Impurities in pure metal fixed points for the calibration of standard platinum resistance thermometers (SPRTs) causes significant variations in the freezing temperature, of the order of sub-mK to several mK. This often represents the largest contribution to the overall uncertainty of the fixed point temperature, and it is therefore of great interest to explore ways of correcting for this effect. The sum of individual estimates (SIE) method, in which the contributions of all the impurities are summed, is the recommended way of determining the correction if one has an accurate knowledge of the impurities present and their low concentration liquidus slopes. However, due to the difficulty in obtaining reliable in-ingot impurity corrections, it remains useful to investigate the influence of impurities on freezing curves using modeling techniques, and ultimately to parameterize the freezing curve by e.g. least-squares fitting to make corrections to the temperature of the freeze. Some success in analyzing freezing curves has been achieved. When parameterizing experimentally determined freezing curves, it is necessary to reliably determine the freezing end-point, and minimize spurious thermal effects. We outline some methods for meeting these requirements. As the influence of impurities is always convolved with thermal influences it is instructive to construct a model which takes into account both heat and impurity transport. We describe the development of more sophisticated models which take both these effects into account.