European Conference on Wood Modification 2009 Color Change in Thermally-modified Wood and its Relationship with Property Changes Marcos M. González-Peña 1,2 and Michael D. C. Hale 1 1 School of the Environment and Natural Resources, Bangor University, LL57 2UW, UK 2 Current address: Centre for Advanced Wood Processing, Faculty of Forestry, The University of British Columbia, Vancouver, V6T 1Z4, Canada [mgonzalez@forestry.ubc.ca] Keywords: CIELab, Fagus sylvatica, Picea abies, Pinus sylvestris, PLS, strength INTRODUCTION In the woodworking industry image analysis is routinely used for quality control and for matching and classification during various processes. An extension of these automated systems for the prediction of physical properties of thermally-modified wood (TMW) is enticing, because to date there is no a generalized procedure for the quality assurance of this material. The purpose of this work was to investigate the relationship between property changes in TMW and heated-induced colour changes. A description of the effect of the temperature and time of treatment on colour variables and its relationship between these and colour changes in Klason lignin is also given. EXPERIMENTAL Small, matched specimens (150 l × 20 r × 10 t mm) of Scots pine, Norway spruce and Beech wood were modified in a N 2 atmosphere. Five residence times (0.33 – 16 h) and four treatment temperatures (190 – 245 °C) were used for thermal modification. After conditioning the modified and untreated specimens for at least six months at 65% RH, 20 °C in darkness, six properties were measured to describe its mechanical behavior, according to BS 373 (BSI 1957) and BS EN ISO 179 (BSI 1997): Janka hardness (H, n=6 per treatment), shear strength (S, n=4), compression parallel to the axis (CA, n=8), compression perpendicular to the axis (CE, n=6), Charpy impact strength (IS, n =10), and three-point bending (n=10). From the latter, MOE and MOR were obtained. Heat- induced weight loss (WL) relative to the initial oven-dry weight of the specimens, and nominal density (ND = oven-dry weight/volume at 65% RH, 20 °C) were obtained from the bending specimens; the anti-swelling efficiency (ASE, n =10) was calculated according to Hill (2006). Color coordinates of the CIEL*a*b* system were determined on the radial plane from scanned images of conditioned bending specimens prior to testing (TAPPI 1994). A cylindrical coordinate system was also tried to find out the best possible predictor of physical changes from color variables. In this, the a* and b* are substituted by the saturation C* and the hue angle h* on the colour circle around the lightness axis L* (TAPPI 1994). Data obtained from the bending specimens were deemed as representative of the full batch -each batch comprised the specimens for all mechanical tests for one treatment. For the modeling of physical properties using colour variables as predictors, simple linear regression was used. Additionally, PLS regression using the unfiltered data of the eleven colour characters (ΔE*, ΔL*, L*, a*, b*, h*, C*, Δa*, Δb*, ΔC* and ΔC* ab ) as independent variables was carried out using SIMCA-P © software (Umetrics AB, Sweden). For constructing each calibration PLS model for MOE, MOR, ND and WL, seven specimens (out of ten) for each treatment were used, while the remaining three specimens from each treatment were left apart for the 181