I think there may be a subtle difference in the sub-lattice positions of both two polymorphs of the compound, which may be relevant to the reorientation of H-bonding or van der Waals interactions crystallization procedure. This is also reflected by the slow II to I transformation as detected by DSC analysis.

But, why it is not reflected in PXRD spectrum, as we have not examined any peak of form -I in a month long studies?

Have you examined the variation of unit cell dimensions of form-II as a time? It may provide an evidence for some phase separation

A 5% may be a LOD of XRPD for this compound. Try other methods, such as solid state NMR.

The polymorphs are not structurally much different, but the DSC curves are showing two peaks corresponding to the melting point of the two forms. Moreover, there are reports of the same compound that form-II being unstable and tranforms to form-I.

I think that the two polymorphs may have crystal symmetries very closer to each other and also due to the slow transformation of polymorph-II, you could not notice any peak relevant to form-I. So if you run the XRD experiment once again at lower scan speed and smaller step interval, you may get some evidence.

Thank you for your suggestion. I will run the PXRD once again as suggested by you.

Did you try variable temperature XRPD? Also, have you tried running the DSC at diffferent rates and are there any differences because of scan rates?

Many times dynamic thermal techniques, such as DSC, gives apparently different result than isothermal techniques. The differences can be attributed to the way the anaysis is done or the changes that occur while the analysis is underway.

Since form II is known to be unstable wrt form I, it is possible that seeds of I have started appearing in the material at ambient consition itself. However, given the limit of detection for usual XRPD methods (unless it is specifically developed as a quantitative method), the presence of such small amounts of seed of I may not show up in XRPD analysis. However, such a material when analyzed by DSC behaves in a different manner. Since the seeds of form I are already present, the heating step in the DSC may provide enough thermal force for acceleration of the conversion to Form I within the timescale of the DSC run. Hence, the final observation are two endotherms. Differing the scanning rate in DSC may indicate whether this hypotheiss is correct or not.

Thank you for your suggestion.

Yes, I have analyzed DSC at different scan rates and found no change in result. We have not observed any peak of form-I initially (for two days), even at very slow rate. So if the heating step in the DSC is providing thermal force, then shouldn't we expect form-I peak from the beginning?

I have done calibration test taking 95% of form-II and 5% of form-I and made a physical mixture and performed XRPD. The result of XRPD of this calibration is of pure form-II. Can we tell from this calibration that 5% is the LOD of XRPD?

Since the rate of transformation is very slow (you said 5% in 40 days), analyzing a sample after 2 days may not show any perceptible difference from the original sample. Because of the same reason, you may not observe Form I peak in DSC in a sample which is stored for 2 days. Try running the sample which is stored for 40 days at different scan rates, preferrably as fast as possible. If Form I is present only as seed, then its peak is expected to be absent when sample is analyzed under fast heating conditions, and show up during slow scan rates.

To the LoD question, if you prepared a physical mixture of 95-5 and placed the "entire" sample on the holder making sure that the area of X-ray focus captures all the material, and still the resultant XRPD indicated no Form I, then definitely your LoD is greater than 5%., i.e., the method cannot detect form 1 even at 5%. A more involved quantitative method may be required to quantify the other form.

My suggestion is if the XRPDs of the two forms are similar and there are overlapping peaks, then you may want to try chemometric methods rather than conventional characteristic peak method for detection and quantification. If they have different characteristic peak, then definitely traditional appraoches can be used. But ensure homogeneity of the mixture while performing calibration. This is always a problem in solid phase analysis. You may try geometric addition rather than just mixing the two components at the same time to prepare the samples.

Otherwise, if you have the option, try spectroscopic techniques such as FTIR or Raman.

Good luck.

You may anneal you sample for accelerated transformation and then look for the XRD results.

Are you doing insitu XRD or only at room temperature. It may be possible that the transformed phases is reverting back to original one after cooling to room temperature !

It would be only by pure coincidence that two compounds might form crystals having identical XRD patterns due to identical three-dimensional spacing of planes in all directions. One example of accidental isostructure is the trihydrate phases of ampicillin and amoxicillin , but such instances are uncommon.

what‘s more, Is your polymorphism solvatomorph?  When substances formed solvated forms, it is observed that the different solvates could exhibit the same lattice structure, have the same PXRD but different  endothermic/exothermic transition temperature.

Above all based on Harry G. Brittain's monograph 'Polymorphism in Pharmaceutical Solids Second Edition'.