Hi

to find glass transition temperature you can opt DSC test method. It is so powerful than other methods.

Best

You can apply a heating rate of 2-5 C/min. and a frequency of 1 Hz. This is the most widely used procedure for temperature sweep tests in DMA. Thickness can be variable but it is related to the type and behavior of specimen and also the force limit of instrument. You should also be aware of testing geometry depending on the sample type.

DMA is a powerful technique to study the glass transition temperature (Tg) if you have enough amount of sample (say more than 200 mg). If the availability of your sample is on the order of milligrams (2-10 mg), then DSC is appropriate. However, DMA is a few orders of magnitude more sensitive than DSC in terms of Tg study. Especially, if you are interested in studying highly cross linked polymer, DSC may not show you clear Tg data. Thus, if sufficient amount of sample is available, I suggest using DMA. In the literature, you can find the Tg determination using both peak temperature of G" (or E") or tan delta peak. The rigorous definition of the Tg determination should be done with G" (E") peak temperature. However, depending on the sample, the baseline of G" (E") spectrum is not flat enough to determine the temperature precisely. Furthermore, tan delta peak position is not exactly the same (it can be away from G" peak as much as 10-40oC). Therefore, if you can determine the peak position in both G" and tan delta, you should use G" (or report both data). Since G" spectrum is a strong function of the sample geometry, having cracks and dimensional difference will crease data fluctuation. Exactly the same influence is exerted on to the G' data. Since tan delta is a ratio of G" against G', geometrical factor will disappear from tan delta spectrum and gives you very flat baseline which helps you to determine the peak position easier. This is the reason some people prefer using tan delta despite less rigorous definition of the glass transition determination. Temperature ramp rate depends on the thermal conductivity of the sample. If you have thick sample or highly thermally insulating sample, the temperature homogeneity suffers for fast heating rate. For this reason, we use 2oC/min for typical polymer and composite samples. The frequency of 1 Hz is quite typical. For shearing geometry, you will obtain G' (or G") whereas for tensile geometry you will obtain E' (E"). E' (E") value is typically greater than G' (G"). These two can be correlated if you know the Poisson's ratio of your sample.

Concerning heating rates i would advice to use reasonable low values especially if materials with low thermal conduction and thick samples are tested.

Two points to consider:

1. You will inevitably get a temperature gradient across your sample which can have an influence on your test result. Here the test mode has an additional influence as e.g. in bending the temperature of the outside layer has a stronger influence as e.g. in tension or compression.

2. If the evaluation of the measurement uses several single oscillations (e.g. 20) for a single datapoint to reduce noise of the data the thermal conditions change already considerably during these 20 oscillations for a heating rate of 5K/min. At 1Hz the temperature change would be already 1.2K for lower frequencies the influence becomes much bigger.

Both effects could be accepted if always the same (comparable) samples are measured for comparison (quality control).

But for the same material with and without a filler (e.g. carbon black in a polymer) there can be already an influence which could only be evaluated if different heating rates are used and compared.

The same applies for measurements of the same material with different thickness values.

There is one international standard that I am aware of, ASTM D5023, but I would also comment that I have not seen it refered to very often.

I would agree with much of the replies above, but would emphasise that DMA is much more sensitive than DSC and so DMA is very valuable for filled systmes where the polymer content is lower or for crosslinked systems where any Tg is less obvious using DSC.

I would add that using 2 frequencies as a standard method (such as 1 and 10Hz) can be useful since changes such as melting are not frequency dependant whereas Tg and beta transitions are frequency dependant, possibly giving useful additional informaiton for polymer research.