Prediction of boron content in wood pellet products by near-infrared spectroscopy.
A rapid method assessed the potential of near-infrared spectroscopy (NIRS) to estimate boron content of wood pellet products. Based on a comparison of NIR spectra data in the 1,100- to 2,200-nm wavelength region of Eastern black spruce (Picea mariana var. mariana) wood pellets treated with preservative concentrations ranging from 0 to 2 percent and from 0 to 20 percent glycol borate-based disodium octaborate tetrahydrate (DOT), the minimum level of boric acid equivalent required to protect wood from biodegradation was revealed. Borate was indicated in the 1,700- to 1,900-nm wavelength region and the visible-NIR absorbance trended to a proportion higher for the lower borate concentrations and lower for the higher borate concentrations. These were correlated by projection to the latent structures--partial least-squares regression method and the sample-specific standard error of prediction method. Calibration sets achieved [R.sup.2] values from 0.7 to 0.95, root mean square error (RMSE) ranging from 0.3 to 1.61 percent, and relative percent difference (RPD) ranging from 1.8 to 4.4, whereas validation statistics achieved [R.sup.2] values from 0.64 to 0.94, RMSE ranging from 0.33 to 1.65 percent, and RPD ranging from 1.7 to 4.3. These preliminary results indicate that NIRS should be able to provide a greater quantitative and qualitative technique of predicting boron content in wood products for the preservation industry.
Wood waterborne preservative solutions are commonly used for most applications in the preservation industry to achieve wood durability by protecting wood materials from weathering conditions and biological degradation agents (Diouf et al. 2006, Mounguengui et al. 2007, Gerardin et al. 2008, Ondo et al. 2013). In the same context, boron-based preservative compounds have significant environmental and health advantages, as they are inexpensive, highly effective wood preservatives. Additionally, boron formulations are effective against both fungal and insect attack and they also have flame retardant properties (Feldhoff et al. 1998, Thevenon and Pizzi 2003, Hettipathirana 2004, Salman et al. 2014). The chemical active agent in boron-based preservative formulations remains intuitively the borate, which contains oxyanions derived from boric acid (Mikeal and Ingegard 2010, Poaty et al. 2010). Because the cost of fossil fuel--based heating and petroleum-based energy production are increasing, the demand for residential construction and the preservation industry to use alternative composites as wood pellet-treated boron-based preservatives has experienced a large increase during the last decade (Forato et al. 1998; Mounanga 2008, 2015; Stevens and Gardner 2010). Also, the majority of the existing methods used to accurately estimate boron-based preservative concentrations of treated wood products are destructive, time-consuming, and require elaborate analytical techniques (Engstrom et al. 1998, Awoyemi et al. 2009, Leblon et al. 2013). Wood samples must be ground to powder and the reflux extracted; then the resulting solution has to be analyzed by colorimetric titration, potentiometric titration, or by other methods such as inductively coupled plasma emission spectroscopy (ICPAES) or again by ICP mass spectroscopy/spectrophotometric analysis. Using and understanding the variables influencing the chemical composition of wood pellets is also of interest (American WoodPreservers' Association [AWPA] 2000, Mounanga et al. 2008, Tsuchikawa and Shwanninger 2013). Hence, near infrared spectroscopy (NIRS) appears as a good alternative technique to estimate boron concentration in treated wood products. NIRS has been coupled with multivariate statistical methods to quantitatively determine boron-based preservative concentrations of treated sapwood board (So et al. 2004, Taylor and Lloyd 2007, Koumbi-Mounanga et al. 2015b) and to successfully predict wood moisture content (MC), surface color, contact angle, and stiffness and also chemical components on the wood surface and adhesive bond strength of wood composite products for wood preservation in the industry (Meder et al. 2002, Fackler et al. 2007, Carneiro et al. 2010, Hein et al. 2011, Schwanninger et al. 2011, Watanabe et al. 2011, Stirling 2013).
The objective of the present work was to evaluate the potential of NIRS to estimate boron-based preservative concentrations of Eastern black spruce (Picea mariana var. mariana) pellet products. NIR spectra data were secondary derivative transformed and then succinctly related to boronbased preservative concentrations through projection to the latent structures--least-squares (PLS) regression method and sample-specific standard error of prediction method. If NIRS has been utilized for the estimation of MC as well as for the quantification of the chemical changes that occurred on the wood surface, it is therefore suggested that it should preferentially be used as the rapid, easy-to-operate, and portable analytical tool for characterization of the derived wood products.
Materials and Methods Sample preparation
Samples of Eastern black spruce (Picea mariana var. mariana) wood were acquired green from a local lumber supply. The wood was selected in order to gather enough samples from one source of lumber that was ground to powder under drying conditions (relative humidity [RH], >2%) with a Thomas-Wiley mill grinder (Laboratory Model 4; Thomas-Scientific). About one-half kilogram of the resulting sawdust was weighed and oven-dried overnight at 103[degrees]C with a digital programmable controller oven (Despatch; maximum temperature, 260[degrees]C) and then screened with a 2-mm mesh Tyler equivalent filter (W.S. Tyler, series equivalent No. 18). The sawdust was treated for borate using quantities of proportional equivalent in weight (wt/wt [g/g]) of sawdust and borate formulations, which were clear solutions of 20 percent disodium octaborate tetrahydrate (DOT) in propylene glycol (glycol borate-based DOT [[Na.sub.2][B.sub.8][O.sub.13] x -4[H.sub.2]]) available commercially as Boracol 20-2 from Sansin Corporation (Ontario, Canada). Some portions of sawdust were left as an untreated control sample following the method described in AWPA standards A19-93 (1996). In order to understand the nature of the analytical detection limit or the minimum required to prevent decay from starting, borate formulations were gradually prepared to eight dilutions ranging from 0 to 2 percent and from 0 to 20 percent DOT glycol borate following the method described in Saadat and Cooper (2013). A quantity (in grams) of formulation per one equivalent gram of sawdust was then weighed and the resulting treated sawdust was properly mixed with a Heidolph mixer (Cole-Palmer Canada Inc.). The mixture was conditioned (24 h) in a plastic bag to minimize drying during the test and stored at room temperature (approaching
24[degrees]C) with >2 percent RH in a Drytech klin chamber (Series 3900MC). The mixture was kept in storage for another 24 h at ambient temperature prior to pelletizing. For each disk of wood pellets made, about 200 mg of the treated sawdust was weighed with a Mattler Toledo Balance (Mattler Toledo; max. = 220 g, d = 0.1 mg). In total, six wood pellets of black spruce for each of the eight dilution regimes in the two ranges (i.e., 0% to 2% and 0% to 20%) were made over applied load under 3,000 lb/in2 (10,000 lb; 4.6 metric tons) during 5 minutes with a Carver compressor (Model 4350.L, S/N 110121) prior to scanning. Each treatment concentration was replicated on three wood pellets. This process follows the method described in Faix (1992) and Stansbury and Dickens (2001).
NIR spectra data were acquired for 1.3-cm (0.5-in.) diameter, 1-mm (0.0394-in.) thick black spruce wood pellet disks with an Ocean Optics NI[R.sup.2]56-2.5-[HL-2000-FHSAHP-LL-RS232] Vis-Near-infrared spectrometer (Ocean Optics Inc.) equipped with an optical probe positioned on the top of the wood pellet samples in a 2-mm-diameter beam (RH, >0% to 2%). The instrument has a spectral resolution of 5 nm (boxcar width, scan time = 80 ms for 10 scans) and was calibrated manually for white/dark after every measurement obtained at different concentration levels as described in Mounanga et al. (2015). Each black spruce wood pellet sample was scanned on both faces three times at randomly different positions from the flat surface.
The data processing was done using the Unscrambler 9.8 (CAMO Software, Inc.). All of the absorbance spectra acquired over the whole wavelength of the 1,100- to 2,400nm region were smoothed by applying a second derivative, 9-point Savitzky-Golay transformation, as suggested by the Unscrambler software, and then related to DOT glycol borate concentrations by the projection to the latent structures--PLS regression method and the sample-specific standard error of prediction method as described by Geladi and Kowalski (1986) and Faber and Bro (2001). PLS regression models were developed with full cross-validation because the number of scans recorded to the Unscrambler program were sufficient to be identified as calibration and validation data sets that were calculated with a number of factors of two principal components (PCs) determined based on the root mean square error (RMSE) of performance values as described in Esbensen et al. (2002).
The accuracy of PLS regression models and sample-specific standard error of prediction models were computed using six statistical metrics: (1) the coefficient of determination ([R.sup.2]) between measured and predicted MC values; (2) the root mean square error of prediction (RMSEP) or of calibration (RMSEC); (3) the bias, which is a mean difference between the measured and predicted values; (4) the standard error of performance (SEP), which is the standard deviation (SD) of the residuals values; (5) the P value or significance level, which measures the probability that an [R.sup.2] should be as large as it is, with a real value of zero; and (6) the relative percent difference (RPD), which is a ratio of the SEP and the SD of the reference data (Boulesteix and Strimmer 2006, Natsuga and Kawamura 2006, Rosipal and Kramer 2006).
The visible-NIR absorbance spectra data trended proportionally with the different concentrations ranging from 0 to 2 percent and from 0 to 20 percent DOT glycol borate. The highest was the concentration; the lowest was the absorbance at all wavelengths of the 1,100- to 2,400-nm regions (Fig. 1). The effect of borate concentration on the NIR spectra was more evident by applying a second derivative, Savitzky-Golay transformation (Fig, 2). The visible-NIR absorbance spectra were an average of six scans for corresponding concentrations of black spruce pellets. The results showed that increasing the DOT glycol borate concentrations consequently resulted in increasing the basicity into wood pellet products as demonstrated by the decreasing intensity of NIR absorbance spectra. It seems that the effect of DOT glycol borate on the wood pellets appears more pronounced in the 0 to 2 percent range compared with the 0 to 20 percent range of borate concentrations.
The PLS regression models of DOT glycol borate in both ranges were built using the average of six absorbance spectra data (Fig. 3). The related statistics are shown in Table 1. Both models were able to provide good correlation coefficient (R) ranging from 0.79 to 0.97 between the measured and predicted values. The calibration regression of the first model (0% to 2%) of DOT glycol borate values yielded an [R.sup.2] of 0.696 and the validation model yielded an [R.sup.2] of 0.64. Thereby, RMSE yielded 0.30019 percent for calibration and 0.33065 percent for validation. The calibration in the second model (0% to 20%) of DOT glycol borate yielded an [R.sup.2] of 0.948 and the validation [R.sup.2] was 0.947. RMSE successively yielded 1.6109 and 1.6511 percent for calibration and for validation. For both concentration ranges (0% to 2% and 0% to 20%), the PLS regression models were significant at P < 0.0001 (Table 1).
[FIGURE 1 OMITTED]
Sample-specific standard error of borate prediction
The sample-specific standard error of borate prediction results are shown in Figure 4. The sample-specific standard error of prediction methods had shown a better estimation of the measurements conducted with the second model (0% to 20%) of DOT glycol borate than with the first model (0% to 2%). Therefore, the average margin standard deviation (SD) for prediction in both of the models was 1.13. The highest values obtained above the average SD value were 1.91, 1.45, and 1.38 for 0.1, 5, and 20 percent DOT glycol borate in relation to the second model. For about 0.02 and 2 percent DOT glycol borate, the SDs were 1.4 and 1.32, respectively. In addition, sample-specific standard error of borate predictions presented strong statistics with reasonable RPDs, which were 1.7 and 4.3 for the first and second model of DOT glycol borate, respectively.
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
The NIR absorbance spectra of Eastern black spruce (Picea mariana var. mariana) pellets showed that the heights of the main peaks decreased with boron retention in the wood pellets as described to predict MC of Douglas-fir (Pseudotsuga menziesii menziesii) veneers and trembling aspen (Populus tremuloides Michx.) strand products by Mounanga et al. (2012) and Koumbi-Mounanga et al. (2015a), respectively. The main peaks are located in the 1,250-, 1,620-, 1,820-, and 2,200-nm spectral regions, which is related to the overtones of fundamental vibrational transitions in the infrared region, mainly CH (hydrocarbons. ..), CO (cellulose, hemicelluloses...), OH (water, alcohol, phenol...), and NH (lignin...) functional groups (Felix and Gatenholm 1991, Cozzolino et al. 2003, Schwanninger et al. 2004). At around the 1,700- to 1,900nm and 2,000- to 2,200-nm spectral data regions, the absorption bands might be induced by the boron-based preservative concentrations owing to the basicity influence of borate on the reaction with phenolic compounds of wood pellets. This is similarly observed in the degradation of the cell wall polymers as likely as not, the phenomenon occurring in the case of thermal treatment of wood (Thygesen and Lundqvist 2000, Koumbi-Mounanga et al. 2015c). The second derivative--transformed spectra were properly described as overlapping absorbance bands of the first and second overtone of OH and CH at around 2,250-nm and 1,900-nm spectral data, respectively. One hypothesis signalled to recalculate the second model of concentration range (0% to 20%), without a total of six outlier scans observed in the visible-NIR absorbance at 0.1 percent DOT glycol borate. There were similar observations in the overtone absorption from the 1,200- to 1,300-nm and the 1,400- to 1,600-nm spectral data regions for both models. This might be caused by oxyanion reactions involved in the boron-based preservative treatment into the wood pellet components that were also identified in different investigations as the cause of carbohydrate degradation and deacetylation reactions of polyoses in the case of thermal treatment of wood. However, future estimation should involve and incorporate more parameters, e.g., the temperatures within more factors, for an easy interpretation of the data (Lee and Luner 1972, Mitsui et al. 2008, Bachle et al. 2010). The present study showed results equivalent to those obtained on frozen and thawed black spruce logs by Hans et al. (2013), who reported that the overtone absorption could be attributed to differences in the MC, wood permeability between sapwood and heartwood portions, and partial drying of wood pellet samples during storage and handling. Indeed, borate concentration influences on wood products, especially chemical reactions involved in pellets, are very complex phenomena. They depend on MC distribution, hygroscopic properties of wood, pH, temperature distribution, pressure, duration of treatment, type, size, and source of the samples (Bums and Ciurczak 2007, Guo 2013).
[FIGURE 4 OMITTED]
Taylor and Lloyd (2007) had similar observations with wood cubes from southern pine (Pinus spp,) sapwood samples that were treated with glycol borate-based DOT with a boron content ranging from 0.01 to 15 percent. They had obtained good (calibration and validation) correlations between measured and predicted values with [R.sup.2] values ranging from 0.96 to 0.86 and RMSE ranging from 0.92 to 2.12 percent. These results were better than those on aspen and Douglas fir veneers obtained by Koumbi-Mounanga et al. (2013) who had [R.sup.2] values ranging from 0.46 to 0.86 and RMSE ranging from 4.47 to 23.35 percent. Adedipe and Dawson-Andoh (2008) assessed different results on a prediction of MC of yellow poplar (Liriodendron tulipifera L.) veneer, with SE ranging from 0.27 to 0.39 percent and bias ranging from 0.27 to 0.37 percent.
Sample-specific standard error of borate prediction
The resulting statistics of the sample-specific standard error of prediction method, when applied to boron retention in wood pellet samples, were in agreement with the results of Faber et al. (2003), who assessed an SD of Gaussian peaks of about 5. Similarly, Koumbi-Mounanga et al. (2015a), in the prediction of MC for aspen strands, obtained different biases that ranged from -6.507E-07 to 4.337E-07 percent for calibration and -0.0254 to 0.0093 percent for validation. This included sample-specific standard error of MC predictions that were evident with a mean margin of error prediction of 5.25 for the SD obtained over the three drying cycles and for all MC ranges. The SE yielded 2.5565 to 5.92 and 2.7197 to 6.0123 percent for calibration and validation, respectively. It is noted that the resulting statistics of the sample-specific standard error of borate prediction performed above the level of 0.3 percent DOT glycol borate were better than those obtained below 0.2 percent DOT glycol borate concentration regimes. Similar observations were found in the study of Faber et al. (2003) with regard to a classification of sample-specific standard error of prediction for partial least-squares regression methodology.
The RPD presently computed clearly differentiated the two models of concentration ranges (0% to 2% and 0% to 20%). The second model (i.e., 0% to 20%; RPD of 4.3) was defined as a better model than the first model (i.e., 0% to 2%; RPD of 1.7). The RPD gave a better indication of the performance of PLS regression models, similar to those computed by Williams and Norris (2001) and Natsuga and Kawamura (2006) over an implementation of NIR technology study.
NIRS has the potential to quantitatively estimate boron-based preservative concentrations of Eastern black spruce (Picea mariana var. mariana) pellets. The overtone vibration of main peaks of NIR absorbance spectra were decreased in relation to boron retention in the black spruce pellets treated with glycol borate-based DOT. We found significant PLS models that linked absorbance spectra to boron content in the wood pellet products from 0 to 2 percent and from 0 to 20 percent DOT glycol borate. The best prediction model was performed for the model ranging from 0 to 20 percent DOT glycol borate, suggesting that NIRS should constitute an easy-to-operate analytical tool to cover the prediction of the largest range of DOT glycol borate for the preservation industry. We found some preliminary results in the prediction of the boron-based preservation formulation in the wood pellet products. Such an approach is still empirical and further work still needs to be developed as physics-based models to explain boron-based preservation influences on the spectra. We tested the method over only one borate formulation and one wood derivative product. Many other investigations on various formulations and wood treatment samples are needed.
The authors greatly appreciate the input of Dr. Daniella Tudor and Nicolas Tanguy from the Faculty of Forestry (University of Toronto, Toronto), Dr. Delphine Dufour from the Faculty of Dentistry (University of Toronto), Frank Rinker from FPInnovations (Vancouver), and Armand LaRocque from the Faculty of Forestry and Environmental Management (University of New Brunswick, Fredericton) for their support with the technical laboratory assistance, and the financial assistance of the NSERC Strategic Network on Innovative Wood Products and Building Systems (NewBuildS), Canada.
Adedipe, O. E. and B. Dawson-Andoh. 2008. Predicting moisture content of yellow poplar (Liriodendron tulipifera L.) veneer using near-infrared spectroscopy. Forest Prod J. 58(4):28. http://connection.ebscohost.com/c/articles/31972096/ predicting-moisture-contentyellow-poplar-liriodendron-tulipifera-l-veneer-using-near-infra redspectroscopy.
American Wood-Preservers' Association (AWPA). 1996. Sample preparation for determining penetration of preservatives in wood. AWPA Standard A19-93. AWPA, Selma, Alabama.
American Wood-Preservers' Association (AWPA). 2000. Standard methods for the analysis of wood and wood treating solutions by inductively coupled plasma emission spectrometry. AWPA Standard A21-00. AWPA, Selma, Alabama.
Awoyemi, L., T. Y. Ung, and P. A. Cooper. 2009. In-treatment cooling during thermal modification of wood in soy oil medium: Soy oil uptake, wettability, water uptake and swelling properties. Eur. J. Wood Prod. 67(4):465--470. DC)I:10.1007/s00107-009-0346-9
Bachle, H., B. Zimmer, E. Windeisen, and G.Wegener. 2010. Evaluation of thermally modified beech and spruce wood and their properties by FT-NIR spectroscopy. Wood Sci. Technol. 44(3):421-433. DOI:10. 1007/s00226-010-0361-3
Boulesteix, A.-L. and K. Strimmer. 2006. Partial least squares: A versatile tool for the analysis of high dimensional genomic data. Brief. Bioinform. 8(l):32-44. D01:10.1093/bib/bbl016
Burns, D. A. and E. W. Ciurczak. 2007. Handbook of Near-Infrared Analysis. 3rd ed. Pratical Spectroscopy Series 35. CRC Press, Boca Raton, Florida. 834 pp.
Cameiro, M. E., W. L. E. Magalhaes, G. I. B. de Muniz, and L. R. Schimleck. 2010. Near infrared spectroscopy and chemometrics for predicting specific gravity and flexural modulus of elasticity of Pinus spp. veneers. J. Near Infrared Spectrosc. 18(6):481-489. DOI:10. 1255/jnirs.911
Cozzolino, D., H. E. Smyth, and M. Gishen. 2003. Feasibility study on the use of visible and near-infrared spectroscopy together with chemometrics to discriminate between commercial white wines of different varietal origins. J. Agric. Food Chem. 51(26):703-708. DOI: 10.1021/jfD34959s
Diouf, P.-N., A. Merlin, and D. Perrin. 2006. Antioxidant properties of wood extracts and colour stability of woods. Ann. Forest Sci. 63(5):525. DOI: 10.1051/forest:2006035
Engstrom, B., B. Johnson, M. Hedquist, M. Grothage, H. Sundstrom, and A. Arlebrandt. 1998. Process modeling system for particleboard manufacturing, incorporating near infrared spectroscopy on dried wood particles. In: Proceeding of the Eleventh Biennial Southern Silvicultural Research Conference, K. W. Outcalt (Ed.), March 20-22, 2002, Knoxville, Tennessee; USDA Forest Service, Southern Research Station, Asheville, North Carolina, pp. 180-187.
Esbensen, K. H., D. Guyot, F. Westad, and L. P. Houmoller. 2002. Multivariate Data Analysis--In Practice: An Introduction to Multivariate Data Analysis and Experimental Design. 5th ed. CAMO Process AS, Oslo. 598 pp.
Faber, N. M. and R. Bro. 2001. Standard error of prediction for multiway PLS. 1. Background and simulation study. Chemometr. IntelI. Lab. Syst. 61(1-2):133-149. DOI: 10.1016/S0169-7439(01)00204-0
Faber, N. M., X.-H. Song, and P. K. Hopke. 2003. Sample-specific standard error of prediction for partial least squares regression. Trends Anal. Chem. 20(5):I-5. D01:10.1016/S0165-9936(03)00503-X
Fackler, K., M. Schmutzer, L. Manoch., M. Schwanninger, B. Hinterstoisser, T. Ters, K. Messner, and C. Gradinger. 2007. Evaluation of the selectivity of white rot isolates using near infrared spectroscopic techniques. Enzyme Microb. Technol. 41 (6):881-887. DOI: 10.1016/j.enzmictec.2007*07.016
Faix, O. 1992. Fourier transform infrared spectroscopy. In: Methods in Lignin Chemistry. Springer Series in Wood Science. Springer-Verlag, Berlin, pp. 83-109. http://link.springer.com/chapter/10.1007/978-3642-74065-7_7#page-1.
Feldhoff, R., T. Huth-Fehre, and K. Cammann. 1998. Detection of inorganic wood preservatives on timber by near infrared spectroscopy. J. Near Infrared Spectrosc. 6(1): 171-173. DOl:10.1255/jnirs.189
Felix, J. M. and P. Gatenholm. 1991. The nature of adhesion in composites of modified cellulose fiber and polypropylene. J. Appl. Polym. Sci. 42(3):609-620. DOI: 10.1002/app. 1991.070420307
Forato, L. A., R. Bernardes-Filho, and L. A. Colnago. 1998. Protein structure in KBr pellets by infrared spectroscopy. Anal. Biochem. 259(1): 136-141. DOI: 10.1006/abio. 1998.2599
Geladi, P. and B. Kowalski. 1986. Partial least-squares regression: A tutorial. Anal. Chim. Acta 185:1-17. DOI: 10.1016/00032670(86)80028-9
Gerardin, C., T. Koumbi Mounanga, and P. Gerardin. 2008. Effect of amphiphilic antioxidant alkyl ammonium ascorbate on inhibition of fungal growth: Application to wood preservatives formulation. IRG/ WP 08-30466. In: Proceedings of the 39th Annual Meeting, May 25-29, 2008, Istanbul, Turkey. 20 pp.
Guo, W. 2013. Self-heating and spontaneous combustion of wood pallets during storage. PhD dissertation. University of British Columbia, Vancouver, British Columbia, Canada. 214 pp. DOI: 10.14288/1. 0073583
Hans, G., B. Leblon, R. Stirling, J. Nader, A. LaRocque, and P. Cooper. 2013. Monitoring of moisture content and basic specific gravity in black spruce logs using a hand-held MEMS-based near-infrared spectrometer. Forestry Chron. 89(5):607-620. DOI:10.5558/tfc2013112
Hein, P. R. G., A. C. M. Campos, R. F. Mendes, L. M. Mendes, and G. Chaix. 2011. Estimation of physical and mechanical properties of agro-based particleboards by near infrared spectroscopy. Eur. J. Wood Prod. 69(3):431-442. DOI:10.1007/s00107-010-0471-5
Hettipathirana, T. D. 2004. Simultaneous determination of parts-per-million level Cr, As, Cd and Pb, and major elements in low level contaminated soils using borate fusion and energy dispersive X-ray fluorescence spectrometry with polarized excitation. Spectrochim. Acta Part B: At. Spectrosc. 59(2): 223-229. Bibcode: 2004AcSpe.59.223H. DOI: 10.1016/j.sab.2003.12.013
Koumbi-Mounanga, T., K. Groves, B. Leblon, G. Zhou, and P. A. Cooper. 2015a. Estimation of moisture content of trembling aspen (Populus tremuloides Michx.) strands by near infrared spectroscopy. Eur. J. Wood Prod. 73:43-50. DOI: 10.1007/s00107-014-0856-y
Koumbi-Mounanga, T., P. I. Morris, J. L. Myung, N. M. Saadat, B. Leblon, and P. A. Cooper. 2015b. Prediction and evaluation of borate distribution in Eastern black spruce (Picea mariana var. mariana) wood products. Wood Sci Technol. 49:457-573.
Koumbi-Mounanga, T., T. Ung, P. Cooper, B. Leblon, and K. Groves. 2015c. Surface quality sensing of trembling aspen (Populus tremuloides Michx.) veneer products by near infrared spectroscopy. Wood Mater. Sci. Eng. 10(1): 17-26. http://dx.doi.org/10.1080/17480272. 2014.923936.
Koumbi-Mounanga, T., T. Ung, K. Groves, B. Leblon, and P. Cooper. 2013. Moisture and surface quality sensing of Douglas-fir (Pseudotsuga menziesii var. menziesii) veneer products. Forestry' Chron. 89(5):646-653. D01:10.5558/tfc2013-116
Leblon, B., O. Adedipe, G. Hans, A. Haddadi, S. Tsuchikawa, J. Burger, R. Stirling, Z. Pirouz, K. Groves, J. Nader, and A. LaRocque. 2013. A review of near infrared spectroscopy for monitoring moisture content and density of solid wood. Forestry Chron. 89(5):595-606. DOL10. 5558/tfc2013-111
Lee, S. B. and P. Luner. 1972. The wetting and interfacial properties of lignin. Tappi 55(1): 116-121.
Meder, R., A. Thumm, and H. Bier. 2002. Veneer stiffness predicted by NIR spectroscopy calibrated using mini-LVL test panels. FIolz RohWerkst. 60(3): 159-164. D01:10.1007/s00107-002-0296-y Mikeal, K. and J. Ingegard. 2010. Surfactants from Renewable Resources. 1st ed. John Wiley & Sons, New York. 303 pp.
Mitsui, K., T. Inagaki, and S. Tsuchikawa. 2008. Monitoring of hydroxyl groups in wood during heat treatment using NIR spectroscopy. Biomacromolecules 9(1):286--288. DOL10.10217008069 Mounanga, T. K. 2008. Original antioxidant amphiphilic compounds for wood preservative formulations. PhD dissertation. University of Lorraine, Faculty of Sciences and Technology previously, University Henri Poincare (UHP), Nancy, France. 223 pp. (In French.) http://
Mounanga, T. K. 2015. Antioxidant amphiphilic compounds and wood preservative formulations. Book No. 3879. Presses Academiques Francophones. 236 pp. (1279 of 1456). (In French.) https://www. presses-aeademiques.eom/catalog/details//store/fr/book/978-3-84163096-4/ tensioactifs-antioxydants-et-produits-de-pr%C3%A9servation-du-bois.
Mounanga, T. K., P. Gerardin, B. Poaty, D. Perrin, and C. Gerardin. 2008. Synthesis and properties of antioxidant amphiphilic ascorbate salts. Colloids Surf. A Physicochem, Eng. Asp. 318(1-3)134. DOI: 10. 1016/j.colsurfa.2007.12.048
Mounanga, T. K., T. Ung, K. Groves, P. Cooper, and B. Leblon. 2012. Moisture and surface quality sensing for improved manufacturing control of composite wood products. In: Proceedings of the 11th Pacific Rim Bio-based Composites Symposium (BIOCOMP), November 27-30, 2012, Shizuoka-City, Japan. 15 pp.
Mounanga, T. K., T. Ung, R. Shafaghi, P. A. Cooper, and B. Leblon. 2015. Estimation of bending stress in earlywood and latewood growth rings of oil thermally treated wood by near infrared spectroscopy. J. Mater. Sci. Appl. 1 (3): 114-123.
Mounguengui, S., S. Dumargay, and P. Gerardin. 2007. Investigation on catechin as a beech wood decay biomarker. Interact. Biodeter. Biodegrad. 60(4):238-244. D01:10.1016/j.ibiod.2007.03.007
Natsuga, M. and S. Kawamura. 2006. Visible and near-infrared reflectance spectroscopy for determining physicochemical properties of rice. Trans. ASABE 49(4): 1069-1076.
Ondo, J. P., L.-C. Obame, T. Andzibarhe, G. Nzi Akoue, E. Nzi Emvo, and J. Lebibi. 2013. Phytochemical screening, total phenolic content and antiradical activity of asplenium africanum (Aspleniacea) and fruit of megaphrinium macrostachyum (Marantacea). J. Appl. Pharmaceut. Sci. 3(08):092-096. DOE10.7324/JAPS.2013.3816
Poaty, B., S. Dumargay, P. Gerardin, and D. Perrin. 2010. Modification of grape seed and wood tannins to lipophilic antioxidant derivatives. Ind. Crops Prod. 31(3)509. D01:10.1016/j.indcrop.2010.02.003
Rosipal, R. and N. Kramer. 2006. Overview and recent advances in partial least squares. In: Subspace, Latent Structure and Feature Selection. Lecture Notes in Computer Science Series Vol. 3940. C. Saunders, M. Grobelnik, S. Gunn, and J. Shawe-Taylor (Eds.). Springer, Berlin, pp. 34-51. D01:10.1007/11752790_2. http://link. springer.com/chapter/10.1007/11752790_2#.
Saadat, N. and P. Cooper. 2013. Factors affecting distribution of borate to protect building envelope components from biodegradation. NewBuilds. Project code: T4-5-C10. 2 pp. http://newbuildscanada.ca/wpcontent/uploads/2013/06/Tech-Note 1-FACTORS-AFFECTING DISTRIBUTION-OF-BORATE-TO-PROTECT-BUILDINGENVELOPE-COMPONENTS-FROM-BIODEGRADATION.pdf.
Salman, S., A. Petrissans, M. F. Thevenon, S. Dumargay, D. Perrin, B. Pollier, and P. Geradin. 2014. Development of new wood treatments combining boron impregnation and thermo modification: effect of additives on boron leachability. Eur. J. Wood Prod. 72:355-365. DOI: 10.1007/s00107-014-0787-7
Schwanninger, M., B. Hinterstoisser, N. Gierlinger, R. Wimmer, and J. Hanger. 2004. Application of fourier transform near infrared spectroscopy (FT-NIR) to thermally modified wood. Holz Roh-Werkst. 62(6):483-485.
Schwanninger, M., J. Rodrigues, and K. Fackler. 2011. A review of band assignments in near-infrared spectra of wood and wood components. J. Near Infrared Spec. 19(5):287-308.
So, C.-L., S. T. Lebow, L. H. Groom, and T. G. Rials. 2004. The application of near infrared (NIR) spectroscopy to inorganic preservative-treated wood. Wood Fiber Sci. 36(3):329-336.
Stansbury, J. W and S. H. Dickens. 2001. Determination of double bond conversion in dental resins by near infrared spectroscopy. Dental Mater. 17(1):71 79. DC>I:10.106/S0109-5641(00)00062-2
Stevens, J. and D. J. Gardner. 2010. Enhancing the fuel value of wood pellets with the addition of lignin. Wood Fiber Sci. 42(4):439-443. Stirling, R. 2013. Near-infrared spectroscopy as a potential quality assurance tool for the wood preservation industry. Forestry Chron. 89(5):654-658. DOI: 10.5558/tfc2013-117
Taylor, A. and J. Lloyd. 2007. Potential of near infrared spectroscopy to quantify boron concentration in treated wood. (Technical note.) Forest Prod. J. 57:414-423.
Thevenon, M. F. and A. Pizzi. 2003. Polyborate ions' influence on the durability of wood treated with non-toxic protein borate preservatives. Holz Roh- Werkst. 61(6):457-464. DOL10.1007/S00107-003-0421-6
Thygesen, L. G. and S. O. Lundqvist. 2000. NIR measurement of moisture content in wood under unstable temperature conditions. Thermal effects in near-infrared spectra of wood. Part I. J. Near Infrared Spectrosc. 8(3): 183-189. DOI:10.1255/jnirs.277
Tsuchikawa, S. and M. Schwanninger. 2013. A review of recent near-infrared research for wood and paper. Part 2. Appl. Spectrosc. Rev. 48(7):560-587. DOI:http://dx.doi.org/10.1080/05704928.2011.621079 Watanabe, K., S. D. Mansfield, and S. Avramidis. 2011. Application of near-infrared spectroscopy for moisture-based sorting of green hem-fir timber. J. Wood Sci. 57(4):288-294. DOI: 10.1007/s 10086-011-1181-2 Williams, P. and K. H. Norris. 2001. Near-Infrared Technology in the Agricultural and Food Industries. AACC International, St. Paul, Minnesota. 296 pp.
The authors are, respectively, Researcher, Professor Emeritus, and Professor, Faculty of Forestry, Univ. of Toronto, Toronto, Ontario, Canada (email@example.com [corresponding author], firstname.lastname@example.org, email@example.com); Senior Scientist, FPInnovations, Wood Products Lab., Vancouver, British Columbia, Canada (Kevin.firstname.lastname@example.org); Researcher, Faculty of Forestry, Univ. of Toronto, Toronto, Ontario, Canada (email@example.com); and Professor, Faculty of Forestry and Environ. Manag., Univ. of New Brunswick, Fredericton, New Brunswick, Canada (firstname.lastname@example.org). This paper was received for publication in May 2014. Article no. 14-00048. [C]Forest Products Society 2016. Forest Prod. J. 66(1/2):37-43.
Table 1.--Regression parameters of the linear relationship of Figure 3 over the partial least-square regression (calibration and validation) models for boron concentrations.3 Replicate Slope Intercept RMSE (%) SE (%) Calibration a 0.9484 0.3423 1.6109 1.62 b 0.6957 0.09938 0.30019 0.301 Validation a 0.9389 0.4138 1.6511 1.6569 b 0.67437 0.10476 0.33065 0.3314 Replicate Bias (%) [R.sup.2] P value RPD n Calibration a 0.0087 0.948 0.0001 4.4 144 b 0.0016 0.696 0.0001 1.8 249 Validation a 0.0086 0.947 0.0001 4.3 144 b -0.0016 0.636 0.0001 1.7 249 (a) Replicates: a = 0 to 20 percent concentration range (second model); (b) = 0 to 2 percent concentration range (first model). RMSE = root mean square error; SE = standard error; RPD = relative percent different; n = number of objects.
|Printer friendly Cite/link Email Feedback|
|Author:||Koumbi-Mounanga, Thierry; Groves, Kevin; Cooper, Paul A.; Ung, Tony; Yan, Ning; Leblon, Brigitte|
|Publication:||Forest Products Journal|
|Date:||Jan 1, 2016|
|Previous Article:||Characterization of residue and bio-oil produced by liquefaction of loblolly pine at different reaction times.|
|Next Article:||Characteristics of cotton stalk torrefaction catalyzed by magnesium chloride.|