ATR-FTIR chemometrics for biological samples : a thesis presented in partial fulfilment of the requirements for the degree of Master of Science in Nanoscience at Massey University, Manawatū, New Zealand

This project has used the analytical infrared reflectance technique of Attenuated Total Reflectance Fourier Transform Infrared (ATR-FTIR) spectroscopy, for the prediction of chemical components in a range of biological samples. Data collection was carried out on 40 hyperaccumulator samples, 56 chicken feed samples, 54 lamb faecal samples and 188 forage feed samples. Predictions were made using several different regression methods including: Ridge, Least Absolute Shrinkage and Selection Operator (LASSO), Elastic Net, Principal Components (PCR) and Partial Least Squares (PLS). The best methods were identified as LASSO, Elastic Net and PLS. Several spectral data pre-treatments were explored, the best of which combined Standard Normal Variant scaling (SNV) with a first-order Savitzky – Golay (SG) spectral derivative and smoothing filter. Several of the resulting models illustrated high quality predictions (R2 > 0.8, Relative Performance Deviation (RPD) ≥ 2). The SNV and SG pre-treatment almost completely reduces the contribution of strong water-based signals to the regression model, allowing the possibility of in situ prediction of forage feed composition with minimal sample preparation. ATR-FTIR spectrometers are available in a hand-held form, and the results of this research suggest that in situ forage quality analysis could be performed using mid – infrared (MIR) reflectance spectroscopy