ATR-MIR Spectroscopy Predicts Total Phenolics and Colour for Extracts Produced by Microwave-Assisted or Conventional Thermal Extraction Methods Applied Separately to Mixtures of Grape Skins from White or Red Commercial Cultivars

2020 
Attenuated total reflectance mid-infrared (ATR-MIR) spectra were collected from 480 grape skin extracts produced in a wide range of conditions. The conditions were created at different phases in parallel response surface RS-optimisation experiments aiming to maximise the total phenolics in red and white grape skin extractions by microwave-assisted (MAE) or conventional thermal (CTE) methods. Skins mixtures for six white and six red Australian commercial cultivars were prepared from grapes collected at veraison and harvest. The total phenolics were measured for individual extracts with the Folin-Ciocalteu method. Partial least squares (PLS1) regression was employed on the spectra to predict the total phenolics (TP) and colour (CIELAB chroma C*) of individual extracts. Models based on the raw spectra and their second derivative (Savitzky-Golay, 20 points) were examined using the full range of spectra 4000–400 cm−1, and sub-ranges: 4000–1100 cm−1, 1500–400 cm−1, 1500–1100 cm−1 and 1457–1168 cm−1. The PLS1 models based on the second derivative of the range of 4000–1100 cm−1 provided the best results for both TP and C*, for each skin colour. Overall, the residual predictive deviation (RPD) exceeded 3.0, indicating the PLS1 models’ ability for fair classification, suitable for screening. A comparative analysis was performed for the range of 4000–1100 cm−1 between the spectra of white and red skin mixture extracts: the peaks at 3840 cm−1, 2940 cm−1, 2358 cm−1, 2197 cm−1, 2136 cm−1, 1725 cm−1, 1100 cm−1 and 1148 cm−1 are flagged to be investigated in further research. ATR-MIR spectroscopy can reduce the time and costs of TP and C* analyses in grape skin extracts.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    40
    References
    2
    Citations
    NaN
    KQI
    []