| Neosinocalamus affinis Keng is a kind of special economic bamboo species in the Southwest China, which has the characteristics of rapid growth, short growing period, renewability, good fiber morphology, and so forth. N. affinis has been used to prepare natural bamboo fibers because of its superior properties. As a new kind of plant fibers used for textile, the fabrics made of natural bamboo fibers are receiving more and more attention. However, to date, there has been relatively few research on predicting bamboo properties and identifying natural bamboo fiber. Furthermore, little attention has been paid to assessing N. affinis bamboo properties by near-infrared(NIR) and mid-infrared(MIR) spectroscopy, and discriminating natural bamboo fiber by two-dimensional infrared(2D-IR) correlation spectroscopy. Therefore, in this research, NIR and MIR spectroscopy together with chemometrics were applied to determine N. affinis bamboo properties, including anatomical properties, physical and mechanical properties, and chemical properties, through developing models. In addtion, MIR combined with 2D-IR correlation spectroscopy were used to distinguish natural bamboo fiber from flax, jute, hemp, cotton, and bamboo pulp fibers according to their IR spectral features. So this paper would provide scientific theoretical basis to the oriented cultivation of N. affinis and its high-addition value utilization.The main results and conclusions of this research were summarized as follows:(1) NIR spectroscopy was used to rapidly predict anatomical properties of N. affinis, including microfibril angle(MFA), fiber length, and fiber width. The results showed that the preprocessing methods of noise and orthogonal signal correction(OSC) could improve the predictive ability of PLS models. And the calibrations of MFA, fiber length, and fiber width, based on noise combined with OSC spectra, gave the strongest correlation presenting coefficient determination(R2) of 0.93、0.97, and 0.74, root mean square error of calibration(RMSEC) of 0.0997、0.0506, and 1.3825. Predictions were very good, with R2 of 0.80, 0.98, and 0.99, root mean square error of prediction(RMSEP) of 0.2920, 0.1460, and 1.1741. Therefore, it is concluded that MFA, fiber length, and fiber width of N. affinis can be rapidly predicted by NIR spectroscopy with sufficient accuracy.(2) The prediction of air density based on different part spectra(the inner culm wall, the outer culm wall, and the middle layer) using NIR spectroscopy was studied. The results indicated that the calibration models built with PLS using the middle layer spectra displayed stronger correlation than those models using the inner and outer culm wall spectra. And the thinner the samples became, the better predictive ability of the calibration model had. Best calibration model for the quantification of density was proposed showing R2 of 0.89, RMSEC of 0.0310, and the prediction was good with R2 of 0.77, RMSEP of 0.0438.(3) NIR spectroscopy was applied to rapidly predict mechanical properties of N. affinis, including modulus of rupture(MOR), modulus of elasticity(MOE), and tensile strength parallel to grain. The results demonstrated that compared with PLS, backward interval partial least squares(biPLS) could effectively find the optimal spectrum regions and improve the predictive ability of models. The optimal calibrations of MOR, MOE, and tensile strength parallel to grain were obtained through biPLS, with R2 of 0.79, 0.87, and 0.72, RMSEC of 0.0202, 1.1349, and 0.0313. And the predictions had R2 of 0.78, 0.73, and 0.77, RMSEP of 0.0185, 1.8561, and 0.0292. Therefore, it is concluded that NIR spectroscopy is promising for predicting bamboo mechanical properties.(4) The crystallinity of N. affinis was analyzed through NIR spectroscopy and X-ray diffractometry. The results showed that the models built with three improved PLS methods(synergy interval partial squares(siPLS), partial least squares(iPLS), and biPLS) had better predictive ability than that of PLS model. The optimal calibration model was obtained by siPLS, with root mean square error of cross validation(RMSECV) of 0.0135, and the prediction had R2 of 0.77, RMSEP of 0.0117. The correlation between NIR predicted and X-ray measured was good.(5) NIR and MIR spectroscopy were applied to quantitatively analyze chemical composition in N. affinis, including holocellulose, α-cellulose, Klason lignin and 1% NaOH extractives. The results showed that the best calibrations developed by both methods for chemical properties all had satisfactory correlations with R2 values ranging from 0.81 to 0.98, RMSEC between 0.0050 and 0.0147. When applied to prediction sets, the correlations were good with R2 above 0.55. Therefore, it is concluded that both spectroscopic techniques combined with chemometric strategies can rapidly predict the chemical components in N. affinis.(6) The IR spectra and second derivative IR spectra of natural bamboo fiber and other textile fibers(jute, flax, hemp, cotton, and bamboo pulp fiber) were compared and analyzed. The results showed that there were subtle spectral differences between IR spectra of natural bamboo and other plant fibers(jute, flax, hemp, and cotton), which had been processed by hydrogen peroxide and glacial acetic acid. With the applying second derivative IR spectra, the resolution of the differences was improved and the tiny differences between natural bamboo fiber and other plant fibers emerged. In addition, natural bamboo and bamboo pulp fibers can be easily identified through IR and second derivative IR spectra. That is because natural bamboo and bamboo pulp fibers had different chemical composition and crystal lattice types of cellulose(Ⅰ, Ⅱ).(7) To accurately identify natural bamboo fiber, the 2D-IR spectra of natural bamboo fiber and other textile fibers were analyzed. The results indicated that in the range of 800-1200 cm-1 and 1425-1750 cm-1, 2D-IR spectra of natural bamboo fiber had eight and two auto-peaks separately, and all the cross peaks were positive. The 2D-IR spectra of natural bamboo fiber were obviously different from those of other textile fibers. Moreover, flax, jute, hemp, cotton and bamboo pulp fibers had their own features in the 2D-IR spectra. So it is demonstrated that MIR combined with 2D-IR spectroscopy can become a new approach for natural bamboo fiber identification.(8) According to the spectral characteristics of natural bamboo fiber in the IR and 2D-IR spectra, unknown thread and fabric samples were identified. The results showed that within the fibers analyzed in this study, the textile sample made of natural bamboo fiber could be accurately identified through IR and 2D-IR spectra. And this method is of characteristic of easy operation, less sample consumption, and low cost. |
PDF Full Text Request