Gradient variations of cellulose supramolecular structures in moso bamboo culm: from nano- to microhorizons

The variation of cellulose microfibril orientation within the bamboo has been confirmed to be one of the critical factors exerting a strong influence on the mechanical properties of bamboo fibers, which is regarded as the response of the cell wall to the internal and external stimulus. In this paper, the radial gradient variation of the cellulose supramolecular structures of bamboo culm was systematically studied by XRD and the linear polarized confocal Raman microspectroscopy to deepen the understanding of the origin of bamboo micromechanics and its functionally graded properties. XRD analysis indicated that bamboo yellow (By) had the isotropic crystallite arrangements, while bamboo timber (Bt) and bamboo green (Bg) displayed preferred orientation of crystals. Moreover, both the crystallinity and crystallite sizes notably grew from the inner By to the Bg. At cell wall level, the variations in the distribution of microfibril orientation were visualized by Raman imaging, with the fiber secondary wall areas adjacent to compound middle lamella displaying higher Raman intensity. Furthermore, Raman band ratio (I1095/I2939) was used to predict the microfibril angle (MFA) in different cell wall types and morphologically distinct cell wall layers semi-quantitatively. The results showed that the ratio was the highest in parenchyma, followed by narrow layer of fiber wall, and the lowest in the broad layer, which indicated the high MFA in the parenchyma. Interestingly, the ratio decreased along the successive and alternating broad and narrow lamellae of fiber wall, in accordance with cell wall micromechanical trend.


Introduction
Bamboo, a natural functionally graded biocomposite material, is attracting attention for the researchers in the exploration of engineering biomaterials owing to its excellent flexibility and fracture toughness . Along the radial direction, the longitudinal tensile modulus of elasticity and the longitudinal tensile strength for the outermost layer are 3-4 times and 2-3 times as high as that of the innermost layer, respectively (Yu et al. 2008). The graded mechanical performance of bamboo culm is influenced to a large degree by its basic structural units (fiber and parenchyma). The mechanical properties of a cell are, in turn, closely related to its dimensions and internal structure (Rasheed et al. 2020;Yang et al. 2021). One of the factors exerting a strong influence on the mechanical properties of fibers is the cellulose supramolecular structure, including cellulose crystal structure and microfibril orientation, which has been shown to affect the tensile strength and elastic modulus (Abe and Yano 2009;Ahvenainen et al. 2017;Zou et al. 2009).
Normally, the cell wall in plants is made up of four basic building blocks: cellulose skeleton (the main structural fiber of the plant kingdom) and matrix substances (hemicelluloses, lignin and trace amounts of pectin) (Cosgrove 2005;Voiniciuc et al. 2018). Cellulose is believed to be the main load-bearing component of plant cell walls, and how bamboo regulate and organize its orientation in response to various environmental perturbations is a key question in plant biology. The ability of the plant cell wall to vary the orientation of cellulose microfibril (CMF) may be induced in response to maturation stress of cell wall , tensile and compressive stress (Thomas et al. 2021), which is regarded as an adaptive physiological response for plant (Polko and Kieber 2019;Emons 1994). It has been proved that the compression wood had the greatest microfibrils angle (MFA), whereas the opposite wood showed the smallest MFA (Purusatama et al. 2019). Unlike typical wood cell wall layering structure, bamboo possesses unique cell wall architecture. It exhibits a polylamellate structure with alternating broad and narrow lamellae arising from the alternation in the CMF orientation (Suzuki and Itoh 2001). This alternating hierarchical structure is considered to be one of the factors that contributed to the high tensile strength and elastic modulus of bamboo culm. Although topochemical and ultrastructural investigations have been reported comprehensively, the research on the radial variations in the cellulose supramolecular structures and distribution of microfibril orientation of bamboo culm, especially associated with multilayered structure of bamboo cell wall, is limited.
To better interpret the origin of flexibility and toughness of bamboo, many different techniques have been used to visualize and analyze the cellulose deposition, supramolecular structures, interaction between cellulose skeleton and matrix substances and orientation of CMF Chen et al. 2020;Hao et al. 2018;Hu et al. 2017;Huang et al. 2018;Li et al. 2020;Lin et al. 2020;Sahlberg et al. 1997;Sun et al. 2016;Thomas et al. 2015;Wang and Shao 2020;Zhang et al. 2017;Zhu et al. 2021). However, these methods to reveal the CMF orientation have suffered from the classical ensemble average limitation presented by analysis of these mixtures of complex biomass tissues. By a novel LC-PolScope imaging system and transmission electron microscopy, the variation in CMF orientation in the bamboo fiber and parenchyma wall has been reported (Hu et al. 2017). Meanwhile, ordered or well-aligned arrangements of cellulose chains can be visualized by polarized microscopy. Considering the cellulose crystallinity in bamboo timber is lower than 66% (CrI by XRD peak height method), the orientation information on amorphous portion of cellulose microfibrils would be neglected when using traditional polarized light microscopy and XRD approach. Recently, by combining the XRD and microspectroscopy techniques, Ren et al. (2021) have revealed average CMF angle in the broad and narrow layer of bamboo species, but limited information on how the bamboo species organize its cellulose microfibrils along the radial direction of bamboo culm and at consecutive cell wall sublayers is available.
Due to the higher spatial resolution (theoretical lateral resolution 0.3-0.8 μm depending upon the wavelength of excitation), confocal Raman microspectroscopy has been successfully used to investigate the cellulose organization, cellulose distribution and CMF orientation (Agarwal 2006;Agarwal et al. 2012;Gierlinger et al. 2010). In addition, organized and crystalline phases, water-induced reorganization of cellulose were reported, which provided deeper horizon in cellulose research by Raman microscopy (Agarwal et al. 2016(Agarwal et al. , 2018(Agarwal et al. , 2021. It has been reported that the Raman intensity of many bands assigned to cellulose is significantly influenced by the orientation of the polymer chain relative to the electric vector (EV) direction of the laser. Variations in orientation can be worked out by choosing selected wavenumber regions for various bands. The MFA describes the angle between the CMF orientation and the fiber axis. An angle of 0° is defined as an alignment of CMF parallel with the fiber axis. In this case, if the EV is perpendicular to the cellulose chain direction, the C-O-C asymmetric stretching band (1095 cm −1 ) becomes smaller, and the CH and CH 2 stretching band (2897 cm −1 ) has higher intensity at the same time. Higher MFA values contribute to the increased intensity of 1095 cm −1 band, and the 2897 cm −1 band becomes relatively less intensive (Agarwal and Atalla 1986). Although confocal Raman microscopy has been employed to qualitatively and semi-quantitatively investigate the MFA (Wang et al. 2014), we still have much to learn about how bamboo fiber and parenchyma organize their CMF orientation relative to the cell axial direction and in the successive multilayered cell wall. Moreover, the micro-scale observation on the varying structural features of bamboo cellulose in different culm regions is less reported.
In the presented work, we firstly study the cellulose supramolecular structures and the spatial CMF orientation in bamboo fiber and parenchyma by the combination of X-ray diffraction (XRD) and linear polarized Raman microspectroscopy analyses. The information presented here will contribute to fundamental understanding of the micromechanics nature of bamboo species as well as expanding to the wider range of lignocellulose biomass.

Materials
Bamboo culm (Phyllostachys pubescens) was obtained from the forestry station in Anhui Taiping experimental center, ICBR. Series of 15-μm-thick cross sections were collected by a rotary microtome (LEICA RM2165, Germany). Moreover, 40-60 mesh bamboo powder was obtained from bamboo green (Bg), bamboo timber (Bt) and bamboo yellow (By) for compositional analysis according to National Renewable Energy Laboratory (NREL) protocol (Sluiter et al. 2010). The Bg presented a white glossy appearance covering the outer surface of the bamboo culm about 20-30 μm. An innermost compact layer of 40-50 μm thickness was designated as By. The middle portion of the bamboo culm was designated as Bt. For the separation of fiber and parenchyma, the bamboo chips were ground to pass a 40-mesh sieve. The particle was soaked in a beaker with 1000 mL of distilled water for 10 min under vigorous stirring. After standing for 5 min, the fiber deposited on the bottom of the beaker and the parenchyma cells in the upper side was separated based on the various density. For further separation of the cells, the short parenchyma cells were repeatedly sieved using a 200mesh sieve (aperture of 75 μm) and the long fiber cells were collected by repeated sieving through a 30-mesh sieve (aperture of 500 μm).

Supramolecular structures characterization
To explore the crystal structures of the bamboo cellulose microfibrils with different CMF orientations, X-ray diffraction (XRD) analysis was performed by a Rigaku Ultima IV X-ray diffractometer equipped with Cu Kα radiation (λ = 0.15419 nm). The acquired XRD patterns were fitted using the pseudo-Voigt peak shape with Maud Rietveld software (version 2.7). Several parameters were modified in the Maud software during the refining process: the scale factor, quadratic background and unit cell length a (layer I: the phase atom); cellulose Iβ: cell length a, scale factor, crystal size and cell angle gamma (March-Dollase model: background and crystal size). The CrI of the samples was calculated based on the fitted results via the equation below (Ling et al. 2019): where A crystal is the area below the calculated crystalline cellulose and A amorph is the area below the calculated amorphous content.
The d-spacings were calculated by Bragg equation (Pope 1997), and crystallite sizes perpendicular to each lattice plane (L hkl ) of cellulose allomorphs were calculated by Scherrer equation (Eq. 2) (Holzwarth and Gibson 2011): where λ is the X-ray wavelength (0.15419 nm), B hkl is the angular FWHM (full width of half maximum) of calculated crystalline peaks, and θ is the scattering angle. The average MFA distribution was measured also using X-ray diffractometer. The tube voltage and tube current of the scan test were 40 kV and 40 mA, respectively, and the step was 0.5°. The sample scan range was 1-359°. The raw data of the intensity were imported into Origin 2019 for Gaussian single peak fitting, and the MFA was calculated by using the 0.6 T method (Wang et al. 2016).

Confocal Raman microspectroscopy analysis
A 15-μm-thick cross section was put on a glass slide with a drop of distilled water, covered by a coverslip (0.17 mm, Thermal Fisher Scientific). Raman spectra were acquired with a confocal Raman microscope (LabRam HR Evolution, Horiba Jobin Yvon). Measurements were taken with an Olympus 100 × Oil objective (UPLSAPO 100×, Oil, NA = 1.25) and a 532-nm linear polarized laser. For mapping, an integration time of 0.5 s, a scanning area of 70 × 50 μm and a scanning step size of 0.5 μm steps were chosen. Half-wave plate was used to align the laser direction parallel to the radial direction of cell wall. The Labspec6 software was used for spectral and image processing and analysis. Prior to a detailed analysis, the baseline of the Raman spectra was corrected by using the linear least squares algorithm. To determine the Raman band intensity of morphologically distinct areas, the resulting spectra were deconvoluted at defined Raman spectra regions. The deconvolution of the spectra by band fitting was performed using Gauss-Lorentz mixture function. The three different sets of spectra from the same cell wall layers were extracted separately. Different chemical images were generated by a cosmic ray removal filter, and default software sum filters (Labspec6, Horiba) were applied to integrate the defined band area in the Raman spectra. To compare the relative intensity of cellulose orientation-sensitive Raman band, the spectra were normalized at band 2939 cm −1 .

Compositional analysis
The compositional analysis for fiber and parenchyma separated from the bamboo green (Bg), bamboo timber (Bt) and bamboo yellow (By) is summarized in Table 1. The carbohydrates analysis revealed that cellulose and xylan-type hemicellulose were the dominant constituents in all cell types. The relative content of cellulose in fiber was higher than that in parenchyma, whereas parenchyma displayed higher xylan content in the corresponding locations. In addition, the fibers located at bamboo timber (Bt-F) displayed the highest cellulose content of 48.6%, while parenchyma had the highest cellulose content (46.5%) located at bamboo yellow (By-P).
(2) L hkl = 0.9 B hkl cos The content of xylan increased from 20.7 to 24.6% for fiber and 25.2 to 27.0% for parenchyma from Bg to By, respectively. Moreover, it was found that the total lignin content decreased from 31.4% to 27.5 (w/w) and 30.0% to 25.3 (w/w) for fiber and parenchyma along the radial direction of bamboo culm.

Crystal structures
XRD patterns revealed typical cellulose Iβ allomorphs for all regions of the bamboo feedstock. The diffraction peaks located at around 2θ = 14.8°, 16.6°, 22.5° and 34.5° (Fig. 1a) were, respectively, assigned to (1-10), (110), (200) and (004) lattice planes of cellulose Iβ (Nishiyama et al. 2002). The two even peaks for (1-10), (110) planes were mixed into a band due to the presence of amorphous lignin and hemicelluloses. Bt and Bg presented higher intensity for (1-10) plane than (110) plane, suggesting the preferred orientation of crystals, which would further disturb the CMF in the cell wall at corresponding regions (French 2014). Crystallinity of the samples was further calculated by XRD via Rietveld refinements (Ling et al. 2019, also see supplementary material) as presented in Fig. 1b. By, Bt and Bg had crystallinity values of 51.9%, 67.0% and 83.0%, respectively, exhibiting an increasing trend from the inner to the outer region. The phenomenon agreed with previous studies on nanostructure of moso bamboo, and the value of Bg was also similar to former research on bamboo cellulose fibers (Ling et al. 2020;Wang et al. 2012). Compared to above compositional analysis, By part had the highest ratios of amorphous xylan-type hemicellulose, followed by outer Bt and Bg, no matter the fiber and parenchyma locations. This phenomenon might be responsible for the increased crystallinity from inner By to Bg. It also suggested that hemicelluloses had more influence on materials crystallinity than lignin, as confirmed by Raman and Segal-XRD methods (Agarwal et al. 2013;Chen et al. 2021). Considering the growing process of cell walls, higher cellulose crystallinity was revealed in more mature outer region, which would favor the oriented arrangements of microfibrils as well as the higher mechanical properties for biomass recalcitrance (Cheng et al. 2015). Table 2 provides the crystal structures   Table 1 Compositional analysis fiber and parenchyma located at the bamboo green, bamboo timber and bamboo yellow Bg-F fiber located at bamboo green, Bt-F fiber located at bamboo timber, By-F fiber located at bamboo yellow, Bg-P parenchyma located at bamboo green, Bt-P parenchyma located at bamboo timber, By-P parenchyma located at bamboo yellow information of the samples based on XRD patterns. There was limited variation of the d-spacings for different regions, suggesting the similar gaps between cellulose crystalline chains. However, the crystallite sizes notably grew from the By to the Bg, which also agreed with above trends of increasing crystallinity. It supported the common result that crystallite sizes and crystallinity vary in the same trends (French and Santiago Cintrón 2013), and more oriented microfibrils and higher mechanical strength derived from crystalline cellulose are existent at Bg region of the bamboo culm.

Average microfibrils angle distribution at tissue level
MFA is an important factor affecting the mechanical properties of bamboo and wood. For bamboo culms in the present work, the average MFA detected by XRD gradually increased radially from the outer to the inner (Fig. 2a) and the average values were 9.96°, 10.57° and 11.25° for Bg, Bt and By, respectively (Fig. 2b). It has been demonstrated that the MFA has a particularly remarkable effect on the cell  1 3 wall micromechanics. The increase in the average MFA will cause a decrease in the elastic modulus and longitudinal tensile strength (Yu et al. 2011). The lower average MFA in the Bg area indicated higher elastic modulus and longitudinal tensile strength, which was confirmed in the previous work (Huang et al. 2016).

Microfibrils orientation distribution at cell wall layers
Raman spectroscopy has been used for detecting structure, dynamics and function of biomolecules. Several organic compounds and functional groups can be identified by their unique spectral pattern, and the intensity of the bands may be used for the calculation of the relative concentration in the sampled entity. Besides, both the distribution of carbohydrates and cellulose orientation can be mapped simultaneously at cellular and sub-cellular level by selecting Raman bands that are specific to these cell wall components (Chen et al. 2017;Ma et al. 2014). Average Raman spectra extracted from the bamboo fiber and parenchyma wall are presented in Fig. 3d-g. The strong band observed at 2897 cm −1 and a dominant shoulder located at 2939 cm −1 were assigned to the stretching of the C-H and C-H 2 groups as well as C-H stretching in OCH 3 groups, respectively. The bands at 1095 cm −1 and 1122 cm −1 were attributed to the asymmetric and symmetric stretching of the C-O-C linkages, respectively. A band at 380 cm −1 was considered typical of the heavy atom stretching of the cellulose (Agarwal and Ralph 1997;Agarwal 2006). It has been revealed that the intensity change of the band at 1095 and 2897 cm −1 was largely dependent on cellulose orientation, while the band at 380 cm −1 was also sensitive to cellulose orientation (Wiley and Atalla 1987) despite a potential intensity contribution from the adjacent band at 370 cm −1 , assigned to S-lignin.
It has been widely accepted that both the bamboo fiber and parenchyma display a concentric layering structure. In the present work, the morphologically distinct regions of the fiber and parenchyma wall were clearly visualized by TEM imaging (Fig. 3a). In particular, the fiber wall displayed thick polylamellate secondary walls, which consisted of alternating broad (B1 and B2) and narrow layers (N1 and N2) (Fig. 3b). Similarly, the parenchyma was characterized by cell wall with a polylamellate structure (Fig. 3c). For an overview of the measurement areas, the Raman images with all cell wall structures were calculated by integrating over the -CH stretching from 2789 to 3045 cm −1 . The layering structure of fiber located at Bg, Bt and By was discernible (Fig. 4a, c, e), while the sublayer of parenchyma could not be differentiated by Raman image (Fig. 4g), mainly due to the limitation of Raman spatial resolution.
Moreover, polarized Raman spectroscopy can be used to analyze the CMF orientation at cell wall sublayers. The ratio of Raman band intensity has been used to predict the CMF orientation in a previous study (Gierlinger et al. 2010). In the linear cellulose chains the intensity of the 1095 cm −1 (glycosidic C-O-C bond) gives strongest band when it is aligned parallel to the polarization of incident light, whereas the intensity of the band representing CH stretching at 2897 cm −1 in this case is at its lowest. Therefore, by comparing the relative intensity of these two bands a semi-quantitative result for the CMF orientation can be obtained.
It can be clearly seen that the intensity of the cellulose orientation-sensitive band at 1095 cm −1 and 2897 cm −1 was changing with the change of cell wall CMF alignment with respect to the laser polarization direction (Fig. 3d-g). The band area ratio of I 380 /I 2939 , I 1095 /I 2939 and I 2897 /I 2939 calculated from various cell layers of fiber and parenchyma located at Bg, Bt and By is shown in Table 3. It was found that the ratio of 380-2939 cm −1 kept constant, indicating less dependence of this band on CMF orientation. By comparison, the ratio of I 1095 /I 2939 was higher in parenchyma, followed by narrow layer of fiber wall, and the lowest in the broad layer, which indicated higher MFA in the parenchyma. Conversely, the cell wall regions with high ratio of I 1095 /I 2939 had a lower value of I 2897 /I 2939 . The higher MFA in parenchyma has also been reported at multi-scale. At the tissue level (more than 400 μm), Ahvenainen et al. (2017) have demonstrated that fiber component was dominated by a high degree of orientation corresponding to small MFA, while the parenchyma component showed significantly lower degree of orientation with larger angles. Furthermore, at subcellular level (0.1 μm), Hu et al. (2017) have revealed that the secondary wall of parenchyma displayed higher MFA values than that of fiber. Based  on the geometrical correlation between the actual and apparent dimensions of cellulose microfibril aggregates, Ren et al. (2022) have revealed that the cellulose microfibrils in the broad sublayers of the bamboo fibers exhibited a relatively small MFA of about 10°, while those in the narrow sublayers were nearly oriented in the transverse with a MFA of about 80°. For parenchyma cells, the MFA of the broad and narrow sublayers was estimated to be 50-70° and 70-80°, respectively. Interestingly, in the present work for the fiber located at various positions of bamboo culm, the ratio of I 1095 /I 2939 was almost constant in all the successive broad lamellae, while the ratio in the adjacent narrow layer changed obviously, with the maximum value (0.82) in the outer narrow lamellae of Bt, indicating higher MFA values in this area. Our previous work has shown that along the radius of bamboo culm the elastic modulus had a gradient decreasing trend for the broad layer of fiber located at Bg, Bt and By (Jin et al. 2021) in accordance with tendency of CMF orientation.
Raman images by integrating over the narrow area of the C-O-C asymmetric stretching region (1083-1100 cm −1 ) are shown in Fig. 4 to visualize cell wall regions with high MFA in the direction of laser polarization. Here, the narrow layers of fiber were highlighted, indicating that they had a high MFA. Characteristically, the CMF in the broad layers were oriented almost parallel to the longitudinal axis of the fiber whereby there was a gradual but only slight increase in the angle from compound middle lamella to lumen. By comparison, the narrow layers consisting of CMF are oriented more perpendicular to the cell axis (Fig. 4b, d, f). Moreover, the higher and homogeneous intensity distribution was visualized in the whole parenchyma wall (Fig. 4h). Similarly, for the CMF orientation in the parenchyma, Ahvenainen et al. (2017) indicated a lower degree of orientation with the preferred orientation perpendicular to that of the bamboo fibers; meanwhile, Hu et al. (2017) reported a higher average MFA for parenchyma than fiber. In the present work, whether the uniform distribution of Raman intensity in the bamboo parenchyma wall is associated with the uniform CMF orientation in the successive lamellae needs further investigation by the Raman system with enhanced lateral resolution. This is the first time that the distribution of CMF orientation within the multilayered fiber wall is clearly visualized. The present work provides a direct approach to visualize the CMF orientation, although the results are semi-qualitatively.

Conclusion
In this study, XRD and linear polarized confocal Raman microspectroscopy were used to interpret the cellulose supramolecular structures along the radius of bamboo culm complementarily, which is a fundamental property determining mechanical strength of cell walls. It was found that By had the isotropic crystallite arrangements, while Bt and Bg presented preferred orientation of crystals. Moreover, both the crystallinity and crystallite sizes notably grew from the By to the Bg. The linear polarized Raman images revealed the variations in the distribution of CMF orientation for the successively alternating broad and narrow layer of fiber. Raman spectra 1 3 analysis indicated the higher MFA in parenchyma and narrow layer of fiber. The findings will deepen our understanding of the origin of bamboo micromechanics and its functionally graded properties.