One of the least appreciated and most important parts of thermal analysis is the preparation of the sample. Everyone teaches it in basic thermal classes but it seems to be the most forgetten part of the class. Some people do get it and take considerable care: most just seem to toss something into the instrument and assume that one run will be accurate. So let’s quick look at sampling and sample prep for the 4 main thermal techniques.
Sampling is a bit outside of this discussion but let’s state that the common approach to thermal samples is wrong. For some reason, labs that routinely run triplicates of GC or Karl Fischer will run one DSC or DMA sample and call it good. This is really short-sighted as the same concerns with sampling and statistics that effect those methods effect thermal analysis. In general, we need 3 specimens from a representative selection of the material.
Let’s start with mechanical specimens – those for TMA, DMA, or physical testing – first. Samples need to be of uniform sizes, with the same aspect ratio, parallel sides, 90 degree corners, and no burrs or lips on the specimens. Uniform size is important because the thermal lag in these specimens can be considerable at higher heating rates. In general, I prefer to heat at 2-3 C/min, althro one can go as high as 5-10 C/min depending on sample type and size. Remember we have to heat the clamps too and depending on the geometry, they can act as a heat sink.
For DSC, the normal rule of thumb is small pieces not filling the pan more than half way. This can be varied but remember the key is good thermal contact with bottom, where the sensor contact the pan. So films need to be squashed flat in many cases so they don’t curl away, large pellets need to be divided, etc. Keeping good thermal contact and a reasonable sample height to avoid thermal lags is the common theme. As DSC designs all have the sensors on the bottom of the furnace, keeping the bottom of the pan flat and in good contact is vital to decent data. Anything that moves too far away – large sized pieces, curled films, etc – decreases the quality of the data.
For TGA, it’s a little more forgiving. Unless one has a TGA/DSC (also called an STA), the weight is measured on the whole sample and larger samples with less thermal contact can be used. This allows accurate results with weights as large as five grams (if your balance design allows that to be handled accurately.) Larger samples may require slower heating rates to get good resolution. However, if one is just doing residual ash, it also may not. The general rule of thumb here is don’t fill the crucible more than 50% full.