Application of a Fast Deconvolution Method to HPIC Peaks

2 Introduction Resolution is one of the principal aims of all chromatographic techniques. Therefore, a major goal of chromatographic method development is to obtain an adequate separation of all components of interest in a reasonable time. However, in spite of ongoing instrumentation improvements, peaks are often poorly resolved and measurements must be made using deconvolution programs. In some circumstances area evaluation is still made by tracing a vertical line across the minimum point between two peaks and using this line as an integration boundary. Unfortunately, this method only gives good results if both peaks are symmetrical and have approximately equal areas. In ion chromatography (IC), peaks generally show some skewness. The form and extent of this skewness depend on both intra-column and extra-column effects, such as the injection volume and the time constant of the detector.1,2 Various methods of peak restoration and deconvolution have been proposed, the most recent of which will be considered here. The H point standard additions method3 uses areas obtained at two previously selected wavelengths as analytical signals. In order to use this method it is necessary to prepare different solutions containing the sample and the sample spiked with known amounts of analyte. Cao4 extracted information from overlapped peaks by comparing them with peaks of standard substances using a spreadsheet program (Microsoft Excel). The profiles of the standard peaks were magnified or reduced until the sum of their heights matched the convoluted profile at three selected points, corresponding to the retention times of the standards and to the intersecting point of the two overlapping peaks. Papoff et al.5 investigated three separate models: an exponentially modified Gaussian, a bi-Gaussian (two half Gaussians with different standard deviation), and the convolution of two Gaussian peaks. Calculations were performed using a commercial curve fitting program (Peak Solve v1.02, Galactic Industries). The parameters obtained allowed these authors to determine thermodynamic information, such as the stability constant of a Cd/Cl– complex. Torres-Lapasió et al.6 presented a program for the general treatment of chromatographic data. As it concerned the analysis of skewed or overlapping peaks, the algorithm was based on a Gaussian function with a polynomial standard deviation. The overlapped signal was considered as the sum of two deformed Gaussian functions and the problem was resolved in a non-linear fashion using a combination of the Powell7 and Newton Raphson Marquardt8 methods. Application of a Fast Deconvolution Method to HPIC Peaks