Terpy (MW = 233.27, >98.0%) was purchased from Tokyo Kasei and was purified by column chromatography followed by recrystallization using toluene/hexane (1:1 vol/vol) mixed solvent. Structure of thus prepared crystals was evaluated as orthorhombic form and was used for following polymorphic transformation experiment.

The starting terpy crystals, orthorhombic form, were ground for a given time manually using a mortar and pestle. The pre-determined grinding time is 0 - 90 min. About 10 mg of the sample powder was placed in an aluminum pan and differential scanning calorimetry (DSC) analysis was conducted on an EXSTER X-DSC7000 (Hitachi High-tech Science) under nitrogen atmosphere with a heating rate of 5˚C/min. Solid state transition temperature and melting temperature were determined by the intersection method. Powder XRD patterns of the crystals, sampled before and after a transition peak observed in the DSC, were measured by RINT2200 (Rigaku CuKα, λ = 1.5418 Å). Structure of the crystals was determined by comparing the measured XRD patterns with theoretical XRD patterns (orthorhombic and monoclinic) calculated using reported crystal structure data [2] [3] .

Figure 3 shows DSC profiles of terpy crystals, initially orthorhombic form, as a function of grinding time. Unground terpy crystals (see Figure 3(a)) reveal only one large endothermic peak at around 87˚C, showing melting of the orthorhombic terpy crystals at this temperature. In this case, no transformation from orthorhombic to monoclinic form was occurred. However, when the orthorhombic terpy crystals were ground for a given time, the peak patterns were drastically changed. For example, terpy crystals ground for 1 min (b) showed a broad and weak endothermic peak at 75.4˚C (estimated by intersection point method). This peak must be indicative of transformation from orthorhombic to monoclinic form. For confirmation, let us check the structures of the crystals sampled before and after the temperature (75.4˚C) by XRD. For XRD measurement of polymorphic substances, care must be taken for temperature change of crystal structures. Our XRD equipment does not have

temperature control unit and only measurement at room temperature is allowed. Since the high temperature phase (monoclinic in this case) has possibility to transform back into the lower phase (orthorhombic). Thus care must be taken especially when the higher temperature polymorph are measured at lower temperature. For terpy, fortunately, monoclinic form (high temperature phase) do no transformed into orthorhombic form just by cooling. Therefore room temperature XRD measurement is appropriate for evaluation of temperature change of structure of terpy crystals. Figure 4(a) and Figure 4(b) show typical XRD patterns of crystals sampled before (a) and after (b) the transformation temperature (75.4˚C). It may be clearly seen that XRD pattern (a) is agreed with orthorhombic form (d), while that of (b) agreed with monoclinic (c). From this experimental results, the small DSC peak appeared before the melting temperature may be concluded as transformation from orthorhombic to monoclinic form. It is interesting in Figure 3, the transformation temperature shift to lower temperature with increase of the grinding time. It is clear that the transformation temperature may be correlated with grinding time as plotted as Figure 5.

As shown in Figure 3 and Figure 5, the transformation temperature shifted to lower temperature with increase of grinding time. There are some possibilities to explain this result. One is crystal size effect. Crystal size distribution of the ground terpy crystals were measured and number averaged size of the ground crystals were calculated. The results are plotted against transformation temperature in Figure 6 (the numeric value in bracket refer to grinding time). In Figure 6, data were taken from another set of experiments with the same transformation procedure. Thus the transformation temperature in Figure 6 disagreed with that of Figure 5. At a glance, the results of Figure 6 looks simple, however, some interesting points may be found. For larger size crystals, e.g. larger than 5.7 μm, it must be clearly said that size of the crystals influences on transition temperature. In literature, size effect has been explained as surface effect [4] [5] . In case of nano-level particles, surface effect may be understand. However, our case is micrometer level. In such a large scale, surface energy effect is not so sensitive

to crystal size. Thus, the size effect found in this study for crystals larger than 5.7 μm must be caused by heat transfer.

Interesting point is found in smaller size region. In smaller size region smaller than 5.7 μm, increase of grinding time is less effective for particle size. However, the transformation temperature decreased with increasing grinding time. In this region, size effect may not be applied. Another factors must be existed. For another possibility, changes in crystallinity caused by grinding must be considered. As well-known, grinding of particles leads to formation of amorphous. Figure 7 shows typical XRD patterns for terpy crystals with different three

grinding time. Comparing the three patterns, no apparent difference is found in width of any peak. This suggests us that crystallinity difference is not occurred among the crystals.

Finally, two possibilities are remains. One is introduction of internal stress and/or defect that is believed to influence the transformation temperature. The other possibility is polymorphic transformation caused by grinding. In our previous study, we show existence of small amount of seed crystals of monoclinic form accelerate the transformation of orthorhombic form. Which factor is correct? To answer the question, we need more detailed examination. Anyway, small amount of internal stress and/or another polymorph that can not be detected by XRD measurement must influence the bulk transformation temperature. These results suggest us possibility of tuning a polymorphic transformation temperature without a help of additives.