Volume 5 Issue 3
Jul.  2020
Article Contents
Turn off MathJax

Citation:

Development of Chemometric Model for Characterization of Non-wood by FT-NIR Data

  • Corresponding author: Mohammad Nashir Uddin, m2nashir@yahoo.com
  • Received Date: 2020-02-13
  • In this study, a model for prediction of lignocellulose components of agricultural residues has been developed with Fourier Transformed Near Infrared (FT-NIR) spectroscopy data. Two calibration techniques (Principal Component Regression (PCR) and Partial Least Square Regression (PLSR)) were assessed for prediction of lignin, holocellulose, α-cellulose, pentosan and ash, and found the PLSR better for lignin, holocellulose and α-cellulose. The PCR also produced better results for quantification of pentosan and ash. Spectral range (7000-5000cm-1) showed more informative than other parts of the spectral data. The PLSR showed maximum value of R2 (R2=0.91%) for prediction of holocellulose. For the prediction of pentosan, the PCR was better (R2=0.68%). The PCR also showed better results (R2=86%) for quantification of ash. To determine amount of lignin, the PLSR was the best (R2=0.83%) when the spectral data were de-trained and smoothed with Savitzky-Golay (S-G) filtering simultaneously. For prediction of α-cellulose, the PLSR was the best model (R2=0.94%) when the data were pretreated with mean normalization. Considering the best alternatives inNear Infrared (NIR) data preprocessing and calibration techniques, methods for quantification of lignocellulose components of agricultural residues have been developed which is rapid, cost effective, and less chemical intensive and easily usable in pulp and paper industries and pulp testing laboratories.
  • 加载中
  • [1]

    Akhtaruzzamen, A.F.M., Shafi, M., 1993. Pulping of jute. Tappi J. 78, 106.
    [2]

    Alves, A., Santos, A., da Silva Perez, D., Rodrigues, J., Pereira, H., Simões, R., Schwanninger, M., 2007. NIR PLSR model selection for Kappa number prediction of maritime pine Kraft pulps. Wood Sci. Technol. 41, 491-499. doi: 10.1007/s00226-007-0130-0
    [3]

    Colares, C.J., Pastore, T., Coradin, V.T., 2015. Exploratory analysis of the distribution of lignin and cellulose in woods by Raman imaging and chemometrics. J. Braz. Chem. Soc. 26, 1297-1305.
    [4]

    Fahey, L.M., Nieuwoudt, M.K., Harris, P.J., 2019. Predicting the cell-wall compositions of solid Pinusradiata (radiata pine) wood using NIR and ATR FTIR spectroscopies. Cellulose 26, 7695-7716. doi: 10.1007/s10570-019-02659-8
    [5]

    Ferdous, T., Quaiyyum, M.A., Bashar, S., Jahan, M.S., 2020. Anatomical, morphological and chemical characteristics of kaun straw (Seetaria-ltalika). Nord. Pulp Pap. Res. J. DOI:10.1515/npprj-2019-0057.
    [6]

    Fiserova, M., Gigac, J., Russ, A., Maholanyiova, M., 2012. Using NIR analysis for determination of hardwood kraft pulp properties. Wood Research, 57, 121-130.
    [7]

    Hattori, T., Murakami, S., Mai, M.K., Yamada, T., Hirochika, H., Ike, M., Tokuyasu, K., Suzuki, S., Sakamoto, M., Umezawa, T., 2012. Rapid analysis of transgenic rice straw using near-infrared spectroscopy. Plant Biotechnol. 29, 359-366. doi: 10.5511/plantbiotechnology.12.0501a
    [8]

    He, L., Xin, L.P., Chai, X.S., Li, J., 2015. A novel method for rapid determination of alpha-cellulose content in dissolving pulps by visible spectroscopy. Cellulose 22, 2149-2156. doi: 10.1007/s10570-015-0652-9
    [9]

    Huang, C., Han, L., Liu, X., Ma, L., 2010. The rapid estimation of cellulose, hemicellulose, and lignin contents in rice straw by near infrared spectroscopy. Energy Sources Part A:Recover. Util. Environ. Eff. 33, 114-120. doi: 10.1080/15567030902937127
    [10]

    Hussain, M.A., Huq, M.E., Rahman, S.M., Ahmed, Z., 2002. Estimation of lignin in jute by titration method. Pak. J. Biol. Sci. 5, 521-522. doi: 10.3923/pjbs.2002.521.522
    [11]

    Jahan, M.S., 2009. Studies on the effect of prehydrolysis and amine in cooking liquor on producing dissolving pulp from jute (Corchoruscapsularis). Wood Sci. Technol. 43, 213-224. doi: 10.1007/s00226-008-0213-6
    [12]

    Jahan, M.S., Chowdhury, D.A.N., Islam, M.K., 2005. Alkalinesulphite anthraquinone methanol (ASAM) pulpig of jute. IPPTA J. 17, 37-43.
    [13]

    Jahan, M.S., Uddin, M.N., Akhtaruzzaman, A.F.M., 2016. An approach for the use of agricultural by-products through a biorefinery in Bangladesh. For. Chron. 92, 447-452. doi: 10.5558/tfc2016-080
    [14]

    Kothiyal, V., Jaideep, Bhandari, S., Ginwal, H.S., Gupta, S., 2015. Multi-species NIR calibration for estimating holocellulose in plantation timber. Wood Sci. Technol. 49, 769-793. doi: 10.1007/s00226-015-0720-1
    [15]

    Li, X.L., Sun, C.J., Zhou, B.X., He, Y., 2015. Determination of hemicellulose, cellulose and lignin in moso bamboo by near infrared spectroscopy. Sci. Rep. 5, 17210. doi: 10.1038/srep17210
    [16]

    Nuruddin, M., Chowdhury, A., Haque, S., Quaiyyum, M.A., 2011.Extraction and characterization of cellulose microfibrils from agricultural wastes in an integrated biorefinery initiative. Cellulose Chemistry and Technology. 45, 347-354.
    [17]

    Poke, F.S., Raymond, C.A., 2006. Predicting extractives, lignin, and cellulose contents using near infrared spectroscopy on solid wood in Eucalyptusglobulus. J. Wood Chem. Technol. 26, 187-199. doi: 10.1080/02773810600732708
    [18]

    Rajesh, K., Ray, A.K., 2006. Artificial neural network for solving paper industry problems:a review. Journal of Scientific & Industrial Research. 65, 565-573.
    [19]

    Roy, T.K., Mohindru, V.K., Behera, N.C., 1998. Jute for specialty pulp. IPPTA J. 10, 81-86.
    [20]

    Saijonkari-Pahkala, K., 2001. Non-wood plants as raw materials for pulp and paper. Agricultural and Food Science in Finland. 10, 1-97.
    [21]

    Santos, A.J.A., Anjos, O., Pereira, H., 2016. Prediction of blackwood Kraft pulps yields with wood NIR-PLSR models. Wood Sci. Technol. 50, 1307-1322. doi: 10.1007/s00226-016-0837-x
    [22]

    Silva, J.C., Nielsen, B.H., Rodrigues, J., Pereira, H., Wellendorf, H., 1999. Rapid determination of the lignin content in sitka spruce (Piceasitchensis (bong.) carr.) wood by Fourier transform infrared spectrometry. Holzforschung 53, 597-602. doi: 10.1515/HF.1999.099
    [23]

    Uddin, M.N., Ahmed, S., Ray, S., Islam, M.S., Quadery, A.H., 2019. Method for predicting lignocellulose components in jute by transformed FT-NIR spectroscopic data and chemometrics. Nordic Pulp & Paper Research Journal 34, 1-9.
    [24]

    Uddin, M.N., Ray, S.K., Islam, M.S., Nayeem, J., Jahan, M.S., 2017. Development of method for rapid prediction of chemical components of dhaincha using FT-NIR spectroscopy and chemometrics. J. Sci. Technol. For. Prod Process. 6, 22-28.
    [25]

    Yeh, T.F., Yamada, T., Capanema, E., Chang, H.M., Chiang, V., Kadla, J.F., 2005. Rapid screening of wood chemical component variations using transmittance near-infrared spectroscopy. J. Agric. Food Chem. 53, 3328-3332.
  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(8) / Tables(2)

Article Metrics

Article views(216) PDF downloads(8) Cited by()

Related
Proportional views

Development of Chemometric Model for Characterization of Non-wood by FT-NIR Data

    Corresponding author: Mohammad Nashir Uddin, m2nashir@yahoo.com
  • a. BCSIR Laboratories, Dhaka, Bangladesh Council of Scientific and Industrial Research (BCSIR), Dhaka-1205, Bangladesh
  • b. Department of Applied Chemistry and Chemical Engineering, University of Dhaka, Dhaka-1000, Bangladesh

Abstract: In this study, a model for prediction of lignocellulose components of agricultural residues has been developed with Fourier Transformed Near Infrared (FT-NIR) spectroscopy data. Two calibration techniques (Principal Component Regression (PCR) and Partial Least Square Regression (PLSR)) were assessed for prediction of lignin, holocellulose, α-cellulose, pentosan and ash, and found the PLSR better for lignin, holocellulose and α-cellulose. The PCR also produced better results for quantification of pentosan and ash. Spectral range (7000-5000cm-1) showed more informative than other parts of the spectral data. The PLSR showed maximum value of R2 (R2=0.91%) for prediction of holocellulose. For the prediction of pentosan, the PCR was better (R2=0.68%). The PCR also showed better results (R2=86%) for quantification of ash. To determine amount of lignin, the PLSR was the best (R2=0.83%) when the spectral data were de-trained and smoothed with Savitzky-Golay (S-G) filtering simultaneously. For prediction of α-cellulose, the PLSR was the best model (R2=0.94%) when the data were pretreated with mean normalization. Considering the best alternatives inNear Infrared (NIR) data preprocessing and calibration techniques, methods for quantification of lignocellulose components of agricultural residues have been developed which is rapid, cost effective, and less chemical intensive and easily usable in pulp and paper industries and pulp testing laboratories.

1.   Introduction
  • Non-wood plants have been used as raw material for pulp and paper industries in many countries of the world (Saijonkari-Pahkala, 2001). Being an agricultural country, Bangladesh is generating substantial number of agricultural residues such as rice straw, wheat straw, bagasse, which could be a potential raw material in pulp and paper industries in Bangladesh (Jahan et al., 2016).

    The chemical compositions of lignocellulosics determine its suitability for pulping. The chemical characteristics of agricultural wastes generated in Bangladesh like rice straw, wheat straw, corn stalks, Dhaincha, were determined in order to assess the suitability of these non-wood raw materials for pulp and cellulose industries (Nuruddin et al., 2011; Ferdous et al., 2020). Cellulose is the prime factor of a pulping raw material, which determines the pulp yield (Ferdous et al., 2020), on the other hand, lignin and ash are undesirable parts of the raw materials. Pentosan is also important part in selecting pulping the materials, which facilitates fiber bonding. Moreover, many studies also have been done on pulping of agricultural residues (Akhtaruzzamen and Shafi, 1993; Roy et al., 1998; Jahan et al., 2005; Jahan, 2009). Based on the available literature, chemical components are determined by traditional wet laboratory method (Hussain et al., 2002), which are tedious, time consuming and involves of hazardous chemicals. Therefore, the possibility of using easy, rapid, cost-effective techniques which could predict chemical characteristics of agricultural residues was tempting.

    Recently interest is growing onnon-destructive methods for chemical characterization of pulp, wood and non-wood using spectroscopy and chemometric (Rajesh and Ray, 2006; Hattori et al., 2012; Fahey et al., 2019). Application of infrared spectroscopy as an alternative approach allows for fast and reliable analysis by non-destructive means (Fiserova et al., 2012). This technique requires much smaller samples than those necessary for analysis by conventional wet methods. An advantage to the method is also the fact that in most cases no sample preparation prior to analysis is needed.

    In recent years, spectroscopic instruments such as Near Infrared (NIR), Raman, Ultra-violate (UV) spectroscopy are used to predict these parameters in different wood samples with the assistance of multivariate techniques (Silva et al., 1999; Poke and Raymond, 2006; Alves et al., 2007; Colares et al., 2015; He et al., 2015; Li et al., 2015; Kothiyal et al., 2015; Santos et al., 2016; Uddin et al., 2017, 2019). For lignin content determination in Sitka Spruce, a chemometric method was developed by Silva et al. (1999) by Fourier Transform Infrared (FT-IR) Spectrometry and Principal Component Regression (PCR). Poke and Raymond (2006) developed a chemometric method with NIR spectral data and linear calibration for determination of lignin and cellulose. Another method was developed by Li et al. (2015) with Artificial Neural Network (ANN) in NIR spectroscopic data for determination of hemicelluloses, cellulose and lignin in Moso Bambo. A few researches have been conducted to quantify these components in non-wood samples using chemometric models (Huang et al., 2010; Uddin et al., 2017, 2019).

    Now it is necessary to develop universal models for quantification of lignin, holocellulose, α-cellulose, pentosan and ash which would be applicable for all type of agricultural residues, potential raw materials for pulp industries. But none of the research has been carried out to develop such spectroscopic method for this purpose by chemometric techniques.

    Therefore, the present study is attempting to develop cost effective, non-destructive and rapid method for quantification of lignocellulose components in agricultural residues based raw materials of pulp industries with NIR spectroscopic data of agricultural residues, potential non-wood raw materials for pulp and paper industries and chemometric modeling techniques.

    In order to develop models for quantification of lignocelluloses components (namely, lignin, holocellulose, α-cellulose, pentosan, extractives and ash), efficiency of two calibration techniques, PCR and Partial Least Square Regression (PLSR), have been compared. At the same time, most informative region of spectral data has been chosen by calibrating model with different segments of the full spectral data as a part of dimension reduction of data matrix. Finally, different de-noising techniques have been assessed to get the best performing model with pretreated data.

2.   Materials and methods
  • Twenty-two distinct non-wood samples, mostly agriculture waste were collected from different parts of the country. The samples are: 1) bamboo (Bambusoideae); 2) bagasse (Saccharum); 3) rice (Oryza sativa) straw; 4) wheat (Triticumaestivum) straw; 5) corn (Zea mays) stalks; 6) mustard (Brassica juncea) stalks; 7) banana (Musa cavendish) pseudo stem; 8) banana leaf; 9) banana; 10) eggplant (Solanummelongena) stalks peduncle; 11) chia (Salvia hispanica) stalks; 12) kash (Saccharumspontaneum) stalks; 13) kaun (Seetaria-ltalika) straw; 14) okra (Abelmoschus esculentus) stalks; 15) jute (Corchorus) fiber; 16) jute sticks; 17) cotton (Gossypium) stalks; 18) red lentil (Lensculinaris) stalks; 19) dhaincha (Sesbaniaaculeata) stalks; 20) pineapple (Ananascomosus) stalks; 21) mulberry (Morus) plant; and 22) cassava (Manihot esculenta) plant.

  • The agricultural residues were collected from Mymensingh region of Bangladesh. These residues were chopped to thesize of 2-3 cm. Then they were ground in a Wiley mill of about 40-60 mesh size which were used for chemical analysis.

    The chemical compositions of these agricultural residues were carried out by following Technical Association of Pulp and Paper Industries (Tappi) Test Methods: the extractive (T204 om88), klason lignin (T211 om83), ash content (T211 om93) and pentosan (T223). The holocellulose was prepared by treating extractive free wood meal with NaClO2 solution (Rajesh and Ray, 2006). The pH of the solution was maintained at 4 by adding CH3COOH-CH3COONa buffer and cellulose was determined by treating holocellulose with 17.5% NaOH (T203 om93).

  • The FT-NIR spectroscopy was performed using a PerkinElmer FT-NIR spectrometer (Model: Frontier, Perkin Elmer, USA) with Indium Gallium Arsenide (InGaAs) detector. The spectral range used was 10, 000-4000 cm-1. For each sample, 32 scans were collected at a spectral resolution of 16 cm-1 with an interval of 4 cm-1, then the 32 scans were averaged and stored as reflectance percentage (%R). Here PerkinElmer Spectrum (Version 10.4.4) software was used for spectral data processing. Reflectance percentage of light by certain sample against each wave number was the main consideration for spectral data.

    In each spectrum there was huge number of reflectance value for each wave point. Each wave point or data point was considered as spectroscopic variable. These variables were huge in number and were mutually correlated. Therefore, they required dimension reduction and preprocessing.

  • After acquisition of spectroscopic data, they had been be preprocessed with different de-noising techniques commonly used in chemometric techniques.

  • Three segments of the full spectral data were taken to find the most informative part of the dataset. They were 8000-7000, 7000-5500 and 5200-4060 cm-1. Calibration was done with full range (8000-4000 cm-1) and other three parts of the dataset.

  • The data were smoothed and denoised with mean normalization, de-trending, de-trending and smoothing with Savitzky-Golay (S-G) filtering as well asde-trendingand second derivative together.

  • Two most popular chemometric calibration techniques namely Principal Component Regression (PCR) and Partial Least Square Regression (PLSR) were assessed for quantification of properties of pulp like lignin, holocellulose, α-cellulose, pentosan and ash with NIR spectral data, and selected the best performing one for method development.

    Developed models for quantification of lignin, holocellulose, α-cellulose, pentosan and ash were validated by creating 6-fold cross validation dataset in each case. All the samples were presented in different colored line diagram in the Fig. 1. Data of reflectance percentage are plotted against wavelength.

    Figure 1.  Raw Fourier Transformed Near Infrared (FT-NIR) spectral data.

3.   Results and discussion
  • The PCA, a multivariate statistical tool that allows us to summarize the information content in large dataset by means of smaller set of summary indices called Principal Components. It has been widely used for pattern recognition, signal processing and factor analysis. In practical, the PCA has been used in this study to find multicollinearity, outliers in the dataset and finally, to development models with this principal component.

    From the PCA it could be found that the first principal component 69% and second one 23%, and both together expressed 92% of the total variation in the data (Fig. 2).

    Figure 2.  Score plot of FT-NIR data of agricultural residues.

    Score plot of the PCA, also known as map of samples, showed that samples were scattered without following any patter which was the indicator that there was no multicollinearity in the data (Fig. 2). Therefore, the spectroscopic data were suitable for model development.

    Influence plot is popularly used to diagnose whether there is any outlier in the dataset which might have adverse impact on the developed models. From Fig. 3, it could befound that there was no outlier in the FT-NIR spectroscopic data of agricultural residues as there was no sample plotted beyond the read lines.

    Figure 3.  Influence plot of FT-NIR data of agricultural residues.

    All parts of spectral data of a sample are not equally informative. Therefore, in order to find the most important section of the spectral data, full spectra were divided into three parts and calibration models were developed with data of full spectral data, and the segmented ranges. Efficiency of the two calibration techniques PCR and PLSR were assessed with the full range data (8000-4000 cm-1) and partial datasets 8000-7000, 7000-5500 and 5200-4060 cm-1. Calibrations were also done separately with calibration dataset and 6-fold cross validation (CV) datasets. Coefficients of multiple determination (R2) is one of the parameters to test the model efficiency. Here, comparisons were shown in terms of the values of R2.

    Among full range and all three segmented ranges of spectral data most information range is 7000-5500 cm-1 (Table 1) for prediction models of lignin, holocellulose, α-cellulose, pentosan, extractive and ash in different agricultural wastes, potential raw materials for pulp industries. In another study, NIR range 7502-4246 cm-1 was found the most informative region for predicting holocellulose in plantation timber (Kothiyal et al., 2015).

    ParameterDatasetFull range (8000–4000 cm-1)8000–7000 cm-17000–5500 cm-15200–4060 cm-1
    PCRPLSRPCRPLSRPCRPLSRPCRPLSR
    LigninCalibration0.2420.2620.2280.2370.4720.8060.6630.778
    CV0.1410.0920.1500.0960.3010.3730.1780.404
    HolocelluloseCalibration0.7460.8820.0060.0160.7060.9050.9090.909
    CV0.2150.4730.0000.0000.5030.6610.6570.691
    α-celluloseCalibration0.7810.7450.1930.5690.7410.9210.8780.879
    CV0.2710.2080.0000.0570.4990.7640.3780.426
    PentosanCalibration0.0140.0160.0010.0190.6780.6280.5610.527
    CV0.0000.0000.0000.0000.3610.2510.1750.078
    ExtractiveCalibration0.0060.0320.0050.0220.2630.2790.0160.035
    CV0.0000.0000.0000.0000.0620.0940.0000.000
    AshCalibration0.5800.5490.3920.3990.8560.8560.7200.834
    CV0.3290.1900.1410.2030.4930.5020.4250.476
    Notes: PCR, Principal Component Regression; PLSR, Partial Least Square Regression; CV, cross validation.

    Table 1.  Range selection of FT-NIR data.

    Between two most popular calibration techniques PCR and PLSR, the later one showed better prediction efficiency in all cases with calibration and cross-validation datasets except pentosan. In case of the pentosan, the values of R2 were very close, but a slight better result could be found for the PCR (R2 = 68%) over the PLSR (R2 = 63%). Similar result could be noticed for Ash prediction model.

  • Spectral data contain some noise or unwanted background information which should be reduced or eliminated in order to improve the robustness of calibration techniques used in this study. Therefore, the FT-NIR spectral data had gone through some pretreatment processes, namely, mean normalization, de-trending, de-trending and smoothing with Savitzky-Golay (S-G) technique combined and finally, de-trending and 2nd derivative together. Like earlier, efficiency of the PCR and PLSR are assessed both for calibration and cross-validation datasets.

    As shown in Table 2, calibration model for prediction of α-cellulose in agricultural residues is the best when the FT-NIR spectral data are preprocessed with mean normalization with calibration dataset (R2 = 94%) and calibration dataset (R2 = 79%). For lignin prediction model, detrending and smoothing with Savitzky-Golay (S-G) techniques showed the best results (R2 = 83%) when they were used simultaneously. However, a severe effect could be observed when the data were pretreated with detrending and 2nd derivative combined. Here R2 was tremendously high (R2 ≈ 99%) for calibration dataset but for cross-validation data, and the figure was horrifically low, which made the model very unstable for using to predict these parameters from FT-NIR spectral data. Calibration performance was found very poor in terms of R2 for prediction of α-cellulose (65%-82%) and lignin (52%-95%) in solid wood without any pre-treatment (Yeh et al., 2005).

    ParameterDatasetMean normalizationDe-trendingDe-trending+ smoothing (S-G)De-trending+ 2nd derivative
    PCRPLSRPCRPLSRPCRPLSRPCRPLSR
    LigninCalibration0.4080.8420.1360.8310.1360.8290.3400.412
    CV0.1480.3920.0610.2430.0520.3380.1410.120
    HolocelluloseCalibration0.7580.7720.6330.9150.6330.9120.7890.998
    CV0.2750.4880.2100.5250.2350.5010.1110.123
    α-celluloseCalibration0.9210.9370.8410.9030.8410.9020.8630.998
    CV0.7890.7850.7340.7240.7720.7860.1990.292
    PentosanCalibration0.6390.6580.5110.4660.5100.4660.5660.976
    CV0.4130.3600.2530.2010.2720.2540.0850.096
    ExtractiveCalibration0.2500.2590.1240.1570.1240.1570.0720.346
    CV0.1230.1640.0090.000.0000.0000.0000.000
    AshCalibration0.6040.5830.7920.8030.7930.4720.4430.999
    CV0.3630.2830.4120.4220.4190.3520.3430.340
    Notes: PCR, Principal Component Regression; PLSR, Partial Least Square Regression; CV, cross validation.

    Table 2.  Pre-treatment of FT-NIR spectral data.

  • For prediction of lignin, the PLSR calibration model showed better performance than the PCR with FT-NIR spectral data detrended and smoothed with Savitzky-Golay (S-G) filtering simultaneously (R2 = 83%) (Fig. 4). It requiredseven factors to reach the destination of getting the best performing model.

    Figure 4.  Model for prediction of lignin (Blue line indicates calibration model; red line indicates validation model; and the black line indicates target line. This connotations are same in other figures as well (Figs. 5-8). Here 'Y' is the data matrix of dependent variable (lignin).

    Holocellulose could be predicted with raw spectral data of range 7000-5500 cm-1 with the PLSR (R2 = 91%) (Fig. 5). Here, model neededseven factors for getting the best performing model.

    Figure 5.  Model for prediction of holocellulose.

    To predict α-cellulose non-wood samples especially agricultural residues, the PLSR performed better than the PCR with FT-NIR spectral data of range 7000-5500 cm-1when they were preprocessed with mean normalization (R2 = 94%) (Fig. 6). The model requiredseven factors to perform best. The result obtained under this study for predicting α-cellulose was very close to another study conducted by Uddin et al. (2017) with Dhaincha sample (R2 = 95%) and Jute sample (R2 = 99%) where FT-NIR spectral data were pre-processed with standard normal variate (SNV) and calibrated with PLSR and arterial neural network (ANN) respectively.

    Figure 6.  Model for prediction of α-cellulose.

    The PCR performed better than the PLSR with raw spectral data of range 7000-5500 cm-1 (Fig. 7). The model consisted of first six principal components (PCs) but the model was not recommendable one to use in practice as its predictive performance was not up to the mark (R2 = 68%).

    Figure 7.  Model for prediction of pentosan.

    Lastly, the performance of the PCR and PLSR were same for prediction of ash in the sample, even though slightly better results could be obtained from the later one with raw spectral data of range 7000-5500 cm-1 (R2 = 86%) (Fig. 8). Here the PLSR model was developed with first six factors. The model was more stable than models with other conditions.

    Figure 8.  Model for prediction of ash.

    The results obtained in the study were much better than those of Huang et al. (2010) who used the NIR for rice straw samples. In that study, R2 > 0.80 for cellulose, 0.60 < R2 < 0.70 for lignin, and R2 < 0.70 for hemicelluloses by the PLSR. Another study involving Eucalyptus globules predicted by multivariate linear regression (MLR) showed R2 ranged from 0.67 to 0.87 (Poke and Raymond, 2006).

    A study conducted by Li et al. (2015) on moso bamboo found at best R2 of 0.91 for hemicelluloses, 0.98 for cellulose, and 0.94 for lignin, where the NIR was used for spectral data processing and the ANN for calibration.

4.   Conclusions
  • Among the segmented parts of the spectral data, the range 7000-5000 cm-1 showed as most informative region. The PLSR (calibration model) showed maximum R2 value (0.91%) for prediction of holocellulose. For prediction of pentosan, the PCR was better (R2 = 0.68%). The PCR also showed better results (R2 = 86%) for quantification of ash. However, some preprocessing treatment was required for quantification of lignin and α-cellulose. In case of lignin, the PLSR is the best (R2 = 0.83%) when the spectral data were de-trained and smoothed with Savitzky-Golay (S-G) filtering simultaneously. For prediction of α-cellulose, the PLSR was the best model (R2 = 0.94%) when the data were pretreated with mean normalization. Calibration models for quantification of lignin, holocellulose, α-cellulose, pentosan and ash were developed with chemometric techniques and FT-NIR spectroscopic data. Finally, these models could be used by pulp and paper industries with minimum cost and time as no further chemicals is needed for quantification of these parameters by producing no chemical waste and keeping the environment clean thereby.

Reference (25)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return