Improving Accuracy in Graphite Furnace Atomic Absorption Spectrometry Through Peak Shape Monitoring

Due to the good detection limits of graphite furnace atomic absorption spectrometry (GFAAS), situations frequently arise when an analyte is determined in the presence of a large excess of concomitants. It is in these cases when the improvements made in furnace design, heating rate optimization, Zeeman background correction, platform atomization and application of matrix modifiers are still not sufficient to provide the analytical chemist with reliable data. On the other hand, large gains in mechanistic insight have not yet yielded a model that can successfully cope with atomization characteristics in samples with a large matrix/analyte ratio. Typical accounts of the current state are found in [E] for explicit differential equations formulations and in 12] for the Monte Carlo approach, but there is no real difference in the predictive abilities of the two. Nevertheless, it is clear that a wide variation of peak shapes are observed and qualitatively monitored by everyone practicing GFAAS. In this work we describe a characterization of peak shapes by various parameters, such as appearance time, mean, mode, peak width, area/height, skewness and kurtosis. This is done by attempting to alter the peak shape systematically through deliberate addition of concomitant elements in a fractional factorial experiment [3,4]. The observed shape parameters are then linked to the characteristic mass after discarding those that vary with analyte mass. A Perkin-Elmer 5000 with HGA 500 and AS 40 autosampler was used throughout. The computations were run on a Perkin-Elmer 7500 data station and on an IBM PC-XT. All data given are Zeeman background corrected and all samples are atomized from a platform. Two different platforms were used. One was the vendor-supplied pyrolytic graphite platform, and the other one was produced from the same type by physical vapor deposition of TaC [5]. A general account of the analytical properties of this TaC-coated platform for ETAAS can be found elsewhere 163. All reagents were of analytical grade. Doubly distilled water and subboiled nitric acid were employed. Sn, Se, and Rh served as model analytes to study the behavior of elements different in both chemistry and volatility. Ten micrograms of each palladium and magnesium as nitrates were used as matrix modifiers for Se, and 200 gg ammonium dihydrogen phosphate was used for Sn. First, temperature programs were optimized for standard solutions and not altered for any of the subsequently described experiments. Generally, atomization temperatures from the TaC-coated platform tended to be higher than for the ordinary total pyrolytic graphite platform [5]. Calibration curves were run to establish the linear range, but-more importantly-the dependence of peak shape parameters on analyte mass. From this the concentration levels were chosen for the interference experiments to be well within the linear range. Cr, Fe, Na, Zn, Mg and Al were chosen as potential interferents and added as nitrates in a 29000 (Sn) to 40000 (Se) fold molar excess over the analyte. A fractional factorial design in these interferents was set up which was replicated once, giving a total of 64 profiles [41. These were recorded in random order with 2 atomizations of the standard solution and 2 atomizations of a blank solution before and after the run of 64.