Rapid identification of carpet face fiber

Polymer type All of the polymers have very distinct, easily distinguished NIR spectra except N66 and N6 which are very similar. The use of second-derivative spectra is necessary to reliably distinguish the small spectral differences between N66 and N6. 

Construction effects The effects of construction were less pronounced than the effects of color. In general, cut pile carpets had larger absorbances across the entire spectrum than did loop pile carpets, but the positions of the characteristic absorption peaks were not affected. As for color, accurate calibration models could be developed on second-derivative spectra as long as sufficient variation in carpet construction was included in the library. 

Measurement speed effects The effect of reducing the number of scans per spectrum, in order to increase measurement speed, was found to be minimal over the range from 20 scans down to 5 scans. 

Sample type Polymer type Sources Samples % correct (absorbance model) % correct (second-derivative model) 

Sampling mode effects The goal was to be able to measure at-line, with no sample preparation at all, using a fiber-optic probe or a remote sampling head. However, the area sampled by the hber-optic probe is much smaller than for the sample transport module. It was found that the remote (probe) spectra were very similar to the static (sample transport) spectra, but the baselines were shifted significantly higher and the absorbance peaks consequently reduced in intensity as before, the characteristic peak positions were not affected. Calibration models developed using spectra obtained with the hber-optic probe performed equivalently to those developed with the sample transport module. 

There were two classes of technical challenges faced in this work. The first challenge was how to prevent variability in carpet color (shade and darkness), carpet coloring method (dyes vs. pigments), and yam and carpet construction (e.g. cut vs. loop pile) from 

Sampling mode effects The goal was to be able to measure at-line, with no sample preparation at all, using a fiber-optic probe or a remote sampling head. However, the area sampled by the fiber-optic probe is much smaller than for the sample transport 

---