 134 in the NBS Tables, while the ranges on the Q I -scale were 10 6 , 1000, 500, and 170, respectively. The work has shown (a) the overall usefulness of the “ Q-concept ” as a quantity for expressing “intensities” of interfering lines, (b) the validity of tranferring Q -values from the one ICP to the other, and (c) the appropriateness of the model and the associated data for simulating spectral scans for spectrometers of “whatever” type and bandwidth. On the whole, the work has demonstrated that the approach is entirely viable. It has provided an extensive data set and a reliable model for manipulating the data in such a way that faithful simulations can be obtained for a multiplicity of “spectroscopic situations”. This, in turn, permits case studies without the need of time-consuming experimental work.

Detection limits of about 350 prominent lines of 65 elements observed in 50 and 27 MHz inductively coupled plasmas (ICP): effects of source characteristics, noise and spectral bandwidth-“Standard” values for the 50 MHz ICP

Abstract This paper links up with a previous publication by the same authors [ Spectrochim. Acta 41B , 1235 (1986)] dealing with the measurement of the effective line profiles of about 350 prominent lines at high spectral resolution and the determination of the physical widths of these lines. In the present work the line widths are used for a breakdown of the detection limits obtained with these lines using different ICPs and different spectrometers. This breakdown takes into account the separate effects of source characteristics, noise, and spectral bandwidth. The availability of the numerical values of the physical widths of a large number of lines permitted a more rigorous approach than in a previous work [ Spectrochim. Acta 40B , 1437 (1985)]. The present approach was applied to detection limits obtained in this work with a 50 MHz ICP at high spectral resolution and to results reported by W inge et al. [ Appl. Spectrosc. 33 , 206 (1979)] and WOHLERS [ ICP Information Newslett. 10 , 601 (1985)]for 27 MHz ICPs. The 50 MHz ICP was shown to have an advantage in source signal-to-background ratio (SBR) with respect to either of the two 27 MHz ICPs. This SBR advantage was a factor of 3–15 with respect to the “Winge ICP” and a factor of 2–6 with respect to the “Wohlers ICP”. The approach was also used to convert detection limits measured in the 50 MHz ICP at high resolution into values for 15 pm spectral bandwidth and a relative standard deviation of the background signal equal to 1%. These values are recommended as standards of performance for the conventional argon ICP. The paper comprises a tabulation of the complete results for the 350 prominent lines and includes four sets of detection limits for these lines.

Mutual spectral interferences of rare earth elements in inductively coupled plasma atomic emission spectrometry. II: Approach to the compilation and use of pseudo physically resolved spectral data

Abstract This paper is the second part of a series of three papers dealing with mutual spectral interferences of rare earth elements (REE) in inductively coupled plasma atomic emission spectrometry (ICP-AES). The present paper describes an approach to compiling spectral data and modelling spectra in such a way that spectral scans of interferents in 80-pm wide windows centred about prominent analysis lines can be simulated over a broad range of spectral bandwidths (BW). Experimental data are collected at spectral BWs between 4 and 8 pm and reduced to “pseudo physically resolved data”, whereby hyperfine structure (HFS) composites are modelled as trapezia with “Voigt shoulders”, characterized by a “HFS width”. In convolutions with the spectral instrumental function of the spectrometer, the pseudo physically resolved data can be formally treated as truly physically resolved data, provided that the spectral BW is not substantially smaller than that used in the present measurements. The approach characterizes the peak intensities of interfering lines in terms of sensitivity ratios with respect to the analysis lines (in Part I termed “Q-values”). The dependence of these Q-values on both the BW and the physical widths of the interfering and analysis lines is taken into account. It is shown that both the scans from which the basic data were derived and scans at essentially poorer resolution (about 17 pm BW) can be well simulated. In particular, it is demonstrated that Q-values at the wavelengths of the analysis lines, crucial for the calculation of true detection limits, can be predicted with a most reasonable accuracy ( i.e., to within a factor of 1.0 to 1.5 with respect to measured values). The paper discusses the theory and its application using typical situations, including pitfalls, for illustration. It is finally shown that the measuring technique in its own right permits the identification of typically 34 lines of 13 REEs in an 80-pm spectral window, the sensitivity of the faintest line being 1 20000 ofthat of the analysis line. The numbers 34 and 1 20000 compare with 4 and 1 80 for the NBS Tables, 8 and 1 100 for the MIT Tables, and 11 and 1 1000 for a recent in-house compilation for the ICP.

Line selection in atomic emission spectrometry using physically resolved spectra

Abstract Defining a quantitative criterion for line selection as a starting-point this paper discusses the type of data needed (a) in principle and (b) concretely. In this light, the shortcomings of available tabulations are pointed out. It is argued that an adequate approach to fulfil the concrete needs is to compile and store the data as peak wavelengths, peak sensitivities, and physical line widths, along with some mathematical functions that describe the essential line shapes. An approach to achieve this using high-resolution dispersive spectrometry is reviewed and discussed.

Spectral interferences in inductively coupled plasma atomic emission spectrometry-I: A theoretical and experimental study of the effect of spectral bandwidth on selectivity, limits of determination, limits of detection and detection power

Abstract This article links up with recent work on high resolution spectroscopy in this laboratory [1] and primarily deals with the effect of the spectral resolution on the “selectivity” in the case of samples that emit line-rich spectra. The concept of selectivity, as developed by KAISER [8] on the basis of the set of calibration equations for a multicomponent system, is considered as a useful starting-point but is rejected as a meaningful analytical figure of merit. Instead the concept of “line selectivity” is introduced as a criterion and related to the analytical error. This approach leads to a definition of the limit of determination such that its dependence on the spectral resolution can be clearly and unambiguously revealed in any concrete situation, that is, once the sample type has been specified. Such a specification is necessary since the numerical values of quantities related to selectivity are inherently linked with sample composition. Thus the theory is illustrated with practical examples including the results of a multiplicity of simulated line overlap situations using representative experimental line profiles measured at two extreme levels of resolution, referred to as “medium” and “high” resolution. It is shown that in the case of line overlap the limit of determination may exceed the limit of detection by one or even two orders of magnitude, unless line selection is based on a selectivity criterion so that the limit of detection is inherently coupled to the limit of determination. It is also shown that the prime benefit of high resolution spectroscopy is the reduction of the limit of determination, not that of the limit of detection. This benefit is found only if the spectral resolution can improve the selectivity, thus if there exist situations of line overlap.

High-resolution spectroscopy using an echelle spectrometer with predisperser. II: Analytical optimization for inductively coupled plasma atomic emission spectrometry

Abstract This work is primarily concerned with the optimization of the slit width (and thus the practical resolving power) of a new type of echelle spectrometer coupled to a 50-MHz ICP operated with a pneumatic nebulizer, as described in Part I of this article series (Spectrochim. Acta 39B , this issue (1984)). The optimization is carried out under “ICP compromise conditions” and uses detection power as criterion. With a “pure water” matrix, the effects of slit width on net line and background signals, signal-to-background ratio (SBR), relative standard deviation (RSD) of background signal and detection limit were evaluated for a set of prominent ICP lines spread over wavelengths between 190 and 500 nm. The detection limits eventually attained under optimum conditions were an order of magnitude better than “standard” values reported in the literature (winge et al., Appl. Spectrosc. 33 , 206 (1979)). The optimization was extended to a Ni-Co matrix, the latter serving as an example of samples that emit line-rich spectra. In this context, a detailed analysis was made of the background enhancements associated with the presence of major elements that emit line-rich spectra. Accordingly the effects of slit width on SBR, background RSD and detection limit were differentiated in dependence on whether the background enhancement was due to quasicontinuous background, due to complete coincidence of the analysis line with a line of the matrix, or due to partial line overlap. The quasi-continuous background was attributed to the wings of strong lines of the matrix, as described in Part III ( Spectrochim. Acta 39B , this issue (1984)). It was established that with pure line wing interference the gain in detection power achieved by improving the practical spectral bandwidth from, say, 0.015–0.005 nm is approximately similar to that found for pure water, that is, a factor of 2–3. In the case of partial line overlap, larger improvements can be achieved depending on the physical widths of the lines involved and the wavelength distance between them. The paper includes a description of a computer technique for manipulating experimental spectral scans stored on floppy disks. This technique is used to simulate scans of solutions of matrices spiked with analytes using only the scans for the pure matrix, the pure analyte and the pure solvent. Various examples are detailed in connection with the question whether and to which extent high resolution does improve the accuracy of trace analysis in the case of partial line overlap. Preliminary tests led to the thesis that high resolution is capable of providing higher accuracy than medium resolution at the same ratio of the concentration present to the detection limit. If this thesis can be definitely substantiated in future work, this means that high resolution not only provides for better detection limits, but will also yield higher accuracy at lower concentrations than medium resolution, the latter advantage being associated in particular with samples than emit line-rich spectra.

High-resolution spectroscopy using an echelle spectrometer with predisperser—I. Characteristics of the instrument and approach for measuring physical line widths in an inductively coupled plasma

Abstract This work is the first part of a series of three articles published in the same issue of this journal. This part describes (a) the characteristics of a novel type of echelle monochromator with predisperser in parallel slit arrangement and (b) the application of this instrument to the measurement of physical line widths in an inductively coupled plasma (ICP). The characterization of the instrument includes an outline of the principle and a brief description of the electronic and mechanical control, the detection, readout and software, and the wavelength calibration using 6 neodymium lines emitted from the ICP. This brief characterization is followed by a detailed treatment of the optical properties of the instrument: dispersion, theoretical resolving power, spectral slit, spectral instrumental function and spectral bandwidth. The experimental part deals with the determination of the practical instrumental bandwidth and the coverage of an instrumental broadening effect attributed to aberrations. This contribution from aberrations is determined with the aid of spectral lines emitted from a hollow cathode lamp and is accounted for by a correction of the theoretical spectral bandwidth. The corrected instrumental function is subsequently used to derive physical line widths from measured effective line widths. The results agree well with those recently published by other authors using different methods. It is concluded that the present approach yields values of physical line widths sufficiently accurate to make an extension of the measurements to a larger number of spectral lines interesting. The article is concluded with a discussion of the extent to which the practical resolving power of the instrument matches the physical characteristics (line widths) of ICP spectra, that is, whether the instrument is capable of the complete physical resolution of the spectra under analytically usable conditions. This discussion borrows an essential result from Part II as to the minimum slit widths compatible with the available radiant flux reaching the detector, which, in view of the associated shot noise, limits the resolving power that can be actually exploited under practical analytical conditions. It is argued that a practical resolving power of 50 000–70 000 in the low u.v. and of 60 000–100 000 in the visible can be realistically used in line-rich spectra (with an inherently enhanced background [Parts II and III], while practical resolving powers of up to about twice these values would be ideally required. It is further shown that the mechanical resolution links up well with the analytically usable practical spectral bandwidth.

Spectral interferences in inductively coupled plasma atomic emission spectrometry—II: An experimental study of the effect of spectral bandwidth on the inaccuracy in net signals originating from wavelength positioning errors in a slew-scan spectrometer

Abstract This article deals with an evaluation of the error in net analyte signals resulting from wavelength positioning errors in a slew-scan monochromator in the event that the analysis line experiences line overlap. The theoretical part concerns the development of an error function and the introduction of the concepts of “10% inaccuracy limit” (c10/cL) and “find peak transition” (cFPT), where cL is the detection limit, c10 the concentration at which the maximum error is 10 % and cFPT the concentration at which the analysis line shows just resolved from the interfering line so that a find peak routine can be used. The experimental part encompasses case studies for two levels of spectral resolution, designated “high resolution” (HR) and “medium resolution” (MR) as defined in a companion paper (Spectrochim Acta40B, 1085 (1985)) for a 1.5-m echelle monochromator with predisperser (Spectrochim. Acta39B, 1239 (1984)). The case studies refer to situations in which prominent lines of Al, Ca, V, Nb, Bi and Sn experience overlap with one or more OH band components. It was shown that HR can bring an advantage over MR amounting to a factor of 15 in c10, e.g. for Al I 309.271 and Al I 308.216, but may also entail only a marginal profit, e.g. for Bi I 306.765/306.774 nm. The effects of the distance between analysis and interfering line and the signal-to-background ratio of the interfering line were studied systematically by applying computer techniques such us shifting of experimental line profiles along the wavelength axis. This study also implied the consideration of the relationship between the 10% inaccuracy concentration, c10, and the limit of determination, cD, defined in the companion paper.