| Abstrakt: |
L'vov and co-workers developed a theoretical model and computational procedure (B.V. L'vov, L.K. Polzik and N.V. Kocharova, Spectrochim. Acta Part B, 47 (1992) 889 and B.V. L'vov, L.K. Polzik, N.V. Kocharova, Yu.A. Nemets and A.V. Novichikhin, Spectrochim. Acta Part B, 47 (1992) 1187) that linearized calibration curves in Zeeman graphite furnace atomic absorption spectrometry by taking into account the presence of stray light. The calculations of L'vov and co-workers were based on three parameters: the rollover absorbance Ar, Zeeman sensitivity ratio R, and the original background corrected peak absorbance values Az. In order to simplify the calculations, R was assumed to be unity. In the studies reported here, this simplification is shown to be unsatisfactory because the slope obtained in the upper portion of the calibration curve, after linearization, is found to be different from the slope obtained in the normal linear region. Deviations between these slopes were found to be as high as 30%. The present work also shows that the theoretical model of L'vov and co-workers does not have a mathematical solution at high values of Az. This failure of the model prevents its use at high Az values. The physical nature of this failure is still unclear, which points to the necessity for further work to understand the inadequacies of the present theory. In the present studies, calculations based on the Newton method of successive approximations (A.I. Yuzefovsky, E.G. Su, R.G. Michel, W. Slavin and J.T. McCaffrey, Spectrochim. Acta Part B, 49 (1994) 1643), allow incorporation of the experimental value of R at the rollover point R′, which better linearizes the calibration curves. By use of this approach, a satisfactory result is obtained for lead (R′ = 0.67) up to the point of failure of the model at high values of Az. |
Full Text from ScienceDirect