Isolation and Characterization of Cellulose Nanocrystals from Rejected Fibers Originated in the Kraft Pulping Process

In the final process of the bleached kraft pulp there are some cellulose fibers that are separated from the main fibers stream; these fibers are rejected and considered as a low quality fibers, these fibers are known as rejected fiber (RF). In the present work the potential use of these fibers for Cellulose Nanocrystals (CNCs) synthesis was studied. The physical and chemical properties of synthesized CNCs were characterized through different techniques such as Atomic Force Microscopy (AFM), Scanning Electron Microscopy (SEM), X-Ray Diffraction (XRD), Fourier-Transform Infrared Spectroscopy (FTIR), and Thermogravimetric Analysis (TGA). Results demonstrate the feasibility of CNCs synthesis with a yield of 28.1% and 36.9%, and crystallinity of 73.5% and 82.7%. Finally, the morphology and synthesis conditions suggest that this industrial reject fiber (RF) could be used as a source for the CNCs production, thus adding value to the kraft process and opening new possibilities for innovation in the pulp industry.


Introduction
The main process for the production of cellulose pulp worldwide is the kraft process. Only in 2016, approximately 138 million Air Dry ton (ADt, cellulose pulp with 10% of wet) were produced using this method, of which 26 million were produced in South America [1]. During the kraft process, various wastes are obtained; these can be considered as by-products according to their later utility. Among these we can mention (A) Lignin, used mainly as fuel for electricity, and also as a source of low content urea-formaldehyde adhesive resins [2], and of carbon fiber [3]; (B) Tall oil, obtained from coniferous extractives, that can be used to obtain phytosterols, as well as distilled tall oil in different fractions, such as Tora, Tofa, and Pitch [4]; (C) Effluent sludge that can be burned to obtain concentrated energy through pyrolysis [5] or through hydrothermal carbonization processes [6].

Preparation of CNCs
CNCs from RF were prepared by sulfuric acid hydrolysis [17,28]. According to preliminary test for RF and studies reported to obtain CNCs, two conditions of hydrolysis were evaluated. Table 2 shows treatment conditions: sulfuric acid concentration, reaction temperature, and reaction time utilized. Five grams of RF sample were added to 200 mL of the acid solution, the resulting solutions were mechanically stirred during the reaction. The reaction was stopped with cold distilled water and the reaction products were washed three consecutive times by centrifugation (10 min, 12,000 rpm). The resulting suspensions were dialyzed against distilled water to constant pH. Finally, the samples were sonicated in an iced-water bath for 30 min and centrifuged at 9000 rpm for 5 min. Thus, the resulting supernatants containing the isolated CNCs were kept at 4 • C for further analysis. Their respective residues were cellulose fibers that were not completely hydrolyzed, called cellulosic solid residues (CSRs), that were dry weighed for yield estimations as described in Section 2.3.

CNCs and CSR Yields
Chemical oxygen demand (COD) measurements were used to determine CNCs yield according to procedure by Wang et al. [17]. The method of analysis is done by closed reflux and the samples are subsequently analyzed colorimetrically. First, the linearity and repeatability of a calibration curve of a potassium hydrogen phthalate solution was verified in the COD range of 25-1500 mg L −1 O 2 . Avicel ® samples were used for calibration and further calculations. A proper aliquot was taken and added in the COD reaction tubes (Spectroquant ® COD cell test). The tubes were thoroughly mixed in a vortex mixer and incubated for 2 h at 148 • C in a Spectroquant ® 320 Thermoreactor (MERCK, Darmstadt, Germany). Once the samples were cooled down to room temperature, the absorbance at 600 nm was measured on a spectrophotometer Shimadzu UV-vis 1603, and the COD was determined by interpolation with avicel calibration curve samples. On the other hand, the separated CSRs were measured using the gravimetric method that consists in overnight oven drying at 105 • C, finally the dry weight was used to calculate the CSR yield for each sample.

CNCs and CSR Images
After hydrolysis, the CSR products were examined using a scanning electron microscope (SEM, JEOL JSM6610LV, JEOL Ltd., Tokyo, Japan). Initially, samples were coated with gold by dc-sputtering technique using a Denton Vacuum Desk V-Sputter/etch unit, with 20 mA of current for 30 s. Then, the samples were observed with an acceleration voltage of 10 kV.
Surface morphology for CNCs samples was examined using an atomic force microscope (NaioAFM, Nanosurf, Liestal, Switzerland). The equipment was operated in phase contrast mode using PPP-FMAuD Gold Coated Force Modulation AFM Probes (Nanosensors, Neuchâtel, Switzerland) at a resonance frequency of 75 kHz, spring constant of 2.8 N m −1 , and tip radius of about 7 nm. For this purpose, the original CNCs solution was diluted fivefold with deionized water and sonicated for 30 min. Then, one drop of the diluted solution was deposited on a regular microscope slide and put it in a vacuum drying oven al 60 • C and 0.5 bar of air pressure for 30 min. After this procedure the samples are ready for AFM data acquisition, where the time of each essay was 310 ms per line and each micrograph has 1024 lines of resolution. From the AFM images obtained, 150 diameter measurements of the CNCs were taken, and average values were determined by using the ImageJ software (Fiji distribution, open-source) [20].

Sulfur Content Analysis
The sulfur content of each CNC was analyzed using an ICP-AES (Optima 800, PerkinElmer, Waltham, MA, USA) according to Chen et al. [28]. An aliquot of 5 mL of CNC suspension was transferred to a Teflon digestion flask and digested at 150 • C for 30 min in a microwave (Ethos One, Milestone, Sorisole, Italy) with 5 mL of 70% HNO 3 before ICP-AES analysis. The time ramp digestion at 150 • C was approximately 40 min.

Fourier Transform Infrared (FTIR) Spectroscopy
FTIR spectroscopy was used to examine the changes in the functional groups of CNCs obtained. The FTIR spectra were collected by using a Nicolet 380 FT-IR Spectrometer (Thermo Fisher Scientific, Hampton, NH, USA) in the transmittance mode. Five mg of powder samples of RF and CNC were dispersed in a matrix of KBr to be mixed and pressed into a pellet. The samples were analyzed in a spectral region between 4000 and 400 cm −1 with a 2 cm −1 resolution, while the obtained spectra were averaged over 32 scans.

X-ray Diffraction (XRD) Analysis
The X-ray diffraction (XRD) analysis was performed to determine the crystallinity of the samples from by-products of pulp mill and CNCs obtained. The samples were examined using WAXS (wide angle X-ray scattering) equipment, in the X-ray diffractometer Rigaku Smartlab ® . Approximately 0.05 g was used for each experiment on the sample holder. Angular scanning was conducted from 5 • to 50 • with 5 • min −1 with Cu Kα radiation (λ = 0.154059 nm), and the generator was working at 45 kV and 200 mA. The analysis of cellulosic material was performed according to reported crystallographic data by Ford et al. [29] and Nam et al. [30], using this information as an initial step to start and optimize the deconvolution of X-ray diffraction patterns.
The crystallite size τ (nm), perpendicular to the lattice plane (002) cellulose I was calculated in Equation (1) by the Scherrer equation [31], where K is the Scherrer constant (0.94), λ is the wavelength of the X-ray radiation (0.154 nm), and β is the full width at half maximum of the diffraction peak (in radians) and θ is the diffraction angle of the peak.
Also, the fractional variation in the plane spacing ∆d/d for the (002) planes was calculated following the Equation (2), according to Cullity and Stock [32].
The Segal CI was calculated according to Equation (3), where it is the total intensity of the (002) peak for cellulose I, and Ia is the amorphous intensity [33].
The thermogravimetric analysis of sample CNCs and RF was carried out on a TGA Q50 (TA Instruments, New Castle, DE, USA) under nitrogen atmosphere, with a gas flow of 50 mL min −1 from 25 to 600 • C at a heating rate of 10 • C min −1 . The weight-loss rate was obtained from derivate thermogravimetric data. For this analysis, about 2.0 mg of powder sample RF and CNCs were used.

Morphological Analysis
Morphology and size distribution of CNCs and its corresponding cellulosic solid residue (CSR) synthesized are shown in Figure 2. Figure 2a,b corresponds to images SEM of the RF that were not completely hydrolyzed, thus generating CSR, derived from experiments 1 and 2 (CSR-1, CSR-2) respectively. Figure 2b, shows a higher fractionation of cellulose fibers for the case of CSR-2 compared to CSR-1 (Figure 2a), such difference on fractionation might be due to the different hydrolysis conditions, resulting in different CNCs size distributions. On the other hand, AFM images shown a bigger size distribution for CNCs obtained in CSR-1 compared to those in CSR-2 (Figure 2c,d). The CNCs obtained are morphologically similar to those reported from by Cranston and Gray [34], derived from cotton fibers, and to those, reported by Leung et al. [35], derived from Flax.

Yields and Sulfur Content
The results of CNCs and CSR yields are shown in Table 3. Yields for CNC-1 and CNC-2 were 28.1% and 36.9% respectively (determined by COD). The production of CSR can be visually estimated from the precipitated cellulosic fibers in the post dialysis sample and quantified by the final solids remaining after overnight oven dry (105 °C). CNC-1 yield was 28.1%, which is less than the value reported by Wang et al. [17] that had a CNC yield of 39.5%, under the same conditions for highquality bleached kraft pulp as a starting material. On the other hand, CNC-2 yield was similar to previous reports using an acid concentration of 64.8 wt % with CNC yield of 38%, using pure microcrystalline cellulose as the starting material [14].  There are factors that can affect the CNCs production and their properties for instance; the cellulose source could impact in the size distribution and morphology of nanocrystals, besides other factors such as temperature and hydrolysis time [36]. The results shown that CNCs can be produced from RF for the experimental conditions evaluated, though differences in the size distribution of their diameters were found, CNC-1 presents a high percentage of nanoparticles with a diameter greater than 40 nm ( Figure 2e) and CNC-2 presents a homogeneous distribution of nanoparticles with diameter values between 10 and 35 nm (Figure 2f).

Yields and Sulfur Content
The results of CNCs and CSR yields are shown in Table 3. Yields for CNC-1 and CNC-2 were 28.1% and 36.9% respectively (determined by COD). The production of CSR can be visually estimated from the precipitated cellulosic fibers in the post dialysis sample and quantified by the final solids remaining after overnight oven dry (105 • C). CNC-1 yield was 28.1%, which is less than the value reported by Wang et al. [17] that had a CNC yield of 39.5%, under the same conditions for high-quality bleached kraft pulp as a starting material. On the other hand, CNC-2 yield was similar to previous reports using an acid concentration of 64.8 wt % with CNC yield of 38%, using pure microcrystalline cellulose as the starting material [14]. Other parameter that has been considered of importance for CNCs acid hydrolysates is the sulfation [37], the use of sulfuric acid for hydrolysis induces grafting of negatively charged sulfate groups on the nanocrystals surface, thus contributing to the stabilization of colloidal suspensions by repulsive interparticle forces [38]. The sulfur content determined for CNC-1 and CNC-2 was of 8.4 and 12.4 mg g −1 respectively (Table 3), similar values to those reported in literature [14,17]. From Table 3, is possible to notice that the sulfur content in CNC-2 is higher than in CNC-1, this due to the fact that sulfation is affected mainly by high acid concentration. Nevertheless, it worth noticing that the presence of sulfate groups on the nanocrystals surface might be disadvantageous for some applications, since this favors its decomposition at various temperatures [20].

FTIR Analysis
Following the chemical characterization, fourier-transform infrared spectroscopy (FTIR) was used to identify chemical group into the raw material (RF) and the mains samples CNC-1 and CNC-2, Figure 3 shows the FTIR spectra for those samples in the range of 600-3800 cm −1 .  From Figure 3, it is possible to notice the following characteristics: the absorbance peaks in the 3400-3300 cm −1 regions are attributed to the stretching and bending vibrations of the OH groups of cellulose, respectively. The peaks around 2900-2800 cm −1 correspond to CH stretching [39]. The CH stretching region is little analyzed, because cellulose, hemicelluloses, and lignin have similar peak shapes in this band [40,41]. The band at 1642 cm −1 of de RF and CNCs obtained is associated to From Figure 3, it is possible to notice the following characteristics: the absorbance peaks in the 3400-3300 cm −1 regions are attributed to the stretching and bending vibrations of the OH groups of cellulose, respectively. The peaks around 2900-2800 cm −1 correspond to CH stretching [39]. The CH stretching region is little analyzed, because cellulose, hemicelluloses, and lignin have similar peak shapes in this band [40,41]. The band at 1642 cm −1 of de RF and CNCs obtained is associated to adsorbed water in cellulose [16,41,42] The peaks observed in the range of 1420-1430 cm −1 were attributed to the symmetric bending of CH 2 [41] and were also related to cellulose I [41,42], the band at 1330-1380 cm −1 corresponded to the bending vibrations of the C-H and C-O groups of the polysaccharides [16,43]. The absorbance peaks observed in the 1161 cm −1 range were attributed to C-O-C asymmetric stretching vibrations associated with cellulose I and cellulose II [44]. Finally, the obtained spectra showed a slight increase in the peak intensities around 900 cm −1 which could be attributed to interactions between glycosidic linkages and glucose units of the cellulose [45]. Finally, the FTIR spectra of reject fiber and CNCs samples indicate that CNCs were successfully extracted from the hydrolysis treatment without any further degradation or secondary products formation. No significant changes were observed in the FTIR spectrum of the CNC obtained.

XRD Analysis of CNCs
The crystallinity of the CNCs obtained was investigated by XRD ( Figure 4). The results indicated that the CNCs (1 and 2) correspond mainly to cellulose type I, with a high index of crystallinity, and both have the particularity that they show the peak around 20.5 • . This is a characteristic for crystals with miller index (021) in cellulose type I polymorph, and this is particular because this peak does not always appear in all type I cellulose samples [27,28]. In general, the same type of crystalline planes observed for CNC-1 were observed for CNC-2 samples, with the difference that for CNC-1 samples, the peak with Miller indexes (101) and (10∑1) become smaller and the amount of cellulose in amorphous state become larger. This information is confirmed when crystallinity of the samples is calculated.  As can be observed in Table 4, crystallinity in CNC-2 samples is higher compares to those in CNC-2 samples. Besides, smaller crystallite size is found in CNC-1, with a high fractional variation in its interplanar distance, this last magnitude is used to analyze the strain and the microstresses between crystalline planes [32]. In this sense, we found that CNC-2 planes have less strain between them than CNC-1, and this could be related with a better crystal quality of CNC-2 in comparison with CNC-1. The fractional variation of the interplanar distance is a measurement for the dispersion of the crystalline planes values, within the crystallite; therefore, a greater value is associated with higher microstresses on the CNC [46]. This microstress could finally influence the physical properties of As can be observed in Table 4, crystallinity in CNC-2 samples is higher compares to those in CNC-2 samples. Besides, smaller crystallite size is found in CNC-1, with a high fractional variation in its interplanar distance, this last magnitude is used to analyze the strain and the microstresses between crystalline planes [32]. In this sense, we found that CNC-2 planes have less strain between them than CNC-1, and this could be related with a better crystal quality of CNC-2 in comparison with CNC-1. The fractional variation of the interplanar distance is a measurement for the dispersion of the crystalline planes values, within the crystallite; therefore, a greater value is associated with higher microstresses on the CNC [46]. This microstress could finally influence the physical properties of future materials or nanocomposite that can be developed using these CNCs as raw material.

Thermogravimetric Analysis
Thermal analysis of the CNC was investigated, and Figure 5 shows the TGA profiles for RF and CNCs samples. All the samples present similar thermal behavior in agreement with reported studies [16,20]. The weight loss below 140-180 • C could be attributed to water evaporation; additionally, two degradation stages were evident around 230 and 350 • C. These findings indicate that the thermal stability of cellulose nanocrystals prepared by sulfuric acid hydrolysis were lower than those for the RF sample, this due to the great number of free end chains in CNC surface derived from the small particle size, those end chains regularly start to decompose at low temperature [47].
It is well known that there is a "size effect" of the nanometersized crystallites in the in-plane direction on the thermal conduction, and that the cellulose crystallites transport heat more effectively with increasing cross-sectional area, being this a dominant factor with respect to the in-plane thermal conductivity [48]. This collected information agrees with the results of this study where CNC-2 exhibited higher crystanility and crystallite size, and lower interplaner distance compared to CNC-1 samples (see Table 4).

Conclusions
Cellulose nanocrystals (CNC) were synthesized from industrial secondary products (Rejected Fiber-RF) by acid hydrolysis with sulphuric acid. The yield obtained and the method for its measurement was discussed in order to present a consistent methodology for CNC production and a proper characterization of it. The results obtained reinforce the value of the degree of sulfation as a valuable indicator of the performance of NCC yield by sulphuric hydrolysis, and demonstrate the feasibility of obtaining thoroughly characterized CNCs of well-defined structure using a kraft industry by-products. The properties of the crystallites of the CNCs obtained reveal an important potential for the RF as a source for the industrial production of CNCs, offering a product of high crystallinity and of similar characteristics to those from sources of higher cost, such as pine or eucalyptus wood.

Conclusions
Cellulose nanocrystals (CNC) were synthesized from industrial secondary products (Rejected Fiber-RF) by acid hydrolysis with sulphuric acid. The yield obtained and the method for its measurement was discussed in order to present a consistent methodology for CNC production and a proper characterization of it. The results obtained reinforce the value of the degree of sulfation as a valuable indicator of the performance of NCC yield by sulphuric hydrolysis, and demonstrate the feasibility of obtaining thoroughly characterized CNCs of well-defined structure using a kraft industry by-products. The properties of the crystallites of the CNCs obtained reveal an important potential for the RF as a source for the industrial production of CNCs, offering a product of high crystallinity and of similar characteristics to those from sources of higher cost, such as pine or eucalyptus wood.