Determination of strongly curved calibration graphs in flame atomic-absorption spectrometry: comparison of manually drawn and computer-calculated graphs

Abstract

The non-linear portion of the calibration graphs in flame atomic-absorption spectrometry (AAS) offers more precise analytical results provided that the calibration graph can be determined reliably. Experimental data for three elements (Cu, Ni, Pb) and synthetic data were submitted to nine analysts and to three mathematical expressions (polynomial, cubic spline and rational function). The quality of hand-drawn and computercalculated graphs was judged for goodness of fit and the accuracy of determining unknowns. It is concluded that results within 1% can be obtained in either instance, provided that 5-8 selected reference solutions are used and weighting factors are applied in the least-squares fit. With a simple criterion to guide the selection of the correct degree, the ordinary polynomial is the preferred expression to fit flame AAS data. With these recommendations it is possible to extend the calibration graph at least to the point where the slope is reduced to 25% of its value at the origin.

Validation and Quality Control with Atomic Absorption Spectrometry for Environmental Monitoring

“…In an effort to detect 'good' or 'bad' calibrations the use of a calibration quality coefficient has While the use of such a coefficient is not as yet available in any commercially supplied software package for an AAS instrument, it can only be a matter of time. The use of such a CQC allows the instrument to decide automatically between a 'good' and 'bad' calibration and choose an alternative calibration method such as a robust regression method [39]. However, while these techniques and the subsequent iterations that they require [40], will undoubtedly improve the calibration, they do require a considerable increase in the number of calculations required.…”

Weighted least squares in calibration: The problem with using “quality coefficients” to select weighting formulas

“…This definition was utilized by some later workers [20,21]; however, others [22,23] preferred a definition equivalent to…”

Fitting Straight Lines with Replicated Observations by Linear Regression. III. Weighting Data

“…In atomic absorption spectrophotometry (AAS), the initially linear relation according to Beer's law breaks down at higher absorbance owing to instrumental imperfections and the influence of atomic spectral line profiles (33). A commonly observed phenomenon in AAS is the ending of the calibration graph towards the concentration axis at elevated concentrations (34).…”