In this technique (TMA), dimensional changes in a sample are the primary measurement, while the sample is heated, cooled, or studied at a fixed temperature. A simple schematic figure of a typical TMA instrument is shown below.

The sample sits on a support within the furnace. Resting upon it is a probe to sense changes in length, which are measured by a sensitive position transducer, normally a Linear Variable Displacement Transducer (LVDT). The probe and support are made from a material such as quartz glass (vitreous silica), which has a low, reproducible, and accurately known coefficient of thermal expansion, and also has low thermal conductivity, which helps to isolate the sensitive transducer from the changing temperatures in the furnace. A thermocouple near the sample indicates its temperature. There is usually provision for establishing a flowing gas atmosphere through the instrument, to prevent oxidation for example, and also to assist in heat transfer to the sample. Helium is effective in this respect.

The load may be applied by static weights, as above, or by a force motor. This latter method gives the advantage that the applied load can be programmed to allow a greater range of experiments. The instrument is calibrated for position measurements by heating a sample whose expansion coefficient is accurately known. Aluminium is commonly used, in the form of small machined blocks or cylinders. Sample sizes are commonly around 5-10mm in height and width. It is important to prepare samples with clean, flat and parallel faces to avoid artefacts in the recorded curves.

There are a number of approaches to temperature calibration. One method uses small flat pieces of pure metals, e.g. indium, sandwiched between thin discs of quartz glass. Melting of the metal sandwich under the probe with a moderate load results in a sharp displacement of the probe. With care, a multi-layer sandwich can be built up using "fillings" of different metals, so that a multi-point calibration can be obtained in one experiment. The TMA curve for such an experiment is shown below.

When the sample carries a zero, or negligible load, the experiment follows the free expansion or contraction of the material, and accurate coefficients of thermal expansion can be routinely determined. Under an appreciable load, softer materials will be compressed, or slump at a glass transition, for instance, and the expansion measurements will be in error. Fortunately, modern instruments are able to apply loads that are sufficient to keep the probe in good contact with the sample, but small enough to avoid sample deformation. For troublesome samples, the ability to measure the underlying true expansion, is one of the advantages of the recent technique of Modulated Temperature TMA (MT-TMA).

Sometimes the required information is about the deformation of the sample, and several types of probe are used for different types of study. A range of typical probes, available with most instruments, is shown below.

The materials studied are usually rigid or nearly rigid solids, as implied by most of the experimental arrangements shown above. Liquids can be studied in the dilatometer-type accessory, which can also be used to measure the volume changes in an irregularly shaped samples, or powders, which are then submerged in an inert liquid such as a silicone oil. Powders can also be studied as a layer with a loosely fitting lid on top, or when pressed into a pellet. Solid-state transitions can be followed in this way.

The major application areas of TMA are in the polymer field. A list of the main types of experiment and the information obtainable is given below.

While many other techniques exist for studying the important and informative region of the glass transition (Tg), TMA offers advantages for certain types of study. The indentation and penetration probes for instance, can follow transitions in very thin films, such as lacquer coatings on metals. Although quantitative mechanical properties (modulus) can in principle be derived from TMA measurements, this may be difficult in practice, and Dynamic Mechanical Analysis may be preferred. TMA is better suited to comparative measurements on a range of materials, and for measurements of transition temperatures and expansion coefficients on relatively small samples, in a conveniently short time.

The following figures show some results obtained from a filled epoxy resin. As well as expansion data, which can be determined with high accuracy, TMA is seen to be a sensitive method for determining the glass transition temperature of such materials. The change in heat capacity at Tg for a well-cured resin is small, and conventional DSC gives generally poor results in this region. MT-DSC is far better, (and indeed a valuable complementary technique) but the experimental time is much greater.

The curve A shows the dimensional changes around Tg when too great a load is used. The sample slumps after entering its softer state, making measurements doubtful. Curve B was recorded under a negligible load, and allows Tg to be measured as the intersection of the portions of the curve above and below Tg.

The next figure shows data for a sample of the same material that has been physically aged, by being maintained at a temperature close to, but below Tg. The ageing of the epoxy in its glassy state results in shrinkage. On going through Tg on heating, the shrinkage is reversed, and the material shows a sudden expansion. On cooling the sample, the curve follows that for the unaged condition, and the difference in length allows the shrinkage on ageing to be quantitatively measured.