https://www.euro-online.org/enog/inoc2007/Papers/mac-slots.html https://www.euro-online.org/enog/inoc2007/Papers/m https://www.euro-online.org/enog/inoc2007/Papers/mac-slots.html

Getting hot with nylon

Nylon may always be synonymous with stockings and tights, but this polyamide is used to produce everything from carpets to conveyor belts to pipes to machine parts. In some of these uses, such as conveyor belts and machine parts, the nylon is likely to heat up, raising the risk that it might fail if it gets too hot.

Thus it would be useful to be able to monitor the temperature of these nylon parts, but inserting a thermometer into a working piece of machinery is usually not very practical. So chemists at the University of Iowa, US, led by Gary Small, decided to see whether NIR spectroscopy offered a way to monitor the temperature of nylon remotely.

To test whether this approach would work at all, Small and his colleagues first constructed a special instrument for heating sheets of nylon. This instrument comprised a brass plate with a square indent in one side, into which the nylon sheet is placed, and a large circular hole through it, exposing part of the sheet. Another plate, containing three thermocouple thermometers is attached to the back of the brass plate. The brass plate and nylon sheet are then heated while the exposed area is analysed by NIR spectroscopy and the temperature of the sheet monitored by the three thermometers.

When Small and his colleagues used this instrument to heat samples of a type of nylon known as nylon 6,6 to temperatures of between 21⁰C and 105⁰C, they found that the generated NIR spectra did vary in line with temperature. This was particularly the case for absorption at two wavelengths that most likely corresponded to the stretching of hydrogen bonds and changes in the structure of the hydrocarbon chains.

As reported in the Journal of Applied Polymer Science, they next used this data to build a partial least squares regression model to predict the temperature of nylon 6,6 from the spectral data. When using raw spectral data, however, they found that the resultant model wasn’t very accurate, especially over the long term. But if they processed the spectral data first to remove noise and unavoidable variation, they found that the models were much more accurate, able to predict the temperature to within 1.5⁰C even after several weeks.

Indeed, although these models were based on nylon 6,6, Small thinks they should be able to predict the temperature of any kind of nylon, as they all have the same basic chemical structure.

Blog tags: 
Techniques: