Development of the partial least squares models for the interpretation of the UV resonance Raman spectra of lignin model compounds

Partial least squares (PLS) modeling was utilized for the interpretation of the ultraviolet resonance Raman (UVRR) spectra of various lignin model compounds. Lignin model compounds presented different structural characteristics of the naturally existing lignin biopolymer within wood materials. The model compounds were grouped according to their monomer type, condensed and conjugated structures. Condensed structures were defined as those having a side group at C5 carbon of the aromatic units. Conjugated model compounds were further subdivided into C=C and C=O as well as stilbene structures. PLS modeling was applied to find the characteristic Raman signals of each structure: p-hydroxyphenyl showed characteristic Raman bands at 1488, 1405, 1338, 1215, 1164, 1094, 861–841, 644 cm , guaiacyl at 1520, 1285–1270, 1186, 1124, 1078, 1024, 920, 784, 761, 711 cm , syringyl at 1588, 1510, 1331, 1228, 1148, 1108, 964, 810, 741 cm 1 condensed structures at 1618, 1558, 1492, 1375–1355, 1328, 1223, 975 cm , conjugated C=C at 1658, 1538, 1475, 1311, 1215, 1130–1114, 974 cm , conjugated C=O at 1695–1662, 1084 cm 1 and stilbenes at 1638, 1515, 1435–1421, 1263, 1191, 877, 768 cm . The loadings line spectra of different PLS models were further compared with the UVRR spectra of model compounds and hence, the feasibility of PLS modeling was verified. To the best of our knowledge, this is the first time when PLS modeling has been utilized in this manner for the interpretation of such spectra. The defined characteristic Raman bands can further be utilized for the interpretation of UVRR spectra of polymeric lignin samples. PLS models introduce new possibilities for the interpretation of Raman spectra since the UVRR spectral data of complex lignin samples contain broad bands which are cumbersome to fully identify. # 2004 Elsevier B.V. All rights reserved.