Data smoothing in stripping voltammetry by simplex function fitting

Stripping voltammetry (SV), in common with all trace voltammetric analytical techniques, suffers from poor signal-to-noise (S/N) ratios due to external interference pick up and inherent generation of noise in the electrochemical system. In addition to other methods developed to smooth noisy data (such as time and frequency domain filtering), optimisation methods can also be used to reject the noise present in electrochemical measurements. In this paper, an optimisation method, based on a modified Simplex procedure, is presented. The method is based on minimising the deviation (in terms of the mean square error) between the recorded data and a theoretical model of the data. In this case, the model was a response surface consisting of one or more peak-shaped functions of one of the following types: Gaussian (G), exponentially modified Gaussian (EMG), Lorentzian (L), combined Gaussian-Lorentzian (GL) and theoretical voltammetric peak (TVP). The method was successfully applied to stripping signals obtained for the trace quantification of riboflavin and for the simultaneous determination of zinc, cadmium lead and copper by SV.