Analysis on Impact of Acid Hydrolysis on the Hydrogen Bonding Network in Cellulose to Obtain Microcrystalline Cellulose: A Statistical Approach

Hydrolysis of a cellulose biomass results in breaking down the cellulose microbrils into microcrystalline cellulose (MCC) or nanocrystalline cellulose (NCC) depending on the reaction conditions. Cellulose microbrils are established robustly due to the synergistic interaction of van der Waals, inter- and intramolecular hydrogen bonds and glycosidic bond between glucan moieties of cellulose polysaccharide. The hydrogen bonding network plays a crucial role in conforming cellulose chains into crystalline and amorphous region thereby determining its degree of crystallinity. The knowledge of hydrogen bonds in cellulose hence becomes indispensable to understand the crystallinity of cellulose before and after a hydrolysis reaction. However, the nature of hydrogen bonds after hydrolysis and how they contribute to the mechanical properties of resultant MCC/NCC are yet to be realized. This paper is therefore intended to discuss the degree of crystallinity of cellulose particles obtained after hydrolyzing waste cotton bers (WCF) in two parts: part I, obtaining MCC with maximum total crystallinity index (TCI) by acid hydrolysis of WCF using Box Behnken Design; part II, comparing degree of crystallinity of MCC sample exhibiting highest TCI with that of WCF using analytical tools like X-ray Photoelectron Spectrometer, X-ray Diffractometer and Fourier Transform Infra- Red spectrometer. The physical dimension of MCC particle with maximum TCI has been veried using Field Emission Scanning Electron Microscopic images. Behnken


Introduction
Conversion of raw cotton to fabric involves a great deal of mechanical and chemical processes wherein textile wastes are generated abundantly. These wastes are either degraded to compost or upcycled to high end ameliorated products. Many authors have reported upcycling of waste cotton leftovers to produce attractive products such as raw materials for biofuels (Chaturvedi & Verma, 2013), micro brillated cellulose (MFC), microcrystalline cellulose (MCC) and nanocrystalline cellulose (NCC) for ller applications (Farahbakhsh et al., 2014). Of all these applications, manipulation of MCC/NCC as llers has attained exceptional attention due to its low cost, low health risk, biodegradability, recyclability and its function as property modi er (Ling et al., 2019) (Farahbakhsh et al., 2014). Both acid and enzymatic hydrolysis are widely used techniques to isolate MCC from raw materials, but the former technique using mineral acids is the mostly preferred method because enzymatic hydrolysis presents few obstacles such as di culty in access to crystalline cellulose and requirement of pretreatment. The important parameters that affect the acid hydrolysis of cellulose biomass are acid concentration, temperature, reaction time and solid: liquid ratio ( ber (g): acid (ml)). These parameters could be optimized to achieve required properties of MCC which includes high yield, particle size, crystallinity index etc.
In order to choose MCC as a reinforcement to enhance the mechanical property of a polymer matrix, it should possess better mechanical properties. (Ilyas et al., 2018) has pointed out that higher crystallinity indices of the bers result in higher tensile strength of the bers. The main aspect that contributes to its mechanical strength and chemical inertness is the hydrogen bonding network that holds the cellulose micro brils together. (Nelson & O'Connor, 1964) denoted this fraction of OH groups which are hydrogen bonded in a regular crystalline pattern as Total Crystallinity Index (TCI). It has been proposed that the absorptivity ratio of the 1372 and 2900 cm − 1 peaks derived from IR data is congruous in assessing the TCI of the cellulose samples. The basis for selection of these of two peaks are: 1372 cm − 1 (C-H bending) is sensitive to degree of crystallinity of cellulose and insensitive to adsorbed water while the absorbance at 2900 cm − 1 (C-H stretching) does not change signi cantly with respect to changes in the crystallinity.
A complete elucidation on crystalline faction and molecular alignment in crystalline faction along with the hydrogen bonding system of cellulose Iβ (prepared from cellulosic mantles of tunicate) was given by (Nishiyama et al., 2002). The comparison between the synchrotron X-ray and neutron ber diffraction results stated that the cellulose Iβ consist of two parallel chains with slightly different conformations and arranged in "parallel up" fashion. The monoclinic unit cell parameters was recommended as follows: a = 7.784 Å, b = 8.201 Å, c (chain direction and unique axis) = 10.38 Å and γ = 96.5º. A detailed description of the preferred and random orientation of cellulose Iα and Iβ, cellulose II, cellulose III I and III II based on simulation using Mercury 3.0 program is presented by (French, 2014). Accordingly, apart from the polymorphic nature of cellulose samples, the re ections and intensities of diffraction peaks is dependent on the sample preparation (a pressed pellet or sprinkled onto sticky tape) and sample orientation to the incident x-ray beams. If the cellulose Iβ crystallites are randomly oriented when they are placed in the sample holder, the shoulder peak at about 20.5º 2θ for the (012) and (102) re ections are observed.
Whereas they do not appear with crystallites having preferred orientation about the ber axis that lie parallel to the sample surface. The moderate peak at 34.5º 2θ is a composite of several re ections and hence (004) is not a dominant contributor to the diffraction peaks of cellulose Iβ. The empirical formula proposed by (Segal et al., 1959) to estimate the degree of crystallinity of native cellulose using XRD pattern is widely used wherein the crystallinity index (CI) is expressed as a ratio of difference between maximum diffraction intensity corresponding to (200) lattice diffraction and the intensity of diffraction at 18 º 2θ to that of maximum diffraction intensity corresponding to (200) lattice diffraction.
Cellulose polymer could be studied using XPS as the carbon atoms respond to the surrounding chemical environment by exhibiting changes in the binding energy. Though the elemental composition of cellulose is primarily carbon, oxygen hydrogen and XPS could only detect carbon and oxygen atoms until the depth between 10 and 20 nm, it helps to achieve better insight on the effect of hydrolysis on the cotton ber surface. In cellulose, the carbon atoms are either bonded to single oxygen atom (C-O) or to two oxygen atoms (O-C-O) and to carbon and hydrogen (C-H). These different environments are re ected as signals in the XPS spectra of cellulose polymer. Theoretically, the O/C ratio of pure cellulose is found as 0.83. Upon deconvolution, C1s signal of pure cellulose is supposed to exhibit two sub-peaks due to (i) C-O or C-OH from alcohol bonding and (ii) O-C-O from acetal bonding. Conversely, most cellulose exhibit an additional peak which may be due to C-C bonding from contaminants or any other impurities in the sample (Ly et al., 2008). (Fras et al., 2005) has reported that the outer laminar layers of waxes, proteins and pectin in the cotton ber consisted of C1 type carbon.
Design Expert is a statistical software that offers a wide range of tools for screening, analyzing and optimizing the vital factors that affect a process. Response Surface Methodology (RSM) is a tool in Design Expert that helps to determine the ideal process settings and thus achieve optimal performance of the process. The most common methods in RSM used for optimization of a process are Box Behnken Design (BBD) and Central Composite Design (CCD). BBD has more advantages over CCD due to the following reasons: (a) less expensive as it provides minimum experimental runs with same number of factors, (b) it runs with fewer design points than CCD. Optimization of acid hydrolysis of WCF to obtain MCC particles with maximum possible TCI using BBD is a novel work. Hence, in this communication, the authors report statistical optimization of the same and the aftermath of hydrolysis on hydrogen bonding network. Since FTIR spectroscopy is a simple yet robust technique as the IR data is in uenced by the background of the molecules involved in the vibration, FTIR was used to evaluate TCI. The diffraction studies were also used to reiterate the degree of crystallinity of the MCC sample with maximum TCI. The purity of the same has been validated using XPS analysis.

Materials
Waste cotton bers was obtained from neighboring textile industry (Coimbatore, Tamil Nadu).
Hydrochloric acid, Iodine, Potassium Iodide, Zinc Chloride and acetone were all laboratory grade reagents purchased from HiMedia.

Methodology-Extraction of Cellulose Particles
WCF was oven dried at 100°C for 30 mins prior to the hydrolysis and cellulose particles were extracted using hydrochloric acid based on these experimental runs provided by Box-Behnken Design. The hydrolysis process resulted in the depolymerization of the cotton bers thereby reducing it into particles which were believed to be microcrystalline in nature. Cellulose particles were then centrifuged at 5000 rpm for 15 mins and neutralized to pH 7 after multiple washes with double distilled water. The resulting aqueous cellulose was ltered using Whatman lter paper and oven dried at 80°C for 30 mins. The dry cellulose powder was stored in airtight vials and were used for further studies.

Model Development
Box-Behnken model from Design Expert software v7.0.0, Stat EaseInc., Minneapolis, USA was used in the present work to optimize output surface response with respect to the input reaction parameters. Four reaction parameters that strictly affect the acid hydrolysis process viz., acid concentration, solid: liquid ratio, temperature and reaction time was opted to optimize TCI of the cellulose powders. A 3-level BBD, low (-1), high (+ 1) and mean level (0) for the input parameters are given below: Acid Concentration (AC) -2.5/5/3.75 N, solid: liquid (S:L) ratio -1:15/1:25/1:20 (g/ml), Temperature (T) -80/90/85°C and time (t) -60/90/75 mins. Combinations of experimental runs with 3 central points as suggested by the BBD model in terms of coded and actual variables are shown in Table 1.

Characterizations
Initially, WCF and all cellulose powders obtained from BBD experimental runs were subjected to ATR-FTIR analysis to evaluate TCI as proposed by (Nelson & O'Connor, 1964) using the ratio of absorbance at 1372 and 2900 cm − 1 . WCF and cellulose sample that resulted in maximum TCI (MTCI) were then subjected to the following characterizations: FE-SEM, XPS and XRD. Colour identi cation test was also performed on MTCI to con rm its microcrystalline nature qualitatively.

Field Emission Scanning Electron Microscopy
WCF and MTCI were assessed using MIRA3 XMU (TESCAN) FESEM microscopy in order to understand and determine the morphology and particle size respectively. The two samples were sputtered with gold prior to observation under FESEM.

X-Ray Photoelectron Spectrometry
X-ray Photoelectron Spectrometry (XPS) was performed using a PHI5000 VersaProbe III spectrometer (ULVAC-PHI, Inc, USA), equipped with a monochromatic X-ray source (Al Ka, hn ¼ 1486.6 eV) operating at 150 W. During the analysis, the samples were ooded with low-energy electrons to counteract surface charging. The hydrocarbon component of the C1s peak with binding energy 285.0 eV was used as a reference for charge correction. Survey spectra were recorded at 1eV/ step while the region spectra were taken at 0.05 eV/ step.

X-Ray Diffractometry
The X-Ray diffraction pattern for WCF and MTCI samples were acquired using Schimadzu, Japan/600 diffractometer with Cu Kα (1.5406 Å) as radiation source. The lattice parameters of the samples were calculated using the following equation.

ATR-Fourier Transform Infra-Red Spectroscopy
The IR spectrum of the WCF and MTCI was obtained using Perkin Elmer ATR-FTIR spectrometer (Version-Spectrum Two). The waste cotton bers were pressed into a small patch so that it could be placed on the Zinc Selenide plate whereas the cellulose powders were ground with KBr in order to collect the spectrum.
The spectra were obtained in the range of 4000 cm − 1 -500 cm − 1 with a resolution of 4 cm − 1 . FTIR data was used to evaluate parameters like hydrogen bond energy (HBE), hydrogen bond length (HBL) and lateral order index (LOI) using the following equations (4), (5) and (6) respectively as described by

Box-Behnken Statistical Analysis
TCI of WCF was found to be 0.367 before hydrolysis. TCI is the key parameter that helps in understanding the hydrogen bonding networks of cellulose. Therefore, TCI of the cellulose powders was chosen as the response to the input parameters that affect its hydrolysis. The experiment was conducted based on the runs provided by the BBD model and the results are tabulated as shown in Table 1.  Table 2 provided the data associated with the sequential sum of squares, mean squares (MS), F-value and p-value. For a statistically signi cant model, the F-value should be the highest and the p-value should be lowest in the SSS table. Based on the response obtained and statistical analysis between the response and the independent variables, the design suggested quadratic model as the best t model as it has the lowest p-value. In a model under consideration, the error that has occurred could be accounted on two bases: (i) lack of t and (ii) random error. A model exhibits lack of t when (a) it does not describe the relationship between the input factors and the response factor satisfactorily or (b) replicate data are displayed or (c) the model has omitted the important terms. For a model to be suggested, it should have insigni cant lack of t which means it should have low F-value and high P-value > 0.1. From Table 3, it was evident that the quadratic model has insigni cant lack of t.   The equation with coded factors and actual factors for the quadratic model as suggested by the design is given below in equations 7 and 8 respectively: The nal equation in terms of coded factors helps in plotting the diagnostic graphs and optimizing the data. The nal equation in terms of actual factors helps in predicting the response based on the actual factors. According to Eq. 8, the TCI response which was predicted by the design for the design points, the actual response and the residual term are shown in Table 6. The residual term is obtained as a difference between actual and predicted value.

Optimization of Process Parameters
Optimization of TCI was done by choosing process parameters within the selected range of low and high level and maximizing the response parameter. The software furnished 25 optimum solutions with desirability value 1. One solution with parameter combination of AC-4.33 N, S/L ratio-1:16.76 g/ml, T-90°C and t-76.24 minutes that provided the maximum TCI of 0.518 was opted for the desirability test. Three trials were performed using above condition and the average TCI has been compared with suggested solution in Table 7. Consequently, a combination of AC-4.3 N, S/L ratio-1:16.7 g/ml, T-90°C and t-76 minutes was considered to be optimum for obtaining maximum TCI cellulose powders from waste cotton bers.

FESEM Analysis
The micrographs of WCF and MTCI acquired from FE-SEM are depicted in Fig. 3.a and b respectively. Apparently, the continuous lament network of WCF was ruptured down after hydrolysis and appeared to be rod-like particles of irregular size and shape. The inset in Fig. 3(b) showed the asymmetrical cleavage of a cotton ber which lead to the irregular size and shape of MTCI particles. The average diameter of WCF was found to be 19.22 microns while the average diameter of MTCI was reduced to 8.84 microns after hydrolysis. The average length of WCF was measured to be 4.5 cm and that of MTCI was reduced to 24.6 microns. This reduction in particle size of MTCI sample has caused the particles to coalesce as seen from the micrograph of MTCI due to increase in the number of contact points per MTCI particle (Metzger et al., 2020).

Colour Identi cation Test
The cellulose samples obtained from cotton ber was almost white in colour, odorless, tasteless with free owing powdery texture. In order to con rm microcrystalline nature qualitatively as observed from FESEM results, the MTCI sample was subjected to colour identi cation test. Iodinated zinc chloride solution was prepared as described by (Ejikeme, 2008). The test involved treating 0.01g of MTCI powder with 2 ml iodinated Zinc Chloride solution. A change in colour from white to violet-blue was inferred which con rmed that the MTCI particles were microcrystalline.

XPS Analysis
XPS analysis was performed in the current study to ascertain the signi cant removal of contaminants and effect of hydrolysis on cellulose polymer. A comparison of wide scan spectrum of WCF and MTCI is portrayed in Fig. 5.a and the presence of carbon and oxygen in the surface of both samples was clearly discernible. The atomic percentage of C and O as derived from wide scan spectra and that of carbon components from high resolution spectra for WCF and MTCI are provided in the Table 8. The O/C ratio has increased from 0.47 to 0.57 in MTCI particles which could be attributed to the removal of noncellulosic components considerably.
The high resolution spectra of C1s peak of WCF and MTCI after deconvolution are displayed in the Fig. 5 (b) and (c) respectively. The deconvoluted peaks are represented as C I, C II and C III starting from the lower binding energy. As seen from the Figs. 5.b and c, the intensity of C I deconvoluted peak has decreased by 7.4 % in MTCI spectrum which could be due to the removal of contaminants or impurities.
The same was re ected in the improved O/C ratio from the survey spectra of MTCI sample. The intensity of C II peak increased while that of C III peak decreased in the deconvoluted spectrum of MTCI when compared to WCF. The reason could be attributed to the rupture of hydrogen bonds and glycosidic bonds that exist between the cellulose molecules (Pan et al., 2013).

XRD Analysis
The structure of the cellulose chain, its arrangement and the bonding between these chains are the prime factors that contribute to the crystallinity of the cellulose polymer. These factors could be inferred from Xray diffraction studies by determining the unit cell parameters, crystallite size and crystallinity index of the material under consideration. X-Ray patterns of WCF and MTCI are illustrated in Fig. 6. In the current study, the cellulose Iβ model as suggested by (Nishiyama et al., 2002) was followed wherein the c-axis direction was considered the ber axis, the intramolecular H-bonds orient along a-axis and the intermolecular H-bonds orient along b-axis direction of the unit cell. Presence of peaks at 22.7º, 14.8º and 16.5º 2θ from the re ections of (200), (1-10) and (110) respectively con rmed cellulose Iβ crystal structure before and after hydrolysis. In line with the discussions of (French, 2014), it has been understood that these re ections are the main peak contributors in the XRD pattern of cellulose Iβ. The low intense peak at 34.5º is a composite of other neighboring re ections and hence is not a dominant contributor. During x-ray analysis, WCF was pressed onto the sample holder such that the ber axis and the plane of surface of sample holder lies parallel to each other giving rise to preferred orientation. Whereas, the MTCI particles were sprinkled onto the sample holder and hence a shoulder peak appears at 20.6º corresponding to (102) plane is divulged indicating the random orientation pattern. Both preferred and random orientation of crystallites of the WCF and MTCI are visible in Fig. 6.
Lattice constants of WCF and MTCI as calculated from the inherent re ections of cellulose Iβ viz., (1-10), (110), (200) and (004), the crystallite size and Segal CI are listed in Table 9. It was evident that the intramolecular H-bonds contracts while the intermolecular H-bonds stretches as perceptible from the decrease in the a-dimension and increase in the b-dimension of the MTCI crystallites. after hydrolysis.
The contraction of intra H-bonds and stretching of inter H-bonds might counterbalance any change in the C1-O-C4 glycosidic bond length. This notion was justi able due to the absence of change in the cparameter of the MTCI crystal. Further, the reason for decrease, increase and null change in a, b and c lattice parameter values respectively is discussed at length using FTIR spectral analysis. Secondly, the crystallite size of MTCI has increased from 4.7 nm to 5.4 nm. Dissolution of amorphous segment during hydrolysis would have allowed the cellulose molecules in the crystalline portion to relax and rearrange the cellulose chains in lateral direction such that the average crystallite size increases in MTCI. Next, the CI of MTCI has improvised by 8% which a rmed the increase in crystalline segments by removing the amorphous segments due to hydrolysis. All these outcomes from XRD analysis will be correlated with FTIR discussions in the following section.

ATR-FTIR analysis
It is believed that almost all OH bonds in a cellulose crystal are engaged in highly coupled and delocalized intra and inter chain hydrogen bonds (Lee et al., 2015). During acid hydrolysis of cotton bers, the amorphous segments gets solvated leaving behind the crystalline segment intact depending on the reaction conditions. This process results in change in the nature of hydrogen bond linkages which could be understood by subjecting the samples to FTIR analysis before and after acid hydrolysis. Since the hydrogen's ability to scatter X-rays are weak, both XPS and XRD were not capable of elucidating the role of hydrogen bond network completely. Yet, FTIR spectroscopic technique bestowed individual hydrogen bonding frequencies upon deconvolution of the spectrum with better precision. Peak positions and assignments to the peaks derived from the spectra of WCF and MTCI were displayed in Table 10. The spectra was divided into three segments viz., 3500-3000 cm − 1 , 3000 − 2800 cm − 1 and 1700 − 600 cm − 1 for the purpose of analysis.  Fig. 7. Profound differences were observed in the region such as relative peak broadening, peak symmetry and increase in peak intensity which offered interesting facets pertaining to the hydrogen bonding system of cellulose after acid hydrolysis. The resolved spectrum of WCF showed peaks at 3482, 3395, 3340, 3275 and 3132 cm − 1 . As observed from the spectrum, the inter-chain vibration 6OH···3O (3275 cm − 1 ) contributed the most to the peak whereas the coupled hydrogen bond interactions 2OH···6OH···3OH···5O (3340 cm − 1 ) contributed the least. This led to conclusion that the intermolecular H-bonds plays the dominant role in holding the cellulose chains to bundle up into micro brils. Taking into account that these peaks arise due to the combined contributions from crystalline and amorphous segments, the peaks 3395, 3340 and 3132 cm − 1 which are present in WCF vanished in MTCI. This showed that the 3OH···5O intra-chain (3395 cm − 1 ) and the coupled hydrogen bond networks (3340 cm − 1 ) of the cellulose chains are more prone to protonic strike of the acids. In addition, the peak 3132 cm − 1 which correspond to the stretching vibration of 2,3 and 6 OH groups that are actively engaged in bonding are also more prone to acidic attack as inferred by the disappearance of this peak. These peaks could perhaps be the interfacial bonds between crystalline and amorphous segments and the solvation of most part of amorphous segments could have affected these bonds. This would result in dangling or free OH bonds in the cellulose molecule of the crystalline portion which was in line with the increase in intensity of C II sub-peak as seen from XPS analysis. Figure 8 demonstrates the structure of cellulose molecule before and after hydrolysis.
In  Table 11. HBE and HBL values obtained was in accordance with the above explanation. An increase in LOI indicated transformation of cellulose bers towards higher crystallinity in MTCI which was corroborated by XRD. Figure (8 to C5 positions and methylene group at C6 position of cellulose unit cell. The C6H 2 asymmetric stretching peaks 2965, 2932 and 2898 cm − 1 in WCF are shifted to higher frequencies in MTCI while the C6H 2 symmetric stretching peak 2857 cm − 1 was shifted to low frequency as seen from the Table 10 and Fig. 9.
The increased frequency of CH 2 asymmetric stretching denoted the strengthening of C-H as a result of weakening of inter-sheet hydrogen bond due to acid hydrolysis. Hence it was concluded that the acid hydrolysis affected the intersheet hydrogen bonding the most. Contrarily, the red shift of CH 2 symmetric increase in the intensity of the 896 cm − 1 peak which could be attributed to the changes around C1 and the surrounding valence atoms. This denoted that these bonds were strengthened after hydrolysis. From the above discussions, it could be concluded that MTCI had improved intra-and inter-molecular H-bonds which could reason out for the high TCI of the sample as indicated by BBD. Also, it was justi ed that at places where H-bond weakened, the associated covalent bond strengthened and vice-versa.

Conclusions
In the present study, Box Behnken design was used to optimize the process parameters of the acid hydrolysis of waste cotton ber to obtain maximum TCI in the given parameter range. The design proposed quadratic model as the best model and the optimum condition to obtain MCC with maximum TCI from waste cotton ber is acid concentration-4.3 N HCl, solid: liquid ratio-1g:16.7ml, temperature-90 o C and time-76 mins. FESEM studies aided to scrutinize the rupture of cotton bers while the colour identi cation test helped to con rm qualitatively that the cellulose particles obtained after hydrolysis were microcrystalline in nature. XPS studies helped to understand the purity and effect of hydrolysis at the surface level of MTCI. XRD analysis con rmed the crystalline make-up of cellulose was not affected in the given conditions of acid hydrolysis process whereas ATR-FTIR analysis revealed breakage of interchain hydrogen bonds and rearrangement of intra-chain hydrogen bonds to form microcrystals. On the whole, the involvement of H-bond and covalent bond was improved after the acid hydrolysis reaction proposed by the BBD and the synergism between these bonds helped to attain maximum TCI in MCC particles. Structure of (a) WCF before hydrolysis and (b) MTCI after hydrolysis FTIR spectra of WCF and MTCI in 3000-2800 cm-1 region