ON THE USE OF THE EDGEWORTH-CRAMER SERIES TO OBTAIN DIFFUSION-COEFFICIENTS FROM TAYLOR DISPERSION PEAKS

Three methods of determining the parameters required for the calculation of diffusion coefficients in Taylor dispersion experiments from quasi-Gaussian output signals have been examined, the height-area method, moment calculation by direct integration, and moment calculation from a nonlinear least squares fit to a truncated Edgeworth-Cramer series. The method using moments determined by integration has been found to be unsatisfactory, as it yields diffusion coefficients very heavily dependent on the choice made for the width of the peak base. Those derived from the height-area method can deviate systematically and significantly from those determined from the variance obtained from the least squares series fit where, as is usual, the peak varies from a true Gaussian. The series fit method, within our reproducibility of 1-2%, yields values consistent with accurate and precise literature data, and is recommended for precise work.