Thermogravimetric measurements are influenced by various factors, such as the following:

1. Method parameters – heating rate, atmosphere (air, nitrogen, argon; pressure, humidity).

2. Sample preparation – sample size, homogeneity and morphology of the sample: coarse crystals, fine powder.

3. Choice of crucible

4. Instrumental effects such as buoyancy and gas flow. These effects can be reduced or eliminated by performing blank curve subtraction.

5. Changes in the physical properties of the sample during the measurement. For example, a change in emissivity (which affects the heat transfer within the sample and from the furnace to the sample) or the volume (which leads to a change in buoyancy).

6. This sample may ‘spit’ or move and artefacts caused by such events can be minimized by grinding the sample or covering with a platinum mesh.

Influence of heating rate

The systematic deviation between the true sample temperature and the measured temperature, which is heating rate dependent, can be determined and corrected through temperature calibration and adjustment. This is usually done using pure metals that have good thermal conductivity properties. Real samples (e.g. polymers) can of course exhibit quite different thermal conductivity behaviour. For this reason, the measured sample temperature will still be expected to show a slight dependence on the heating rate, even if the instrument has been properly adjusted. The effect is very small with the onset temperature, but is more pronounced with the peak temperature.

If the sample undergoes chemical reactions, the temperature region in which the reaction occurs is very much dependent on the heating rate. In general, higher heating rates cause reactions to shift to higher temperatures. The choice of the heating rate is particularly important if secondary reactions occur with starting temperatures that differ only slightly from each other. If unsuitable heating rates are used, the reactions may overlap and remain undetected. It is, however, often possible to separate different reactions by choosing favourable heating rates (in general lower, sometimes higher).

A quite different approach for separating overlapping reactions makes use of rate of change in sample weight to automatically control the heating rate: the faster the change in mass, the slower the heating rate. Nowadays, software is available that can control the heating rate in this way. This technique is generally known as sample controlled or constrained rate thermal analysis (SCTA); see Figure 1. Care must be taken when using these techniques not to produce artificial steps in the data.

Figure 1. Influence of the heating rate on the resolution of partial reactions. In the inserted diagram on the right, the dotted and solid TGA curves of copper sulphate pentahydrate were measured conventionally at 5 and 25 K/min, whereas the dashed curve was recorded using the sample controlled heating rate. In this presentation of mass against temperature, the steps in the curve appear to be nearly vertical because, at low heating rates, the reaction takes place almost isothermally. In contrast, in the mass against time presentation (main diagram), the shapes of the three curves at first sight appear similar. On closer inspection, the better separation obtained using sample controlled heating rates – especially in the first two steps – becomes apparent.

Influence of crucible

During the TGA measurement, crucibles must of course be ‘open’ to the atmosphere. It can be important, however, to seal the sample hermetically first, before the actual measurement is performed in order to prevent it from coming into contact with air. The lid of the crucible is then pierced immediately before the start of the measurement (e.g. in the sample changer).

Reactions in the gas phase proceed more rapidly in completely open crucibles than in a so-called self-generated atmosphere. In a sealed crucible with a very small hole in the lid, or in a crucible with a lid without a hole placed loosely over the sample, the weight loss is shifted to a higher temperature.

The material of which the crucible is made must not influence the reaction of the sample. In general, alumina (aluminium oxide) crucibles are used for TGA measurements. These have the advantage that they can be heated to over 1600◦C. Sapphire crucibles are even more resistant and are especially suitable for the measurement of metals with high melting points, such as iron, which partially dissolve and penetrate ordinary alumina crucibles at high temperatures.

Platinum crucibles have the advantage of good thermal conductivity, which improves DTA performance. Finally, platinum is not always inert. It has a catalytic effect and can, for example, promote combustion reactions.

Damage to platinum crucibles due to alloy formation with metal samples can be prevented by covering the bottom of the crucible with a very thin layer of α-aluminium oxide powder before inserting the metal sample.

Influence of furnace atmosphere

Clearly, the mass of a sample in a closed system remains constant and cannot be a function of temperature or time. Thermogravimetric measurements are only possible if the sample is free to exchange material with its immediate surroundings. An important requirement is therefore that the gas atmosphere surrounding the sample can be changed to suit the experimental requirements.

First, a protective gas is required to protect the balance against any corrosive gases that may be evolved. Typically, dry inert gases such as nitrogen or argon at flow rates of 30 mL/min are used but users should always refer to the manufacturer’s instructions. Besides the protective gas, a purge gas and/or reactive gas can be led into the furnace chamber via separate gas lines. The purge gas removes the gaseous reaction products from the furnace chamber. If helium is used as purge gas, heat transfer from the wall of the furnace to the sample improves, especially at temperatures below about 700◦C. Reactive gases can be delivered to the sample in order to observe the interaction of the reactive gas with the sample. Examples of reactive gases are air or oxygen (oxidation) or hydrogen (catalysis, reduction) diluted with argon (usually 4% hydrogen with 96% argon) to prevent the possibility of an explosion. Typically, flow rates of 30 mL/min are used for reactive and purge gases.

If high concentrations of hydrogen are being used then special care must be taken to remove the risk of explosions. Instruments where the furnace does not come into direct contact with the atmosphere can be useful for these measurements.

Influence of residual oxygen in inert atmosphere

Very often the question arises as to the amount of residual oxygen in the system. This can very easily be determined by measuring the combustion rate of activated carbon at 700◦C in the thermobalance.

For example, given a purge gas flow rate of 100mL/min a 5 ug/min weight loss corresponds to 90 ppm concentration of oxygen. A routine check might require the weight loss to be less than, say, 10 ug/min; see Table 1.

Table 1. Sources of residual oxygen and precautions

Influence of reduced pressure

Mass losses through vaporization or evaporation often occur at the same time as a decomposition reaction and are therefore difficult to distinguish from one another. The separation of the effects can often be improved by reducing the pressure in the measuring cell. A typical example is shown in Figure 2. The measurement at normal (atmospheric) pressure shows that mass is lost from about 320◦C onwards and the process possibly occurs in two steps. When the measurement is performed under reduced pressure at 1.5 kPa (15 mbar), the separation of the two steps is greatly improved.

Figure 2. Effect of pressure on the decomposition of an elastomer: sample A is measured at normal (atmospheric) pressure, and B under reduced pressure, 1.5 kPa (15 mbar). Under reduced pressure, the vaporization of the volatile components (additives, 1) is clearly separated from the decomposition of the elastomer (2,3). At normal pressure, the step height evaluated (on an expanded scale) from the stable region of the baseline at 270◦C to next DTG maximum at 380◦C does not correspond to the true additive content.

Even at reduced pressure, a purge gas must still be used to protect the microbalance against the condensation of possibly corrosive decomposition products. In vacuum operation, the vacuum pump is normally in continuous operation because the reaction products, air from possible leakages, and the purge gas have to be removed in order to achieve a constant vacuum. To obtain realistic values for the pressure in the furnace chamber, the pressure meter should be installed close to the furnace chamber and not in the vacuum line leading to the vacuum pump. The working pressure is typically in the range 0.1–10 kPa.

Use of reduced pressure may require recalibration of the temperature scale if high temperature accuracy is required.

Influence of humidity control

The interaction of water with many materials can have a significant impact on the material properties, for example plasticization of the glass transition temperature and stability. TGA measurements under controlled humidity are useful in studying the adsorption and desorption of water and for distinguishing bound and unbound water. Dynamic vapour sorption (DVS) is a similar technique and has its own class of dedicated instrumentation.

Special points in connection with automatic sample changers

Increasingly, automatic sample changers are being used. This poses particular problems since the samples on the turntable waiting for measurement must be protected against loss of volatile components such as moisture, and against the uptake of water or oxygen from the surrounding air. Ideally, the crucible should only be opened just before it is actually measured.

This can be done in a number of ways:

1. The sample is protected against direct contact with the surrounding atmosphere by placing an aluminium lid on top of the crucible. The lid is then removed and held by the sample robot’s gripper during the measurement. In this case, alumina or other high-temperature crucibles can be used.

2. An aluminium crucible is sealed with a perforable aluminium lid and then automatically pierced, or opened, immediately before insertion into the measuring cell. In this case, the maximum temperature of the experiment is limited to 640◦C to prevent melting of the crucible.

3. The carousel may be purged with a protective gas.

Inhomogeneous samples and samples with very small changes in mass

If the volatile content of the material is very low, or if the material is inhomogeneous, then clearly a large sample must be used. An idea of the sample mass required to detect small changes in mass can be obtained by considering the following imaginary experiment.

You want to determine an ash residue of 1% with an accuracy of about 1%. If the reproducibility of the blank curve is about 10 ug, then an ash residue of about 1 mg is required to obtain 1% accuracy. It follows that the quantity of sample required is 100 mg.