Determination of Iodine Value Using FT-NIR Analysis

Iodine value (IV) is a measure of the total number of unsaturated double bonds present in fats and oils. It is normally expressed in terms of “number of grams of iodine that will react with the double bonds in 100 grams of fats or oils”. It must be noted that a high IV value in oil consists of a greater number of double bonds than a low IV oil. Fats with high iodine value are normally soft or liquid and are less stable to oxidation.

Advantages of NIR IV Determination

The AOCS recommended method D1959-97 (Wij’s procedure) is broadly used for determination of IV. The method involves adding halogen in the presence of potassium iodide and the titrating the released iodine with sodium thiosulfate using a standard starch solution as the indicator.

The time taken for this analysis is at least 30 minutes and has to be performed on samples taken out of the process line. On the other hand, near infrared (NIR) spectroscopy is quick with an analysis time of the order of less than 1 minute and does not need the use of any solvents or reagents. Fiber optic probes are very useful and help perform real time, on-line measurement of IV resulting in huge time and cost savings.

Laboratory measurements were performed by AOCS Cd1e-1 method using a Bruker Optics FT-NIR spectrometer with 8 mm disposable vials as shown in Figure 1. The oils and fats of different types used to generate the calibration model consisted of IV’s ranging between 0 and 185. Three spectra per oil sample were collected at a resolution of 8 cm^-1.

Figure 1. Laboratory measurements using FT-NIR spectrometer with 8 mm disposable vials.

Figure 2 shows three measured NIR spectra between 7800 and 9000 cm^-1. The intensity of the spectral feature at ~8580 cm^-1 can be correlated with the known oil IV as shown in the figure.

Figure 2. Three measured NIR spectra between 7800 and 9000 cm^-1.

A partial least squares (PLS) calibration model formulated using the known oil samples resulted in an R² of 99.9 and a Root Mean Square Error of Cross Validation (RMSECV) of 1.05 as shown in Figure 3. The RMSECV is an indication of the error to be expected when the model is used for prediction of unknown samples.

Figure 3.Comparison between NIR value% and reference value %

The model is then validated with a set of independent samples for checking the predictability of the universal IV model. R² =99.99 %, and the Root Mean Squared Errors of Prediction (RMSEP) is 1.06 IV number. The standard deviation of differences of reproducibility of the validation samples was found to be 0.34 IV.

Automation and User Interface

A dedicated user interface was developed to obtain and evaluate samples using the Bruker OPUS software package and the quantitative analysis model developed. The prediction result is automatically output to a spreadsheet for QC lab record.

Results and Discussion

On-line monitoring of Iodine Value of edible oil has been shown using a Bruker Optics FT-NIR spectrometer coupled with specially designed fiber optic probes. This type of process implementation offers real-time data that speeds up analysis, minimises production down time and achieves product consistency.

This information has been sourced, reviewed and adapted from materials provided by Bruker Optics.

For more information on this source, please visit Bruker Optics.