4.1. Hydrodynamic Characteristics and Zeta Potential of the Investigated CNC Samples
The potential difference between the stationary liquid layer affixed to the dispersed particle and the dispersion medium is known as the zeta potential. It is a potential for colloidal particles in an electric field in the shear plane [
28]. The high zeta potential will provide stability for very small molecules and particles, which means the solution or dispersion will be resistant to deposition. When the potential is low, the attractive forces can overcome the thrust, breaking the dispersion, and causing the particles to clump together. As a result, low zeta potential colloids tend to clot or agglomerate, as illustrated in
Table 4, whereas high zeta potential (negative or positive) colloids are electrically stabilized.
The measurement of zeta potential is a straightforward method for estimating the stability of CNC suspensions, which is critical for nanocomposite materials.
Figure 2a displays the average zeta potential of three different types of CNC or CNF particles (HCl hydrolysis, H
2SO
4 hydrolysis, and enzymatic hydrolysis), ranging from −20.2 mV to −42 mV, with the highest stability registered for 1S
1 and 2S
1 samples (PZ between −42 mV and −32.8 mV, respectively). It has been determined from previous studies that increasing acid concentration, duration, and temperature causes hydrolysis to occur more effectively. When the temperature of the acid is around 70 °C and more concentrated sulfuric acid is used (55% and more), the color of the product turns black [
30]. In the same research, it was found that the zeta potential of CNC is −29.7 mV by using 65% sulfuric acid, during 50 min at 55 °C, but in our study, we used only a 10-min hydrolysis time at 70 °C by using a 60% sulfuric acid solution. Considering the zeta potential of CNCs extracted by the HCl hydrolysis method, these samples show fewer stable suspensions (PZ ranges between −20.2 and −20.3). The main cause of this is the lack of negative charges on the surface of CNCs produced by HCl hydrolysis. Similar results were found by Sánchez et al. [
31].
On the other hand, 1E
3 and 2E
3 samples (−26.4 and −25.4, respectively) exhibit slightly higher stability than 1C
3 and 2C
3 samples. This is due to the citrate ions used as a buffer solution during the enzymatic hydrolysis of cotton. This work showed very similar results to De Aguiar et al. [
25] who reported a zeta potential value of CNCs, which were obtained from sugarcane bagasse and straw, of −25 mV, showing a colloidal suspension with moderate stability, whereas in another work, Chumchoochart et al. [
32] reported a high ZP value of −45.8 mV by using the enzymatic hydrolysis method.
Figure 2b demonstrates that the mean hydrodynamic diameter sizes of the 6 CNC samples range from 77 nm to 539 nm, indicating a broad diversity of lengths. The samples from H
2SO
4 hydrolysis (1S
1 and 2S
1) had the smallest particles, measuring 97.91 and 77.17 nm, respectively. (
Figure 2b). The main reason for the larger particle sizes in the HCl hydrolysis samples (1C
3 and 2C
3), which are 350.8 and 243.1 nm, respectively, would be the following: 1S
1 and 2S
1 samples are prepared using 60% H
2SO
4 (the highest concentration among S samples), whereas 1C
3 and 2C
3 samples are prepared using 10% HCl (lowest concentration among C samples). In both samples, the 2nd method shows smaller particle size proving that the NaOCl bleaching method is more effective than the H
2O
2 bleaching method in terms of hydrodynamic diameter sizes. Similar size distributions were also observed in the work of Sánchez et al. [
31] who obtained larger CNC-C sample particle sizes (300–800 nm), compared to CNC-S samples (100–500 nm). Further, 1E
3 and 2E
3 samples have the biggest particle sizes—380.2 nm and 538.7 nm, respectively.
Figure 2c displays the images that demonstrate the stability of all 6 samples in aqueous suspension after being sonicated for 20 min at 40% amplitude. The suspensions were photographed at 0 h, 12 h, and 72 h. At first, all samples are colorless and homogenous, and no sedimentation was observed at the bottom of the bottle; but, over a considerable period (72 h), sedimentations happen in the 1E
3 and 2E
3 samples (due to large mean diameter and reduced ZP values), whereas the good stability for samples 1S
1 and 2S
1 are maintained even after 72 h. Among the six samples, 1S
1 and 2S
2 samples showed the best homogeneous suspension due to the electrostatic repulsion forces between sulfate groups in CNC surfaces, in accordance with zeta potential and DLS results.
4.2. FTIR Analysis
FTIR was used to examine the chemical structure of the synthesized CNC samples and raw cotton and the corresponding spectra are presented in
Figure 3.
The broad absorbance peaks at 3400–3300 cm
−1 areas that are present in all the recorded spectra are due to the stretching and bending vibrations of the cellulose OH groups. The absorption peak at that area is known as the crystalline band. The peaks at 2900–2800 cm
−1 correspond to CH stretching, while the peak around 1400 cm
−1 may be assigned to CH
2 bending vibration [
33]. The asymmetric C-O-C glycosidic band stretching vibrations caused the peak at 1164 cm
−1 [
34], while the peaks seen at 1119 cm
−1 and 1061 cm
−1 were caused by the stretching of the C-O bond [
35,
36]. The recorded spectra of the investigated samples show certain structural variations in addition to their remarkable similarity. For instance, the FTIR spectra of the S and E samples show a second peak at about 1640 cm
−1, which may be the result of the bending of adsorbed water, while the absence of this signal in the spectra registered for C samples and raw cotton indicate that the water was efficiently removed from these materials.
The lack of peaks at 1726 (C=O), 1550, and 1533 cm
−1 (carboxylate doublet) [
37,
38] suggests that the samples are free of hemicellulose and lignin units, and the FTIR spectra support the purity of cellulose in all the analyzed samples.
4.3. Raman Spectroscopy
Raman spectroscopy is a chemical analysis method that may be used to identify and measure the materials by the scattering of visible or near-infrared light from a laser source. Based on the molecular structure and forces, a sample’s vibrational spectrum may be assessed in Raman spectroscopy as a distinct “fingerprint”. Cellulose nanomaterials can be analyzed and characterized through Raman spectroscopy, which offers very important information about the crystallinity of the cellulose nanomaterials, and the detection of cellulose II or sulfate esters in CNC surfaces [
39].
Figure 4a–c shows the Raman spectra of all the examined materials as well as raw cotton.
The Raman spectra of CNC samples obtained through H
2SO
4 hydrolysis are shown in
Figure 4a. C-O-C glycoside linkage and C-C-C ring deformation bending are seen in samples 1S
1 and 2S
1 at 520 cm
−1 and 530 cm
−1, respectively [
40], whereas the signal around 1000 cm
−1 suggests stretching, C-C, and C-O: in-plane rocking (CH
2) [
41]. Cellulose ethers are identified by the intensities of peaks around 1100 cm
−1 in the region of the C-C and C-O single bonds [
42]. H-C-C, H-C-O, and H-O-C stretching and bending in-plane scissoring is indicated by vibrations between 1337–1378 cm
−1 (CH
2) [
43]. CH deformation intensities are seen in samples 1S
1 and 2S
1 at 1434 cm
−1 and 1436 cm
−1, respectively [
44]. The amide bond [
45] corresponds to the signal at 1664 cm
−1, which may be found in the amorphous region of CNC samples in the form of pectin residue. C-H and O-H stretches are seen in the 2500–3700 cm
−1 area of the Raman spectra of CNC samples [
46].
The Raman spectra of samples produced by HCl hydrolysis are shown in
Figure 4b. Samples 1C
3 and 2C
3 showed vibrations at around 435 cm
−1 and 379 cm
−1, respectively, which were caused by the deformation of the C-C-C and C-C-O cellulosic rings [
47]. The same peak intensities at 999 cm
−1 in both samples suggest stretching, and C-C and C-O in plane rocking (CH
2) [
41]. Peak intensities at 1095 cm
−1 can be linked to cellulose ether’s unique properties [
42]. The signals from 1374–1380 cm
−1 indicate stretching HCC, HCO, and HOC bending in-plane scissoring (CH
2) [
48]. CH deformation intensity is recorded at 1438 cm
−1 for both samples [
44]. The amide bond is indicated at 1663 cm
−1 [
45]. Both in Raman and FTIR spectra CH stretching (2800–3000 cm
−1) intensities are observed around 2900 cm
−1 [
44].
The Raman spectra of samples produced by enzymatic hydrolysis are shown in
Figure 4c. The C-S stretch of two methionine groups, which are present as enzyme residues in cellulose samples, is responsible for the weak band in the 1E
3 and 2E
3 samples at 700 cm
−1 [
49]. We also found vibrations in the E samples that were comparable to those in the S and C samples at roughly 1000 cm
−1, indicating stretching between C-C and C-O: in a rocking plane (CH
2) [
41]. The characteristics of cellulose ethers, which are formed by the reaction of cotton with an alkaline solution such as those seen in S and C samples, can be attributed to the vibration at 1095 cm
−1 [
42]. We observed Raman shifts at about 1367 cm
−1, which are caused by vibrations of CH
2 [
50]. CH deformation intensities are seen in both samples at 1443 cm
−1 [
41]. For samples 1E
3 and 2E
3, we noticed vibrations at 1666 cm
−1 and 1667 cm
−1, like in earlier samples, suggesting the presence of an amide bond in the amorphous portion of both samples. Vibration at 2901 cm
−1 indicates CH stretching [
44].
In addition to the existence of tiny quantities of lignin and ether, all spectra in Raman and FTIR confirm the characteristic bond in CNC samples and demonstrate the partial efficiency of CNC extraction from cotton.
4.4. XRD
The X-ray diffraction patterns for raw cotton and samples that were prepared using various techniques are shown in
Figure 5. Due to the phase transition of cellulose during the hydrolysis process, all samples exhibit peaks associated with crystalline cellulose, although the strength of these peaks varies between samples.
The primary crystalline plane (0 0 2) diffraction for nanocellulose samples was identified at 22.5° [
51]. The intensity of the peaks for the (0 0 2) plane (2θ = 22.5°) for all CNC samples is higher than that of the raw cotton sample. This indicates that in all samples the crystallinity is higher than that of raw cotton regarding the (0 0 2) plane. The same phenomena are observed for the (1 0 1) plane (2θ = 14.7°) also. We have used two different methods to calculate the % of crystallinity of the samples.
- (1)
The integration method [
52]: The integration method, which compares the area under the crystal peaks to the area under the whole curve using the following equation, can be utilized to determine the percentage of crystallinity of samples.
- (2)
Segal’s method [
51]: The intensities of the crystalline and amorphous regions can be used in Segal’s method, where I (Crystalline) is the maximum peak intensity of the crystallinity peak at 2θ around 22.5° or peak intensity at the plane (200), and I (amorphous) is the minimum peak intensity at 2θ around 18.2° or minimum peak intensity at the valley between planes (0 0 2) and (1 1 0) shown in
Figure 5 and the values of calculated % crystallinity through the two methods are reported in
Table 5.
In the H
2SO
4 hydrolysis method, the sample 1S
1 has higher crystallinity than 2S
1 (72.4% and 67.8%, respectively), indicating that the 1st bleaching method (20% (w) H
2O
2 and 4% (w) NaOH) has more effect on the crystallinity of the samples than the 2nd bleaching method (16% (w) NaOCl and 4% (w) NaOH) because all the other factors are the same for both samples (concentration of acids, reaction time, and temperature). A higher crystallinity value (77%) was reported for cellulose nanocrystals that were obtained from wood chips [
53], but due to the use of various cellulosic materials, the reported Segal’s techniques was 90% in the same paper, which is also higher than our results (86% and 82%). Nevertheless, in the HCl hydrolysis procedure, sample 2C
3 exhibits somewhat more crystallinity than the sample 1C
3 (76.4% and 73.1%, respectively) based on the integration method which is relevant to Segal’s method (89.1 and 87.9, respectively) and also a similar result has been found by Yu et al. [
51]. In the enzymatic hydrolysis method, sample 2E
3 shows much more crystallinity than sample 1E
3 (75.1% and 63.5%, respectively), but in Segal’s method, we have found around 84% crystallinity, which is very similar to the value that is reported by Camargo et al. [
54] that had 83.7% crystallinity by using the enzymatic hydrolysis method from sugarcane bagasse. All these results show that the 2nd bleaching method is more effective than the 1st bleaching method in C and E samples in terms of crystallinity. It was found that the bleaching methods had an impact on the crystallinity of the samples.
4.5. Scanning Electron Microscopy (SEM)
Figure 6 and
Figure 7 illustrate the morphology of the CNC samples made from raw cotton and the CNC samples obtained from acid and enzymatic hydrolysis. The morphology of raw cotton before bleaching is shown in
Figure 6a,b, and the morphologies of the 1E
3 and 2E
3 samples, which were obtained by enzymatic hydrolysis, are shown in
Figure 6c,d and
Figure 6e,f, respectively. This shows that the size of the sample’s fibrillated structure was reduced and that enzymatic treatment broke down the long fibrillar structure into individual fibrils, though the fibrils are still long.
Figure 6e,f show the 2E
3 sample has a higher fractionation of cellulose fibers than the 1E
3 sample (
Figure 6c,d). Both XRD and SEM images of E samples show that the 2nd bleaching method (NaOCl) has a greater effect on crystallinity and morphology than the 1st bleaching method (H
2O
2).
We may conclude that the cellulases and xylanases did not totally break the fiber length into extremely short pieces throughout the enzymatic process, just changing the smooth surface of raw cotton into a little rough and coarse structure. E samples can be named CNF (cellulose nanofibrils) instead of CNC (cellulose nanocrystals). A similar result was also reported by Zhai, X. et al., 2021 [
55].
Figure 7a–d display the SEM micrographs of the 1S
1 and 2S
1 samples, respectively; both bleaching processes (H
2O
2 and NaOCl) encourage the removal of cotton’s lignin and hemicellulose components, resulting in a surface that is very rough and full of wrinkles. In addition, needle-like structures can be seen on the surfaces of both samples.
Figure 7e–h illustrate the morphology of 1C
3 and 2C
3 samples obtained from hydrochloric acid hydrolysis, showing that the procedure did not totally shatter the structure of cellulose, but on the surface of cellulose fibrils (
Figure 7f,h), we can see rod-like structures rather than needle-like and fibrils are still long like E samples. Similar SEM pictures demonstrating the impact of acid hydrolysis have been published in the scientific literature [
14].
4.6. TGA Analysis
The main thermal parameters from thermogravimetric investigations are reported in
Table 6, while the TGA curve along with its derivative thermogravimetric curve (DTG) of each analyzed CNC product are reported in
Figure 8a,b.
From the graphs, it is seen that all three types of samples (S, C, and E) have their own thermal characteristics. The first weight loss stage at 150 °C in the case of the raw cotton sample was most likely caused by absorbed water, followed by the start of degradation (10%) at 304 °C and the maximum degradation temperature at around 375 °C. The residue at 600 °C of the cotton was less than 0.5%, noting the complete degradation of the sample. For samples 1S
1 and 2S
1, the first mass loss of 2% is most likely caused by water and volatile compounds up to 100 °C, and the highest weight loss (up to 10%) is seen in the onset degradation temperature range of 214 °C to 230 °C for the sample 1S
1 and 202 °C to 227 °C for the sample 2S
1. The residue at 600 °C of the CNC was about 34% for 1S
1 and 24% for 2S
1 due to the inorganic sulfate groups and shows that the NaOCl bleaching sample would have lower residue than the H
2O
2 bleaching sample. Similar outcomes have been reported [
56]; however, the presence of sulfate groups on the CNC surface is undesirable for particular applications since it accelerates decomposition at low temperatures [
57].
For samples 1C
3 and 2C
3, the maximum weight loss (up to 10%) is observed in the onset degradation temperature range from 300 °C to 326 °C for sample 1C
3, and from 302 °C to 328 °C for sample 2C
3. The residue at 600 °C of the CNC was about 1.4% for 1C
3 and 4% for 2C
3, revealing that there were no inorganic acids left behind after the combustion with oxygen gas. Similar work has been conducted [
24] by using 12% HCl acid (~3.3 M) and 120 min hydrolysis time at 100 °C. T
max was found as 342 °C and the initial onset temperature was found at 292 °C, which corresponds to our results despite using different conditions.
For samples 1E3 and 2E3, the maximum weight loss (up to 10%) is observed in the onset degradation temperature range from 241 °C to 291 °C for sample 1E3, and from 241 °C to 290 °C for sample 2E3. The residue at 600 °C of the CNC was about 13% for 1E3 and 12% for 2E3, showing that there can be some buffer solution (sodium citrate) that had been used together with an enzymatic cocktail.
We may conclude that while the thermal stabilities of the 1S
1, 2S
1, 1E
3, and 2E
3 samples are comparable, the sulfate groups had a detrimental impact on the thermal stability of the CNC samples [
58]. The slower rate of weight loss compared to the other samples in samples 1S
1 and 2S
1, however, may indicate that the sulfate groups in these samples may operate as flame retardants [
59].
4.7. DSC Analysis
The thermal parameters calculated from DSC analysis of cellulose nanomaterials at a heating rate of 10 °C/min are presented in
Table 7, and the obtained thermal curves are displayed in
Figure 9.
Because surface water is becoming dehydrated, all thermograms display an endothermic event around 80 °C, which corresponds to the evaporation of absorbed water. This is typical behavior of cellulosic materials due to the fact that there are multiple interactions between the hydroxyl groups from its surface and water molecules [
60].
The enthalpy for the melting transition and dehydration of the samples were calculated with the aid of the following equation.
According to the thermograms registered in
Figure 9, it can be observed that the obtained S, C, and E samples exhibit their own thermal characteristics. Large endothermic peaks around 80 °C for all nanocellulose samples demonstrate the heat absorption phenomenon for the water evaporation process [
61]. Sample 2S
1 has a water-loss temperature of 86.3 °C and an enthalpy of dehydration of 52.1 J/g, showing that there is a stronger interaction of water than in sample 1S
1, which has a water-loss temperature of 81.8 °C and an enthalpy of dehydration of 47.6 j/g. Similarly, the temperatures at which water is lost from samples 2C
3 and 1C
3 are 80.7 °C and 77.4 °C, respectively, and the enthalpy of dehydration are 61.7 j/g and 52.2 j/g, respectively. These results confirm that the NaOCl bleaching method for both acid hydrolysis presents higher water-loss temperature and enthalpy of dehydration than the H
2O
2 bleaching method. According to the melting point (T
m) of the crystalline area of cellulose nanomaterials, an endothermic event is seen at around 237 °C for 1S
1 and 2S
1 samples and at about 333 °C for 1C
3 and 2C
3 samples. These observed differences in T
m in the case of C (T
m = 333 °C) and S samples (T
m = 237 °C) may be explained by the requirement of less activation energy to initiate the thermal decomposition of S sample because of the addition of sulfate groups [
62].
In the case of 1E3 and 2E3 samples, an exotherm peak at 381 °C and 365 °C, respectively, is noted. This crystallization peak can be observed due to the rearrangement of molecules in a tighter network system, resulting in a certain amount of heat being released. Due to the significant number of amorphous areas in the materials or possible contaminants, the thermograms of these samples lack particular melting points and melting enthalpies. Both E samples showed approximately the same dehydration temperature but the enthalpy of dehydration of the sample 2E3 is much higher than that of the sample 1E3 (80 °C and 65.1 °C, respectively) indicating a higher water content than its counterpart. These results implied that different acid and enzymatic hydrolysis methods along with bleaching techniques could greatly affect the thermal degradation of nanocellulose.