Understanding the effects of carboxylated groups of functionalized graphene oxide on the curing behavior and intermolecular interactions of benzoxazine nanocomposites

Abstract

Novel benzoxazine (BOZ)/carboxylated graphene oxide (GO-COOH) composites were prepared via in situ intercalative polymerization. The curing behaviour, morphology and intermolecular interactions of GO-COOH based nanocomposites were investigated and compared with those of a graphene oxide (GO) blend system to clarify the influence of carboxylic groups. Compared to GO, GO-COOH with a large amount of carboxylic groups, relatively higher thermal stability, and exfoliated sheet morphology might be more easily dispersed and reacted in the BOZ matrix. The GO-COOH nanoplatelet based composites possessed a different polymerization path from that of the GO based system, implying that carboxylic groups not only provided catalytic effects but also participated in the grafting polymerization reactions between carboxyl groups of GO-COOH and phenolic hydroxyl groups of BOZ. A significant improvement of both the glass transition temperature (T_g) and crosslinking network density of the GO-COOH blend system further confirmed that covalent bonding occurred between filler and polymer chains, indicating that the GO-COOH nanoplatelets had a stronger influence on the thermal property improvement of the nanocomposites than that of the GO blend system. Surprisingly, a very low amount (1 wt%) of GO-COOH can affect the thermal properties of the composite remarkably, leading to a more than 30 °C increase of T_g in comparison with pure benzoxazine.

---

Introduction

Polybenzoxazines have a number of unique properties such as excellent mechanical properties, high char yield, near zero volumetric shrinkage/expansion upon polymerization, low water absorption, excellent resistance to chemicals and UV light, and high T_g even with rather low cross-link density.^1–7 However, pure benzoxazine-based polymers also suffer many disadvantages, such as high curing temperature (200 °C or higher), difficulty in processing and poor mechanical strength (brittleness). Generally, for performance enhancement, the preparation of novel monomers, copolymers, polymer alloys, inorganic filler and fiber reinforced composites have become increasingly attractive to overcome the shortcomings of the typical polybenzoxazines.^8–13 There is especially interest in the incorporation of nano-sized particles and fillers dispersed into polymeric matrices to improve thermomechanical properties.^14–17

Graphene oxide (GO), a layered atomic thin graphene sheet functionalized with several oxygen-containing functional groups (including epoxide, hydroxyl, carbonyl, and carboxylic groups), has attracted considerable attention due to its unprecedented physical and chemical properties (i.e. high surface area, high thermal conductivity, mechanical stiffness, and excellent electrical conductivity).^18–22 GO has been introduced into various polymers such as poly(vinyl alcohol), poly(methyl methacrylate), polystyrene, poly (ethylene glycol), polypyrrole, polypropylene, epoxy, and poly(arylene disulfide) for the fabrication of high performance nanocomposites.^23–30 Recently, GO reinforced benzoxazine nanocomposites have generated intense interest since we first reported an effective protocol to prepare graphene oxide/benzoxazine nanocomposites by in situ polymerization.^31–36

The basic understanding of the effects of the oxygen-containing functional groups of GO on the curing behavior of benzoxazine monomers, and covalent and noncovalent interactions between GO and benzoxazine is essential for its application. Noted that GO is reactive leading to deoxygenation and possible grafting reaction under a ring opening polymerization window of 150–220 °C for benzoxazine. Benzoxazine is also a reactive monomer with a cationic ring opening mechanism. The self-dissociation of the benzoxazine ring could produce free phenolic structures that may act as catalysts for further ring-opening curing reactions. Moreover, it is well known that both phenolic and carboxylic acids groups are effective benzoxazine polymerization initiators and/or catalysts. Therefore, the polymerization mechanism of benzoxazine is compounded by the addition of GO with phenolic and carboxyl functional groups. At present, there exist different opinions of the influence of the functional groups of GO based on the previous studies. Based on some references, the presence of GO can produce a clear decrease in the polymerization temperature and polymerization enthalpy of benzoxazine.^31,32 It is hypothesized that carboxyl groups of GO acted as weak organic acid which accelerated the ring opening procedure. However, Alhassan documented that the addition of GO to main chain benzoxazine polymer just caused premature ring opening of benzoxazine due to heat evolved from exothermic deoxygenation of GO.²¹ Furthermore, a strong hydrogen bonding interaction is observed, however, there is no evidence to prove the possible chemical reaction between benzoxazine and GO in most cases. More recently, the proposed chemical reactions focusing on the epoxy groups of GO have been investigated in the model blend system with high concentration of GO. The esterification and etherification reactions are occurred between star-like telechelic benzoxazine with GO via carboxyl–epoxy reaction and phenolic hydroxyl–epoxy reaction, respectively.³² However, the expected covalent interactions via other reactive hydroxyl and carboxyl groups of GO should be further confirmed. Therefore, it is necessary to elucidate the effects of functional groups of GO.

We are interested in the influence of carboxyl functional groups of GO. Therefore, we attempted to prepare the carboxylated graphene oxide (GO-COOH) in order to convert hydroxyl groups to carboxylic acid (–COOH) moieties of GO. It is expected that carboxylation of GO not only renders proposed catalytic effects but also offers reactive sites for further covalent and noncovalent interactions. The curing behaviour, morphology and intermolecular interactions of GO-COOH based nanocomposites were investigated and compared with that of GO blend system to clarify the influence of carboxylic groups. To verify the effect of GO-COOH on the polymerization of BOZ, the purified BOZ monomer was used in the experiments to delete the catalytic influence of phenolic groups of dimers and oligomers. The novelty of the current study is concentrated on the effects of carboxyl functional groups of functionalized graphene on the curing behavior and intermolecular interactions of benzoxazine composites.

Experimental

Preparation of benzoxazine (BOZ) and graphene oxide (GO)

Benzoxazine monomers (bis(3-phenyl-3,4-dihydro-2H-1,3-benzoxazinyl)isopropane, BOZ) were prepared based on Ishida's report and described in our previous work.^31,37 The details of synthesis and purification methods are available in ESI 1.† The purified product: white crystal, yield: 86%. Melting point: 109 °C (DSC). Purity: 99% (HPLC). ¹H-NMR (CDCl₃, ppm): 5.30 (N–CH₂–O), 4.55 (O–CH₂–Ar), 1.57 (–CH₃). FTIR (KBr, cm⁻¹): 1233, 1028 (C–O–C), 950 (benzoxazine ring).

Graphene oxide was synthesized via a modified Hummers method described in our previous work.³¹ The details of synthesis and purification methods are also available in ESI 2.† Noted that the size of natural graphite is 23 μm used in this study, which is smaller than that we reported before (48 μm).³¹ The product: black powder. FTIR (KBr, cm⁻¹): 1720 (COOH), 1400–1060 (C–O–C/C–OH), 1615 (C=C). XPS (eV): 284.8 (C–C), 286.5 (C–O), 288.4 (C=O).

Synthesis of carboxylated graphene oxide

Carboxylated graphene oxide (GO-COOH) was prepared according to the method reported by Park et al.²⁰ Briefly, to generate fully exfoliated GO sheets, 0.40 g of graphene oxide powder was dispersed into 200 mL of deionized water, followed by an ultrasonic treatment for 60 min to form a stable GO solution (yellow-brown color, 2 mg mL⁻¹). For carboxylation, 48 g NaOH (analytical grade, Shanghai Chemical Reagents Corp., China) and 40 g chloroacetic acid (Cl–CH₂–COOH, analytical grade, Shanghai Chemical Reagents Corp., China) were added to the GO solution. The mixed solution was stirred, and then treated by ultrasonication in water bath for 120 min to convert the –OH groups to –COOH via conjugation of acetic acid moieties. The resulting GO-COOH solution was neutralized. Finally, the resultant solid was recovered by centrifugation, and then purified by repeated rinsing and filtration. The obtained product was vacuum-dried and stored in a desiccator.

Preparations of carboxylated graphene oxide/benzoxazine, graphene oxide/benzoxazine and graphite/benzoxazine composites

The BOZ powder was mixed mechanically with the GO-COOH particles, and then milled completely at room temperature. For later comparisons, benzoxazine based composites with different contents of graphite and graphene oxide were also prepared in the same ball milling method to act as controlled samples. Besides, pure BOZ sample was subjected to the same process. The in situ intercalation polymerization processes of carboxylated graphene oxide/benzoxazine (GO-COOH/BOZ), graphene oxide/benzoxazine (GO/BOZ) and graphite/benzoxazine (G/BOZ) composites were as follows. The milled blends were casted in a rectangular mold and removed solvent and water in vacuum oven at 80 °C for 2 h, and then cured in ordinary oven at 150 °C for 6 h, 160 °C for 4 h, 170 °C for 2 h, 180 °C for 2 h, and 190 °C for 6 h, separately. Finally, the cured samples (length: 35 mm, width: 10 mm, thickness: ca. 1.5 mm) were obtained when the temperature was cooled down naturally. By changing the weight ratios (1 and 3 wt%) of graphite and functionalized graphene in the blends, three series of blends were prepared and coded as GO-COOH/BOZ-x, GO/BOZ-x, and G/BOZ-x, where the x represents the weight ratios of fillers. The pure sample of benzoxazine was cured in the same condition and coded as BOZ.

The chemical structures of monomer and polymer of BOZ, graphite, graphene oxide and carboxylated graphene oxide were summarized in Scheme 1.

Scheme 1 The chemical structures of BOZ monomer (a), polybenzoxazine (b), graphite (c) graphene oxide (d), and carboxylated graphene oxide (e).

Characterizations

Proton nuclear magnetic resonance (¹H-NMR). ¹H-NMR spectra were taken on a Bruker TD-65536 NMR (400 MHz). All samples were dissolved in deuterated chloroform and tetramethylsilane (TMS) was used as an internal standard.

High-performance liquid chromatography (HPLC). High-performance liquid chromatography (HPLC) analyses were conducted with a HPLC system consisting of a Shimadzu-CBM20A equipped with a VP-ODS column (5 μm, 250 mm × 4.6 mm, Shimpack, Japan) set at 30 °C. Tetrahydrofuran/water (1 : 1 v/v) was used as the eluant at a flow rate of 0.5 mL min⁻¹. The eluate was monitored at 254 nm.

X-ray photoelectron spectroscopy (XPS). X-ray Photoelectron Spectroscopy (XPS) analysis of the sample was conducted using a model XSAM800 instrument (KRATOS Product, Britain). An Al Kα target at 1486.6 eV and 15 mA × 12 kV was used in the experiment. The sample was detected under 2 × 10⁻⁷ Pa. The binding energies were calibrated against C1s peak at 284.8 eV. The curve fitting was carried out with a mixed Gaussian–Lorentzian function.

Field emission scanning electron microscopy (FESEM). The cured samples were fractured after immersing them in liquid nitrogen for the purpose of analyzing the internal morphology (transversal view) of the samples. The samples were gold-coated in a JUC-500 Magnetron Sputtering Device (JEOL, Tokyo, Japan), and a Field Emission Scanning Electron Microscope (INSPECT F, FEI, Netherlands) was used for the observation of the cross section of the samples.

Fourier transform infrared spectroscopy (FTIR). The Fourier Transform Infrared Spectroscopy was carried out on a spectrometer (Nicolet 6700, Thermo scientific, USA). Spectra in the wavenumber range of 4000–400 cm⁻¹ were collected over 40 scans with a resolution of 2 cm⁻¹. The test specimens were vacuum-dried and then prepared by the KBr-disk method.

Wide-angle X-ray diffraction (WXRD). Wide-angle X-ray diffraction (WXRD) was recorded on an X-ray diffraction (X'Pert pro MPD, Philips), by using Cu Kα radiation (λ = 15.405) at 40 kV and 30 mA with a scan rate of 4° min⁻¹. The diffraction angle ranged from 4 to 40°.

Differential scanning calorimetry (DSC). A Differential Scanning Calorimeter Q20 (TA, USA) was used to study the curing behaviors of the benzoxazine monomer and blends, and exothermal behaviors of GO-COOH and GO. The powder samples were analyzed directly, approximately 4 mg each time. Indium was used for temperature calibration and nitrogen as the flushing gas. Non-isothermal experiments of samples were carried out at 10 °C min⁻¹ from 30 up to 300 °C. Experiments were always performed below 300 °C to prevent any possible degradation reactions inside the chamber. Sample weights were taken after each non-isothermal test. The weight losses, if there were any, were negligible.

Dynamic thermal mechanical analysis (DMA). The thermo-mechanical properties of the cured samples were conducted using a dynamic mechanical analyzer (DMA/SDTA861^e, Mettler Toledo, Switzerland) with the shear mode. The samples (length: 8 mm, width: 7 mm, thickness: ca. 1.5 mm) were measured with a vibration frequency of 1 Hz, maximum force amplitude of 5 N, maximum displacement amplitude of 10 μm, and a heating rate of 5 °C min⁻¹ from 30 to 250 °C under air atmosphere.

Thermogravimetric analysis (TGA). The TGA was performed on the SDT Q600 instrument (TA, USA) thermogravimetric analyzer, and an evolved gas analysis (EGA) furnace. The samples of GO-COOH and GO were carried out under a nitrogen atmosphere from 25 to 800 °C at a heating rate of 10 °C min⁻¹.

Results and discussion

Structure and morphology of carboxylated graphene oxide

Fig. 1 showed the FTIR spectra results for graphite, GO and GO-COOH. In the FTIR spectrum of GO, the characteristic absorption peak of R–OH was in the region of 3000–3500 cm⁻¹, and other C–O functionalities such as –COOH (1720 cm⁻¹) and C–O–C/C–OH (1400–1060 cm⁻¹) were clearly visible. From the FTIR spectra, the spectrum of graphene oxide was in good agreement with previous work, which proved the successful oxidation of graphite to graphene oxide.^31,38,39

Fig. 1 FTIR spectra of graphite (a), GO (b), and GO-COOH (c).

After reaction of GO with chloroacetic acid, the spectrum of GO-COOH was shown in Fig. 1(c). Compared to the spectrum of GO, the characteristic vibration bands due to C–O–C (1052 cm⁻¹) decreased, meanwhile, the absorption peaks at 1349 cm⁻¹ corresponding to the characteristic of in-plane OH bending mode appeared. Besides, the absorption band of C=O groups (around 1620 cm⁻¹) became broader and stronger, and was overlapped with the absorption peaks at 1720 cm⁻¹ (–COOH). The phenomenon was attributing to the skeletal vibrations of oxidized graphitic domains, indicating further carboxylation of hydroxyl groups of GO.^18,20 Moreover, the OH absorption band (3000–3500 cm⁻¹) became stronger and shifted to lower wavenumber, which was also attributed to the presence of large amounts of carboxyl groups on the graphene surface.

The surface morphologies of graphite, GO and GO-COOH were observed by FESEM and shown in Fig. 2. The surface of pristine graphite was very flat and smooth, and each layer connected together closely due to the strong inter-layer van der Waals interactions. Like graphite, GO still had the obvious lamellar structure. It is noted that GO occupied an enlarged layer-to-layer distance, attributing to that GO was exfoliated during the oxidation process. In addition, the surface morphology of GO was not smooth, suggesting that the intercalating oxide functional groups were covalently bound to graphite. Compared to GO, the general structure of GO-COOH is greatly changed. GO-COOH showed a rough surface and appeared a large number of irregular wrinkles on the surface, which might resulted from numerous carboxyl functional groups introduced by the oxidation and sonification process. Furthermore, it is difficult to find the lamellar structure for GO-COOH. During the carboxylation of hydroxyl groups of GO process, GO was exfoliated and intercalated further, which should be emphasized in here.

Fig. 2 FESEM images of graphite (a), GO (b), and GO-COOH (c, d).

The inner structures of graphite, GO and GO-COOH were corroborated by WXRD measurements and shown in Fig. 3. The powder WXRD patterns determined the changes of interlayer distance of the samples graphite and GO. The diffraction peak in the WXRD pattern of graphite appeared to be 27°, corresponding to the layer-to-layer distance of 0.336 nm. Compared to graphite, the diffraction pattern of graphene oxide showed a peak at 11° corresponding to the layer-to-layer distance of 0.845 nm. It is significantly greater than that of graphite due to the introduction of oxygenated functional groups on the graphite sheet, which resulted in an increase in the d spacing.^31,39

Fig. 3 WXRD patterns for graphite (a), GO (b), and GO-COOH (c).

The WXRD patterns of GO-COOH exhibited a broad and dwarf peak at 15–35°, indicating that the original lamellar structure of graphite was delaminated into the disordered flakes of GO-COOH to some extent.^18,20 However, the interlayer distance of GO-COOH decreased to 0.360 nm which was close to graphite's. Meanwhile, the diffraction peak at 11° for GO were completely disappeared after chemical modification and ultrasonic treatment. The phenomenon could be explained as follows. The hydroxyl groups were transformed into carboxyl groups, which made the GO-COOH become more compact. In addition, in the preparation of GO-COOH, the effects of NaOH were not only alkalizing GO, but also leading a reduction to some parts of GO, which resulted in a recovery of the electronic conjugation to a certain degree.^40–42

The thermal stability of graphite, GO and GO-COOH were analyzed by TGA and shown in Fig. 4. Data related to the temperatures corresponding to 5 wt% (T_d5), 10 wt% (T_d10), 50 wt% (T_d50) and maximum degradation temperature (T_max) of the initial weight, as well as residues at 300 and 800 °C were summarized in Table 1.

Fig. 4 TGA thermogram of graphite, GO, and GO-COOH.

Table 1 Thermal properties of samples obtained from TGA

	Tmax/°C	Td5/°C	Td10/°C	Td50/°C	W300°C/%	Char yield/%
Graphite		—	—	—	100	98
GO	201	85	163	579	84	42
GO-COOH	368	170	222	740	58	44

---

In the figure, the graphite occupied the excellent thermal stability even when the temperature increased to 800 °C. However, the graphene oxide displayed two stages of decompositions. The first weight loss occurred below 200 °C, which was mainly due to the evaporation of the adsorbed moisture. The second weight loss was attributed to the thermal exfoliation, degradation of oxide functionality groups, and release of carbon dioxide (CO₂) and carbon monoxide (CO). The previous studies reported that the thermal decomposition of instable oxygen-containing functional groups occurred at a temperature of 200 °C. This investigation is in agreement with literature.^21,32 However, the results in Table 1 indicated that the maximum degradation temperature of neat GO-COOH was 368 °C, which was much higher than that for GO (201 °C). The enhanced thermal stability could be due to the relatively compact inner structure and partial recovery of electronic conjugation of GO-COOH as mentioned in the part of WXRD.^40–42 Therefore, GO-COOH had a relatively higher thermal stability than that of GO, resulting from further chemical modification. All of the above results demonstrated that the carboxylation of hydroxyl groups of GO was successful.

Morphology and chemical structure of carboxylated graphene oxide/benzoxazine

The dispersion of nanofillers in a polymer matrix is one of the key points to the successful development of graphene-based nanocomposites and to the improvement of their performance. In the case of nano-particles such as clay, multiwall carbon nanotubes and polyhedral oligomeric silsesquioxane (POSS), the functionalization of the nanoparticles is necessary to achieve the best dispersion into polybenzoxazine matrix by maximizing the interfacial adhesion between the polymeric matrix and the nanoparticles.²¹ Therefore, it is desirable to achieve a good dispersion performance of functionalized graphene in the polymer matrix. The cured BOZ, GO/BOZ-1 and GO-COOH/BOZ-1 samples and their morphologies observed by optical microscopy were shown in Fig. S1.† All these specimens were optically transparent with no visual evidence of phase separation, suggesting a good compatibility between functionalized graphene and BOZ. Therefore, the oxygen-containing groups on the surface or at the edge of functionalized graphene sheets played a critical role in improving the compatibility between functionalized graphene and the polymer matrix.

The FESEM photographs gave important information of the dispersion of graphite, GO, and GO-COOH in BOZ polymer matrix (Fig. 5). The cross-sections were observed with different magnification scales (×5000, ×10 000, and ×50 000). It was clear to observe the graphite with flake or bundle structures were dispersed in polybenzoxazine matrix. According to the enlarged layer distance of graphene oxide, surfaces of GO/BOZ nanocomposites were relatively smooth. However, random distribution of GO sheets can still be observed. Furthermore, there existed the homogeneous ripple morphology for the blends with GO-COOH added. The debris at the edge of the ripple was assumed to be the chemical modification of graphene oxide after close observation with high magnification (×50 000).

Fig. 5 SEM images of BOZ (A), G/BOZ-1 (B), GO/BOZ-1 (C), and GO-COOH/BOZ-1 (D) samples (magnification scales: left, ×5000; middle, ×10 000; and right, ×50 000).

According to the homogeneous morphology, the efficient adhesion was arisen for both GO/BOZ and GO-COOH/BOZ blends due to the hydrogen bonding and π–π stacking interactions between functionalized graphene and polybenzoxazine. The homogeneous morphologies of both GO and GO-COOH blends were also ascribed to the possible chemical grafting between the phenolic hydroxyl groups of benzoxazine prepolymer and carboxyl groups of GO and GO-COOH, which enhanced the interfacial interaction with matrix.³² Moreover, it was attributed to the in situ polymerization protocol, in which it was easy for functionalized graphene sheets to be exfoliated and dispersed during the polymerization process of the intercalated BOZ monomers, and hence led to the formation of nanocomposites. It is noted that relatively homogeneous and smooth morphology was formed for GO-COOH/BOZ blend, resulting from the single or few layer structure and wrinkled morphology of carboxylated graphene oxide. Based on WXRD and FESEM observations of GO-COOH, it was convenient for the single or a few layer of GO-COOH to be dispersed homogeneously in the polymer matrix, which should be emphasized in here.

WXRD is also an effective method to evaluate the exfoliation and dispersion of GO sheets in a polymer matrix. With the advancement of the polymerization, compatible filler disperses into the polymer matrix, affording intercalated or exfoliated structure.³³ As mentioned in Fig. 3, comparing with pristine graphite, the interlayer spacing of graphene oxide was increased from 0.336 to 0.845 nm, indicating the weakening of interlayer van der Waals forces. In Fig. 6(A), WXRD diffractogram of cured GO/BOZ-1 blend showed a typical diffraction pattern of an amorphous polymer with a broad peak around 18°, which was completely different from the sharp peak of GO centered at 11°. The results suggested that GO sheets were crumpled and dispersed in the matrix resulting in the intercalation structure, which was in accordance with the FESEM observation.

Fig. 6 (A) The WXRD patterns of GO and GO/BOZ-1 nanocomposite; (B) the WXRD patterns of GO-COOH and GO-COOH/BOZ-1 nanocomposite.

In Fig. 6(B), the WXRD pattern of GO-COOH exhibited a broad amorphous peak around 25°, indicating the complete carboxylation of hydroxyl groups of GO. This is the prerequisite to successfully exfoliating of GO-COOH and obtaining single-layer or small-layer functionalized GO during the in situ polymerization of BOZ. The cured GO-COOH/BOZ-1 blend showed a relatively wider and lower amorphous peak ranging from 14 to 38°. The almost completely disappearance of the crystalline peak of modified graphene oxide in the blend was an indicative of the disruption of the ordered platelets leading to the exfoliated functionalized graphene composites.³³ The exfoliated GO-COOH sheets were well dispersed into polybenzoxazine network due to the effective intermolecular interaction and intercalating polymerization, which was also in agreement with the FESEM observation.

Three series of the uncured and cured composites (G/BOZ, GO/BOZ and GO-COOH/BOZ) were characterized by Fourier transform infrared spectroscopy (FTIR). In the spectra of BOZ monomer and blends of BOZ monomer and carboxylated graphene oxide (Fig. 7(A)), the peaks centered at 1494 and 950 cm⁻¹ were characteristic of the tri-substituted benzene ring in the benzoxazine structure, corresponding to the in-plane C–C stretching and the out-plane C–H deformation of tri-substituted benzene ring, respectively. There was no clear difference between BOZ and GO-COOH/BOZ uncured composites, implying that GO-COOH should not have the capacity to initiate the ring-opening reactions of monomers at room temperature. Furthermore, the delineation of the FTIR spectra of BOZ/GO-COOH composites were almost similar to those of G/BOZ and GO/BOZ samples (Fig. S2†).

Fig. 7 (A) FTIR spectra of uncured benzoxazine (a), GO-COOH/BOZ-1 (b), and GO-COOH/BOZ-3 (c); (B) cured benzoxazine (a), GO-COOH/BOZ-1 (b), and GO-COOH/BOZ-3 (c).

Three series of composite polymers after curing reactions were also characterized by FTIR. The spectra of cured BOZ and blends of BOZ monomer and carboxylated graphene oxide were shown in Fig. 7(B). The ring-opening reaction can be monitored by the disappearance of the oxazine ring neighboring at 950 cm⁻¹. The integrated intensity of the band at 1494 cm⁻¹ decreased upon curing reaction. These two peaks almost disappeared in the thermally polymerized pure benzoxazine spectrum. Meanwhile, a new band for the tetrasubstituted aromatic ring appeared at 1487 cm⁻¹, representing in-plane carbon–carbon stretching of the tetrasubstituted benzene ring. An intense broad band can also be observed at 3500–3100 cm⁻¹ corresponding to the formation of the phenolic hydroxyl groups. The above results indicated the opening of the oxazine ring and the polymerization of the monomers into oligomers and polymers.^43–46 The similar tendencies were found for GO/BOZ and G/BOZ systems (Fig. S3†).

As for the GO-COOH/BOZ composites, the intermolecular interactions were different from those of the BOZ samples, for the various hydrogen bonding interactions between BOZ and GO-COOH. Three types of hydrogen bonding that could possibly occur in the blends of GO-COOH and polybenzoxazine: (1) intermolecular hydrogen bonding between two hydroxyl groups of GO-COOH and polybenzoxazine, (2) intermolecular hydrogen bonding between the carbonyl group of GO-COOH and hydroxyl group of polybenzoxazine, and (3) intermolecular hydrogen bonding between hydroxyl group of GO-COOH and nitrogen atom on the Mannich bridge of polybenzoxazine. In Fig. 7(B), FTIR result of pure benzoxazine sample showed a broad band (3426 cm⁻¹) of phenolic hydroxyl functional groups. As the amount of GO-COOH increased, the absorption peak of OH groups centered around 3426 cm⁻¹ became broad visibly, and the band maximum of the shoulder peak (around 3385 cm⁻¹) shifted to the lower frequencies. This implies that the original inter and intra-molecular hydrogen bonds in the BOZ networks were destroyed due to the interpenetrating and intercalation (entanglements) between BOZ and GO-COOH. The slightly difference was found in the GO/BOZ system. For GO/BOZ composites (Fig. S3†), the strong absorption peak around 3422 cm⁻¹ caused by the hydrogen bonding between BOZ and GO appeared, compared with that of pure benzoxazine (3426 cm⁻¹) and G/BOZ (3426 cm⁻¹, Fig. S3†) cured composites, suggesting that intermolecular hydrogen bonds were formed.

It was expected that the possible chemical grafting reaction would happen between the carboxyl groups of GO-COOH and the phenolic hydroxyl groups of BOZ prepolymer (Scheme 2). In Fig. 7(B), an obvious shoulder peak of blends at 1720 cm⁻¹ became intense as compared with that of the pure benzoxazine. Zhou et al. reported that esterification reaction occurred between hydroxyl groups of resole and carboxyl groups of GO, as confirmed by the absorption band at 1722 cm⁻¹.⁴⁷ More recently, Ishida observed that the relative increase in the band at 1728 cm⁻¹ was attributed to the esterification reaction between carboxyl groups of benzoxazine and epoxy groups of GO.³² Therefore, the absorption band at 1720 cm⁻¹ was assumed to the esterification reaction between phenolic –OH of BOZ and carboxylic groups of GO-COOH. Moreover, the delineation of the mentioned characteristic peak of GO-COOH/BOZ samples was almost similar to those of GO/BOZ composites (Fig. S3†), although the intensities for latter were a little lower as compared with those for the former. The obtained results further confirmed that ester formation resulting from carboxylic groups of functionalized graphene and hydroxyl groups of BOZ possibly occurred. It could be concluded that carboxylic acid groups of functionalized graphene were incorporated into the polymeric structure based on FTIR observation.

Scheme 2 The proposed grafting reaction between carboxyl functional groups of GO-COOH and phenolic –OH of BOZ.

The above results gave the evidence that there are effective hydrogen bonding and π–π stacking interfacial interactions between BOZ and GO-COOH for the GO-COOH/BOZ composites, leading to the shifting of bandwidths and enhancement of the hydrogen bonding bands. The possible grafting reaction is owing to that the carboxyl functional groups on the chemically modified graphene oxide can be reactive with the phenolic –OH of BOZ chains. There is no clear spectra difference between GO and GO-COOH blends, implying that similar intermolecular interaction between nanofiller and polymer matrix. However, changes in the frequency and intensity of the characteristic absorption bands might be correlated to the degree of structural alteration. A large amount of carboxylic groups and their relative high thermal stability of GO-COOH increase the probability of functionalized graphene to react with phenolic –OH of BOZ. Moreover, GO-COOH with single or few layers of exfoliated sheet morphology might be more easily dispersed and intercalated in the BOZ matrix based on FESEM and WXRD observations, which is also benefit for the intermolecular interaction between nanofiller and polymer matrix. Therefore, carboxyl functionalization of GO is considered to play a critical role in the chemical structure and morphology for BOZ blend system. Furthermore, thermal polymerization of benzoxazine containing carboxylic acid groups has been reported to proceed through an autocatalytic mechanism in which these carboxylic acids are assumed to act as catalysts at the beginning of heating that promote the benzoxazine ring-opening and accelerate the polymerization process. The influence of carboxylic acid groups of functionalized graphene on the curing behavior of BOZ should be discussed in the following section.

Curing behavior of carboxylated graphene oxide/benzoxazine composites

DSC is an attractive technique because it can provide detailed knowledge of the cure mechanism and the preferred cure temperature during the formation of three dimensional networks in materials. It is known that oxazine ring opening is initiated by the presence of acidic media. Weak carboxylic acids are known to catalyze certain polymerization.^9,37 In this study, DSC has been employed to elucidate the effect of carboxylic acid groups of functionalized graphene on the curing behavior of BOZ. The key cure process parameters, such as melting temperature, cure temperature, and reaction rate has also been observed in non-isothermal DSC experiments.^9,37

The typical DSC curing curve as a function of temperature was obtained from 30 to 300 °C in the heating run, where the BOZ monomer was converted to polymer networks. Fig. 8 showed the DSC thermograms of BOZ monomer and the blends of G/BOZ, GO/BOZ and GO-COOH/BOZ. The detailed data were summarized in Table 2.

Fig. 8 (A) The curing behaviors of BOZ (a), GO-COOH/BOZ-1 (b), and GO-COOH/BOZ-3 (c); and (B) the curing behaviors of BOZ (a), G/BOZ-1 (b), and GO/BOZ-1 (c) observed by DSC.

Table 2 Cure process parameters and glass transition temperatures of BOZ and blends

	BOZ	G/BOZ-1	GO/BOZ-1	GO-COOH/BOZ-1	GO-COOH/BOZ-3
a Temperature of the maximum tan δ peak height.
Tm (°C)	109	110	109	109	109
Texo (°C)	261	263	251	252	251
Tg (°C)^a	192	205	207	223	225

---

As shown in Fig. 8, BOZ showed an endothermic peak at 109 °C attached to melting and an exothermic peak at 261 °C due to thermal polymerization. As shown in Fig. 8(A) and (B), incorporation of graphite, graphene oxide, and carboxylated graphene oxide into benzoxazine monomer did not show regular change on melting temperature of the crystalline state of monomer. There was a single peak observed in the DSC curves for pure monomer and all blends, it was assumed that there was a single chemical process occurring.

With the addition of the graphite modifier, the polymerization of benzoxazine was postponed. In Fig. 8(B), the curing exothermic peaks shifted toward a relatively higher temperature for G/BOZ blend. Moreover, the addition of graphite led to an increase in the enthalpy of benzoxazine ring opening polymerization. The observation might be attributed to the flake graphite diluted the concentration of the benzoxazine monomer and delayed the polymerization reaction to a certain extent. The result could be also owing to that the graphite particles were not dispersed well in the BOZ matrix.

However, DSC traces of GO/BOZ and GO-COOH/BOZ samples indicated that the maximum exothermic peaks of the benzoxazine units occurred at comparatively lower temperatures than that of unfilled BOZ. Both GO/BOZ and GO-COOH/BOZ blends studied herein exhibited the similar effects more or less. It is noted that small amounts of GO and GO-COOH produced the clear decrease in the polymerization temperature. The decrease ca. 10 °C and 9 °C of the exothermic peak maximums of the polymerization were found by the inclusion of 1 wt% of GO and GO-COOH, respectively. It was concluded that the polymerization of the benzoxazine monomer was accelerated to some extent in the presence of GO and GO-COOH.

The measured results (Fig. 8(A) and (B)) from nonisothermal experiments also indicated that the addition of GO and GO-COOH led to the clear decrease in the enthalpies of benzoxazine ring opening polymerization. The drop in heat of polymerization was due to the fact that both GO and GO-COOH were highly reactive.²¹ As seen in Fig. 9, both GO and GO-COOH showed only one exothermic peak at 213 and 162 °C, respectively. The exothermic deoxygenation of both GO and GO-COOH appeared earlier before the ring opening polymerization of benzoxazine monomer. It is noted that the GO deoxygenation reaction is a highly exothermic reaction. Therefore, the heat released from exothermic deoxygenation of GO in effect led to the decrease in ring opening polymerization exotherm.^32,47 On the contrary, the GO-COOH deoxygenation reaction is a mild exothermic one that has little effect on the decrease in ring opening polymerization exotherm. The drop in heat of polymerization was mainly due to the possible catalytic effect caused by the carboxyl groups of functionalized graphene for GO-COOH/BOZ composites.

Fig. 9 DSC thermograms of GO and GO-COOH.

It is well known that the ring-opening reactions of BOZ generated free phenol hydroxyl groups that can actually accelerate further ring opening.⁴⁸ Moreover, the auto-catalyst mechanism is carried out only at high temperature, which is usually above 180 °C. The curves of both G/BOZ and GO/BOZ blends (Fig. 8(B)) showed the same shape as those for the purified monomer. It was evident that the curing reactions for both the pure monomer and blends of G/BOZ and GO/BOZ were auto-catalyzed, implying that the addition of the GO was just contributing to the ring-opening process but not polymerization. However, GO-COOH/BOZ blends showed one new exothermic peak in the temperature range from 165 to 230 °C (Fig. 8(A)). It is noted that the onset of the exothermic peak was lower than 180 °C. The observation suggested that polymerization pathway of BOZ/GO-COOH blends was different from the pure monomer and other blends. This is hypothesized to be due to the catalytic effect caused by the carboxyl groups of carboxylated graphene oxide. Based on FTIR, FESEM, WXRD and DSC results of composites, GO-COOH may contribute not only to the ring-opening process but also to polymerization.

In our recent report, the larger the amount of GO presented, the more pronounced was the effect (Fig. S4†). The decrease ca. 25 °C of the exothermic peak maximum of the polymerization was found for the blend with the addition of 3 wt% of GO, indicating that both hydroxyl and carboxyl groups of GO has significant catalytic activity on benzoxazine polymerization. In Fig. 8(A), the exothermic peak maximum of GO-COOH/BOZ-3 appeared at 251 °C, which was similar with that of GO-COOH/BOZ-1. Not only no clear decrease on the maximum of the cure exotherm but also a slight decrease in the enthalpy was observed by increasing the GO-COOH content. The phenomenon could be explained that carboxyl functional groups of GO-COOH not only served to catalyze the polymerization of BOZ, but also participated as reactants and therefore consumed by the reaction.⁴⁹ Based on FTIR, WXRD and FESEM results of composites, a large amount of carboxylic groups, relatively higher thermal stability, and exfoliated sheet morphology might increase the probability for GO-COOH to react with BOZ. Of course, the phenomenon maybe also related to the diffusion problem of increased content of GO-COOH dispersed in the high viscosity of the purified monomers.³⁵

As mentioned above, both GO and GO-COOH could contribute to the decrease of ring opening polymerization temperature and polymerization enthalpy, and readily participate in the grafting polymerization reactions which favored the further exfoliation of the platelets. Compared to GO, GO-COOH occupying a large amount of carboxylic groups, relatively higher thermal stability, and exfoliated sheet morphology might be more easily dispersed and reacted in the BOZ matrix. It is noted that the GO-COOH based composites possessed the different polymerization path from that of GO, implying that carboxylic groups not only rendered proposed catalytic effects but also participated in the grafting polymerization reactions between carboxyl groups of GO-COOH and phenolic hydroxyl groups of BOZ. Therefore, carboxyl functionalization of GO is considered to play a critical role in the polymerization reaction for BOZ blend system.

Thermal properties of carboxylated graphene oxide/benzoxazine composites

To study the influence of GO-COOH on the structure and thermal property of the BOZ blends, the dynamic thermo-mechanical properties of BOZ and their blends were obtained as a function of temperature. DMA investigation on crosslinking network, molecular motion and glass transition temperature provides a possible method to evaluate the intermolecular interaction, chemical structure and thermal properties for functionalized graphene filled BOZ. Fig. 10 showed temperature dependence of the loss factor (tan δ) results of BOZ, GO/BOZ and GO-COOH/BOZ samples.

Fig. 10 The dissipation factor (tan δ) of BOZ (a), GO/BOZ (b), GO-COOH/BOZ-1 (c) and GO-COOH/BOZ-3 (d) samples.

tan δ is the ratio of energy dissipated as heat to the maximum energy stored in the sample. The height and width of the α-relaxation peak can be used to analyze the trends in the crosslink density and molecular motion of crosslinking polymers. According to the relatively lower values of height of α-relaxation peaks (<1.20), all samples occupied the compact crosslinking microstructure. There existed a sharp peak for BOZ, indicating the homogeneity of single polymer networks. Compared to pure BOZ, the slight increasing width of the tan δ peak for GO/BOZ was indicative of heterogeneity and multi-molecular motions with the addition of GO. In addition, the slight increasing height of the tan δ peak for GO/BOZ suggested the decrease of the crosslinking density of polymer networks. The reason for the phenomenon could be explained that the evolving heat from GO deoxygenation might lead to premature ring opening of benzoxazine before crosslinking with adjacent benzoxazine rings.^21,35

It is noted that both GO-COOH/BOZ-1 and GO-COOH/BOZ-3 nanocomposites had relatively broader and lower α-relaxation peaks in comparison with BOZ and GO/BOZ. The presence of carboxyl-groups of the GO-COOH surfaces allowed increased covalent and noncovalent bonding at the interface of functionalized graphene and polybenzoxazine, resulting in further hindered relaxation mobility in polymer segment near the interface leading to broaden glass transition temperatures.⁵⁰ The multiple curing reactions observed in GO-COOH systems, that is, the reaction among benzoxazine monomers and the reaction between benzoxazine and GO-COOH were also believed to cause network heterogeneity in this system.^51,52 The results supported the conclusions of FTIR. Moreover, with the increase of GO-COOH quantity, the height of the α-relaxation peaks of GO-COOH/BOZ decreased accordingly. The results revealed that the crosslinking degree of the GO-COOH/BOZ increased when the amount of GO-COOH increased, suggesting that chemical grafting reaction in effect happened between GO-COOH and BOZ. The results are in accordance with DSC observations.

Generally, the maximum of the tan δ peak (α-relaxation peak) reflects the glass transition temperature of the polymers as well as their composites, and may be analyzed to provide information about the motion of molecules. The glass transition temperatures of cured samples were obtained from DMA thermograms and summarized in Table 2. The data showed that the glass transition temperature (T_g) of polybenzoxazine increased upon incorporation of GO-COOH or GO. The relatively higher T_gs of the nanocomposites were due to the improvement of the damping properties of the polymer matrix with the addition of functionalized graphene sheets. It was noted that the nanocomposites containing 1 and 3 wt% GO-COOH sheets had relatively higher T_gs (around 223 and 225 °C) than those of the pure BOZ (192 °C) and GO/BOZ (207 °C) samples, indicating that effective intermolecular interaction between GO-COOH and polymer chains occurred. It is assumed that hydrogen bonds and possible covalent bonds between functionalized graphene sheets and polymer chains limited the polymer segment motions to result in relatively higher T_gs.^31,32 It is worth noting that hydrogen bonds will be weakened at temperatures higher than 100 °C.¹⁶ Moreover, an extensive literature summary shows that noncovalent bonding is generally incapable of providing enough restriction by interactions between the matrix polymers and fillers for graphene or GO based polymer nanocomposites, resulting in no clear increase of T_g.^16,34 Therefore, the significant improvement of T_gs of the GO-COOH blend system might be due to the covalent bonds between filler and polymer chains because of the sufficient reactive functional groups on the surface of functionalized graphene sheet.

However, as the amount of GO-COOH sheets further increased from 1 to 3 wt%, the T_g was not increased significantly and the reason might be explained as follows. Normally, when the amount of nano-sized fillers increased more, the increased fillers tended to aggregate together, as a result, the interactions between fillers and the polymer matrix would become weak. In addition, the increased nano-sized fillers might also restrict the formation of covalent bonds of the polymer networks during the in situ polymerization.

Based on DSC results of GO filled BOZ system, both phenolic and carboxylic acids on the functionalized graphene were effective benzoxazine polymerization initiators and/or catalysts. Moreover, the larger the amount of GO presented, the more pronounced was the effect. However, the slight increase of T_g inferred that the insufficient interactions between GO and polymer significantly reduced the actual performances of GO in the polymer matrix. Therefore, it is expected the covalently bonded functionalized graphene–polymer composites would be developed to enhance thermo-mechanical properties of polymer nanocomposites.³⁴ Using GO-COOH with benzoxazine polymers has several advantages over using GO directly such as: GO-COOH occupying a large amount of carboxylic groups, relatively higher thermal stability, and exfoliated sheet morphology might be more easily dispersed and reacted in the BOZ matrix. Based on DSC results, the GO-COOH nanoplatelets had the different polymerization path from that of GO resulting from the catalytic effect caused by the carboxyl groups and grafting polymerization reactions between GO-COOH and BOZ. Furthermore, significant improvement of T_gs and crosslinking network density of GO-COOH blend system might also be due to the covalent bonds between filler and polymer chains. It is noted that a very low amount (1 wt%) of GO-COOH can remarkably improved the thermal properties of composites, leading to a more than 30 °C increase of T_g in comparison with pure benzoxazine. Therefore, it could be concluded that the GO-COOH nanoplatelets had stronger influence on thermal properties improvement of the nanocomposites than that of GO obtained in this study.

Conclusions

We prepared the novel benzoxazine based composites with carboxylated graphene oxide via in situ intercalative polymerization. Compared to GO, GO-COOH occupying a large amount of carboxylic groups, relatively higher thermal stability, and exfoliated sheet morphology might be more easily dispersed and reacted in the BOZ matrix. Based on DSC results, the GO-COOH nanoplatelets based composites possessed the different polymerization path from that of GO, implying that carboxyl functional groups of GO-COOH not only served to the decrease of ring opening polymerization temperature and exothermic enthalpy, but also participated in the grafting polymerization reactions between GO-COOH and BOZ. Furthermore, significant improvement of T_g and crosslinking density of the GO-COOH blend system might also be due to the covalent bonds between carboxyl groups of functionalized filler and hydroxyl groups of polymer matrix, suggesting that the GO-COOH nanoplatelets had stronger influence on thermal properties improvement of the nanocomposites than that of GO system obtained in this study. Surprisingly, a very low amount (1 wt%) of GO-COOH can remarkably affect the thermal properties of composites, leading to a more than 30 °C increase of T_g in comparison with pure benzoxazine. Therefore, the as-prepared covalently bonded BOZ/GO-COOH nanocomposites might have potential applications in many areas, such as aerospace, electronics, and packaging.

Acknowledgements

We gratefully thank the SRF for ROCS, State Education Ministry, PR China, the Fundamental Research Funds for the Central Universities, China University of Geosciences (Wuhan) (Contract Grant No. CUGL090223), Hubei Provincial Department of Education (XD2010037), and the grant of the Opening Project of State Key Laboratory of Polymer Materials Engineering (Sichuan University) (KF201106) and Engineering Research Center of Nano-Geomaterials of Ministry of Education (CUG).

Notes and references

1. Y. Yagci, B. Kiskan and N. N. Ghosh, J. Polym. Sci., Part A: Polym. Chem., 2009, 47, 5565–5576.
2. R. Huang, S. O. Carson, J. Silva, T. Agag, H. Ishida and J. M. Maia, Polymer, 2013, 54, 1880–1886.
3. Y. Cheng, T. Qi, Y. Jin, D. Deng and F. Xiao, Polymer, 2013, 54, 143–147.
4. W. Li, T. Wei, Y. Gao, K. Xi and X. Jia, Polymer, 2012, 53, 1236–1244.
5. X. Wu, S. Z. Liu, D. T. Tian, J. J. Qiu and C. M. Liu, Polymer, 2011, 52, 4235–4245.
6. H. D. Kim and H. Ishida, J. Phys. Chem. A, 2002, 106, 3271–3280.
7. H. D. Kim and H. Ishida, Macromolecules, 2003, 36, 8320–8329.
8. H. Y. Low and H. Ishida, Polym. Degrad. Stab., 2006, 91, 805–815.
9. Z. Shi, D. Yu, Y. Wang and R. Xu, J. Appl. Polym. Sci., 2003, 88, 194–200.
10. T. Agaga, H. Tsuchiya and T. Takeichi, Polymer, 2004, 45, 7903–7910.
11. C. K. Chozhan, M. Alagar and P. Gnanasundaram, Acta Mater., 2009, 57, 782–794.
12. Y. Wu, M. Zeng, H. Jin, Q. Xu, L. Lu, J. Wang and S. Hou, Tribol. Lett., 2012, 47, 389–398.
13. Y. Wu, M. Zeng, Q. Xu, S. Hou, H. Jin and L. Fan, Tribol. Int., 2012, 54, 51–57.
14. Y. H. Wang, C. M. Chang and Y. L. Liu, Polymer, 2012, 53, 106–112.
15. T. Agag and T. Takeichi, Polymer, 2011, 52, 2757–2763.
16. K. H. Liao, S. Aoyama, A. A. Abdala and C. Macosko, Macromolecules, 2014, 47, 8311–8319.
17. C. F. Wang, S. W. Kuo, C. H. Lin, H. G. Chen, C. S. Liao and P. R. Hung, RSC Adv., 2014, 4, 36012–36016.
18. H. W. Yang, M. Y. Hua, S. L. Chen and R. Y. Tsai, Biosens. Bioelectron., 2013, 41, 172–179.
19. Y. J. Song, K. J. Qu, C. Zhao, J. S. Ren and X. G. Qu, Adv. Mater., 2010, 22, 2206–2210.
20. W. K. Park and J. H. Jung, J. Power Sources, 2012, 199, 379–385.
21. S. M. Alhassan, S. Qutubuddin, D. A. Schiraldi, T. Agag and H. Ishida, Eur. Polym. J., 2013, 49, 3825–3833.
22. H. C. Schniepp, J. L. Li, M. J. McAllister, H. Sai, M. Herrera-Alonso, D. H. Adamson, R. K. Prudhomme, R. Car, D. A. Saville and I. A. Aksay, J. Phys. Chem. B, 2006, 110, 8535–8539.
23. H. Kim, A. A. Abdala and C. W. Macosko, Macromolecules, 2010, 43, 6515–6530.
24. Z. Xu and C. Gao, Macromolecules, 2010, 43, 6716–6723.
25. X. J. Shen, X. Q. Pei, S. Y. Fu and K. Friedrich, Polymer, 2013, 54, 1234–1242.
26. N. Ning, W. Zhang, J. Yan, F. Xu, T. Wang, H. Su, C. Tang and Q. Fu, Polymer, 2013, 54, 303–309.
27. P. Fabbri, L. Valentini, B. S. Bittolo, D. Foix, L. Pasquali, M. Montecchi and M. Sangermano, Polymer, 2012, 53, 6039–6044.
28. Y. F. Huang and C. W. Lin, Polymer, 2012, 53, 2574–2582.
29. Y. Huang, M. Zeng, Z. Feng, D. Yin, Q. Xu and L. Fan, RSC Adv., 2016, 6, 3561–3570.
30. C. Bora and S. K. Dolui, Polymer, 2012, 53, 923–932.
31. M. Zeng, J. Wang, R. Li, J. Liu, W. Chen, Q. Xu and Y. Gu, Polymer, 2013, 54, 3107–3116.
32. A. A. Alhwaige, S. M. Alhassan, M. S. Katsiotis, H. Ishida and S. Qutubuddin, RSC Adv., 2015, 5, 92719–92731.
33. S. Sahila and L. S. Jayakumari, Polym. Compos., 2015, 36, 1–7.
34. R. Sasi kumar, M. Ariraman and M. Alagar, RSC Adv., 2015, 5, 23787–23797.
35. C. R. Arza, H. Ishida and F. H. J. Maurer, Macromolecules, 2014, 47, 3685–3692.
36. F. B. Meng, H. Ishida and X. B. Liu, RSC Adv., 2014, 4, 9471–9475.
37. J. Dunkers and H. Ishida, J. Polym. Sci., Part A: Polym. Chem., 1999, 37, 1913–1921.
38. X. Zhao, Q. Zhang and D. Chen, Macromolecules, 2010, 43, 2357–2363.
39. Y. Huang, M. Zeng, J. Ren, J. Wang, L. Fan and Q. Xu, Colloids Surf., A, 2012, 401, 97–106.
40. N. Pan, D. Guan, Y. Yang, Z. Huang, R. Wang, Y. Jin and C. Xia, Chem. Eng. J., 2014, 236, 471–479.
41. X. Fan, W. Peng, Y. Li, X. Li, S. Wang, G. Zhang and F. Zhang, Adv. Mater., 2008, 20, 4490–4493.
42. Q. Li, F. Fan, Y. Wang, W. Feng and P. Ji, Ind. Eng. Chem. Res., 2013, 52, 6343–6348.
43. H. Ishida and Y. H. Lee, J. Appl. Polym. Sci., 2001, 81, 1021–1034.
44. H. Ishida and Y. H. Lee, J. Polym. Sci., Part B: Polym. Phys., 2001, 39, 736–749.
45. S. W. Kuo, Y. C. Wu, C. F. Wang and K. U. Jeong, J. Phys. Chem. C, 2009, 113, 20666–20673.
46. S. Wirasate, S. Dhumrongvaraporn, D. J. Allen and H. Ishida, J. Appl. Polym. Sci., 1998, 70, 1299–1306.
47. J. Zhou, Z. Yao, Y. Chen, D. Wei, Y. Wu and T. Xu, Polym. Compos., 2013, 34, 1245–1249.
48. Y. X. Wang and H. Ishida, Polymer, 1999, 40, 4563–4570.
49. S. Rimdusit and H. Ishida, Polymer, 2000, 41, 7941–7949.
50. T. Agag and T. Takeichi, Polym. Compos., 2008, 29, 750–757.
51. S. Rimdusit and H. Ishida, J. Polym. Sci., Part B: Polym. Phys., 2000, 38, 1687–1698.
52. C. Jubsilp, K. Punson, T. Takeichi and S. Rimdusit, Polym. Degrad. Stab., 2010, 95, 918–924.

---

Footnotes

† Electronic supplementary information (ESI) available: Experimental section with preparation methods of benzoxazine monomer (ESI 1) and graphene oxide (ESI 2), and measurement method of optical microscope (OM) (ESI 3), digital photos and OM images of the cured BOZ, GO/BOZ-1 and GO-COOH/BOZ-1 samples (Fig. S1), FTIR spectra of uncured and cured G/BOZ-1 and GO/BOZ-1 (Fig. S2 and S3), and DSC thermograms of BOZ, GO/BOZ-1 and GO/BOZ-3 (Fig. S4). See DOI: 10.1039/c5ra28016h

‡ Present address: Graduate School of Life Science, Hokkaido University, Sapporo 060-0810, Japan.

---