Reliable determination of freeze-concentration using DSC

The objective of this study was to determine the feature of a DSC endotherm that can be most reliably used to determine the composition of a freeze-concentrate. Samples (3-10 mg) of sucrose in water (0-60%, w/v) were frozen and then heated (at 0.2-2.0 °C/min) on a DSC. The peak (Tpeak) and offset (Toffset) temperatures were obtained from the melt endotherms. A freezing point osmometer was also used to determine the freezing point depression of sucrose solutions. Two theoretical models were developed, one that describes melting in a one-component system (ice) with a heterogeneous temperature distribution, and a second model, which describes melting in a binary system (sucrose and ice) with a homogeneous temperature distribution. Modeling Laboratory (MLAB) was used to simulate melt endotherms using the two models. In Gray's theory for analysis of dynamic thermal measurement (1968), the end of the melting occurs at the peak of the endotherm followed by a much sharper recovery to the baseline than that observed experimentally. Our theoretical model for one- and two-component systems predicts that the melting continues beyond the peak resulting in a delayed recovery to the baseline as experimentally observed. Both experimental and simulated Tpeak and Toffset (also defined as the return to the baseline after completion of the peak) increased with an increasing scan rate. Extrapolation of the experimental Toffset to a zero scan rate yielded a value of the equilibrium melting temperature, Tm, which is closer to accepted literature values and osmometry data than did the extrapolation of Tpeak. Neither Tpeak nor Toffset reliably (at a finite scan rate) define the multicomponent freezing point. Rather, an extrapolation of Toffset to a zero scan rate gives the most reliable measure of freezing point using DSC. Simulation of DSC endotherms supports this conclusion and also exhibits the same trends as observed from experiment.