Use of first and second derivatives to accurately determine key parameters of DSC thermographs in lipid crystallization studies

In this study, an objective method to provide sufficiently unbiased analysis of raw DSC data, using the calculated first and second derivatives in combination with error analysis is described and applied to analyze thermograms of lipid samples. A statistical method based on a closer study of the residuals of fit and data correlation techniques with adequate criterion for the goodness of fit has proven to be valuable in evaluating the experimental error. The derivatives were utilized to define unambiguously the key signal characteristics, such as minima, maxima, as well as end and start points of thermal events, with uncertainties directly related to the actual noise of each signal. By estimating the pure experimental error and data correlation length, we have been able to have a better appreciation of the quality of the experimental data and detailed information on our DSC system. The calculated errors using 15 runs for the same sample were consistent with the estimated standard deviations generated by the statistical analysis.