Introduction

Cellulose nanofibers (CNFs) are promising biobased reinforcing materials owing to their high strength [1], high modulus [2], low thermal expansion [3], and intrinsic nanoscale dimensions [4]. In particular, CNFs can be incorporated into transparent polymers without significantly affecting optical transparency because their width is smaller than the wavelength of visible light, and CNFs exhibit negligible absorption in the visible-light region [5]. The optical transparency of CNF-based composites can be maintained by selecting a polymer with a refractive index close to that of CNFs [6]. Thus, CNFs are highly suitable as reinforcing fillers for transparent thermoplastics such as polymethyl methacrylate (PMMA), which are widely used in optical components, including lenses, display panels, and protective glazing. Owing to the high interfacial free energy between the two components, achieving a homogeneous dispersion of CNFs in hydrophobic PMMA is challenging. Consequently, specialized techniques for CNF/PMMA compounding, such as CNF aerogel impregnation [6,7,8,9,10,11,12], the solvent casting of CNF/PMMA mixtures [13,14,15,16,17,18], and surface-initiated polymer grafting [19,20,21], have been reported. However, these methods are often labor intensive.

To mitigate these shortcomings, Pickering emulsions have recently attracted attention. CNFs can function as Pickering emulsifiers and adsorb strongly at oil/water interfaces, enabling the stabilization of emulsions [22]. In a previous study, we reported that by using the hydrophobic monomer styrene as the oil phase, CNF-stabilized emulsions can serve as templates for polymerization, enabling CNFs and polymers to be simultaneously compounded during the polymerization step [23]. Because this approach is conceptually aligned with industrial suspension/emulsion polymerization processes, it could improve scalability relative to conventional CNF/polymer composite fabrication methods.

In this study, we report a simple aqueous process for the preparation of transparent and homogeneous CNF/PMMA composites using a CNF-stabilized Pickering emulsion as a structural template (Fig. 1). CNFs modified with tetrabutylammonium (TBA) were used to stabilize methyl methacrylate (MMA)-in-water emulsions, after which aqueous polymerization and heat molding were used to obtain mm-thick transparent CNF/PMMA composite plates. The resulting composites exhibited uniform CNF dispersion, high visible-light transmittance, and improved thermomechanical performance, confirming the effectiveness of emulsion templating for the scalable fabrication of transparent fiber-reinforced polymer materials.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Schematic of the preparation process for the CNF/PMMA composite plates. a Preparation of Pickering emulsions by sonicating mixtures of CNFs modified with TBA (CNF-TBA), water, and MMA. TBA was ionically bound to the carboxy groups on the CNF surface (COO⁻ TBA⁺). b Heating of the emulsion to initiate the polymerization of MMA. c Recovery of the CNF/PMMA composite powder after HCl was added to the CNFTBA/PMMA composite dispersion. Notably, the addition of HCl removed TBA from the system and converted the surface carboxy groups of the CNFs back to the free acid form (COOH). d Formation of the CNF/PMMA composite plate by hot pressing the CNF/PMMA powder

Materials and methods

Materials

2,2,6,6-Tetramethylpiperidine-1-oxyl (TEMPO)-oxidized pulp with a carboxylate content of 1.78 mmol g−1 was obtained from DKS Co. Ltd. (Kyoto, Japan). The oxidation process is described in a previous study [24], and the carboxylate content was determined using conductivity titration [25]. MMA, 2,2′-azobis(2,4-dimethylvaleronitrile) (ADVN), and all other chemicals were obtained from Fujifilm Wako Pure Chemical Corporation (Osaka, Japan) and used as received without further purification.

Preparation of the CNF dispersion in water

The TEMPO-oxidized pulp was suspended in water (0.5 wt%), and its pH was adjusted to 2 by the addition of 1 M HCl. The surface-free carboxy groups (–COOH) of the TEMPO-oxidized pulp were subsequently neutralized by the addition of a 10% TBA–hydroxide solution and stirring for 4 h to produce –COO− TBA+ groups. The molar ratio of TBA-to-carboxy groups of the CNFs was adjusted to 1:1. The modified TEMPO-oxidized pulp was disintegrated using a high-pressure homogenizer (HJP-25005X, Sugino Machine, Japan) for three passes at a pressure of 150 MPa to obtain the TBA-modified CNF (CNF-TBA). For comparison, TEMPO-oxidized CNFs with the Na ion as the counterion (CNF-Na) were also prepared. The TEMPO-oxidized pulp was neutralized by the addition of 1 M NaOH, and the pH was adjusted to 7.5. The resulting pulp was disintegrated following the same procedure as the CNF-TBA.

Preparation of the CNF-stabilized MMA-in-water Pickering emulsion

To prepare the Pickering emulsion, MMA dissolved in 0.1 M ADVN (initiator) was added to the CNF-TBA aqueous dispersion. The weight ratio of CNF-to-MMA was adjusted to 2% or 5%, resulting in two types of composites with different CNF concentrations. The MMA-to-water ratio was adjusted to 10:90 by diluting the CNF dispersion with distilled water to prepare the 2% CNF sample. The water/CNF/MMA Pickering emulsion was created via ultrasonication for 10 min with cooling in a water bath at 4 °C.

Preparation of the CNF/PMMA composite

The CNF-stabilized MMA-in-water Pickering emulsion was placed in a water bath at 70 °C to polymerize MMA and subsequently stirred continuously overnight. After polymerization, 1 M HCl was added to the suspension until the CNFs and PMMA aggregated. The resulting CNF/PMMA composite was collected, washed via filtration using a PTFE membrane (0.1 µm pore size), and vacuum dried at 40 °C. To mold the composite powder into a plate, the CNF/PMMA composite powder was hot-pressed using a mold with a thickness of 0.5–2 mm. A constant pressure of 9 MPa was applied for 7.5 min at temperatures of 140, 160, and 180 °C. The sample was subsequently transferred to another press machine and placed between the press plates at room temperature without applying pressure for 5 min for cooling.

Optical microscopy

After polymerization, the emulsions and suspensions were observed using an optical microscope (BX51, Olympus Corp., Ltd., Japan). Images were captured using a digital camera (DP 27; Olympus Corp., Ltd., Japan), and the diameters of the droplets were measured using its software (DP2-SAL).

Scanning electron microscopy (SEM) observations

The surface microstructures of the composite powder and cross-sections of the composite plates were observed using SEM (S-4800, Hitachi, Japan) at an acceleration voltage of 1 kV. Before observation, the samples were coated with osmium (~5 nm thick) using a Neo Osmium Coater (Meiwafosis, Tokyo, Japan) at 5 mA for 10 s.

Measurement of the optical properties

To measure the transparency of the composite plate, light transmittance spectra were recorded using a spectrophotometer (V-670, JASCO Corp., Tokyo, Japan).

Measurement of the yellowness index

To evaluate the discoloration of the composite plates after hot pressing, the yellowness index was measured using a spectrophotometer (CR-400; Konica Minolta Japan, Inc., Tokyo, Japan).

Measurement of the thermomechanical properties

The thermomechanical properties of the composite plates were measured on samples at a molding temperature of 160 °C, and their dynamic viscoelasticity was measured using dynamic mechanical analysis (DMA). The measurements were conducted with a dynamic mechanical analyzer (DVA-200, ITK Co. Ltd., Osaka, Japan) in tensile mode at a frequency, heating rate, and applied strain of 10 Hz, 5 °C/min, and 0.05%, respectively. The samples were 5 mm wide, 40 mm long, and 1 mm thick. The measurement temperature ranged from room temperature to 180 °C unless the storage modulus was < 105 Pa.

The thermal expansion behavior of the composite plates was analyzed using a thermal mechanical analyzer (TMA-60; Shimadzu, Japan) in tensile mode with a preload force of 0.1 N/mm2. The temperature ranged from 30 to 130 °C at a heating rate of 5 °C/min, and the samples were 5 mm wide, 20 mm long, and 0.5 mm thick. The coefficient of thermal expansion (CTE) values were calculated at 30–80 °C.

Results and discussion

Preparation of the CNF/PMMA composite via the CNF-stabilized Pickering emulsion

In this study, CNFs modified with TBA, referred to as CNF-TBA, were used as emulsifiers. The incorporation of TBA into the CNFs was confirmed by Fourier transform infrared (FTIR) spectra (Fig. S1). The conversion ratio of TBA was considered 100% because the free carboxy group (COOH, ca. 1720 cm−1) completely disappeared and shifted to the carboxylate groups (COO-, ca. 1600 cm−1).

The CNF-TBA emulsifier successfully stabilized an MMA-in-water emulsion with a uniform and stable droplet dispersion (Fig. 2a, b). In contrast, when conventional CNFs were neutralized with Na ions (CNF-Na), the emulsion immediately collapsed, and phase separation occurred immediately after emulsification (Fig. S2). The difference in emulsification capacity can be explained by the surface free energy and attractive polar interactions between the CNFs and the monomer. In particular, polar interactions are attributed to the balance between the polar components of the surface free energy: the electron-acceptor (γ+) and electron-donor (γ−) components. In γ−-rich oils, emulsion formation failed if the γ+ concentration of the CNFs was not sufficiently high, because the interactions were insufficient and the CNFs detached from the oil surface [26]. The γ+ values for CNF-Na and CNF-TBA were reported to be 1.36 and 2.52 mJ/m2, respectively [26]. Because MMA is a γ−-rich oil, surface modification with TBA increased the attractive interactions with CNFs and MMA, enabling emulsification.

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Optical microscopy images of a, b CNF-stabilized Pickering emulsions and c, d their corresponding suspensions after MMA polymerization. In both emulsions, the volume ratio of the MMA-to-aqueous phase was fixed at 10:90, and the initial weight ratios of CNF-to-MMA were a, c 2 and b, d 5 wt%

Because surface modification with TBA improved the adsorption of CNFs at the MMA/water interface, CNF-TBA was used as the emulsifier for subsequent composite preparation. The diameters of the oil droplets in the emulsions prepared with CNF contents of 2 and 5 wt% relative to MMA were 2.8 ± 0.9 and 2.3 ± 0.7 µm, respectively. In both cases, the emulsions remained stable, without any visible coalescence. The tendency for smaller droplet sizes with 5 wt% CNF was attributed to the increased viscosity in the aqueous phase because of the higher CNF concentration, which likely increased the cavitation-induced shear forces during ultrasonication and hindered droplet coalescence after emulsification.

The emulsion was heated to initiate MMA polymerization. Optical microscopy revealed that μm-sized particles, corresponding to the size of the original emulsion droplets, were dispersed throughout the system after polymerization (Fig. 2c, d). However, the number of such particles decreased, suggesting the fragmentation of oil droplets during the polymerization process [23]. When polymerization did not occur without the addition of an initiator, no fragmentation was observed, whereas the coalescence of the monomer droplets due to heating was observed (Fig. S3).

After polymerization, the suspension was dried, and the CNF/PMMA composite powder was obtained. The average yields of the 2 and 5% CNF/PMMA powder composites were 94.6 ± 0.1 and 95.2 ± 0.8%, respectively, on the basis of the initial dry mass. Polymerization of MMA via a CNF-stabilized Pickering emulsion was successfully achieved. To better elucidate the morphology of the resulting CNF/PMMA composite powder, SEM was used (Fig. 3). Particles with diameters of several micrometers were present, and their surfaces were densely coated with CNFs; thus, the interfacial structures created in the emulsion were preserved even after polymerization (Fig. 3b, e). In addition to these microparticles, finer particles undetectable by optical microscopy were observed. These nanoparticles (~100 nm in diameter) were observed within the CNF network (Fig. 3c, f) and likely precipitated from the droplets into the aqueous phase during polymerization. A similar phenomenon, in which styrene was used as the oil phase in a CNF-stabilized emulsion, has been reported [23]. They reported that the formation of polystyrene particles with diameters of approximately 150 nm occurred predominantly after polymerization, whereas microsized particles were rarely observed. In this study, the selective synthesis of nanoparticles was not achieved, and both nano and microparticles coexisted in the composite powder.

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

SEM images of the a–c 2 and d–f 5 wt% CNF/PMMA powder composites. b, e Enlarged views of a microparticle and c, f nanoparticles

Preparation of the CNF/PMMA composite plates using heat-molding

CNF/PMMA composite plates (2 mm thick) were successfully prepared via heat molding of the CNF/PMMA composite powder (Fig. 4a) at temperatures of 140, 160, and 180 °C. At a CNF content of 2 wt%, the total light transmittance of the plate was >89% for all molding temperatures, whereas that of neat PMMA (refractive index = 1.492 [27]) was 92%, which is consistent with the theoretical value based on the Fresnel reflection (92.0%). Therefore, the transmittance loss owing to the addition of 2 wt% CNFs was ~3%. Because the same heat-pressing conditions were applied to all the plates, the surface smoothness was comparable between these samples. Thus, the transmittance loss in the composite plates was likely caused by the internal structure of the composite.

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

a Images of the 2 and 5 wt% CNF/PMMA composite plates heat-molded at 140, 160, and 180 °C. Optical properties of the composite plates: b total light transmittance, c haze at a wavelength of 600 nm, d density, and e yellowness index. All the samples were prepared with a thickness of 2 mm

Although light scattering at the CNF/PMMA interface is a potential cause, this effect was considered minimal because the TEMPO-oxidized CNFs had nanoscale widths and a refractive index of 1.545 ± 0.002 [28]. The difference in the refractive index between the CNFs and acrylic resin is approximately 0.1, which is small enough to ensure high optical transparency in the CNF/acrylic resin composites owing to the nanosize effect [29, 30]. Thus, the decrease in light transmittance was likely attributable to the presence of internal voids.

The measured densities of the 2 wt% CNF/PMMA composite plates molded at 160 and 180 °C were 1.15 ± 0.03 and 1.17 ± 0.01 g/cm³, respectively. These values were lower than those expected by 2.4% and 2.1%, respectively, when a true density of 1.62 g/cm³ for CNFs was assumed [31]. This finding suggests the possible presence of residual voids even after molding, and these voids likely contributed to the transmittance loss via light scattering.

When the CNF content was increased to 5 wt%, the CNF/PMMA composite exhibited lower total light transmittance and higher haze values (Figs. 4b, c and S4). This was likely attributed to the greater degree of CNF aggregation in the 5 wt% CNF/PMMA composites. As the width of CNF bundles increases owing to aggregation, light scattering becomes more distinct [6].

The density of the composite plates is shown in Fig. 4d. When the samples were pressed at 160 °C, the porosities of the 2 and 5 wt% CNF/PMMA composites were comparable at 2.4% and 2.1%, respectively. Therefore, light scattering within the composites was promoted mainly by CNF aggregation with increasing CNF concentration. The cross-sections of the composite plates are shown in Fig. 5. After hot pressing, the morphology of the microsized particles remained unchanged. Although the 2 wt% CNF/PMMA composite contained a greater quantity of microsized PMMA particles throughout the matrix, the optical transparency of the composite plate was greater than that of the 5 wt% CNF/PMMA composite. As previously discussed, this phenomenon was attributed to the well-dispersed CNF morphology and the similar refractive indices of the CNFs and PMMA. Although the total light transmittance decreased with the addition of CNFs, the maximum transmittance at 600 nm was 80%, which is high for a composite with a thickness of 2 mm. In previous reports, CNF/PMMA composites prepared via aerogel impregnation exhibited relatively low transparency. For example, the transmittance was ~75% at 600 nm for 2 mm thick plates with 1% CNFs [11] and 60–70% at 550 nm for 3 mm thick plates [10]. This difference may be attributed to the drying process prior to PMMA infiltration, which can result in CNF aggregation and increased light scattering. Solvent-casting methods have achieved high transparency (~90%) for films with thicknesses of 100–200 µm [14, 16, 17]. However, these techniques are typically limited to thin films owing to the high viscosity of CNF dispersions. Thus, our emulsion-based approach enables the fabrication of mm-scale transparent composites without affecting optical clarity.

Fig. 5
Fig. 5The alternative text for this image may have been generated using AI.
Full size image

Cross-sections of a, b 2 and c, d 5 wt% CNF/PMMA plates prepared through molding at 160 °C

A decrease in the number of voids within the composite plate further increases the transparency of the composite plate. The presence of voids was also observed as the space between the micro- and nanoparticles in some cross-sectional areas of the composite plates (Fig. S5). Therefore, the following mechanism is proposed for the formation of voids: The core of the microparticles was composed of a soft PMMA matrix, whereas the shell consisted of a stiff CNF network. By hot pressing, the PMMA core underwent thermal deformation, but the rigid CNF shell did not deform to the same extent. Consequently, these voids were likely generated. Identifying effective methods for decreasing microparticles is a key challenge for future research, because these microparticles cause internal voids and light scattering, reducing the transparency of composite plates.

The yellowness index increased with increasing molding temperature (Fig. 4e), likely because of thermal degradation of the CNFs during the heat-molding process. This tendency was more distinct at a CNF content of 5 wt%, suggesting that additional strategies, such as decreasing the molding temperature or the heating time, may be necessary to prevent discoloration. On the basis of the density, total light transmittance, and yellowness index results, the molding condition at 160 °C was the most suitable in the present composite. Therefore, in the following section, the thermomechanical properties of the composites molded at 160 °C were examined.

Thermomechanical properties

The thermal expansion of the CNF/PMMA composites (Fig. 6a) was suppressed below the glass transition region of neat PMMA (~100 °C) (Fig. 6b). However, no significant difference was observed between 2 and 5 wt% CNF loading in this temperature range.

Fig. 6
Fig. 6The alternative text for this image may have been generated using AI.
Full size image

a Temperature dependence of the thermal expansion curves. b CTE values of the CNF/PMMA composite plates prepared through molding at 160 °C

The DMA results of the CNF/PMMA composite plates are shown in Fig. 7. Even when the molding temperature was changed to 140–180 °C, the storage modulus curves of the composite plates almost completely overlapped (Fig. S6), indicating that the molding temperature had a minimal effect on the thermomechanical properties.

Fig. 7
Fig. 7The alternative text for this image may have been generated using AI.
Full size image

Temperature dependence of the storage modulus of the CNF/PMMA composite and neat PMMA plates prepared through molding at 160 °C

Compared with the neat PMMA, the CNF/PMMA composite plates exhibited higher storage moduli, both below and above the glass transition temperature of PMMA.

However, this effect was more distinct above the glass transition temperature. At 140 °C, compared with that of the neat PMMA, the storage moduli of the 2 and 5 wt% CNF/PMMA composites increased by 650 and 4500%, respectively. This increase was attributed primarily to the rigid network structure formed by the CNFs, which effectively hindered the mobility of the polymer chains. Furthermore, the composite with 5 wt% CNF exhibited a higher modulus, suggesting the formation of a denser and more effective percolated network within the PMMA matrix.

Conclusion

CNF/PMMA composites were successfully prepared via an aqueous polymerization process based on CNF-stabilized Pickering emulsions. CNF/PMMA composite plates (mm thick), which were fabricated via heat molding, exhibited high optical transparency, with a total light transmittance of >80% at 600 nm. The incorporation of CNFs significantly improved the thermal dimensional stability of the composites, as confirmed by their lower CTE than that of the neat PMMA. Furthermore, the storage modulus increased, particularly above the glass transition temperature of PMMA, suggesting the formation of a rigid CNF network within the matrix. Increasing the CNF content resulted in higher haze and yellowness indices; however, the thermomechanical properties improved with increasing CNF loading, suggesting the formation of a denser CNF network. These simultaneous improvements in thermomechanical performance and optical transparency highlight the potential of CNF/PMMA composites as sustainable and high-performance transparent structural materials for use in applications such as automotive components and architectural glazing.