SIGNAL ENHANCEMENT OF POTENTIOMETRIC STRIPPING ANALYSIS USING DIGITALSIGNAL-PROCESSING

Four digital signal processing methods, moving average smoothing, poly nomial smoothing, rectangular low pass filtering and exponential low p ass filtering have been compared in this study of potentiometric strip ping analysis (PSA) data. For the filtering methods, a new procedure i s introduced based on treating the potential coordinate as mathematica lly equivalent to a time coordinate to transform data to the frequency domain. After Fourier transformation, each method has yielded signal- to-noise (S/N) improvement over the original data, for example exponen tial low pass filtering providing a 2.3 times S/N enhancement. Peak ar ea measurement was unaffected by the two smoothing methods, while an u ndesirable increase in peak area was observed in the two filtering met hods. A linear peak area-concentration relationship was retained in al l cases. The original peak height calibration is poorly correlated lin early, although the complex curve is reproducible. However, after digi tal signal processing, peak height as the measuring parameter gave a g ood linear relationship to concentration. The filtering approach which allows optimum filtering and best peak resolution was readily establi shed. The resolution of overlapping potential peaks, using copper and bismuth peaks as the example, was best improved using the rectangular low pass filtering method.