High energy oxidation and organosolv solubilization for high yield isolation of cellulose nanocrystals (CNC) from Eucalyptus hardwood

Cellulose nanocrystals (CNC) have been widely used as responsive materials, chiral templates, and tough nano-composites due to its unparalleled properties. Acid and enzyme hydrolyses are extensively employed to prepare CNC. These traditional approaches exhibit inherent limitations of corrosion hazards, time-consuming process, and/or low yield. Herein, irradiation oxidation and organosolv solubilization are conducted to cause rapid degradation with simultaneous crystallization of cellulose to achieve approx. 87% yield of CNC. The morphology, spectroscopic, and stability properties of the as-prepared CNC are characterized through UV-vis spectroscopy, zetal potential, XRD, TEM, DLS, GPC, FT-IR and TGA techniques. The resultant CNC suspension presents unique property with high stability after 9 months storage at 4 °C. Moreover, CNC liquid crystal phase is successfully generated by addition of anions or cations solution to the CNC aqueous dispersion without stirring. The innovative approach in this work opens an avenue to obtain CNC directly from lignocellulosic biomass through irradiation oxidation and organosolv solubilization without acid hydrolysis and washing procedure.

Recently, many new methods have been put forward to produce CNC in the literatures. For instance, Lu and co-workers 10 , Chen and co-workers 11 presented an efficient approach for extracting CNC from bamboo pulp and fiber using fully recyclable organic acids. Kontturi and co-workers 12 reported a novel strategy on CNC preparation in hydrogen choride vapor with 97.4% yield from Whatman 1 filter papers. The authors demonstrated that the use of HCl vapor resulted in the rapid degradation and hydrolysis of cotton-based cellulose fibers without practically any mass loss in the cellulose substrate 12 . Lee and co-workers 13 reported a facile and eco-friendly extraction of cellulosenanocrystals from softwood pulp via electron beam irradiation followed by high-pressure homogenization. However, limited information on CNC isolation directly from lignocellulosic biomass has been available using irradiation oxidation followed by organosolv solublization.
In this present work, using Eucalyptus hardwood as model raw feedstock, a facile environment eco-friendly proposal is described to prepare CNC directly from lignocellulose via irradiation oxidation and organosolv solubilization. The organosolv solubilization can result in a rapid lignin-dissolving and high-yield isolation of CNC from cellulose without adding adscititious acid as catalyst. More importantly, an increase in crystallinity occurred, as observed by X-ray diffraction (XRD), with high zeta potential (ZP) value during hydrolysis process. Moreover, the morphology, spectroscopic, and stability properties of the as-obtained CNC are systematically characterized. In the end, CNC liquid crystals phase is successfully produced by adding monovalent (H + , Na + ) and high valency cations (Co 2+ , Zn 2+ , Fe 3+ ) into CNC suspension at room temperature without stirring. In comparison with the CNC isolation strategies in the literatures, the merits in our work are: (1) CNC isolation directly from lignocellulosic biomass not from pulp and filter; (2) High yeild (~87%) of CNC obtained without acid hydrolysis and time-consumption washing; (3) High stability of the resultant CNC with no precipitates after 9 months storage; (4) Liquid crystals phases formation with low CNC concentration of 0.5 wt% by adding cation solution.
CNC isolation from Eucalyptus hardwood via organosolv solubilization. Prior to solubilization experiments, Eucalyptus wood was irradiated at 800 kGy using 60 Co-γ as radiation source with the intensity of 9.99 × 10 15 Bq in the Hunan Irradiation Center (Changsha city, Hunan province, China). The procedures of irradiation pretreatment on Eucalyptus were detailed in our previous work 3, 15 . The irradiated wood material was milled to <20 meshes particle size and dried at 105 °C for 4 h. CNC was successfully obtained from irradiated Eucalyptus by a three-step process that comprised cellulose fractionation in organosolv mixture solutions, NaClO bleaching and ultrasonic disintegration process. In the first step of fractionation in organosolv mixture solution, residue solid (RS) of cellulose fraction was obtained. In the second step, the separated RS was selected for further bleaching treatment with a mixture solution containing NaClO and NaOH, and the bleached cellulose (BC) was achieved. In the third step, BC samples suspended in distilled water were subjected to ultrasonic disintegration to obtain a uniform dispersion, and finally CNC solution was obtained. During CNC preparation, two types of Eucalyptus, irradiated and untreated feedstocks, were compared.
Cellulose fractionation from irradiated and untreated Eucalyptus in organosolv mixture solutions. 40 g of irradiated Eucalyptus were loaded into 500 mL high-pressure stainless steel reactor equipped with a stirring impeller. Then, 400 mL of organosolv solutions (including GVL/H 2 O (50:50, vol%), THF/H 2 O (60:40, vol%), dioxane/ H 2 O (60:40, vol%)) were subjected to the reactor. Fractionation reaction was started at 170 °C for 1 h by stirring at 400 rpm. After reaction, the reactor was immediately cooled to below 40 °C by flowing cold water through the external heating jacket. The reaction mixture was filtered using 45 μm filter paper to separate the residual solids and the filtrate. The residual solid (RS), assigned as insoluble cellulose fraction, was thoroughly washed with organosolv solution followed with ultrapure water for three times. The RS samples were denoted as I-GRS, I-TRS, and I-DRS corresponding to GVL, THF and dioxane organosolv solutions, respectively.
RS ) as organosolv mixture solutions. The other conditions were similar to those from irradiated Eucalyptus described as above. The RS samples from unirradiated Eucalyptus were expressed as U-GRS and U-ERS related to GVL and ethanol organosolv solutions, respectively.
All as-prepared RS samples were stored at 4 °C for the following CNC preparation.
CNC preparation by NaClO bleaching and ultrasonic disintegration process. CNC was successfully obtained from the RS samples by a two-step process that comprised NaClO bleaching and an ultrasonic disintegration process 16 . The conditions of NaClO bleaching were depicted in Table 1. Specifically, the RS samples were selected for further bleaching treatment with a mixture solution containing NaClO and NaOH. The bleaching experiments were performed in a flask at 50 °C for 2 h. The solid/liquid ratio was 1:20 (w/v). After bleaching, the slurry was centrifuged and washed by deionized water until pH neutralization. Then, white-like solid cellulose was obtained and referred as bleached cellulose (BC), which was expressed as I-GBC, I-TBC, I-DBC, U-GBC, and U-EBC, respectively. U-GBC was hydrolyzed with sulfuric acid to extract high crystalline cellulose. All obtained BC samples were suspended in distilled water (concentration 0.5 wt.%, calculated by the weight percentage of oven-dried sample to solution) with the total volume of 300 mL. The dispersion was subjected to sonication (600 W, a cylindrical titanium alloy probe (20 mm in diameter), 20 kHz ultrasonication frequency, JY98-IIIDN, Ningbo Scientz Biotechnology Co., Ltd, China) for 80 min to obtain a uniform dispersion, resulting in CNC solution. However, U-GBC and U-EBC dispersions were subjected to sonication for 30 and 100 min respectively to obtain a uniform dispersion. The as-generated CNC suspensions were labeled as I-GCNC, I-TCNC, I-DCNC, and U-ECNC, respectively.
Determination of CNC yield. A specified content of the CNC suspension was transferred to a weighing bottle, followed by freeze-dried. Then, the bottle containing the dried CNC powder was weighed. The final CNC yield was obtained through weight difference and calculated as eq. 1.
where, m 1 is the total mass of dried CNC powder and weighting bottle (mg), m 2 is the mass of the weighting bottle (mg), m 3 is the mass of feedstock material (mg), C is the cellulose content of raw feedstock material (%).

Morphology, spectroscopic, and stability properties of the as-obtained CNC. UV-vis spectroscopy
observation. The UV-vis spectroscopy of aqueous CNC suspensions (0.5 wt.%) was tested in the wavelength range of 400-900 nm on a UV-vis spectrophotometer (Shanghai YouKe Instrument Co. Ltd, China).

X-ray diffraction (XRD) analysis.
The XRD patterns were obtained using an X-ray diffractometer (Bruker, Germany) with Ni-filtered Cu Kα radiation at 40 kV and 40 mA. The diffraction data were collected from 2θ = 5−45°, 30 min per sample. The relative crystallinity index (CrI) was estimated using XRD peak height method 17,18 and calculated as eq. 2.
am 002 002 where, I 002 is the peak intensity of the (002) lattice diffraction at 2θ ≈ 22.3°, which represents both the crystalline and amorphous regions, and I am is the diffraction intensity of amorphous fraction at 2θ ≈ 18.3°.
Zeta potential (ZP) determination. All experiments were carried out in dilute CNC suspension (0.5 wt.%). The zeta potential (DLS, Zeta sizer Nano ZS, Malvern instrument) was used to evaluate the colloidal stability (zeta potential and size) of CNC suspensions based on electrophoretic light scatting at 25 °C.
Dynamic light scattering (DLS) measurement. CNC samples were dispersed in distilled water to achieve 0.5 wt.% suspensions. Size distribution experiments of CNC were performed in duplicate with a dynamic light scattering evaluator (DLS, Zeta sizer Nano ZS, Malvern instrument) at 25 °C.
Gel Permeation Chromatography (GPC) analysis. Prior to GPC analysis, cellulose samples were derivatized using anhydrous pyridine and phenyl isocyanate according to the procedures reported by Meng et al. 19 . The derived cellulose samples were dissolved in THF solvent (Baker, HPLC grade) with the final concentration of 10 g/L. The mixture solutions were filtered with 0.22 μm nylon membrane syringe filters. GPC analysis was carried out using an 1515 GPC instrument (Waters, USA) with three columns (Polymer Laboratories, 300 × 7.5 mm) packed with polystyrene-divinylbenzene copolymer gel (10 μm beads) with nominal pore diameters of 10 4 , 10 3 , and 50 Å. THF was used as eluent with the flow rate of 1.0 mL/min. An injection volume was 50 μL. A refractive index (RI) detector was used for measurement. Retention time was converted into molecular weight (Mw) by applying a calibration curve using polystyrene standards (molecular weight is ranging from 500 to 4 × 10 6 Da). Degree of polymerization (DP) of cellulose was calculated using the molecular weight dividing by 519, the molecular weight of cellulose tri-carbanilate monomer 20 .
Fourier Transform infrared spectroscopy (FT-IR) Analysis. FT-IR spectra were recorded using a FT-IR spectroscopy (VERTEX 80 v, Bruker, USA). The spectra were recorded from 600 to 2000 cm −1 at a resolution of 1 cm −1 . Transmission electron microscopy (TEM) images. TEM images of the resultant CNC were observed on a HT7700 Hitachi TEM (Hitachi, Japan). Sample was diluted to 0.01 wt.% and drop-cast onto the freshly glow-discharged carbon-coated copper grids. The sample was negatively stained with 2 wt.% aqueous uranyl acetate for 5 min. After removal of the excess solution, the sample was dried at room temperature prior to TEM imaging. The sizes of CNC were measured using an image analysis system (Image J version: 2.0.0-rc-43/1.50e).
Thermal gravimetric analysis (TGA) curves. The thermal stability of the cellulose samples was characterized with a DTG-60A thermal gravimetric analyzer (Shimadzu, Japan). The samples were heated from 40 to 1000 °C at a heating rate of 10 °C·min −1 in nitrogen atmosphere with the flow rate of 30 mL·min −1 .
Storage stability of CNC suspension. 0.5 wt.% of CNC suspension was stored at 4 °C for 9 months to evaluate its stability through UV-vis observation, DLS and ZP measurements. The as-prepared CNC suspension was expressed as the fresh CNC suspension, while the CNC suspension stored for 9 months was expressed as the mature CNC suspension.

Results and Discussion
Effect of organosolv solubilization on cellulose fractionation from biomass.  Fig. 1.
As seen from Fig. 1a, approx. 50-90% of the total lignin and 80-100% of the hemicellulose were removed from the Eucalyptus via organosolvents solubilization (Fig. 1a). It indicates that the selected three organosolv systems are suitable for solubilization solutions. As expected, the efficiency of delignification and hemicellulose removal seems to vary with organosolvent systems. In comparison with THF and dioxane systems, GVL/H 2 O solubization achieves about 91% of delignification and 100% of hemicellulose removals from Eucalyptus. It is thus speculated that the capacity of solubilization effect is dependent on the used solvents 3 .
In case of unirradiated feedstock, approx. 30% of residual solids (RS) are achieved. For irradiated material, the yield of RS is approx. 40% (Fig. 1b). This phenomenon is attributed to the fact that the irradiation pretreatment generates reactive oxygen free radicals (O 2 − ), which was detected by EPR (Fig. S1). These radicals are responsible for the oxidization of hydroxy groups on the aromatic rings of lignin and parts of hemicellulose to carboxyl groups. The main feature peak of carbonyl groups (C=O) toward irradiated samples at 1730 cm −1 was observed whilst it does not appear in the original sample (Fig. S2). This result is good concision with the XPS pattern of irradiated sample (Fig. S3). The amount of carboxylic groups allegedly by irradiation was quantified by conductometric titration 22 , and the value is 0.4 mmol/g CNC. The extent of irradiation-mediated oxidation is almost equivalent to the TEMPO-mediated 22,23 , electron beam-mediated 11 , and oxalic acid-mediated oxdition 13 . The irradiation oxidation was suitable for isolation of cellulose because the oxidation agent is more aggressive towards lignin and hemicellulose, whereas cellulose is hardly decomposed under the mild conditions 16 . After irradiation pretreatment, the irradiation-mediated carboxyl groups can act as active acid moieties to hydrolyze both amorphous regions of cellulose fraction to C6 (9.6%, 180 mg/L) and hemicellulose to C5 sugars (82.3%, 600 m g/L) in orgnosolv systems. These sugars were detected by HPLC and the data are show in Fig. 1c,d. In our initial works, irradiation pretreatment would decompose the lignocellulosic stubborn structure, resulting in enhancement of the downstream enzymatic hydrolysis, yeast fermentation and ethanol conversion of cellulose 3,15,[24][25][26] . In 2016, Zhou and co-workers reported a novel strategy of lignocellulose fractionation at solid biomass loadings up to 40 wt.% through irradiation assisted self-catalysis in GVL/H 2 O (40:60, vol.%) without adding mineral acids 3 . These articles showed that irradiation pretreatment had benefit effect toward delignification and hemicellulose removal.
Effect of organosolv solubilization on the characteristics of CNC. The as-prepared bulk cellulose was subjected to NaClO bleaching and an ultrasonic disintegration process to achieve CNC. Using I-GCNC as an example, Fig. 2a shows that a substantial change is observed in the dispersion of the CNC in water before and after ultrasonic disintegration (30 mL suspension was sonicated at 600 W for 5 and/or 10 min). The transmittance of CNC suspension is dependent upon sonication time. UV-vis observation indicates that 0.5 wt.% CNC suspension is reaching 95% of transmittance at 900 nm regardless of organosolvent (Fig. 2b). After 9 months storage at 4 °C, the CNC suspension is still transparent via UV-Vis observation and no precipitates are observed. DLS analysis reveals that the sizes of the CNC are 100-250 nm for 9 months storage and is small enough to be highly dispersed in aqueous solutions (Fig. 2c). The surface charges of the fresh CNC (I-GCNCF) and the mature CNC (I-GCNCM) for 9 months storage were detected by ZP measurements, which both show negative ZP values of about -43 mV (Fig. 2d). Obviously, CNC prepared from unirradiated Eucalyptus has lower ZP value (−30~−35 mV) than that from irradiated raw material. Low ZP value suggests weak electrostatic repulsive forces among the CNC fibers, which are easily aggregated together to form large precipitates. This hypothesis is confirmed through the UV-vis observation of CNC suspensions from unirradiated Eucalyptus (Fig. S4). On the contrary, high ZP values (−43 mV) of CNC suspension isolated from irradiated Eucalyptus suggest strong electrostatic repulsive forces among the CNC fibers, hampering to aggregate together of CNC fibers. The reasonable explanation is the fact that hydrophilic carbonyl groups or enol ethers are formed on cellulose structure after irradiation oxidation toward cellulose, these carbonyl groups (C=O) was detected by FTIR and XPS (Figs S2 and S3). Morphology and nano-size observation of the obtained CNC in this work were conducted by TEM. As seen from Fig. 3a-d, the widths of CNC range 4 ~6 nm and its lengths are 150~250 nm. However, if the Eucalyptus was treated with irradiation, the widths and lengths of the resultant CNC (I-GCNC) were slight narrower and shorter than the unirradiated samples (U-GCNC). In comparison with the references, the width of CNC from Eucalyptus hardwood in our work is one fifth smaller than that measured by Elazzouzi-Hafraoui et al. 27 (20~30 nm) and Lee et al. 13 (~26 nm), and similar width with that measured by Peyre et al. 22 (7 nm), and three times larger than that measured by Abushammala et al. 28 (1.9 nm). These differences are explained by supposing that these authors have measured the width of elementary crystalline which is not probably correspond to single crystals, but one particle inherently constituted of several crystallites.
As control, acid hydrolysis method was employed to prepare CNC from filter paper and irradiated Eucalyptus 9 . TEM images show that the widths and lengths of CNC obtained by sulfuric acid hydrolysis filter paper are similar to those CNC prepared by organosolvents solubilization (Figs S5 and S6). It is worthily noticed that acid approach for CNC preparation exhibits inherent limitations of corrosion hazards and environmental issues. Also, acid hydrolysis for CNC preparation is a time-consuming process due to long time dialysis (about 4 days). As seen from Fig. 3e,f, the yield of CNC obtained by irradiated feedstock by organosolv solubilization is calculated to be approx. 87%, it is much higher than those obtained by concentrated sulfuric acid hydrolysis and peroxide solvothermal methods 9,16 . and almost equivalent to those of CNC obtained by TEMPO oxidation 29 and by simultaneous mechanochemical activation and phosphotungstic acid hydrolysis 10 . Some researchers reported much higher CNC yield (97.4%) could be isolated from Whatman 1 filter papers in hydrogen chloride vapor 12 .
The crystallinity and thermal stability of the obtained CNC were analyzed using XRD and TGA methods. As seen in Fig. 4a-c, all XRD patterns exhibit the typical diffraction peaks around 2θ ≈ 18.3° and 22.3°, which confirm that the crystal lattice type I of native cellulose is preserved although raw material is treated through a series of irradiation oxidation and organosolv solubilization. The CrI, as a key factor in influencing the mechanical and thermal properties, was estimated according to the peak height method method 30 . The CrI of the irradiation oxidized samples (I-F) is estimated to be 60.1%, showing 4.1% decrease in comparison with the raw feedstock (U-F, 64.2%). The decrease in the CrI value undoubtedly is caused by the oxidation depolymerization of cellulose, indicating that irradiation oxidation has a destructive effect on the amorphous region and imperfect crystalline region of cellulose. This phenomenon is confirmed through molecular mass weight (Mw) variance after a series of treatments (Table S1). As seen in Table S1, DP w values of cellulose are decreasing from 250 for native cellulose to 52 for irradiated sample. After GVL/H 2 O solubilization, the CrI (69.1%) of CNC (I-GRS) is slight higher than native cellulose (64.2%), indicating that the removal of lignin and hemicellulose are occurred due to solvent solubilization effects. The CrI value of bleached cellulose (I-GBC) increases to 73.3% after NaClO treatment because of further delignification. After ultrasonication, the type I crystal lattice and CrI of the obtained CNC is unchanged (Fig. 4a), but the CrI value (77.8%) only slightly increases by 6.5% in comparison with that of I-GBC sample (73.3%). Fig. S7 show that the tendency of Crl values of the CNC obtained from unirradiated Eucalyptus is similar to those from irradiated feedstock. The TGA curves of the resultant CNC are shown in Fig. 4d-f. Cellulose thermal degradation procedure involves dehydration, depolymerization and decomposition of glycosyl units 10 . In other words, the TGA curve of lignocellulosic biomass has three characteristic regions: volatiles removal was attribute to mass loss, the removal of oxidation and other volatile materials was belonged to the second mass loss, and the third mass loss was attributed to the oxidation of residual coal occurs 31 . In all cases (I-GCNC, U-ECNC, U-GCNC), a small weight loss is found in the range of 25-120 °C due to the humidity evaporation or low molecular weight compounds remaining from the isolation procedures. In cases of all samples, the starting decomposition temperature occurs approx. 220 °C due to the decomposition of residual pectins and hemicellulose. At ca. 350 °C, this is associated with largest mass loss due to cellulose decomposition. At 400 °C almost all cellulose is pyrolyzed, and the solid residuals at 1000 °C are close to 18-25 wt.%. TGA thermograms show evidence of residual non-cellulosic matter 32 . As a result, either irradiation oxidation treatment or NaClO bleaching treatment shows a negative impact on the crystalline region of cellulose. This phenomenon is in good accordance with that reported by Li and co-workers 16 . CNC liquid crystals phase formation and optical properties. CNC liquid crystals phase was produced by diffusing monovalent (H + , Na + ) and high valency cations (Co 2+ , Zn 2+ , Fe 3+ ) into the CNC dispersion, through binding affinity of cations with carboxylate groups of CNC, the CNC gelation process was initiated 33 . As seen from Fig. 5, rapid gelation of CNC is triggered by adding 100 μL of cations solution to the 0.5 wt.% CNC dispersion, which was observed by inverting the vials with the gels. The concentration of cation solution is ranging from 0.1 M to 1.0 M. Unlike the preceding CNC aqueous dispersion, the CNC liquid crystals phase had no fluidity and maintained shape even with slight shaking. Dong and co-workers demonstrated that the driving force of gelation was proposed to be a screening effect of the repulsive potentials on the fibril surfaces by strong cation ion-carboxylate bonding 34 . As observed in Fig. 6a-d, polarized optical microscopy (POM) images show that CNC liquid crystal phase exhibits optical activity with a shift in color from gold to blue through clockwise rotation of 90° for liquid crystal phase. It is attributed to CNC ordered arrangement 35 . In comparison with the references, low CNC concentration (0.5 wt.%) in our work can form liquid crystals phase 36,37 . High polydispersity of CNC rods promote the formation of liquid crystal phase and arrest gel-like glassy states 38 . The strength of the gel depends on the concentration and type of cation ions. From Fig. 5, when 1 M of acetic acid or formic acid are added into the CNC suspension, the strength of CNC gel is weak, however, 1 M of HCl, Na + , CO 2+ , Zn 2+ and/or Fe 3+ are added into the CNC suspension, the CNC gel is much stronger due to the strong repulsive ion strength for gel formation 38 . Therefore, CNC is an ideal material as the chiral nematic liquid crystal template for optical devices, sensors and chiral separation 39 .

Conclusions
This study reports a novel strategy of organosolv solubilization to achieve CNC with the yield of approx. 87% through rapid degradation with simultaneous crystallization of cellulose. The resultant cellulose is 4-6 nm wide and 150-250 nm long, with high polydispersity of 25-63. CNC properties are comprehensively characterized by TEM, DLS, ZP, UV, XRD, XPS, and FT-IR. More importantly, no mineral acids are used through the whole CNC preparation process. Therefore, the approach proposed in our work is a sustainable and environment eco-friendly alternative for CNC preparation. The CNC suspension exhibits high stability, and behaves as a chiral nematic lyotropic liquid crystal with a shift in color. Future work will be done to generate the CNC gels with highly tunable optical properties, mechanical properties, and chemical selectivity.