Many pharmaceutical materials exist in a range of different crystal forms. In this example, a sample of commercially available chlorpropamide has been heated at a slow rate and shows conversion from one form to another. The original crystal form melts, beginning around 120◦C, then recrystallises, as shown by the exotherm around 125◦C, and then the new crystal form melts around 127◦C. In fact, the complexity of the second melt suggests that more than one form is melting in this region.

One of the uses of DSC is to confirm which crystal form of a drug is present. Since the melting point of a particular crystal form is specific to that crystal form, different polymorphs can be identified from their differing melting points. In some cases, melting points may be very close together so that distinguishing them requires good resolution, and one peak may appear as a shoulder on another, but this information is essentially available from observation of the melting profile. However, in a situation where a recrystallisation event has occurred, as with chlorpropamide (Figure 1), it is not possible to be sure whether some of the higher melting forms were present initially or not. Clearly some of the initial form is recrystallising, but it is not possible to tell from the slow scan whether all of the higher melting form resulted from the recrystallisation.

Figure 1. Chlorpropamide heated at 5◦C/min showing melting, recrystallisation and then further melting. This is a classic example of polymorphism in a pharmaceutical material as observed by DSC.

With increasing scan rate however, it may be possible to scan sufficiently fast such that recrystallisation is prevented and the analysis of the original material can be observed. Figure 2 shows chlorpropamide heated at 300◦C/min.

Figure 2. Chlorpropamide heated at 300◦C/min. The heat flow trace and first derivative are shown. This shows one single melt. The fast scanning rate has prevented recrystallisation and allowed confirmation of the purity of the original crystal form.

At this scan rate, the material is unable to recrystallise and only the melting of crystal forms originally present will result. If only one form is found in such a situation it can be concluded that only one form was present to begin with so the material was pure with regard to crystal form. The second derivative trace can help when viewing a melting profile and, in this case, helps to show that there is only one transition.

It is sensible to consider the effect of increased thermal gradients at increased scanning rate, since melting peaks not only get bigger with increasing scan rate, they also get broader, so the question arises as to whether the lack of two or more peaks is just the lack of resolution. The effectiveness of this approach relies on the fact that there is a sufficient difference in the melting point between the forms in order to distinguish them. Decreasing the sample weight in order to improve resolution is an important aspect as it is in all polymorphic studies. Another aspect to look at is the trends that occur as the scan rate is increased, and often the higher melting forms can be clearly distinguished to the point that they are no longer formed. If pure samples of different forms are available it may also be possible to prepare a mixture to see how the different forms can be distinguished. If a particular crystal form is present to begin with, then in general it will become more obvious as the scan rate is increased since the peak size will increase. The same is true with any transition. Figure 3 shows an example with carbamazepine. A sample of a commercially available material was run at increasing scan rate and it can be seen that the lower temperature transitions become much more obvious as the scan rate increases. In general, if more than one form is present it will be distinguishable at high rates, though note that rates fast enough to prevent a particular recrystallisation may not always be achievable.

Figure 3. Commercially available carbamazepine. Increasing scan rate shows increasing size of major transitions.

Different crystal structures can be identified by their melting points, so in principle by scanning a sample in a DSC the different crystal forms present can be distinguished. Whereas in the above application the aim is to prevent recrystallisation, when recrystallisation does occur it may be of interest to characterize the forms produced, and so identify a number of different polymorphs. By scanning at a range of different rates, different forms may be produced depending upon the kinetics of the crystallisation reaction. Therefore, it can be informative to scan a sample from slow to high rates to see what transitions can be found.

In addition, cooling a sample at different rates can also be useful. Not all samples recrystallise, but those that do may form different crystal forms depending upon the cooling rate. These can be examined upon reheating, so the analyst has a range of experimental permutations to choose from for the initial heating, cooling from the melt, and reheating. Figure 4 shows chlorpropamide on reheating after cooling at different rates; note how the ratio of melting peaks changes as a function of cooling rate.

Figure 4. Chlorpropamide reheated at 50◦C/min after cooling at 10◦C/min (solid line), 20◦C/min and 50◦C/min. Rate of cooling can influence the polymorph formed.

Where a material does not recrystallise on cooling or where the cooling rate is sufficient to prevent any recrystallisation, it may cool into a glass, and the Tg can be measured on heating. Subsequent crystallisation may occur with these materials and melting measurements can then be made.

On some occasions, melting and recrystallisation can occur almost simultaneously so that at a slow heating rate neither the melting nor recrystallisation processes may be distinguished, or even noticed. As with the carbamazepine sample in Figure 3 small, apparently unnoticed, overlapping transitions can be separated or enhanced into much larger, clearly measurable, transitions at higher rates. Therefore, it is worth checking materials at high scan rates to see what additional information may be available. The resulting information is more likely to tie in with other techniques, such as X-ray crystallography, because at high rates the sample does not have time to change and true values can be measured.

Measurements of the glass transition have proven to be greatly enhanced at high scan rates due to the increase in sensitivity, and it is possible to study this transition in a wide range of materials where the Tg could not previously be measured; in particular those with only a small percentage of amorphous content where the Tg is very small, possibly as low as 1%. The need to analyze for low amorphous content arises for a variety of reasons: for example, development of materials and the effect of manufacturing processes such as milling, which is used to produce the desired particulate size. There is concern that these processes could increase the non-crystalline content, so there is need to measure it. There are already a number of well-defined methods for doing this, for example vapour sorption or solution calorimetry, but these can be very time consuming and not all methods will work well for a given material so there is need for a further rapid method.

A study has been performed with lactose which shows that low amounts of amorphous content can be quickly detected by HyperDSC. Fast scanning rates greatly increase sensitivity, Figure 5, and by measuring the height of the glass transition a linear relationship has been shown between height of Tg and per cent amorphous content, Figure 6. Thus, if the relative height of Tg can be determined this can be related to the amorphous content.

Figure 5. Tg of spray-dried lactose measured at increasing scan rate from 100◦C/min (bottom curve) to 500◦C/min (top two curves). The faster the scan rate the larger the Tg.

Figure 6. Height of Tg as a function of amorphous content of spray-dried lactose mixed with crystalline a-monohydrate. This shows a linear relationship between height of Tg and amorphous content.

Materials which are 100% crystalline should have no glass transition, so if a sample thought to be 100% crystalline is scanned at high rates by DSC and a Tg is found, this immediately means that there is some amorphous material present which needs further investigation. A measurement like this need only take 2 or 3 min to prepare and complete. To obtain a quantitative value, the height of the Tg needs to be calculated and compared with that of a known standard. A value for a 100% amorphous material can frequently be obtained by cooling the material rapidly from the melt into its amorphous state and then reheating it. This is best done in the analyser since taking a sample out, even to drop into liquid nitrogen, is sufficiently uncontrolled such that crystallinity can occur. The sample needs to be cooled as rapidly as possible through the Tg region and into the glassy state both to avoid crystallinity and enthalpic relaxations on top of the Tg which cause difficulties for accurate measurement. This approach requires that the material does not decompose on melting, nor that any crystallisation occurs. This can be checked by subsequent melting.

A practical approach for measuring small amounts of amorphous content:

1. It is always helpful to know as much about a sample as possible, and in particular the Tg region in question. If this is not known, then cool a sample from the melt rapidly into the glass and reheat it as described above. The sample mass should not be so large that it will seep out of the pan when melted.

2. Sample mass generally will need to be maximized in order to make the best measurements since glass transitions are usually very small. Use a large pan and compress to give good thermal contact. Heat as fast as possible through the Tg region, but do not melt, as large sample masses may cause contamination issues when melted. Make sure that the pan is vented to allow volatiles to escape and to prevent pan distortion under pressure. Make sure that the sample has fully equilibrated in the purge gas atmosphere before the start of the run, this may take a few minutes depending upon the flow rate.

3. To calculate the height of the Tg use a consistent method. One approach would be to normalize the data (this divides by the sample weight) and slope the curve before Tg so that the baseline approaching Tg appears flat. Any variation in the position of the start of the calculation has no effect. Select the height from the baseline to the maximum of the step and compare the height of Tg from the as-received sample with the height measurement from a sample of known content, typically a sample previously cooled from the melt into a glass.

One study looking at the measurement of undissolved drug in a matrix showed that high scan rates are essential for accurate measurement. This study reports that low heating rates allow further solution of the drug in the matrix, whereas fast rates permit measurement of the melting profile of the drug without further solution occurring. The amount of undissolved material can be obtained from the enthalpy of melting. The same approach can be used with foods where, for example, the amount of undissolved sugar needs to be measured.