Two pure Ni cells for thermocouple calibration were fabricated in alumina crucibles with different internal structures to investigate their melting and freezing behaviors. One of the produced crucibles had a smooth surface, and the other was made with internal circular grooves. The melting and freezing processes of each cell were investigated in 10 melting/freezing cycles using a type B thermocouple calibrated with a standard radiation thermometer. The reproducibility of the melting point was better than that of the freezing point by a factor of more than 2, regardless of the internal shape of the crucible. The freezing temperature gradually decreased with the degree of supercooling, and the slope was observed to be about −0.08 • C. It was found that in terms of reproducibility and easiness of realization the melting point was better than the freezing point for use as a fixed temperature point. Therefore, it is proposed to use the melting point as a fixed point for thermocouple calibration. Regarding the difference in internal structure of the crucible, it was observed that a cell with circular grooves was slightly better in terms of reproducibility of melting and freezing points, but it was still unclear whether this was due to the grooves formed inside the cells, and, thus, additional research is needed. The average melting temperature and uncertainty of the Ni cells fabricated in this study was evaluated as (1454.1 ± 0.9) • C.