This paper discusses two main aspects of spectral interferences: (1) the handling of the vast amounts of data needed for adequately coping with the lack of selectivity of atomic emission spectroscopy (AES), and (2) the concept of selectivity and the use of selectivity as a major analytical performance characteristic in AES. The main discussion centres about the problem of correctly determining the magnitude of interfering signals in the spectral windows of analysis lines, thus background correction in the widest sense. The approaches are classified into two groups depending on whether the sample concomitants or the gaseous atmosphere in the source contribute unstructured or structured background. The latter situation is emphasized in the subsequent discussion. In this light, the paper reviews data compilations published since in 1884 Hartley laid the foundations of quantitative AES. These compilations are featured as (1) “hardware only” (plates, charts, chart recordings, plots, tables, atlases), (2) “software/hardware conversion” (tape or disk with computer retrieval for data visualization) or (3) “software” only (tape or disk with computer retrieval and processing in combination with data on samples). It is concluded that up to the present most approaches must be categorized as (1) or (2), which means that either the operator looks up and the operator decides or the computer looks up and the operator decides. Only in a few instances are the capabilities of the computer fully exploited and is it the computer which looks up the data and also makes the decisions (“artificial intelligence”). The future should and will see the balance tip into that direction. The second part of the discussion deals with the concept of selectivity and in that context considers the effects of the practical resolving power (“spectral resolution”) on analytical performance in the case of line overlap. A review of early and recent work covers such topics as physical line widths, required spectral resolution, trade-off between spectral resolution and background radiant flux at the detector, and the extent to which modern spectrometers fulfil the conditions for exploiting the benefits of high resolution in ICP-AES. These benefits are discussed in terms of limits of detection, selectivity, and limits of determination. It is shown that, in the case of line overlap, the conventional determination of limits of detection yields irrealistic values, which, in addition, entirely veil the actual benefits of high resolution spectroscopy. The vital profits are related with the selectivity and can be quantitatively expressed in terms of the limit of determination, as defined in this paper. It is concluded that, in the case of line overlap, analytical performance gains much more from an increase in selectivity than from a reduction of conventional detection limits. The full recognition of the implication of this conclusion may still substantially stimulate the further improvement of the analytical performance of AES by the exploitation of high resolution spectroscopy, the optimization of the gaseous atmosphere and the operating conditions of the source, and selective volatilization, where appropriate.
Read full abstract