The Control of Crystal Structure and Mechanical Property in Regenerated Cellulose Film by Coagulation Conditions

Cellulose lms regenerated from aqueous alkali-urea solution possess different properties depending on coagulation conditions. However, the correlation between coagulant species and properties of regenerated cellulose (RC) lms has not been claried yet. In this study, RC lms were prepared from cellulose nanober (CNF) and microcrystalline cellulose (MCC) under several coagulation conditions. Cellulose dissolved in aqueous LiOH/urea solution was regenerated using various solvents at ambient temperature to investigate the effects of their polarity on the properties of RC lm. The crystal structure, mechanical properties, and surface morphology of prepared RC lms were analyzed using X-ray diffraction (XRD), tensile tester, and atomic probe microscopy (AFM), respectively. It is revealed that the preferential orientation of (110) and (020) crystal planes, which are formed by intra- and inter-hydrogen bonding in cellulose crystal regions, changed depending on coagulant species. Furthermore, we found out that tensile strength, elongation at break, and crystal structure properties of RC lm strongly correlate to the dielectric constant of solvents used for coagulation process. This work, therefore, would be able to provide an indicator to control the properties of RC lm depending on its application and to develop the detailed research on controlling the crystal structure of cellulose.


Introduction
The development of a replacement for petroleum-based plastics is becoming more necessary in the face of global warming and plastic pollution. In particular, single-use plastics (SUPs), such as plastic bags, packaging materials, and straws, are substantial sources of plastic pollution (Schnurr et al. 2018). In recent years, many types of bio-based and biodegradable plastic have been developed. Regenerated cellulose (RC) lm, which is composed solely of cellulose, have attracted attention as an environmentally friendly lm, especially in food and medicine packaging application due to its biodegradability, gas barrier property, and high mechanical strength (Hyden 1929 Zavrel et al. 2009). Among these solvent systems, aqueous alkali/urea solution has received considerable attention owing to its simplicity, low toxicity, and rapid dissolution. This solvent system can rapidly dissolve cellulose at low temperature without producing any hazardous byproducts and signi cant degradation of cellulose (Zhou and Zhang 2000). It has also been reported that the solubility of cellulose in LiOH/urea aqueous solution is higher than that of NaOH/urea solution (Cai and Zhang 2005). In alkali/urea aqueous solution, 15 N and 23 Na NMR measurements clari ed, adding urea can effectively improve the stability of the alkali-cellulose complex owing to the strong interaction between alkali hydroxides and urea, and low temperature can accelerate the breakdown of intermolecular hydrogen bonding among cellulose, thus prevent the agglomeration of alkali-ureacellulose inclusion complexes (Jiang et al. 2014).
In general, coagulation/regeneration of cellulose from alkali/urea solution is carried out with acidic aqueous solution or polar organic solvents, and it has been discovered that the properties of RC materials, such as crystal structure, mechanical properties, and surface wettability can be varied widely with types of coagulant (Bingbing et al. 2008;Yamane et al. 1996; G. Yang et al. 2007). For instance, regenerated cellulose bers coagulated with acid aqueous solution, which possesses smaller selfdiffusion coe cients, showed more uniform and ner nano bril structure, and a superior mechanical strength (Zhu et al. 2018). Besides aqueous solution, the effects of organic coagulants, such as methanol, ethanol, and acetone, on the properties of regenerated cellulose have also been discussed.
Isobe et al. have reported the pore size distribution of regenerated cellulose hydrogel is not affected by the type of coagulant, while regenerated cellulose gel coagulated using aqueous solution and organic solvents show different surface morphology and wettability because of the changes in crystal structure (Isobe et al. 2011). Up to now, the effects of coagulation conditions have been widely discussed. Nevertheless, there is currently no literature that clari es the correlation between the properties of RC lms and the types of coagulant. A central question to reveal this correlation would be how the crystal structure of cellulose changes depending on coagulant species during the regeneration process, and how the changes in crystal structure have an effect on mechanical strength and exibility of RC lms. It is well known that RC has a crystal structure of cellulose composed of three crystal planes, which are (1-10), (110), and (020) planes (Yamane et al. 2006). In addition, Yamane et al. revealed that (110) and (020) formed by the arrangement of the glucan chains via inter-and intra-hydrogen bonding, while the planar glucan chains were arranged through weak hydrophobic interactions in (1-10) plane. Based on our knowledge, we presumed the polarity of coagulant would have an effect on the interactions among cellulose during the regeneration process, and so the composition of crystal planes, leading to the changes in mechanical properties of RC lms. composition of these three crystal planes, and so mechanical strength of RC lms. Therefore, the purpose of the present contribution is to investigate the crystal structure, especially the composition of crystal planes and mechanical properties of RC lms prepared with various coagulation conditions, and to clarify and visualize the correlation between properties of RC lms and types of coagulant.
In this study, RC lms were prepared from cellulose dissolved into LiOH/urea aqueous solution and coagulated using acid aqueous solution and organic solvents, which have different polarity represented by dielectric constant. The crystal structure of prepared RC lms was analyzed using X-ray diffraction (XRD) and we correlated the changes in composition of crystal planes with the polarity of coagulants successfully. In addition, it is clearly revealed that the occupancy of (110) and (020) crystal planes which are formed by strong hydrogen bonding is one of the factors that change the mechanical properties of RC lms. This study will expand the detailed research on tuning the crystal structure of cellulose, and provide an indicator to control the properties of RC lm depending on its application.

Materials
The water slurry of cellulose nano ber (CNF) with the concentration of 5.0 wt% was provided by Sugino Machine Limited (Toyama, Japan), and stored in a refrigerator before use. Microcrystalline cellulose (MCC) with a particle size of 20 µm and lithium hydroxide (LiOH, ≧98%) were purchased from Sigma-Aldrich. Urea and all solvents were supplied from Wako Pure Chemical Corporation. All reagents and solvents were laboratory grade and were used as received.

Preparation of regenerated cellulose lms
In this study, regenerated cellulose (RC) lms were prepared following a slightly modi ed procedure of LiOH/urea aqueous system reported elsewhere (Q. Yang et al. 2011). The solvent mixture comprising LiOH, urea, and H 2 O with a weight ratio of 4.6:15:80.4, containing the desired amount of cellulose, was cooled at -14 ºC for an hour. Concentration of CNF was xed at 3 wt%. The cooled mixture was then vigorously stirred for 30 min. Cooling and stirring processes were repeated at least twice in order to make cellulose completely dissolve in LiOH/urea aqueous solution. The resulting transparent and viscous solution was degassed by centrifugation at 5 ºC and 8500 rpm for 3 min, spread on a glass plate using a spin coater, and immersed in a coagulation bath at ambient temperature for regeneration. RC lms composed of MCC were also prepared from 9 wt% of MCC/LiOH/urea aqueous mixture following the same procedure as CNF.

Characterization
Chemical structures of pristine CNF, MCC, and RC lms were characterized using FT-IR/ATR spectroscopy Nicolet iS5 (Thermo Fisher Scienti c K.K., Japan). Transmittance spectra was collected in a wavenumber range of 400-400 cm − 1 at a resolution of 4 cm − 1 . X-ray diffraction (XRD) was performed on a diffractometer MiniFex series (Rigaku Co., Japan) and operated at 40 kV and 15 mA in re ection mode with Cu Kα radiation (λ = 0.154 nm), a scanning speed of 10 degree/min along with 2θ range from 3 to 70˚. The re nement of XRD spectra including the breakdown between crystalline and amorphous phases was carried out by using the MAUD Rietveld program (Lutterotti 2010). In this study, crystal information les (.cif) for cellulose structure (Langan et al. 2001) was used for performing the Rietveld method. The preferential formation and orientation of crystal plane formed by intra-and inter-hydrogen bonding among regenerated cellulose crystal region was expressed as a peak-are ratio calculated following equation: Where A (110) , A (020) , and A (1)(2)(3)(4)(5)(6)(7)(8)(9)(10) are the peak area of crystalline phases corresponding to (110), (020), and (1-10) planes respectively. Mechanical properties of the RC lms were evaluated on a tensilecompressive tester IMC-18E0 (Imoto Machinery Co., Ltd, Japan) at a tensile speed of 5 mm/min. For each RC lm, at least three specimens were tested. Atomic probe microscopy (AFM) AFM5300E (Hitachi High-Tech Co., Japan) equipped with silicon cantilever was utilized to characterize the surface morphology and roughness of regenerated cellulose lms. The images were recorded in dynamic force mode (DFM).

Results And Discussion
Preparation process and conditions of regenerated cellulose lm In this study, regenerated cellulose (RC) lms were prepared from microcrystalline cellulose (MCC) and cellulose nano ber (CNF) using various types of solvent as coagulant to study the effects of the polarity of coagulants on the crystal structure and mechanical properties of RC lm. The preparation process and conditions including regeneration bath solutions, coagulation time, and dielectric constant of each solvent were shown in Fig. 1  The challenge in preparation of RC lm was to prevent the shrinkage and bending of lm during the drying process. In this study, to overcome this issue, hydrogel-like regenerated cellulose was dried gently under tense state at ambient condition. Figure 2 shows images of RC lm prepared from a 3 wt% CNF  Chemical structure of regenerated cellulose lms FT-IR/ATR spectra of RC lms prepared from CNF and MCC through regeneration with methanol, dried mixture composed of cellulose, LiOH, and urea, and the neat MCC and CNF are shown in Fig. 3. The broad peak at 3000-3700 cm − 1 was assigned to the O-H stretching vibrations arising from hydrogen bonding in molecular chains of cellulose. In addition, the spectrum of neat cellulose showed characteristic peaks at around 2900, and 1000 cm − 1 corresponding to the stretching vibration of C-H and C-O-C, respectively. In the spectrum of dried mixture containing cellulose, LiOH, and urea, three sharp peaks, which are attributed to N-H stretching, bending vibration of N-H, and C-N stretching of urea were con rmed at around 3400, 1600, and 1460 cm − 1 , respectively. All peaks suggesting the presence of urea disappeared and the same spectrum as neat cellulose was obtained after regeneration. These results provide evidence that cellulose was regenerated without derivatization after the removal of urea and LiOH through the coagulation and washing process. A comparison of the native cellulose and RC lm, the transmittance peak corresponding to the stretching vibration of O-H became more gradual after dissolution and coagulation. This means that the regeneration process has an effect on the intra-and inter-hydrogen bonding in cellulose. RC lms coagulated other solvents also showed the same FT-IR spectra as that regenerated with methanol, implying that the functional groups of cellulose did not change depending on the species of coagulants (Fig. S1).
Crystal structure of regenerated cellulose The crystal structure of RC lms prepared with different types of coagulant was investigated by means of XRD analysis. Figure 4 depicts the XRD pro les and RC lms showed the structure of cellulose , having peaks at around 2θ = 12.5˚, 20˚, and 22.5 ˚ attributing to the (1-10), (110), and (020) crystal plane respectively, while pristine cellulose is classi ed as cellulose I α or I β .
Based on resulting XRD pro les, the relative intensities showed differences for each RC lm. The intensity of peak corresponding to (1-10) plane, which is formed by the arrangement of the planar glucan chains via hydrophobic interactions, became stronger than other peaks, as RC lms were coagulated with H 2 O, ethylene glycol, which possess high polarity and the low polar solvents such as 1-butanol, acetone, and nhexane. In contrast, RC lms regenerated using ethanol and methanol as coagulant showed that the intensity of the peaks corresponding to (110) and (020) planes, which are formed by inter-and intrahydrogen bonding, were higher. These results may indicate that the preferred orientation of the crystals was changed in the process of regeneration of cellulose depending on the types of coagulant. The similar behavior was observed for regenerated cellulose gels (Isobe et al. 2011). To demonstrate that preferred orientation is the cause of intensity variation, the RC lm sample was cut into small particles with a mill, sieved to obtain a ne powder, and presented to the diffractometer. The XRD pro les of RC lm and powdered RC lm are shown in Figure S2. The result showed that the diffraction pattern approaches the ideal patterns of cellulose described in previous work (French 2014).
For further investigation, XRD spectra of each RC lm was analyzed by using Rietveld method and the peak areas composed of (1-10) and overlapping peaks of (110) and (020) were calculated and the changes in preferred orientation of the crystal planes were evaluated as a peak-area ratio. Figure 5 depicts the result of re nement performed on XRD pattern of RC lm prepared from MCC through coagulation using ethanol as an example. The calculated peak-area ratio of the crystal plane formed by intra-and inter-hydrogen bonding in regenerated cellulose coagulated with polar organic solvents, such as ethanol and methanol were much higher than those coagulated with other solvents (Table 2).

Mechanical properties of regenerated cellulose lms
Given the changes in crystal structure of RC lms depending on the species of coagulant, the obtained RC lms were expected to be different in mechanical properties. That is why a tensile test was conducted to evaluate the mechanical properties of RC lm prepared with various coagulants. Figure 6 depicts the stress-strain curves for each RC lm prepared from CNF and MCC, and the average of tensile strength, elongation at break, and young's modulus were summarized at Table S1. The mechanical properties of RC lms changed signi cantly depending on the solvent used as coagulants. In particular, the mechanical properties of MCC-based RC lms were strongly affected by the types of coagulant. For instance, the RC lm, prepared from 9 wt% MCC in LiOH/urea/H 2 O through regeneration with ethanol, showed excellent tensile strength of 81.31 MPa and elongation at break of 4.15 %. On the other hand, MCC-based RC lm coagulated with nonpolar solvents, such as 1-butanol and n-hexane was very brittle and fragile. As compared with CNF, the elongation at break of RC lms prepared from MCC resulted in lower as a whole, due to the difference of aspect ratio between brillated cellulose and crystal. However, the tensile strength of MCC-based RC lms regenerated with polar organic solvents was as high as that prepared from CNF. These results prove that MCC was packed and formed a ne structure through crystallization during dissolution and coagulation process.

Correlation between polarity of coagulants and properties of regenerated cellulose
To control the properties of RC lm by preparation condition, clari cation and visualization of correlation must be crucial. In this study, polarity of coagulants, which is able to be represented in dielectric constant, was adopted as a parameter to investigate the correlation with the mechanical properties and crystal structure of RC lms. Figure 7 depicts the peak-area ratio of (110) and (020) lattice planes formed by intra-and inter-hydrogen bonding to (1-10) plane, tensile strength, and elongation at break as a function of dielectric constant of solvents used for coagulation. Both RC lms prepared from CNF and MCC show superior mechanical properties, as methanol which has dielectric constant of 32.6 is used for coagulation. Moreover, the peak-area ratio of (110) and (020) crystal planes formed in RC lms regenerated with methanol is much higher than RC lms coagulated using other solvents. Therefore, there is an apparent correlation between the crystal structure varied in the process of regeneration, mechanical properties of RC lms, and dielectric constant of coagulants. These results clearly revealed the correlation that polarity of solvent has a strong effect on the interactions among the molecular chains of cellulose during the regeneration process. In addition, the correlation for RC lms prepared from CNF is less apparent compared to MCC. This may be because that brillated cellulose molecular intertangle physically during dissolution and coagulation process, due to its high aspect ratio and molecular chain length of CNF.

Surface morphology and roughness of regenerated cellulose lms
It is well known that the property of coagulants affects not only mechanical properties but also surface morphology of regenerated cellulose. In this study, AFM observation was conducted to evaluate the surface morphology and roughness of RC lms prepared from 3 wt% CNF solution in LiOH/urea/H 2 O through coagulation with mixed aqueous solution comprising 5 wt% H 2 SO 4 and 5 wt% Na 2 SO 4 , and methanol. Both of RC lms show nanoporous structure, however, there is an obvious difference between the RC lms coagulated using aqueous acid solution and organic solvent, which is the former have an agglomerated structure of cellulose ber (Fig. 8a and c), while the latter shows brous structure ( Fig. 8b  and d). The arithmetical mean height of RC lm coagulated using 5 wt% H 2 SO 4 and 5 wt% Na 2 SO 4 aqueous solution and methanol was calculated based on AFM images, and resulted in 8.05 and 5.77 nm respectively. This indicates that polar organic solvent penetrates slowly into the cellulose solution to remove LiOH and urea, and then cellulose molecular chains self-align su ciently. On the other hand, aqueous acid solution diffuses into cellulose solution rapidly, and destroys the LiOH-urea complex immediately. Therefore, cellulose chains cannot form the perfect brillated structure and agglomerate each other. These differences of morphology between RC lms prepared through coagulation by acid aqueous solution and methanol may support the result that RC lm prepared using methanol possess higher tensile strength and elongation than acidic aqueous solution.

Conclusions
The correlation between the dielectric constant of solvents used for regeneration and the crystal structure as well as mechanical properties of regenerated cellulose lms was revealed successfully. It is found that the types of coagulant effect on the occupancy of crystal planes formed by intra-and inter-hydrogen bonding, which is described as (110) and (020) lattice planes. The experimental results demonstrated that RC lms prepared from 3 wt% CNF and 9 wt% MCC solution through regeneration using high-polarity organic solvents, such as methanol and ethanol showed higher occupancy of hydrogen bonding based crystal, tensile strength, and elongation compared to aqueous solution and nonpolar solvents. The ndings will be applicable as a parameter to control the properties of regenerated cellulose lm depending on its application.

Declarations
Acknowledgments Authors would like to thank Prof. Teruhisa Ohno from Kyushu Institute of Technology for technical support in sample analysis.
Funding This research received no external funding.
Con icts of interest The authors declare that they have no con icts of interest.
Availability of data and material All data and materials support the claims herein and comply with eld standards.
Code availability Not applicable. Ethical approval Ethical approval is not applicable for this article.
Compliance with ethical standards