Page 26 - An Introduction to Analytical Atomic Spectrometry - L. Ebdon

Page 11

Statistical theory tells us that, provided sufficient readings are taken, 68.3% of the actual readings lie
within the standard deviation of the mean, and that the mean ±2s and the mean ±3s will contain 95.5%
and 99.7% of the readings, respectively. Hence there is only a 0.3% chance that, if the readings are
larger than the mean of the blank readings by three times the standard deviation, this is merely due to an
unusually high blank reading. Thus, the limit of detection may be defined 'as that quantity of element
which gives rise to a reading equal to three times the standard deviation of a series of at least 10
determinations at or near the blank level'. This assumes a 'normal' distribution of errors, and may
consequently result in more or less optimistic values.

The limit of detection is a useful figure which takes into account the stability of the total instrumental
system. It may vary from instrument to instrument and even from day to day as, for example, mains-
borne noise varies. Thus, for atomic absorption techniques, spectroscopists often also talk about the
characteristic concentration (often erroneously referred to as the sensitivity—erroneously as it is the
reciprocal of the sensitivity) for 1% absorption, i.e. that concentration of the element which gives rise
to 0.0044 absorbance units. This can easily be read off the calibration curve. The characteristic
concentration is dependent on such factors as the atomization efficiency and flame system, and is
independent of noise. Both this figure and the limit of detection give different, but useful, information
about instrumental performance.

Q. Define the limit of detection and characteristic concentration?

Q. What information is given by the limit of detection, and how does this differ from that given by the
characteristic concentration?

1.5 Interferences and Errors.

1.5.1 Interferences

Interference is defined as an effect causing a systematic deviation in the measurement of the signal
when a sample is nebulized, as compared with the measure that would be obtained for a solution of
equal analyte concentration in the same solvent, but in the absence of concomitants. The interference
may be due to a particular concomitant or to the combined effect of several concomitants. A
concomitant causing an interference is called an interferent. Interference only causes an error if not
adequately corrected for during an analysis. Uncorrected interferences may lead to either
enhancements or depressions. Additionally, errors may arise in analytical methods in other ways, e.g.
in sample pretreatment via the