FTIR spectral analysis combined with chemometrics in evaluation of composite mixtures of coconut testa flour and wheat flour

Fourier transform infrared (FTIR) authentication coupled with chemometric evaluation is widely used in qualitative and quantitative analysis of food and agricultural products. This study aimed to characterize coconut testa flour (CTF) of five different Sri Lankan cultivars and to develop statistical models between FTIR data and parameters of CTF, wheat flour (WF) and their admixtures. Flour samples were subjected to FTIR analysis within the range of 4000-500 cm(-1) and proximate data were recorded. Principal component analysis (PCA) was applied for FTIR data and models were developed using both univariate and partial-least-squares (PLS) regression to predict fat, protein, carbohydrates, moisture and level of admixture from FTIR spectra. Results showed that the spectral bands of both CTF and WF were indicative of organic functional groups associated with fat (similar to 1745 cm(-1)), protein (similar to 1655, similar to 1550 cm(-1)), carbohydrates (similar to 3300, similar to 1065 cm(-1)) and moisture (similar to 3300 cm(-1)). Spectral studies on the admixtures confirmed that proportional differences of the biomolecules had a direct influence on the variations occurring in the spectral bands. Application of PCA resulted in a clear discrimination among flour samples with a total variance of 97%. PLS regression showed more precise results than univariate regression. The precision of the developed PLS-regression predictive models was found to increase with the number of selected relevant spectral points and these models resulted in a good R-2 value and low RMSEC and RMSEP values. The results of this study would be helpful for further chemical studies on authentication, product development and nutrition of CTF and WF.

Automated summary

Fourier transform infrared (FTIR) authentication coupled with chemometric evaluation is widely used in qualitative and quantitative analysis of food and agricultural products. This study characterized coconut testa flour (CTF) of different cultivars and developed statistical models to predict parameters of CTF, wheat flour (WF), and their admixtures using FTIR data. Spectral bands were indicative of organic functional groups associated with fat, protein, carbohydrates, and moisture. PCA and PLS regression were used to discriminate among samples and predict parameters. The results have implications for authentication, product development, and nutrition studies of CTF and WF.