The identiPol produces a number called the Quality Index Score (QIS). This is a score based on comparing the test sample against a previously produced reference set.

If, as a user, you have a material that you know produces acceptable quality goods, this can be used to start your initial reference set. In effect, you train the unit to recognise this material. The recognition that is occurring is based on comparing thermo mechanical data generated typically on your material from just above room temperature through the softening point or melt of your material. This effectively encompasses the mechanical and thermal behaviour of your material both in the solid form all the way through to the melt or plastic flow region of your material. As well as the QIS value (an effective pass/fail test), the unit can also give you a Tg (glass transition temperature), Tm (melting point) or an end temperature where the material as reached the flow point (useful for amorphous polymers without a melting point).

Typical display available to the user is shown here;
The reference set in the above example can be identified as those samples tested and shown with a green circle encompassing the point. Also shown is a running mean and the Tm of this material. What is behind the QIS value?
The above is typical of the ‘patterns’ obtained from the unit. This particular output is from the mechanical thermal spectrum. Shown are tan delta values vs temperature.

What is tan delta?

Tan delta is a nice way that scientists use to summarise the modulus and damping information from instruments known as Dynamic Mechanical Analysers (DMA). The identiPol has a strong relationship to these instruments, having been modified to cope with a QA / factory situation. In essence tan delta reflects the viscous and elastic properties of the thermoplastic as you increase temperature. In the above example, in effect we are seeing a region where the viscous nature of the plastic increases. At the same time, the stiffness of the material will also be decreasing. If required, this can also be inspected on the identiPol.

There are many choices of ‘patterns’ that the identiPol can use to produce a QIS, though the above is the most common and the normal default. In most cases, this reflects most behaviour required to differentiate samples. Other common outputs are a baseline corrected delta T. This is reflecting the energy required to alter the material as it is heated. Can be useful if the melting area is of particular interest. Stiffness can also be used. This should be used with a little more care but will reflect a good end temperature if amorphous material is examined and is useful for virtually all thermoplastics for this purpose. It is used as standard in the identification mode as is the baseline corrected delta T.

Now we have a set of patterns from our samples, what is happening?

Triton Technology Ltd., with assistance from Bristol University’s Chemometric Department, have developed special algorithms that effectively reduce the patterns to produce the QIS. This is done using a Chemometric technique especially developed for this type of data known as Distribution of Pairwise Intersample Distances (DPID).

In effect, we take, in the above example, every tan delta value at every temperature of the test and compare this with the comparable values from all the other tests in the reference sets. In effect an array of differences are produced and these differences are statistically assessed and converted into a QIS distribution. When an unknown sample is run, it is compared to this distribution and if it is outside at any point, it will score low compared to the reference set.

Other Chemometric ‘tools’ are available in the software for the more advance user/operator, allowing visualisation and comparisons of different batches/grades/polymers as required. See example below;