Page 64 - An Introduction to Analytical Atomic Spectrometry - L. Ebdon

Page 46

2.4.1 Spectral Interferences

These are the only type of interference that do not require the presence of analyte. For AAS the problem
of spectral interference is not very severe, and line overlap interferences are negligible. This is
because the resolution is provided by the 'lock and key' effect. To give spectral interference the lines
must not merely be within the bandpass of the monochromator, but actually overlap each other's
spectral profile (i.e. be within 0.01 nm). West [Analyst 99, 886, (1974)] has reviewed all the reported
(and a number of other) 'spectral interferences' in AAS. Most of them concern lines which would never
be used for a real analysis, and his conclusion is that the only 'real' problem is in the analysis of copper
heavily contaminated with europium! The most commonly used copper resonance line is 324.754 nm
(characteristic concentration 0.1 µg cm ) and this is overlapped by the europium 324.753 nm line
-3
(characteristic concentration 75 µg cm ).
-3
Spectral interferences from the overlap of molecular bands and lines (e.g. the calcium hydroxide
absorption band on barium at 553.55 nm) cannot be so easily dismissed. Lead seems to be particularly
prone to such non-specific absorption problems at the 217.0 nm line (e.g. sodium chloride appears to
give strong molecular absorption at this wavelength). This type of problem is encountered in practical
situations, but can sometimes be removed by the technique of background correction (see Section
2.2.5.2).

Q. Why can it be said that spectral interferences can be virtually eliminated in AAS?

Q. How can non-specific absorption problems in AAS be overcome?

2.4.2 Ionization Interferences.

Also called vapour-phase interferences or cation enhancement. In the air-acetylene flame, the
intensity of rubidium absorption can be doubled by the addition of potassium. This is caused by
ionization suppression (see Section 2.2.3), but if uncorrected will lead to substantial positive errors
when the samples contain easily ionized elements and the standards do not. An example is when river
water containing varying levels of sodium is to be analysed for a lithium tracer, and the standards,
containing pure lithium chloride solutions, do not contain any ionization suppressor.

The problem is easily overcome by adding an ionization suppressor (or buffer) in large amount to all
samples and standards.