TY - JOUR
T1 - Determination of cellulose crystallinity by terahertz-time domain spectroscopy
AU - Vieira, Francisco Senna
AU - Pasquini, Celio
PY - 2014/4/15
Y1 - 2014/4/15
N2 - Terahertz-time domain spectroscopy (THz-TDS) has the ability to probe the crystallinity of several materials, due to the interaction of THz radiation with optical phonons in crystal lattices. In this work, THz-TDS has been used to quantify the degree of crystallinity of microcrystalline cellulose (MCC) samples. The THz spectra of cellulose present absorption features which could be directly correlated with the crystallinity index (CI) obtained by means of the well-established powder X-ray diffraction (PXRD) technique. The effect of THz time-domain signal processing was investigated, and both univariate and multivariate, based on partial least-squares (PLS), regressions were carried out with the signal in the frequency domain to correlate the THz spectra with CI. Results show that the multivariate regression models based on spectral data, collected with the sample displaced from the focal plane of the THz optics to improve representativeness and measurement repeatability, present the best performance with external validation achieving an absolute root-mean-square error of prediction (RMSEP) of 4% for CI. This result compares well with the PXRD technique.
AB - Terahertz-time domain spectroscopy (THz-TDS) has the ability to probe the crystallinity of several materials, due to the interaction of THz radiation with optical phonons in crystal lattices. In this work, THz-TDS has been used to quantify the degree of crystallinity of microcrystalline cellulose (MCC) samples. The THz spectra of cellulose present absorption features which could be directly correlated with the crystallinity index (CI) obtained by means of the well-established powder X-ray diffraction (PXRD) technique. The effect of THz time-domain signal processing was investigated, and both univariate and multivariate, based on partial least-squares (PLS), regressions were carried out with the signal in the frequency domain to correlate the THz spectra with CI. Results show that the multivariate regression models based on spectral data, collected with the sample displaced from the focal plane of the THz optics to improve representativeness and measurement repeatability, present the best performance with external validation achieving an absolute root-mean-square error of prediction (RMSEP) of 4% for CI. This result compares well with the PXRD technique.
UR - http://www.scopus.com/inward/record.url?scp=84898852759&partnerID=8YFLogxK
U2 - 10.1021/ac4035746
DO - 10.1021/ac4035746
M3 - Article
C2 - 24654843
AN - SCOPUS:84898852759
SN - 0003-2700
VL - 86
SP - 3780
EP - 3786
JO - Analytical Chemistry
JF - Analytical Chemistry
IS - 8
ER -