Near-infrared reflectance spectroscopy represents a radical departure from conventional analytical methods, because a single sample is characterised in terms of its absorption properties, rather than separate batches being treated with various chemicals to isolate specific components.1 In the present work the potential of FT-NIRS in predicting total organic carbon (TOC) of hybrid mesoporous silica, functionalized by a MW-assisted method,2 was evaluated in order to obtain a rapid, inexpensive and non-destructive samples technique. A calibration model was built by partial least-squares regression (PLS-R), centred on a selected spectral range, using forty-nine functionalized silica samples. The model was both cross-validated by leave-one cross-validation and PRESS analysis.3 Moreover, the model was validated by measurement of the error in prediction by an external validation set of fifteen samples. The RMSEC value, RMSECV value, goodness of calibration (R2) the goodness of validation (Q2) and ratio of prediction to deviation (RPD) were calculated. The R2 value was 0,98 for the calibration and the Q2 value was 0,96 for the validated model. The RMSEC value for the calibration was 0,129 and the RMSECV for the validation was 0,190. The RPD value of 3,58 confirms the goodness of the model, since in the prediction of soil properties included total organic carbon,4 good models have RPD greater than 2. The study revealed the robustness and stability of the FT-NIRS model and the goodness of predictivity, especially for silica sample characterized by the same functional group. A calibration model for each functional group must be developed. Further studies to investigate the possibility to develop a method able to select the organic carbon contribution divers functionalized silica have to be performed.
|Titolo:||DETERMINATION OF TOTAL ORGANIC CARBON OF MESOPOROUS SILICA FUNCTIONALIZED SAMPLES BY FT-NIR SPECTROSCOPY|
|Data di pubblicazione:||2015|
|Appare nelle tipologie:||4.2 Abstract in Atti di convegno|