Estimation of Brazilian charcoal properties using attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectrometry coupled with multivariate analysis

The aim of the present work was to estimate fixed-carbon, volatile matter and ash contents in Brazilian commercial charcoal by using attenuated total reflectance-Fourier transform infrared (ATR-FTIR) spectroscopy together with multivariate calibration methods. Several multivariate calibration techniques, including partial least squares (PLS), interval partial least squares (iPLS), and genetic algorithm (GA), were compared and validated by establishing significance testing. Charcoal samples (n = 72) were divided into calibration (n = 52) and validation sets (n = 20) by applying the classic Kennard-Stone (KS) selection algorithm to the ATR-FTIR spectra. For the fixed-carbon content, the results obtained using PLS-GA for the root mean square error of cross-validation (RMSECV) and prediction (RMSEP) were 3.77% and 4.29%, respectively. For volatile matter, RMSECV and RMSEP of 4.36% and 4.65% were achieved by the PLS model using seven latent variables (LV). Finally, for ash, RMSECV and RMSEP of 0.58% and 0.38% were achieved by the PLS model using eight latent variables (LV). A t-test and quantile-quantile (Q-Q) plot were performed to compare the results of the models with each other and with a reference method. These results suggest that ATR-FTIR spectroscopy and multivariate calibration can be used effectively to determine fixed-carbon, volatile matter content and ash content in Brazilian charcoal.