Dependence of Chromatogram Peak Areas Obtained by Curve-Fitting on the Choice of Peak Shape Function

Abstract

Seven different chromatographic peak shape functions are evaluated for use with nonlinear least-squares curve-fitting for asymmetric peak area measurements. The four-parameter functions that are evaluated are unable to adapt to peaks exhibiting substantial tailing, which results in peak area underestimates. Peak tailing is more readily represented by using six- and eight-parameter functions; however, the use of these more flexible functions can result in peak area overestimation and inaccuracies for overlapping peaks. The exponentially modified Gaussian function is advocated for general purpose chromatogram curve fitting because it adapts to extended peak tailing better than the other four-parameter functions evaluated and provides better reproducibility when large digitization intervals are employed or when poor initial curve-fitting parameters are chosen than the six- and eight-parameter functions that were evaluated.