Differential scanning calorimetry (DSC) is the most widely used of the thermal techniques available to the analyst and provides a fast and easy to use method of obtaining a wealth of information about a material, whatever the end use envisaged. It has found use in many wide ranging applications including polymers and plastics, foods and pharmaceuticals, glasses and ceramics, proteins and life science materials; in fact virtually any material, allowing the analyst to quickly measure the basic properties of the material. Many of the application areas are dealt with in greater depth within DSC, and the principles involved extend to many other materials that may not be mentioned specifically. It is in fact a fascinating technique and the purpose of this introduction is to provide an insight into this method of measurement, to provide the necessary practical guidance a new user will need to go about making measurements, and to give understanding about the information that can be obtained and how to interpret the data.

A DSC analyser measures the energy changes that occur as a sample is heated, cooled or held isothermally, together with the temperature at which these changes occur. The energy changes enable the user to find and measure the transitions that occur in the sample quantitatively, and to note the temperature where they occur, and so to characterise a material for melting processes, measurement of glass transitions and a range of more complex events. One of the big advantages of DSC is that samples are very easily encapsulated, usually with little or no preparation, ready to be placed in the DSC, so that measurements can be quickly and easily made.

The main property that is measured by DSC is heat flow, the flow of energy into or out of the sample as a function of temperature or time, and usually shown in units of mW on the y-axis. Since a mW is a mJ/s this is literally the flow of energy in unit time. The actual value of heat flow measured depends upon the effect of the reference and is not absolute. What matters is that a stable instrumental response or baseline is produced against which any changes can be measured. The starting point of the curve on the y-axis may be chosen as one of the starting parameters, and it should be set at or close to zero.

Two different conventions exist for the display of the heat flow curve: one shows endotherms in the downward direction, the other upward. The operator has a choice with most software packages. Traditionally, with heat flux systems endotherms are shown as going down, since endothermic transitions result in a negative temperature differential, whilst with power compensation systems they are shown as going up since with this principle endothermic transitions result in an increase in power supplied to the sample. The value of measuring energy flow is that it enables the analyst to identify the range of different transitions that may occur in the sample as it is heated or cooled.

The specific heat (heat capacity, Cp) of a material can be determined quantitatively using DSC and is designated Cp since values are obtained at constant pressure. Traditionally, this is done by subtracting a baseline from the heat flow curve, but values may also be obtained using modulated temperature techniques. The subtracted curve referenced against a standard gives a quantitative value of Cp, Figure 1. The accuracy that can be obtained depends upon the instrument and method in use.

Figure 1 Heat capacity of PET obtained using fast scanning techniques showing the three traces required for subtraction. The height of the sample compared to the empty pan is divided by the scan rate and the mass of sample to obtain a value for Cp. This is referenced against a known standard such as sapphire for accuracy. If small heating steps of, for example, 1◦C are used the area under the curve can be used to calculate Cp. This calculation is employed as an option in stepwise heating methods.

In practice the traditional standard test method provides a fairly rapid method for determination of Cp and many manufacturers provide software specifically designed to comply with this. Three runs are required, each consisting of an isothermal period, temperature ramp and final isotherm. This method is applied identically to the succeeding runs:

1. First run: a baseline with uncrimped empty pans placed in the furnace.

2. Second run: as above but adding a reference (typically sapphire) to the sample pan.

3. Third run: replace the reference with your sample.

The three curves are brought up on the screen, isothermals matched, data subtracted and referenced against the standard. Most software packages will do this automatically, and if the differing weight and heat capacity of sample pans are taken into account then the baseline and reference runs may be used for subsequent samples, provided the DSC is stable. In fact, because the procedure is based on a subtraction technique between measurements made at different times, any drift will cause error. The DSC must be very stable and in practice it is best not to use an instrument at the extremes of its temperature range where stability may be compromised. The standard most often used is sapphire, and the mass used should be similar to the sample; in any event the sample should not be a great deal larger or errors will be increased. This method relies on the measurement of the heat flow of the sample compared to that of an empty pan. Whilst there may be a number of factors which dictate the scan rate of choice it should be noted that faster scan rates result in increased values of heat flow giving increased accuracy of measurement, and this also minimises the time of the run and potential drift of the analyser. It has been reported that fast scan rates used by fast scan DSC can give extremely accurate data. A similar principle is employed in stepwise heating methods where the temperature may be raised by only a fraction of a degree between a series of isotherms. This is reported to give a very accurate value for Cp because of the series of short temperature intervals. Specific heat data can be of value in its own right since this information is required by chemists and chemical engineers when scaling up reactions or production processes, it provides information for mathematical models, and is required for accurate kinetic and other advanced calculations. It can also help with curve interpretation since the slope of the curve is fixed and absolute, and small exothermic or endothermic events identified. Overall, it gives more information than the heat flow trace because values are absolute, but it does take more time, something often in short supply in industry.

The enthalpy of a material is energy required to heat the material to a given temperature and is obtained by integrating the heat capacity curve. Again many software packages provide for the integration of the Cp curve to provide an enthalpy curve. Enthalpy curves are sometimes used for calculations, for example when calculating fictive temperature, and can help in understanding why transitions have the shape they do. In the cases where amorphous and crystalline polymer materials exhibit significantly different enthalpies, the measurement of enthalpy can allow an estimate of crystallinity over a range of temperatures as the polymer is heated.

Derivative curves are easily obtained from the heat flow curve via a mathematical algorithm and aid with interpretation of the data. Typically they can help define calculation limits, and can aid with the resolution of data, particularly where overlapping peaks are concerned. The first derivative curve is useful for examining stepwise transitions such as the glass transition, and is very useful for thermogravimetric analysis (TGA) studies where weight loss produces a step. The second derivative of a peak is more easily interpreted than the first derivative. In this case the data are inverted, but any shoulders in the original data will resolve into separate peaks in the second derivative curve. It is particularly useful for examining melting processes to help identify shoulders in the peak shape due to multiple events. An example is shown in Figure 2. The second derivative produces a maximum or minimum for each inflection of the original curve. Shoulders in the original curve show up as peaks in the second derivative. The shoulder in this example is quite clear, but often the second derivative can pick out multiple events when the original data are much less clear.

Figure 2. Indomethacin form 2 scanned at 500◦C/min. The shoulder on the melt resolves into a separate peak in the second derivative, which shows a doublet pointing downwards.

The higher the derivative level the greater is the noise that is generated, so good quality data are needed for higher derivative studies. Curve fitting or smoothing techniques used on the heat flow curve before generating a derivative can also be very useful. In general, sharp events and inflections produce the best derivative curves. Studies at high rate also produce very good derivative curves since the rates of change are increased.