Some typical experiments are detailed below, but first we must check the basics:

l  What is the goal of this measurement?

l  Do you know what the material is?

l  What geometry and sample size should you use for an ideal sample stiffness?

l  What temperature range should be used?

These are the most basic questions. Starting with the first, what data are required? If the value of modulus and tan δ are needed at room temperature, then this is a very quick and simple measurement. For samples >1 mm thick, three-point bending will almost certainly be the best geometry to use. It is easy to satisfy a low applied strain level to avoid any non-linear behaviour (typically keep dynamic strain below 0.1%) and strain levels can be varied, if required, by changing applied amplitude and also sample length (remember that the strain depends on the square of the sample length and thickness). For the second question, if the material is known, then the expected modulus value and possibly damping factor (tan δ) will be known. Therefore, question 3 can be answered: based upon the modulus, choose what stiffness you would like the sample to have and calculate the sample size to achieve this. Most software accommodates this, or advises on the valid modulus range for the geometry entered. Question 4 is easy to answer as room temperature data have been requested.

Now let us assume some different answers. First, say, we do not know what the material is. A guess can usually be made at the modulus, since we may well have an idea of the likely material. Is it a glassy polymer for example, in which case the modulus will be close to 5 GPa (at room temperature)? On the other hand, is it a metal? Here the likely modulus will be around 100 GPa. Alternatively, it may be a rubber and therefore its modulus will lie between 1 and 10 MPa (typical values for industrial rubbers). If you still do not know, measure it in three-point bending with a long free length for stiff samples or single cantilever bending, medium free length for flimsy samples (e.g. rubbers). This should give you a reasonably accurate figure.

Second scenario: now say the purpose of the measurement is to find the Tg and Tm for the material. If the material is known we can answer all of the questions. If not, we can see how to arrive at a suitable sample size from the procedure above. So we can go straight to question 4. Is the sample crystalline, glassy or rubbery? A metal would be crystalline and polymer could be all three depending on the temperature! It should be easy to guess what state the material is in, just from the feel of it. If it does not appear to be stiff, then it is probably a rubber and therefore testing should commence from at least −100◦C. To be on the safe side testing should continue until about 300◦C, as melting points are frequently in excess of 200◦C. Some DMAs allow experiment termination when the stiffness falls to a preset limit. This can automatically stop the instrument once melting has occurred and avoids decomposition of the sample. Once Tg and Tm have been established a sensible range can be set for comparative measurements on other samples. Suggest at least 50◦C below Tg for the start temperature and about 10–20◦C above the melting point. More information can be gathered if two frequencies are used (1 and 10 Hz for example) as this allows easy distinction between Tg and Tm via the frequency dependence of data.

The above is only an illustration of likely scenarios. However, it is often useful to perform a wide-range thermal scan (−100–300◦C, or the melting point, if known) at two frequencies (1 and 10 Hz for example). This allows easy distinction between Tg and Tm via the frequency dependence of data. Generally, the experimental methods are best considered with real data. However, a brief description of experiment types is given below.

Thermal scanning experiments

Thermal scanning experiments are the most common ones performed with DMA and, based on my personal experience with users, I estimate that a simple, single frequency, single strain amplitude test accounts for some 80% of all DMA (Dynamic Mechanical Analysis) measurements made. Such an experiment will yield the key identifying parameters for any material. Figure 1 shows the tan δ curve for LDPE and LLDPE which exhibit the features shown by semi-crystalline polymers. Starting from high temperature and working downwards, we start with the melting point, a crystalline relaxation process α around 90◦C, a β relaxation at −20◦C due to chain branching and finally the γ relaxation at −120◦C. Transitions are shown as peaks in the tan δ trace.

Figure 1. Tan δ for various polyolefins. Transitions are shown as peaks in tan δ which increases as the sample is heated into the melt region.

Isothermal experiments

These experiments are mostly performed where the sample changes as a result of a stimulus. Examples of this may include the drying of a sample, perhaps as studied in a controlled humidity environment, the post-cure of a thermoset resin or the decomposition of a sample at high temperature. There is usually little restriction with regard to data collection for these experiments, as they are made over relatively long time periods. Therefore, they are frequently performed at multiple frequencies and this can be used to evaluate half-lives or relaxation times for the process under investigation.

Frequency scanning experiments

These experiments have a specific utility in the generation of materials data, for example for sound damping applications. In such cases, it is important to know the modulus and damping factor at certain frequencies and temperatures. Considerations must be given to ideal sample size and any errors that may spoil the measurement accuracy. Note that many DMAs cannot measure above the resonant frequency. Some DMAs will have higher resonant frequencies than others; the mass of the driveshaft is the main parameter that will determine this. The lighter the driveshaft the higher the resonant frequency. Some can measure above the resonant frequency anyway, thereby extending the useful range of measurement. However, it is usually best to avoid measurement at or near this value as results can be significantly in error.

Step-isotherm experiments

These are a combination of the isothermal and frequency scanning experiments. They are commonly used for detailed investigation of the glass transition process using time–temperature superposition (TTS) analysis. As such this is the most complicated experiment to perform, especially if using tension mode with thin films, for example. It is only recommended for experienced users of DMA and these experiments frequently last for 24 h. There is also much consideration needed for the sample and whether or not it is suitable for TTS analysis. Only samples that remain in the same state throughout the measurement and have a clearly identifiable single relaxation are suitable. This immediately excludes a large group of materials, in that semi-crystalline polymers rarely satisfy this condition. There are data-quality checks that can be made before TTS analysis, such as the wicket plot. It is highly recommendable to make a simple multi-frequency temperature scan before performing such long tests as this can show problems that occur with the sample and this avoids much lost time if the desired experiment needs modification.

Creep – recovery tests

DMA can be used to make creep, creep-recovery and sometimes stress-relaxation experiments. In some respects these tests defeat the purpose of a DMA, in that generally the dynamic data (E‘ , tan δ) are more useful and give greater insight into the materials’ mechanical behaviour than these constant load tests. However, they can be useful where a component is used at a specific temperature and load, e.g. a fixing lug, and it may be useful to test the material under the exact loading conditions to be used in the application.