Firstly, 24 different fatty oils with various Iodine values were measured (in duplicate) and the NIR spectra used to calculate a calibration curve model. Next, the calibration curve model generated using the PLS method was applied to measured spectra and the calculated Iodine values were correlated with the wet chemistry method. Finally, the Iodine values of 5 actual samples were measured utilizing this calibration curve model.

Figure 3 shows the PLS calibration curve model which gives good correlation between the ‘true’ value (obtained by wet chemical method) and evaluated value (obtained by spectral analysis). The correlation coefficient is 0.9995. The results of evaluating the quantitation reproducibility (n=30) of 5 different fatty oils with different Iodine values using this model are shown in Table 1. The average value and the standard deviation obtained from the spectra are presented in the table. The values in parentheses are Iodine values obtained by wet analysis method.

Figure 3. PLS calibration curve model

The error between the results obtained with the two analysis methods is less than 2% and the standard deviation of the reproducibility is less than 0.173, which can be considered as a excellent result.

The results demonstrate that the an FT-NIR spectrometer with high wavenumber accuracy, fitted with a Peltier thermostated cell holder to control the sample temperature is very effective for measurement of the iodine value. It provides comparable data quality to the wet chemistry method, but is much more convenient, being simpler, faster and nondestructive.