Quantitative Analysis Of Biomass In Polypropylene-based Wood Plastic Composites Using Infrared Spectroscopy Techniques

The raw materials of wood plastic composites(WPCs) are abundant and low-cost. In addition, WPCs not only increase the utilization rate of biomass, but also resolve the environmental problems caused by solid wastes of polymer products. Therefore, they are widely used in construction, packaging, garden, indoor furniture and automobile interior parts. The biomass content in WPCs greatly affects the physical and mechanical properties and price. Fast and accurate evaluation of the biomass content is important for the further development of WPCs. However, because of the compositional complexities of WPCs, the mutual influence of different components is serious. Therefore, the current methods cannot fulfil the requirement of the market.In this paper, Chinese fir(softwood species), poplar(hardwood species) and moso bamboo(bamboo species) were chosen as fillers, and polypropylene(PP) was used as matrix materials. With some additives, WPC samples with different biomass contents, ranging from 30.0% to 60.0 %, were prepared by extrusion pelletizing. Firstly, MIR coupled with simple linear regression(SLR), multiple linear regression(MLR) and partial least squares regression(PLS) were used to build the model for biomass estimation in Chinese fir/PP composites, respectively. Base on this, a mixed model for the prediction of biomass in three types of WPCs was constructed by MIR combined with PLS. The predictability of the optimal model was validated by external validation. Finally, a mixed model was developed to predict the biomass contents in different types of PP-based WPCs by NIR combined with PLS. The purpose of this paper was to establish a rapid and accurate method for prediction of the biomass in WPCs. The main results of this paper are summarized as follows:(1) The MIR method can be used for biomass measurement in WPCs. All of the models based on different calibration methods; including simple linear regression, multiple linear regression and partial least squares regression, have better predictive power. Among the models, the PLS model showed the best performance. The results of external validation showed that the relative prediction deviations for biomass determination were lower than ± 6.0%. The results of the precision test showed that the PLS model showed the best precision, the deviations between the two predicted values were 0.4-3.3%.(2) There are high similarities between the spectra of three WPC species in the range of 1800―800 cm-1. The results showed that it is feasible to develop a mixed model for quantification of biomass in different types of PP-based WPCs. Then mixed models were established using PLS. The results of cross validation showed that the first derivative spectra corrected by standard normal variate(SNV) yielded the optimal model. The results of external validation showed that the predictabilities of the mixed model based on MIR data are excellent for biomass estimation in three WPC species. It is observed that the predicted values of biomass content are in good agreement with reference values. The relative prediction errors were less than 6.0%. The R2, the standard errors of prediction(SEP) and the ratio of performance to deviation(RPD)(ratio of the standard deviation of the reference value to the SEP) were 0.942, 1.376 and 4.189, respectively. The results of the precision test showed that the mixed model has strong repeatability and high precision, and the deviations between the two predicted values were 0.0-3.1%.(3) The results of NIR analysis showed that there are strong similarities among three WPC species. Although the bands of both biomass and PP are all reflected in the WPC spectra, the bands are overlapped and difficult to assign. Therefore, in order to estimate the biomass content, the multivariate methods are necessary to obtained useful information from the WPC spectra. NIR in combination with PLS was used to build the mixed model for biomass determination. The results of cross validation indicated that the best prediction model was developed using baseline correction data processing method. A fully external validation was applied to evaluate the predictive power of the best model. It is found that the predicted contents of biomass in three WPC species were very close to the actual values with R2 value of 0.941. The range of the relative prediction deviations were not above ±7%. The SEP and RPD values were 1.760 and 3.955, respectively. The results of the precision test showed that the mixed model based on the NIR spectral data has strong repeatability, and the deviations between the two predicted values were 0.4-2.8%.(4) The structural characteristics of PP and biomass were obvious in the MIR spectra of WPCs. The position, intensities and shape of the characteristic peaks were obviously different. Thus, the useful information can be directly extracted from the MIR spectra of WPCs. Then it is easy to determine the biomass content of WPCs. The absorption peaks of PP and biomass in the NIR spectra were weak, complex and seriously overlapped. Therefore, chemometric techniques must be employed to obtain useful information from the NIR data of WPCs. Compared to the mixed model based on MIR spectral data, the prediction accuracy and precision of the mixed model based on NIR spectral data is almost the same. Compared to those thermo-analytical methods, the current level of accuracy of infrared techniques is significantly improved. Moreover, infrared techniques are low-cost, fast and simple, easy-handing, and the infrared methods described here have good application effect and promising development.