Initial work with polymer melt performed by Pijpers and Mathot shows that the melting profile of polymers is not only measurable at fast rates, but also it is often different to that obtained on slow heating because of annealing processes. Melting of any material usually gives a fairly large peak since a lot of energy is involved in the bond-breaking process, and there are concerns about how quickly this energy can be transmitted to the sample and the effect that this would have on the resolution obtained. Accordingly, low sample weights may be required if melting processes are to be observed at high rates. This may well be of significant advantage as in the case of films adhering to a substrate. Figure 1 shows a trace from a multilayer film with a large number of peaks all well resolved, indicating clearly the ability of HyperDSC both to heat fast and to retain good resolution.

Figure 1. This shows the ability to distinguish a range of polymers of a multilayer film from their melting profile. In this case, the polymer is a thin film coated on the surface of a substrate from which it cannot easily be removed, heated at 150◦C/min. At low rates sensitivity is insufficient to make these types of measurements.

Not only can the melting points of the constituent polymers be measured accurately, but the fact is that the increased sensitivity obtained at higher rates allows the melting process to be observed easily, whereas at low rates little would be seen.

Intuitively, there is an expectation that the faster a sample is heated the higher will be the observed melting point of the material due to an increase in thermal lags or thermal gradients across the sample during the melting process. However, this is not necessarily what happens. Annealing processes can allow growth of larger, more stable higher melting crystallites at the expense of less stable lower melting structures. This means that during slow heating the crystal structure of a material can change to give a higher melting point and narrower melting range than that of the material originally put into the DSC. The situation is even more evident if a polymer actually recrystallizes on heating as is shown in Figure 2 where nylon 10.6 is analyzed. In this case, the sample undergoes cold crystallization during slow heating; a process that can be prevented by fast heating. Therefore, the lower trace in this figure more accurately represents the melting of the original material.

Figure 2. Nylon 10.6 heated at 10, 100, and 200◦C/min. The faster the rate the less time the sample has to recrystallize and the trace is more representative of the original material put into the pan.

It is not possible to suggest any particular scan rate as being the best to use. The rate needed to prevent crystallization from occurring and the true structure of the sample to be measured may vary from sample to sample. A rate should be sought where the profile does not change as a function of scan rate as this indicates that a sufficiently fast heating rate has been used. In the example in Figure 2 a rate of 200◦C/min is fast enough to prevent an obvious exotherm from being observed, but that does not mean that all changes have been prevented, and it may be of value to examine faster scans to get a complete picture.

The rate of cooling is also important for examination of crystallization processes and determination of material properties. The slower a material is cooled from the melt the more time for crystallization to occur and the greater the extent of crystallization that will result. Thus, an examination of increasing cooling rate and the need for fast cooling rates is important to both characterise and condition a material. For most thermoplastics it is important to heat, cool and reheat. So what cooling rate should be chosen? The answer is that a range of cooling rates may be useful as in the characterisation of PET shown in Figure 3. To obtain these data, the heating rate chosen (300◦C/min) was fast enough to prevent changes occurring during heating. The differences in the melting profiles of PET are due to the different cooling rates employed which permit different levels of crystallinity to develop. Some of these profiles show distinctly binodal characteristics (see the curve following a 5◦C/min cool) something not seen if slow heating rates are chosen because annealing alters the crystalline structure and the binodal characteristic disappears. In this way the material has been fully and accurately characterised. In Figure 4 the height of the glass transition and the heat of fusion of the melt are plotted as a function of cooling rate and this gives a view of the extent of crystallinity of the material. These measurements cannot be made at low rates since the material will recrystallize. Though PET could be considered to be a well-known and well-characterised material, fast heating rates can further accurate information.

Figure 3. PET heated at 300◦C/min after cooling at a range of different rates indicated. Increasingly slow cooling allows increased crystallinity, which can be measured with fast heating rates. Slow heating would allow further crystallisation so that all the traces would become similar.

Figure 4. The height of the Tg and heat of fusion measured on heating PET at 300◦C/min after cooling at different rates. The relative height of the Tg indicates that the sample of PET is still 60% amorphous even after cooling at 1◦C/min.

Not only are the results of a fast or a slow heat of value but so are the trends that are observed as the material experiences different scan conditions. The cooling trace observed may also be of significant value; sometimes blends give greater resolution between crystallisation events during cooling than is found during corresponding melting events.

It should also be evident that measurements of the extent of crystallinity could be affected by slow scanning and the occurrence of annealing. Measurements of the extent of crystallinity are normally made from the heat of fusion of a material, and these may be better made at high rates before the crystalline structure changes. This may also apply to enthalpic methods which seek to give a value for the temperature dependence of crystallinity, by comparing the heat capacity of a material with known standards at a given temperature. These methods require that the analyser be fully quantitative for energy measurement at high rates, which modern power compensation systems have proven to be. Figure 5 shows an example of the measurement of crystallinity in PET. A small amount of crystalline structures can be observed even though the sample is expected to be fully amorphous after quenching in liquid nitrogen. To quench-cool the sample, a pan with molten PET was taken from a DSC furnace and dropped into liquid nitrogen, but this process was clearly not fast enough as the existence of a crystalline melt measured by the fast scan shows.