The surface of the WS is compact and smooth (Fig. 3a), covered with a layer of wax composed of saturated hydrocarbon compounds and their derivatives. The inner part of the WS is mainly consisted of cellulose, hemicellulose and lignin (Liu et al., 2005; Liu et al., 2017). Cellulose plays as the role of skeleton that provides mechanical properties. Hemicellulose is closely connected with cellulose and plays the adhesive role. Lignin is a kind of high molecular compound, which has aromatic characteristic and three-dimensional structure, affecting the performance of the WS. After the alkali treatment, the wax on the surface of the WS was dissolved and the hydrogen bonding between cellulose and hemicellulose was weakened. The saponified ester bonds between hemicellulose and lignin molecules destroyed the structure of hemicellulose and lignin. Therefore, the cellulose fibers were partially separated from the fiber bundle (Fig. 3b). However, there are still a lot of fiber bonding together (Fig. 3b), indicating that the alkali treatment with a low concentration can not completely remove hemicellulose and lignin. These results were similar to the previous study (Liu et al., 2005).
Figure 3. a, b: FE-SEM images of pristine and alkalized WS powder; c: photo of TEMPO-oxidized cellulose jelly from the treated WS powder; d: TEM image of CNF from cellulose jelly
After been alkali treated, the WS was subjected to selectively TEMPO-oxidation followed by a mechanical disintegration treatment, and the jelly like hydrogel was obtained as shown in Fig. 3c. The structures of non-cellulosic materials and partial amorphous cellulose were reacted and destroyed, and most of them were dissolved, mainly remaining the cellulose components.
The results of the TEM (Fig. 3d) revealed that the individual CNF with almost uniform widths of 5–10 nm and length of several micrometer, which is lower than that of the lignocellulosic nanofibrils generated from the WS by a combined method (Bian et al., 2018a). According to the previous literatures, the geometrical dimensions of the CNF are found to be widely various, depending on the source of the cellulosic materials and the extraction methods (Saito et al., 2006b). Typically, the geometrical characteristics of the CNF originating from different cellulose sources are shown in Table 1. The reported width is generally approximate to a few nanometers, but the length of the CNF spans a larger range, from hundreds of nanometers to several micrometer. It is worth noting that the dimensions of obtained CNF from the WS are much higher than those of plant celluloses. While comparing with the TEMPO-mediated CNFs from other native cellulosic plant materials, the width and length of the observed CNF generated from the WS were generally much larger.
Source Length Width (nm) Reference Wheat straw Several micrometer 5–10 This study Cotton plant Several micrometer 5–10 Miao et al. (2016) Jute 100–2000 nm 5–10 Yu et al. (2014) Wood 100–500 nm 3–4 Saito et al. (2006a) Cotton 100–500 nm 3–5 Saito et al. (2006a) Tunicin >5 μm 10–20 Saito et al. (2006a) Bacterial >5 μm 10–20 Saito et al. (2006a) Tunicate ~10 μm ~20 Iwamoto et al. (2009)
Table 1. Length and width of CNFs from various sources obtained by TEMPO-oxidation
Fig. 4a shows the FT-IR spectra of the WS, WS-3% and CNF. The board band of –OH groups stretching vibrations within 3360–3330 cm–1 region and the maximum absorbance band of C–H stretching vibrations in methyl and methylene around 2900 cm–1 were observed in all spectra, indicating the principal functional groups found in lignocellulosic materials (Han et al., 2013). Four strong absorption bands at 1732 cm–1 (C=O stretching vibration of methyl ester and carboxylate groups in hemicellulose), 1512 cm–1 (C=C stretching vibration of aromatic hydrocarbon in lignin) (Rosa et al., 2012; Han et al., 2013), 1454 cm–1 (C–H and –CH2 bending vibration in lignin) (Sinha and Rout, 2009) and 1245 cm–1 (C–O stretching vibration of acetyl in hemicellulose) (Carrillo et al., 2004) in WS FT-IR spectra became unobvious in WS-3% FT-IR spectra, which almost disappeared in CNF FT-IR spectra, indicating that the majorities of hemicellulose and lignin in the WS were dissolved and removed during the processing of the CNF.
Figure 4. a: FT-IR results of pristine, alkalized WS and CNF; b: Zeta potential of CNF aqueous solution; c: 1D SR-WAXS integral curve; d: schematic of crystallite dimension of CNF
In addition, the bands at 899 cm–1 (β-glucosidic linkage), 1425 cm–1 (bending vibration for C6-CH2), 1595 cm–1 (aromatic C=C vibration) and 2899 cm–1 (aliphatic saturated C–H stretching vibration in cellulose and hemicellulose) (Kasyapi et al., 2013) of the WS were shifted to 900 cm–1, 1421 cm–1, 1605 cm–1 and 2901 cm–1 of the CNF, respectively. The reason was that the original intramolecular and intermolecular hydrogen bonds were broken due to the C6 hydroxyls being oxidized to carboxylates. The higher polarity of new carboxylate groups and hydrogen bonds interactions affected the vibration modes of organic functional groups of cellulose Iβ (Bian et al., 2018b).
Zeta potential is a measure of the stability of colloidal suspensions, reflecting the degree of repulsion between adjacent, similarly charged particles in a dispersion (Pelissari et al., 2014). The more highly negative zeta potential values of the CNF suspensions manifest the more highly converted surface C6 primary hydroxyls to carboxylates, indicating the more highly stability of the CNF suspensions. In neutral water, the CNF aqueous solution obtained from the WS exhibited negative zeta potential value of –46.1 mV (Fig. 4b). Similar values have been reported for the CNF isolated from jute (Miao et al., 2016b) and cotton plant (Miao et al., 2016a) by similar preparation methods in our previous works (from –41.0 to –50.1 mV). The oxidized carboxylates on the CNF surface provided enough surface charges to stabilize the suspension and prevented nanofibrils from aggregating. Furthermore, the stable colloidal suspension provided powerful conditions for the preparation of the CNF aerogel and CNF filament.
Crystal information of the WS, WS-3% and CNF has been analyzed by synchrotron radiation wide-angle X-ray scattering (SR-WAXS). It can be seen from Fig. 4c, the scattering curves of WS, WS-3% and CNF are the typical scattering curves of cellulose I, and the characteristic peaks (peaks at 2θ—11.6°, 13.4°, 16.7°, 18.1° and 27.8° corresponding to I(110), I(110), I(021), I(200) and I(004), respectively) representing cellulose I can be clearly found in band resolution results of CNF scattering curve. However, there are no common scattering peaks (peaks at 2θ around 16.1° and 17.5° corresponding to the II(110) and II(021), respectively) representing cellulose II exist in all of these three curves, which indicates that the alkali treatment by using a 3% NaOH solution was unable to transform the cellulose lattice from cellulose I to cellulose II. This result is consistent with our previous conclusion on the alkali treatment of jute fiber (Yu et al., 2014).
The crystal size and crystallinity of the WS, WS-3% and CNF were calculated, respectively, by band resolution results of SR-WAXS scattering curves. Fig. 4d shows a model for the cross section of the average crystalline region of these three kinds of cellulose microfibrils. It can be found that the cross-section dimensions of the microfibrils in the CNF is larger than those in the WS and WS-3%. The possible reason was that the cellulose chains in the WS or WS-3% were mutually combined depending on the hydrogen bonding interaction between C6 hydroxyl groups (C6-OH) in molecular chains, and the repulsion force between these cellulose chains was increased after these hydroxyl groups were selectively oxidized to carboxyl groups by TEMPO. After being washed, the carboxyl groups are presented in the form of carboxylate, as a result, the hydrogen bonding interaction is weakened and the electrostatic repulsion is further increased. Moreover, the isolated CNF are more likely to aggregate and further crystallize after the dissolution of lignin, hemicellulose and other noncellulose materials (Bian et al., 2018b). Therefore, the crystallite dimension in the CNF generated from the WS were stretched. Some of the cellulose amorphous phases, remnant lignin and hemicellulose were oxidized and removed also resulting in a higher crystallinity of the CNF than the WS and WS-3%.
The TGA and DSC curves of pristine WS, WS-3% and CNF films are shown in Fig. 5. All the samples showed an initial weight loss in the region of 50–110 ℃ due to the evaporation of absorbed moisture (Fig. 5a). The moisture contents of WS, WS-3%, CNF were 2.1%, 2.3%, 4.1%, respectively. The higher moisture content of the CNF was attributed to stronger hydrogen bonding interaction, which can combine a greater amount of water. Thermal decomposition (Td) point of the CNF was approximately 215 ℃, which was lower than those of the WS and WS-3% (approximately 250 ℃). The decrease of thermal degradation point was due to the change of hydroxyls on the surfaces of the CNF into the sodium carboxylate groups during TEMPO-mediated oxidation process (Fukuzumi et al., 2010). The char yields for the WS-3% and CNF at 500 ℃ is approximately to 25.2% and 26.1%, respectively, which are higher than that of WS (20.1%). This phenomenon also indicated that the other components such as lignin and hemicellulose were removed during alkali treatment.
The DSC curves are shown in Fig. 5b. A broad peak observed in the temperature range of 60–150 ℃ of all samples corresponded to the heat of vaporization of absorbed moisture in these samples. According to the previous report, the hemicellulose was degraded around 220–315 ℃, cellulose was decomposed around 315–400 ℃, and the lignin was hard to decomposed (Yang et al., 2007). The two exothermic peaks around 360 ℃ of the WS and WS-3% were attributed to the decomposition of cellulose. The exothermic peaks at 238 ℃ and 314 ℃ were assigned to the co-existence of TEMPO-oxidized CNF and partially oxidized original WS microcrystalline cellulose.
The CNF dispersion with a concentration of 0.5 wt% was used to fabricate a CNF film by vacuum filtration using a microfilter followed by being dried in vacuum oven, for details refer to the experimental section. The prepared film showed high smooth, flexibility and remarkable transparency (Fig. 6a). The microstructure of the film was observed by SEM. The representative images are shown in Fig. 6b and 6c. The surface of the film is composed of uniform entangled nanofibrils network. These nanofibrils are closely packed with no large voids or other defects, which may contribute to the excellent mechanical properties. The fractured cross-section of the film reveals a stratified structure, bundles of nanofibrils and individual nanofibrils have being pulled-out from the layers at considerable length scales up to several tens of nanometers. As suggested by Benítez et al., 2013, the layers formation originated from concentration-induced aggregation and floc formation during late stage of the filtration. The pull-out nanofibrils on several length scales were due to the weakened interfibrillar interactions by the presence of humidity, which resulted in achieving toughness and inelastic deformation.
Figure 6. a: photo of CNF film; b, c: FE-SEM images of surface and cross section of CNF film; d: stress-strain curve of CNF film
The tensile behaviors of the WS CNF film were investigated by universal tensile test. Strain-stress curve and tensile mechanical average values are shown in the inset of Fig. 6d. The glassy film displayed high mechanical properties, exhibiting the tensile strength and Young's modulus were 228.13 MPa and 21.40 GPa, respectively. The results were the highest values in comparison with the previous reports of the CNF film generated from crop straw (Aulin et al., 2012; Malho et al., 2012; Sehaqui et al., 2012; Wang et al., 2016). The WS CNF film with outstanding transparency and exceedingly mechanical properties makes it has a great potential application in electrical and optical components as well as other research areas.
The cylindrical freeze-dried CNF aerogel was shown in Fig. 7a. To investigate the microstructure of the aerogel, the surface and fracture cross-section were examined by SEM. As can be seen from Fig. 7b, a highly porous internal network is clearly visible in the surface of the CNF aerogel. The individual nanofibrils have been forced into sheets by the growing ice crystals in the process of freeze-drying, and these sheets are connected to form an open, highly porous structure exhibiting large pores with several micrometers both in length and width. Furthermore, the sheets show a web-like comprising numerous individual nanofibrils (inset of Fig. 7b). The typical web-like structure can be ascribed to the rapid frozen of the CNF suspension by liquid nitrogen. When the cylindrical mold was dipped into the liquid nitrogen, the liquid CNF suspension near the mold wall was rapidly cooled and frozen, and a large number of individual nanofibrils were quickly fixed that prevented their aggregation into compact sheets. These in individual nanofibrils contributed much higher specific surface area of 9.22 m2/g (Fig. 7d) to the resultant aerogel (Mulyadi et al., 2016).
Figure 7. a: photo of CNF-assembled aerogel; b, c: FE-SEM images of surface and cross section of aerogel; d: nitrogen adsorption and desorption isotherms of aerogel
The morphology of the fracture cross-section is similar to that of the surface, which also shows a highly porous internal network structure (Fig. 7c). It can be seen that there were few individual nanofibrils in the presence of the sheets (Fig. 7c). During the process of frozen, the liquid CNF suspension in the core of the mold was frozen only after been put into liquid nitrogen for a period of time that have enough time to self-assemble into sheets, which also may be the reason for that the sheets accumulated in the cross-section look like more compact than the surface.
Wet spinning of the WS CNF suspension was successfully carried out by using the acetone as a coagulation bath. The spinning suspension was dehydrated immediately by acetone, leading the CNF to aggregate into fibrous structures. After been dried to remove the acetone, we have obtained the CNF filament with excellent flexibility (Fig. 8a). The morphology of surface and cross-section of the filament were examined by SEM as shown in Figs. 8b and 8c, respectively. At high magnifications of the filament, a rough surface and irregular cross-section (with an average diameter around 45 μm) were observed, which due to the highly mutual diffusion of acetone and water as well as the fiber-drying induced by acetone volatilization. This result was in good agreement with the filament spun from the wood cellulose nanofibers reported by Iwamoto et al. (2011). Moreover, the WS CNF could be observed on the surface under a higher magnification (inset of Fig. 8b). Fig. 8d shows the typical stress-strain curve of as-peppered filament. The table inserted in Fig. 8d lists the mechanical properties, exhibiting the CNF filament has good mechanical strength and rigidity due to these nanofibrils being arranged along the filament axis. The as-prepared single CNF can endure a counterweight of 20 g as presented in Fig. 8d.