A New Method to Quantitatively Evaluate the Chemical Composition of Waterlogged Wood by Means of Attenuated Total Reflectance Fourier Transform Infrared (ATR FT-IR) Measurements Carried Out on Wet Material. Issue 5 (May 2013)

• Record Type: Journal Article
• Authors: Pizzo, Benedetto; Pecoraro, Elisa; Macchioni, Nicola
• Is Part Of: Applied spectroscopy. Volume 67:Issue 5(2013)
• Page Start: 553
• Page End: 562
• Publication Date: 2013-05
• Subjects: Partial least square (PLS) regression -- Multivariate analysis -- FT-IR spectroscopy -- ATR -- Waterlogged archeological wood -- Lignin -- Holocellulose; Spectrum analysis -- Periodicals; 543.505
• DOI: 10.1366/12-06819
• Languages: English
• ISSNs: 0003-7028

Abstract

Wood degradation in waterlogged conditions, although slow, is relevant in some specific contexts, such as in the case of cultural heritage preservation (e.g., ancient shipwrecks found under the water table). This decay process induces the selective depletion of the biopolymers constituting wood cell walls and mainly of their structural polysaccharides (both cellulose and hemicelluloses). Assessment of residual composition of cell walls is normally carried out using conventional chemical analyses, which, however, require elevated amounts of material. The present work reports on a new approach to quantitatively determine the residual chemical composition of wood (in terms of amount of lignin and holocellulose) and therefore its extent of degradation. This was accomplished by acquiring attenuated total reflectance Fourier transform infrared (ATR FT-IR) spectra on material still in waterlogged conditions, which was simply placed in optical contact with the ATR crystal. Data for the calibration set were obtained by means of wet analyses, and the calibration model was based on multivariate statistical analysis through partial least square regression. Acquired spectra showed signals higher and better resolved than for dry wood. Calibration was selected after comparing two different normalization procedures and after processing signals of two different spectral ranges. Furthermore, the exclusion of some outliers led to a substantial improvement in the error parameters ( R 2 and root mean square error of cross validation) associated with the calibration model, thus allowing for the selection of the best model. The stability of this best model was also established by means of the leave-more-out method. Finally, a procedure of external validation confirmed that the proposed model also could be applied to similar (for instance, belonging to Pinaceae) wood remains from other excavations.