A comparison of curve fitting algorithms for flame atomic absorption spectrometry

Abstract

A comparison of six commercially available curve fitting algorithms for calibration in flame atomic absorption spectrometry (FAAS) has been made. Programs using the algorithms were written in BASIC for a microcomputer and, where possible, tested against the commercial version. Calibration curves for magnesium, chromium and nickel produced by the algorithms were compared, on the basis of the sum of squares of the percentage deviations and its root mean square, with each other and with linear interpolation and manual fitting. The standard deviations of the goodness of fit parameters were calculated to indicate significant differences between the fits obtained. With the exception of the simple parabola, linear interpolation and manual methods, which were significantly poorer, the algorithms performed similarly. In general the errors in curve fitting were well below 5% for the commercial algorithms.