Hybrid composite of Nafion with surface-modified electrospun polybenzoxazine (PBz) fibers via ozonation as fillers for proton conducting membranes of fuel cells

Nafion was investigated for its compatibility in the preparation of hybrid composites with electrospun Polybenzoxazine (PBz) surface-modified fibers by evaluating the effects on the surface and structure of the composite. A PBz fiber mat was first crosslinked by thermal treatment after electrospinning to enhance the mechanical integrity of the fibers prior to modification. Further surface modification via free radical ozonation was carried out by potentiating oxygen-based functional groups of hydroxyl radicals (–OH) onto fibers' exposed surfaces. The sequential modifications by crosslinking and ozone treatment were evaluated by analyzing surface properties using XPS, ATR-FTIR and water contact angle which determined the enhanced properties of the fibers that were beneficial to the target functionality. Electron spectroscopy confirmed that fibers' surfaces were changed with the new surface chemistry without altering the chemical structure of PBz. The presence of higher oxygen-based functional groups on fibers' surfaces based on the resulting atomic compositions was correlated with the change in surface wettability by becoming hydrophilic with contact angle ranging from 21.27° to 59.83° compared to hydrophobic pristine PBz fibers. This is due to electrophilic aromatic substitution with hydroxyl groups present on the surfaces of the fibers endowed by ozonation. The resulting surface-modified fiber mat was used for the preparation of composites by varying two process parameters, the amount of Nafion dispersion and its homogenization and curing time, which was evaluated for compatibility and interaction as fillers to form hybrid composites. The analyses of SEM images revealed the effects of shorter homogenization and curing time on composites with rougher and wrinkled surfaces shown on the final hybrid composite's structure while decreasing the amount of Nafion at the same homogenization time but longer curing time showed its influence on improvement of compatibility and surface morphology.


Introduction
Polymer electrolyte membrane fuel cell (PEMFC) is one of the most promising type of fuel cells, but currently still need further enhancement and innovation to reduce cost, improve durability and optimize performance. 1 PEMFCs are currently operated at relatively low temperature (50-80 C) which is causing limitation in terms of commercial viability. 2 With these technical challenges, it has been proven that low temperature operating conditions for PEMFCs have several disadvantages which are the main reasons for not commercializing it yet on a larger scale. All of these problems can be overcome by operating it at relatively high temperature (>80 C). 3 This entails further innovative approaches on a unit cell-level to overcome the technological issues on cost and performance of the membrane component. 4,5 Material selection, development and improvement are the important aspects of technological advancement for improvement of the fuel cell membrane's performance and reduction in cost. 6 Currently, the proton conducting membrane used in PEMFC is the peruorosulfonic acid (PFSA) membrane or commercially known as Naon, which is the current standard for electrolyte of PEMFC up to the present time. 2,5,7 It is considered as the standard polymer membrane for fuel cell due to its thermal and chemical stability as well as high ionic conductivity. [7][8][9] These notable features of Naon were the results of its structure composed of tetrauoroethylene (TFE) backbone and per-uorovinly ether terminated with sulfonate groups. 5,8 The structure produced a hydrophobic phase from the TFE and a hydrophilic phase of sulfonic acid group which acts as a reservoir for water during transport of protons. 3 The water promotes the dissociation of protons from the sulfonic acid groups and provides highly mobile hydrated protons. 3 Thus, it is essential for Naon to be fully hydrated for good proton conductivity. 7,8 Although Naon is still the commercially used membrane, the problems remain unresolved to make it appropriate for long term operation under more severe conditions without losing its conductivity and stability that will reduce system costs. 10 Increasing the temperature (>80 C) at low relative humidity will affects its performance due to anisotropic membrane swelling which results to irreversible conductivity decay. 3,11 At this condition, Naon has weaker proton conductivity performance due to dehydration of the initial ionic domains. 8 This results to lower power output as well as reduced mechanical strength. 12 There were several investigations conducted on improving Naon by adding functional additives, but in most cases the proton conductivity is lower than pristine Naon membrane. 13 In addition, research efforts in the past were focused on the modication of Naon to address its weaknesses at high temperature and low humidity. These studies were conducted to improve its water management by incorporating hygroscopic components and high conductivity llers such as silica and inorganic materials. 14 Recent development in composite membrane involves incorporating nanobers produced from electrospinning method. 9,13 The concept of ber-reinforced polymeric composite is very simple which combines the ber and the polymer material with properties limited to those of the two components but gained a wide range of applications in aerospace, marine, civil engineering, energy, medicine, electronics and other emerging scientic domains. 15 Incorporation of brous structures of nanostructured materials into a polymer matrix is now widely investigated as potential strategy for the development of proton-conducting membrane. Electrospun nanobers of a proton-conducting polymer embedded in an inert polymer can provide pathways for proton conduction while the inert polymer will serve as the mechanical support. 9 This made composting a viable approach for developing membrane to possibly overcome the limitations of per-uorinated and non-uorinated hydrocarbon polymers. Composite has the main advantage of availability of diverse materials that can be used in various forms through different fabrication methods. 7 It has been established that combining dissimilar organic and inorganic properties within a single material to form a composite produces superior properties compared to their pure components. 7,16 In this study, electrospun bers of polybenzoxazines (PBz), a high-performance thermoset polymer, 17 were modied for its surface properties via free radical ozonation treatment and were utilized as brous llers for the preparation of hybrid composite with Naon. Electrospun bers used in the preparation of composite as reinforcing materials have proven their benets from involvement of more than one component due its various physical, structural, and chemical properties resulting to new materials. 18 The ber mat used was fabricated by electrospinning process which is known to have benecial effects on morphology and surface properties. 18 Naon composite with polymer bers could possibly reduce the humidity-induced stress but also improve the water transport under low or dry condition. 19 Through this theoretical assumption, the modication via ozone treatment was explored in this work to investigate the effect of oxygen-based functional groups of hydroxyl radical (-OH) potentiated onto bers' exposed surfaces. This composite preparation with surface-modied non-woven bers is a potential innovation in nding solution to water transport and hydration of Naon. But the compatibility of PBz surface-modied bers in the preparation of composite with Naon was the preliminary consideration for hybridization, which to the best of our knowledge, was the rst time such an approach has been used for crosslinked and ozonated PBz bers. Thus, this study looked into the possible benets of employing simple preparation method for hybrid composite by only varying signicant process parameters. The relationships between the amount, homogeneity and curing time of Naon prior to combining with bers in hybrid composite preparation were examined as potential strategy for the development of proton-conducting membrane. Through this study, the development of hybrid composite with Naon involving two important aspects, surface modication of bers as llers and blending the polymer-bers, was evaluated for compatibility by analyzing the surface structure and interaction in forming the composite.

Materials
Polybenzoxazines (PBz) was prepared in the lab as reported in the work of Lin et al. 20 Dimethylsulfoxide (DMSO) (ACS grade, Echo, 99.9%), and tetrahydrofuran (THF) (inhibitor free, high purity, Tedia, 99.8%) were used as received. Naon D2020 (Coplolymer Resin, 20-22%, Dupont) were also used as received. Electrospun PBz (ES-PBz) ber mats were prepared based on the procedure described in the work of Parreño et al. 21

Surface modication of electrospun bers by ozonation
Functional groups of hydroxyls (-OH) were incorporated on the samples' surfaces by free radical ozonation of electrospun ber mat using Ozonizer (Three Oxygen Ent. Co. Ltd Taipei, Taiwan) at low temperature. Sample of ber mat (approx. 3 cm Â 3 cm) was immersed completely in 400 mL of deionized (DI) water inside the 500 mL glass reactor vessel. Then, the reactor vessel was placed in a basin with water and ice to improve ozone utilization at low temperature. Aerwards, the Ozonizer was switch on and set to the desired operating conditions with the oxygen (O 2 ) connecting tube placed inside the reactor vessel to supply the oxygen from the Ozonizer to the reactor vessel containing the sample. The oxygen ow rate was adjusted and maintained at around 8 L min À1 to have enough amount of O 2 bubbling inside the reactor for effective treatment. Ozone treatment of the sample was carried out for 30 min. Aer completion, the sample inside the reactor was purged with argon for another 30 min to remove residuals. Then, the sample was set aside and dried at room temperature in a Petri dish.

Characterization of surface-modied electrospun bers
Attenuated total reectance with Fourier transform infrared spectroscopy (ATR-FTIR) (PerkinElmer Spectrometer, MA, USA) was used to investigate the structural composition of the electrospun PBz ber sample before and aer surface modication by ozone treatment. A portion of the ber mat sample was directly analyzed in the ATR accessory without any sample preparation. All measurements were performed at the range of 4000-400 cm À1 with a resolution of 4 cm À1 . The changes in surface chemical composition were analyzed by X-ray photoelectron spectroscopy (XPS) (VG Microtech MT-500 ESCA). Resolution of sub peaks was performed using the least-squares peak analysis soware, XPS PEAK 95 version 3.0. The surface wettability of surface-modied and pristine ES-PBz ber mats were measured by water contact angle meter (First Ten Angstroms (FTA) Model: FTA 1000 B) with water drops of about 5 mL at ambient conditions. The contact angle data were obtained from the average of three replicates using ve measurements of mat samples.

Preparation of hybrid composite membrane
The bers as llers for the hybrid composite were prepared via electrospinning and thermally crosslinked as described in the work of Parreño et al., 21 prior to composite preparation. Physical blending method was employed in the fabrication of hybrid composite where the polymer was directly poured into the ber mat to disperse it into the whole surface of the ber mat, followed by drying to solidify the composite. The amount of Naon and the time for homogenization and curing were varied for the preparation of composites. The amounts of Naon dispersion solution used were 6 mL and 9 mL. Before physical blending, the polymer dispersion was homogenized and cured to remove bubbles and prevent interfacial and surface skin defects. The homogenization and curing time used were 90 min for ultrasonication with varied curing time of 1 h and 24 h prior to composite formation. Then, using different amount of Naon solution that was homogenized and cured, the polymer solution was poured into the ber mat sample and completely dispersed all throughout the sample. Aerwards, it was set aside for 1 h to let the sample absorb completely the Naon solution. The hybrid composite sample was dried in the vacuum oven (Deng Yng, DOV 300, Taipei, Taiwan) at 40 C, and 50 C, each for 1 h, then 80 C, 90 C and 105 C, each for 0.5 h. Then, the sample was cooled down. Aer cooling, the composite was immersed in DI water overnight to further remove residual solvents.

Interfacial compatibility of ozone-treated bers with Naon
The surface morphology and structure were analyzed using Scanning Electron Microscope (SEM) (Phenom XL, Thermo Fisher Scientic, AZ, USA) to examine the interfacial adhesion and compatibility of bers with Naon matrix in the nal hybrid composite. The bers as reinforcing llers were examined for its ability to blend with Naon in terms of surface appearance as indicated from the resulting skin and densication of the composite.

Results and discussion
3.1 Structural composition and surface chemistry of ozonetreated PBz bers Electrospun PBz bers were rst characterized for its structural composition to determine if there were changes in the chemical structure of the PBz aer undergoing the surface modication via ozonation. Attenuated total reectance (ATR) as sampling technique in conjunction with Fourier transform infrared (FTIR) spectroscopy was used to analyze the characteristic absorption peaks of the PBz ber sample to correlate spectral change with corresponding modications in surface structure and chemical composition of the samples. As shown in Fig. 1, the spectra of the pristine electrospun PBz bers showed the characteristic peaks associated with the benzoxazine structure at 1230-1235 cm À1 (asymmetric stretching of C-O-C), at 1330-1340 cm À1 (CH 2 wagging into the closed benzoxazine ring) and at 1495-1510 cm À1 (tri substituted benzene ring). These absorption peaks were also present in the spectra of samples that were modied by ozonation. Three samples (1, 2 and 3) of ozone-treated ES-PBz bers as shown in Fig. 1 were analyzed for their characteristic absorption peaks to better evaluate the spectral changes that occurred aer ozonation treatment. Based on these results, there were no additional peaks in the spectra of the three samples which indicated that no other interactions occurred when the PBz bers undergone ozone treatment.
To further examine the results of surface modication and conrm the endowed -OH functional groups, the spectra were analyzed for both the pristine PBz bers and ozone-treated PBz bers. The hydroxyl groups incorporated onto bers' surfaces have the potential to enhance the water retention capability of the ber mat when used as llers in the Naon composite. According to Ketpang et al., 22 the peaks at wavenumber of 3455 and 1625 cm À1 corresponds to the -OH stretching vibration and -HOH bending vibration, respectively. However, based on the result at wavenumber of 3455 cm À1 , the peak of -OH band was not present in the IR spectra of surface-modied samples which could be due to the limitation in spectral resolution of the IR. Although the broad band for the -OH stretching vibration as additional functional group in the bers' surfaces was not shown in the spectra, another way to examine the result was the -OH that formed as -COOH due to the electrophilic aromatic substitution of -OH in the C-H as adsorption sites or active sites of benzene ring in PBz. Based on a larger image of the IR spectra in Fig. 2, it revealed the -COOH that formed aer ozonation which produced a peak of -OH band at 1415 cm À1 as the hydroxyl bonded to the PBz's aromatic ring. This conrmed the presence of hydroxyl groups onto surfaces of the PBz bers.
The further conrmation on surface modication of ES-PBz bers in relation to -OH functional groups was examined by Xray photoelectron spectroscopy (XPS). As shown in the XPS survey scans of the samples in Fig. 3, the two main elemental components, carbon (C) and oxygen (O), were present in the wide spectra of the samples for both pristine PBz bers and ozone-treated sample. The survey scans in Fig. 3a and d showed that the O/C content ratio of ozone-treated PBz bers was higher than the pristine PBz bers which indicated that the electrospun bers were successfully oxidized. The C 1s peaks for both samples were enlarged into narrow peaks in Fig. 3b and e. These peaks were deconvoluted into three separate peaks indicating the carbon-containing group corresponding to the observed bond energies. The O 1s peaks for both samples were also deconvoluted into two separate peaks as shown in Fig. 3c and f to validate the higher O/C content ratio aer ozonation of the PBz bers. The scans of O 1s conrmed the same ndings in the O/C content ratio of the C 1s scans.
The percentage of each group was calculated according to the tting peak area. The calculated compositions were 80.5% O and 19.5% C for ozone-treated sample while only 16.9% O for pristine PBz bers with higher composition of 83.1% C as shown in Table 1. This result validated the surface modication via ozonation of PBz bers which was in agreement with the results of IR spectroscopy.

Surface wettability of ozone-treated PBz bers
The measurement of water contact angle is another way to assess the surface chemistry of the samples aer surface modication, if there have been changes in the wettability of the samples. As shown in Fig. 4, the water contact angles were reduced for the surface-modied samples which validated the surface chemistry results of XPS that conrmed the surface modication. Electrospun PBz bers are hydrophobic with water contact angle of 130.03 AE 0.27 (Fig. 4a) but aer the modication it became more hydrophilic with water contact angle ranging from 21.27 to 59.83 (Fig. 4b-f) for the ve measurements of one sample. These results indicated that the surface was modied by hydrophilization through electrophilic aromatic substitution in the benzene ring with -OH functional groups from ozone treatment.

Preparation of hybrid composite of PBz bers with Naon
The two factors investigated in the preparation of composite with bers were the amount of Naon dispersion solution and homogenization and curing time. These factors signicantly inuence the compatibility and interfacial adhesion based on the surface morphology and structure of the composite. Initially, the preparation of Naon solution involved homogenization by ultrasonication for 90 min. Then, set aside for 1 h to further dispersed and removed the bubbles before pouring onto the PBz ber mat. The amount of Naon dispersion solution used in the rst trial was 9 mL for the ber mat sample with approximate size of 3 cm Â 3 cm. Aer drying of the nal composite in the vacuum oven, it appeared to have uneven, wrinkled surface with bubbles present in the composite as shown in Fig. 5a. Then, based on the skin appearance, the amount of Naon solution was reduced to 6 mL for the same size of ber mat sample in the second trial. But homogenization time was retained at 90 min for ultrasonication and followed by 1 h curing time. There was a slight improvement observed in the surface appearance of the composite as compared to previous trial as shown in Fig. 5b. The presence of bubbles was still observed but with smoother and less wrinkled composite's skin. For the third attempt, the same amount of Naon solution was used with homogenization by ultrasonication of 90 min but set aside for longer curing time of 24 h before it was poured onto the ber mat sample. It was observed that the composite produced at this condition showed smoother and even surface without the presence of bubbles and lesser prominent wrinkles in the composite's skin as shown in Fig. 5c.
These results showed that reducing the amount of Naon dispersion solution from 9 mL to 6 mL was enough to completely embed the bers without affecting the preparation process. It had similar effects on surface morphology and structure with the presence of bubbles and wrinkles. However, allowing longer time for dispersion and curing of Naon solution aer sonication prior to use in the composite preparation, resulted to better surface appearance as exhibited in Fig. 5c.
Thus, in the nal preparation of composite of Naon with added bers as shown in Fig. 6, the adjusted processing

Compatibility and interfacial adhesion of PBz bers with Naon
Analyzis of the SEM images of the composites were conducted to determine the compatibility and interfacial adhesion between the ozone-treated bers as llers and Naon as polymer matrix in forming the hybrid composite. The surface morphology as well as the skin of the composite were considered as important indicators of well-prepared hybrid composite based on SEM images. The irregularities such as voids, aws, fractures, lines, and defects were also checked to show differences in the skin appearance. For the ozone-treated berreinforced composite as shown in Fig. 7b, the bers were indistinguishable from the structure which showed that Naon was able to completely penetrate between the bers and formed a continuous phase within the composite. This structure was comparable with the skin of pure Naon without ozone-treated bers in Fig. 7a. Based on the work of Sigwadi et al., 23 the result was the same with the SEM surface morphology of Naon 117 having a skin of dark in color and plain surface due to absence of additives or llers. According to Prykhodko et al., 24 interfacial     defects arise from the existence of the space charge layer in addition to the material that lls the polymer space. But it showed that the composite was fully densied with Naon which was an indication of good interfacial adhesion and compatibility between the resin matrix and bers without any effect from the space-charged layer or from the hydroxyl in the bers' surfaces. This also conrmed that Naon adhered rmly to the ber surface and no evident detachment was observed at the Naon/bers interfaces. Although the composite showed densication, slight defects were observed in the skin with the presence of numerous tiny shallow lines. Based on the magni-ed images, the shallow lines were not fractures or cracks in the composite's structure but were just visible on the composite's surface.

Conclusions
Naon is the widely-used commercial membrane and remains to be the standard polymer electrolyte membrane for PEMFC. Many research efforts have been conducted in the last ve years to further enhance its properties and address the current limitations of Naon for fuel cell application. New routes for enhancement were done such as chemical modication of functional properties as well as physical modication by additive enhancement. One of the most promising modication methods is by incorporating brous structures as llers to produce hybrid composite with Naon. This study utilized the second route by incorporating electrospun bers as llers with Naon by investigating the compatibility and interfacial adhesion. The main purpose was to provide enhancement in water transport and hydration of the composite through the effect of surface-modied properties of electrospun bers. The ozonation treatment of electrospun PBz bers resulted to potentiated oxygen-based functional groups of hydroxyl radical onto bers' exposed surfaces. The presence of higher oxygen-based functional groups in bers' surfaces was correlated to the change in surface wettability by becoming hydrophilic from a hydrophobic pristine PBz bers. In the preparation of hybrid composite, the processing parameters that showed signicant effects were lesser amount of Naon but longer curing time prior to composite formation. The production of nal hybrid composite proved the compatibility of llers and polymer matrix. The composite was fully densied with Naon which indicated good interfacial adhesion and compatibility between the resin matrix and bers without any effect from added hydroxyl in the ber's surfaces. This also conrmed that the Naon adhered rmly to the ber surface and no evident detachment was observed at the Naon/bers interfaces. Thus, ozonation can be carried out on crosslinked PBz bers by enhancing its target properties and then utilized as llers for Naon in a composite. This outcome could be used as basis to further study the succeeding route as a result of ozonation by looking into the next step for potential surface-initiated graing onto the bers. This could provide additional functional groups aside from hydroxyls such as sulfonates for polymer electrolyte membrane.