Stress Transfer Analyses In Cellulose Nanofiber/Montmorillonite Nanocomposites With X-Ray Diffraction And Investigation On Effects of Chemical Interaction Between Cellulose Nanofiber And Montmorillonite


 Cellulose nanofiber is one of the promising materials for its eco-friendliness as well as high mechanical performance and high functionalities. Nanocomposites with cellulose nanofiber matrixes and inorganic nanofillers also possess more excellent mechanical properties by the reinforcement effects of the nanofillers. The mechanical reinforcement effects depend in a large part on the interfacial interaction between the nanofillers and the cellulose matrixes and the dispersion of the nanofiller in the nanocomposites. The quantitative evaluation of the reinforcement effects is insufficient, which is desired for the material design of industrial use of the cellulose composites. In this study, we used nanocomposites of cellulose nanofibers and montmorillonite with various surface properties. Their mechanical properties were investigated through tensile tests and the stress transfer to the nanofillers in nanocomposites with various combinations of cellulose nanofibers and nanofillers was analyzed through the X-ray diffraction method. The strong correlation between Young’s modulus and stress transfer coefficients was revealed. In particular, the composites of TEMPO-oxide cellulose nanofiber and ion-exchanged montmorillonite possessed not only the highest Young’s modulus but also the largest stress transfer coefficients. The large mechanical reinforcement effect of the loaded montmorillonite filler was observed and was attributed to the electrostatic interaction of the interface between the cellulose matrix and the montmorillonite filler.


Introduction
Cellulose nanocomposites have received much attention as environmental-friendly "green" materials. textures without the destruction of their wood structure and precursors of polymer resins were penetrated into porous structure of the textures. The incorporation with wood textures and polymers was accessible to high mechanical performance of the composite materials. Moreover, it has been investigated that the nano llers in various cellulose composites provided the further mechanical reinforcements to nanocellulose-based materials with high modulus. The reinforcement effects directly depended on the modulus and the shape of the loaded llers. Isogai, Saito and co-workers have been reported on oxidized CNF/montmorillonite (MMT) composites and investigated on their high Young's modulus and tensile strength and their excellent gas barrier properties. (Wu et al. 2012) MMT with two dimensional structure is originated from cray mineral, and is accepted as one of typical reinforcement llers in nanocomposites. (Usuki et al. 1993;Liu et al. 2011;Wu et al. 2012) Because single nano ber of cellulose itself inherently possesses high modulus even without any reinforcement llers, (Sakurada et al. 1962;Nishino et al. 1995;Cheng and Wang 2008;Zhai et al. 2018) the mechanical properties of CNF/MMT composites were reinforced more largely by loading MMT llers. Therefore, CNF/MMT composites have large advantages from the perspective of excellent mechanical performance as well as environmental friendly "all-green" composite materials.
It is well-known that the reinforcement effect in nanocomposites was provided from not only their high Young's modulus and the dispersibility of the loaded nano llers but also the interfacial interaction between matrix polymers and inorganic llers. (Komarneni 1992;Hussain et al. 2006;Camargo et al. 2009;Miculescu et al. 2016) The reinforcement effect and the stress concentration are evaluated as stress transfer from matrixes to nanno llers. The effective stress transfer is attributed to the strong interfacial interaction between llers and matrixes, which leads to larger reinforcement effects. Therefore, the stronger interaction at interface between llers and matrixes is signi cant. For the investigation on stress transfer in composites, X-ray diffraction measurements Xu et al. 1992;Nishino et al. 1998Nishino et al. , 2000Nishino et al. , 2001Nishino et al. , 2006

Materials
MMT powder "Kunipia-G" was supplied from Kunimine Industries Co., Ltd.. Tetraethylammonium hydroxide solution (Nacalai Tesque, Inc.) was used for exchanging interlayer cation of MMT. The re ned pulp was prepared from Kenaf bast bers (Toyota Boshoku Co., Ltd.), which were accepted as llers of composites of structural materials in motor vehicle industry, through degreasing, Wise method, and alkaline treatment, according to the literatures (Nobuta et al. 2016). All the chemical products were purchased from chemical suppliers, and were used without any further puri cation.

Ion exchange of MMT
In exchange of MMT was performed based on the method reported by Jinnai and co-workers (Janeba et al. 1998). MMT powder 0.6 g was added to 120 mL of distilled water and the aqueous dispersion was stirred for 1 day to obtain 0.5 wt% of MMT dispersion. Aside from this, 0.8 g of 10 wt%

Measurement of thermal decomposition behavior
Thermal gravimetric analyses (TGA) of FCN and TOCN sheets and their composites were performed with Thermo plus EVO2 TG8121 (Rigaku) under nitrogen gas. The ow rate of nitrogen gas was 200 mL/min, the heating rate was 10°C/min, and the scanning of temperature was from room temperature to 500°C.
To exclude water contained in the measurement samples, the samples was dried over under vacuum at 40°C for longer than 6 h and was remained at 120°C for 30 min before measurements. The temperature where the 5 wt% of the sample weight was lost compared to those at 150°C was de ned as the thermal decomposition temperature (T d5 ).
Measurements of X-ray diffraction X-ray diffraction pro les of MMT powders, CNF sheets and the composites were measured with RINT2100 (Rigaku) in the θ/2θ method. The X-ray beam was generated with 40 kV and 20 mA and the wavelength was 1.5418 Å (CuKα). The scanning rate was 2°/min, the step was 0.02°, and the scanning range of 2θ was from 3° to 40°.

Measurements of two-dimentional X-ray diffraction
Two-dimentional X-ray diffraction pro les of the CNF sheets and the composites were measured with RINT2000 (Rigaku) in the transmittance method. The X-ray beam was generated with 40 kV and 20 mA and the wavelength was 1.5418 Å (CuKα). The camera distance was 56.3 mm. The irradiation time of Xray beam was 20 min. The X-ray beam was irradiated into the sample lms from the direction perpendicular or parallel to the surface of the samples and the diffraction was detected with imaging plates. From the measurements with these geometries, the in-plane or through-plane orientation of the crystallites were evaluated.

Measurements of mechanical properties
Tensile tests were performed with autograph AG-X plus (Shimadzu). The samples were trimmed into 30 mm ⋅ 5 mm rectangles, the initial length in the tensile test was 10 mm and the tensile speed was 0.5 mm/min. The densities of the lms were evaluated in the oating method using calcium chloride aqueous solution at 30°C. The cross sectional area was calculated from the densities and the weight and the length of samples. In order to obtain reliable data, we performed tensile tests of more than ve specimens for every sample, and averaged their obtained mechanical properties.
Stress transfer analyses of composites with X-ray diffraction X-ray diffraction measurements for stress transfer analyses was performed with RINT2000 (Rigaku) in the symmetric transmittance method using CuKα beam (1.5418 Å). The X-ray beam was generated with 40 kV and 20 mA and the wavelength 1.5418 Å (CuKα). The scanning rate is 0.1°/min, the step was 0.006°, and the scanning range of 2θ was from 61° to 63°. The samples were trimmed into 30 mm ⋅ 5 mm rectangles. The samples were set in a stretching device and the stress was measured with a load cell LUB-B (Kyowa Electronic Instruments). The tensile apparatus put on the gonio stage of the X-ray diffractometer. Under applying tensile stress to the samples, the X-ray diffraction peaks were evaluated. In the case of the CNF/MMT composites, we focused on the shift of diffraction peaks originated from (060) plane of MMT. The strain of MMT crystallite ε c was calculated from the lattice distance d 0 of (060) plane before loading and the difference of the lattice distance ∆d from the diffraction peaks before and under loading, according to the below Eq. (1). In addition, from the Eq. (2), the stress σ c transferred to the MMT nano ller and the stress concentration coe cient σ c /σ 0 were estimated.
where E (060) is the elastic modulus of MMT crystal for the (060) plane of MMT, and σ 0 is the stress applied to the whole sample. The E (060) of MMT was 400 GPa. (Manevitch and Rutledge 2004) Results And Discussion The FCN and TOCN as matrixes in the composites were prepared from the kenaf pulps through mechanical and chemical nano brillization with micro uidics method and TEMPO oxidation, respectively. The characterization of the obtained CNF are shown in Figure S1 in the Supporting Information. For the investigation of diameters of the CNF, AFM observation was performed. From the height of the CNF in the topological images, it was revealed that the diameters of nano bers of FCN and TOCN were 19 ± 6.8 nm and 2.7 ± 0.5 nm. FCN was a bundle of single cellulose bers, while TOCN stood as a single elementary bril. In the FTIR spectra of FCN and TOCN in Figure S1e in the Supporting Information, the absorption band of carboxyl groups at 6-position of TEMPO-oxidized cellulose were appeared at 1610 cm − 1 , whereas no band of carboxyl groups in the non-brillated re ned pulp and FCN was observed. For the quantitative estimation of the oxidation of hydroxyl groups of their CNF, conductometric titration was performed, as shown in in Figure S1f in the Supporting Information. As this result, TOCN possessed carboxyl groups in the concentration of 1.66 mol/g-cellulose. In contrast, the concentration of carboxyl groups in FCN was only 0.14 mol/g-cellulose. This means that the oxidation of hydroxyl groups at 6-position of cellulose was progressed and 81% of hydroxyl groups at 6-position at the surface of the Isogai and Saito's cellulose crystallite model (Daicho et al. 2018(Daicho et al. , 2020) was converted to carboxyl groups. In X-ray diffraction pro les of both the CNF, the diffraction peaks of type I β of cellulose crystal were observed at 14.9° and 22.5°. This suggested that, even after mechanical and chemical nano brillization, crystalline structure of cellulose in CNF kept unchanged.
Ion exchange of sodium cations to tetraethyl ammonium ions in MMT was performed with the addition of tetraethyl ammonium hydroxide into MMT aqueous dispersion. Figure 1a shows X-ray diffraction pro les of MMT and ion-exchanged MMT (MMT-NEt 4 + ). In the pro les, the 001 re ection of MMT at 7.2°w ere completely shifted to 6.3° after the ion exchange (Stathi et al. 2007), which indicates the interlayer distance was increased by intercalation of tetraethyl ammonium ion larger than sodium ion. The ion exchange converted all the MMT to MMT-NEt 4 + . In the preparation of CNF/MMT composites, the mixture method of CNF and MMT in aqueous dispersion was employed because both the matrixes of CNF and llers of MMT were well dispersed into water. After slow ltration, the composite lms were obtained. For the investigation of thermal stabilities of the composites and the amounts of MMT llers in the composites, their thermal gravimetric behaviors were measured. The decomposition temperature T d5 was increased by loading the MMT llers in the both composites. The FCN/MMT was decomposed at higher temperature than FCN/MMT-NEt 4 + , while T d5 of TOCN/MMT-NEt 4 + was higher than that of TOCN/MMT. It is suspected that the interaction of MMT llers with CNF matrixes would decide their decomposition behavior because the molecular motions of CNF matrixes was restricted by the MMT ller.
From the residual weight at 500°C, the weight fractions of MMT llers in the composites were estimated as around 3 wt%. This value approximately coincided to the mixture ratio in preparation recipe, which suggests MMT in CNF/MMT mixed dispersion passed through the lter under lteration. In addition, from the X-ray diffraction pro les of the prepared composites as shown in Fig. 1b, both the diffraction peaks originated from CNF and MMT were observed in all the composites. The structure of cellulose I β crystallites and layered structure of MMT llers were both remained even after preparation of their composites. Moreover, the 100 re ection of MMT in both FCN and TOCN composites were broadened and weakened relative to those of MMT-NEt 4 + . This result reveals that MMT in composites was exfoliated or intercalated by CNF and the distance between MMT layers was increased. In contrast, the ion-exchanged MMT-NEt 4 + even in composites remained rigid layered structure of MMT-NEt 4 + itself.
For the evaluation of the structural orientation of the composites, two-dimensional X-ray diffraction were measured with two geometries. Figure 2 showed 2D X-ray diffraction images of the CNF/MMT composites when the X-ray beam was irradiated from the "through" direction perpendicular to the surface, and "edge" direction parallel to the surface. In the "through" images of all the composites, Debye-Sherrer rings of 110/1-10 and 200 re ections of cellulose were observed, while, in the edge diffraction images, not only the diffractions of cellulose but also 001 re ection of MMT llers were detected as arcs on the meridional direction. This suggested that, in all the composites, the cellulose nano bers as well as llers were oriented. The ber axis and the MMT plane were oriented to the direction parallel to the surface because these composites were prepared by slow ltration from CNF/MMT aqueous dispersion. Compared with FCN composites, it is revealed that the in-plane orientation of TOCN in TOCN/MMT and TOCN/MMT-N + Et 4 composites were increased judging from sharper arc of re ection of cellulose in their edge diffraction images. Figure 3 shows the schematic structure model of CNF matrixes and MMT llers in the composites observed from in-plane and edge directions. The in-plane structures of all the prepared composites were random and isotropic. In the view of edge side of the composites with FCN matrix, the bers of FCN were mainly oriented but some possessed disordered structure, whereas the TOCN-based composites possessed the highly oriented structure. This enhanced in-plane orientation of TOCN should be attributed to the smaller diameters and higher aspect ratios of TOCN. In addition, from the results of X-ray diffraction pro les of the composites, the 2D plates of MMT in both the CNFs composites were more exfoliated and isolated rather than those of MMT-N + Et 4 and the MMT planes in the all the composites were laid parallel to the surface of the composite lms.
For the investigation on the mechanical properties, tensile tests of the CNF sheets and composites were performed. The strain-stress curves and parameters of the mechanical properties of the CNF/MMT composites were shown in Fig. 4 and Table 2, respectively. All the TOCN lms and composites possessed higher Young's modulus than FCN ones, which were attributed to the lower diameters and higher aspect ratios of TOCN.  To reveal the mechanical reinforcement effects of MMT llers in CNF composites, stress transfer to the MMT llers in their composites were evaluated in the X-ray diffraction method under loading tensile stress. When the mechanical stress was loaded to the CNF/MMT composite, the diffraction peak of (060) plane was shifted to the lower angles, as shown in Figure S3 in the Supporting Information. This means the increase of the lattice spacing with loading tensile stress. Because the MMT llers were dispersed in the CNF composites in the parallel to the composite surface and the (060) plane received tensile stress in perpendicular direction, the lattice strains were directly calculated from the shifts of the diffraction peek and the elastic modulus of the (060) plane of MMTs. In addition, the comparison between the estimated stress of MMT llers and the mechanically applied stress of the composites provided the stress transfer effects from the composite bulks to the MMT llers and mechanical reinforcement effects of the MMT llers with the dependence of interfacial interaction.
The stress on the MMT llers in the four composites were evaluated under applying different tensile stress. Figure 5 shows the plots of the stress on the MMT llers for the various applied stress to the whole composite lms. The inclination of the obtained approximate straight lines to the plots provided stress transfer coe cients σ c /σ 0 from composites to llers. The stress-transfer coe cients σ c /σ 0 of all the CNF/MMT composites are also shown in