A study on interfacial behaviors of epoxy/ graphene oxide derived from pitch - based graphite ﬁ bers

: Graphene oxide ( GO ) is a versatile material with inherent unique properties that can be used in a wide range of applications. GO is produced from graphitic mate rials including graphite, and its properties can depend on the nature of stacking in the graphene structures. In this study, GO was prepared from pitch - based graphite ﬁ bers via the modi ﬁ ed Hummer ’ s method and subsequently incorporated into an epoxy matrix to obtain graphene loaded nanocomposites ( EP/GO ) . Presented experimental results revealed that the addition of 0.6 wt% GO yielded an ∼ 110% increase in the fracture toughness. The corre sponding fracture energies as well as the ﬂ exural strengths and ﬂ exural modulus exhibited similar trends to the fracture toughness. The thermophysical properties of the EP/GO, to further demonstrate the reinforcing e ﬀ ectiveness of GO, were also observed. Collectively, these results indicate that GO investigated in the study can be a viable reinforce ment candidate to develop next - generation nanocompo sites with multifunctional properties.

Many academic and industrial researchers are struggling in manufacturing graphene with high quality and large quantities, which is divided into two categories: one is the bottom-up method that synthesizes from hydrocarbon sources using chemical vapor deposition for single or a few layers with the minimum defect area [8,9]. This method cannot be used for most applications that require graphene in large quantities. The other is the top-down method that applies chemical and/or mechanical exfoliation of graphite, which is much advantageous to produce 2-D graphene oxide (GO) in a scalable manner. In particular, GO produced by the top-down method has attracted increasing attention as a nanofiller for polymer-based nanocomposites [10][11][12].
Until now, it was considered that graphite (synthetic or natural) is the primary source for mass production of GO or reduced GO using the top-down method. Jiao et al. reported that GOs synthesized from different types of natural graphite show a remarkable difference in the crystalline morphologies, chemical properties, etc. because of different graphitization degree [13]. Understanding the characteristics of precursors and controlling parameters such as medium and temperature during the synthesis process is crucial to obtain GO with desirable properties for use in a wide range of applications [14][15][16][17].
Along with these, other sp 2 -carbon sources are widely used for GO production, such as graphite fibers (GFs) [18], carbon nanotubes [19], coal [20], and biomass [21]. Among them, GFs have abundant crystalline graphite with a high carbon content of more than 99%. This is because they are manufactured by the graphitization process at approximately 3,000°C. GFs are thus extensively used in applications that require high strength and high modulus, including applications in the railways and the automotive, aerospace, drilling, and consumer product industries [22][23][24][25][26][27]. GFs with considerable amounts of 2-D conjugated sp 2 domains are acceptable, which can be converted to graphene and/or graphene derivatives for mass production [18,28,29].
In this study, GO was prepared from pitch-based GFs using a typical oxidation and subsequent exfoliation process, called the Hummer's method, and the chemical and morphological properties of GO were investigated. Moreover, GO-incorporated epoxy (EP/GO) nanocomposites as a function of GO loading amount were fabricated to investigate the interfacial interactions between GO and epoxy matrix. The present study aimed concretely at evaluating the effects of GO on the interfacial properties and mechanical properties (e.g., surface free energy, interfacial adhesion, fracture toughness, flexural strength, and flexural modulus) of epoxy nanocomposites. Furthermore, we demonstrate that GO derived from pitch-based GFs can be an effective reinforcement as an alternative 2-D material for polymeric matrix, and present perspective to meet the requirements of light weight and strong mechanical properties for automotive and aerospace applications [30,31].

Preparation of GF-derived GO
The synthesis procedure is shown in Figure 1(a). The asreceived GFs were cut into fragments and then ground in a ball mill at 250 rpm for 4 h to obtain GF particles (GFPs). Oxidation was performed using the modified Hummer's method [32]. Then, 3 g of GFPs was added to a 500 mL three-necked round-bottom flask charged with a mixture of concentrated H 2 SO 4 (360 mL) and H 3 PO 4 (40 mL). Next, 21 g of KMnO 4 was slowly added to the flask, and the temperature of the solution was maintained at 0-4°C for 6 h before increasing it to 60°C for 18 h. The solution was poured into 2,000 mL of ice water, followed by dropwise addition of H 2 O 2 until no new bubbles emerged. The solution was treated by the following steps: static precipitation for 24 h, centrifugation to remove most of the acid solution, and infiltration to remove the remaining acids. The final GO sample was obtained using the freeze-drying process, and 4.8 g of GO was collected to be used as a reinforcement. Notably, the rotational speed should not exceed 250 rpm because the GO structure would be destroyed under the high shear stress [33]. The mixture was then degassed at 80°C for 48 h to remove the solvent completely. DDM (20 wt% of EP) was then slowly added to the solution and mixed with a planetary mixer, followed by further degassing at 60°C for 30 min. The prepared EP/GO samples were cured by the following three steps: 110°C for 1 h, 140°C for 2 h, and 170°C for 1 h.

Characterization
X-ray diffraction (XRD; D2 Phaser, Bruker, Germany) was used to investigate the structure of the GO in the 2θ range of 4-60°. Fourier-transform infrared spectroscopy (FT-IR; Jasco PS-4000, JASCO, Japan) was used to obtain the infrared spectra of the GO. The qualities of the fillers before and after thermal oxidation were analyzed by Raman spectroscopy (Raman; LabRAM Revolution, HORIBA, Japan) equipped with a charge-coupled device detector. X-ray photoelectron spectroscopy (XPS; ESCA LAB MK-II, VG Scientific, UK) was conducted to analyze the surface chemistry of the GO. High-resolution scanning electron microscopy (HR-SEM; SU 8010, Hitachi, Japan) was used to observe the cross-section morphology of the prepared nanocomposites. Field-emission transmission electron microscopy (FE-TEM; JEM-2100F, JEOL, Japan) was used to examine the structure of the GO. The dynamic mechanical properties were measured using rectangular specimens with dimensions of 20 mm × 5 mm × 2 mm, held in a dual-cantilever clamp, using a dynamic mechanical analyzer (DMA; Q800, TA Instruments, USA).
The total surface free energy, which is the sum of London dispersive (γ L ) and specific components (γ SP ), can be calculated using the following equation [34]: where γ L and γ SP result from intermolecular interactions of London force of van der Waals attraction and the specific force, respectively. According to Fowkes' suggestion [34], the surface free energy of a solid material can be examined by measuring the contact angle using the following geometric mean, known as the work of adhesion, W A : where θ is the contact angle; subscripts L and S represent liquid and solid, respectively; superscripts L and SP represent London force and specific force, respectively. Equations L should give a straight line, enabling easy calculation of the surface free energy [35].
The critical stress intensity factor (K IC ) of the EP/GO nanocomposites was examined via a three-point flexural test, which was performed using a universal test machine according to the standard ASTM E-399. A span-to-depth ratio of 4:1 and a crosshead speed of 1 mm min −1 were considered. The K IC value of the cured epoxy and its nanocomposites was calculated using the following equations: where a, P, d, L, and b represent the crack length, critical load for crack propagation, thickness of specimen, length of the span, and the width of specimen, respectively. The critical strain energy release rate (G IC ) of the epoxy and its nanocomposites was calculated using the following equation: where E is the Young's modulus obtained from the fracture test and ν is the Poisson's ratio for the nanocomposites (0.3). The flexural properties were tested with an Instron 1125 mechanical tester according to the standard ASTM D-790. To determine the mechanical properties, five experimental measurements were conducted for each specimen and the average value was obtained.
3 Results and discussion

Characterization of the GO
HR-SEM images of the GFs used in the study were observed to verify the presence of lots of stacked 2-D carbon sheets along with the long axis of the GF, as shown in Figure 2. Figure 3 displays the FT-IR spectra of the GF and GO derived from the GF. As shown in Figure 3(a), the FT-IR spectra of the GF demonstrated a small absorption peak at 3,380 cm −1 corresponding to hydroxyl groups (−OH). For the GO, new peaks were observed at 1,732 cm −1 , which can be attributed to stretching vibrations of the carboxylic groups (−COOH). The functional groups of the prepared GO are consistent with those reported in previous studies, demonstrating the feasibility of oxidation [36][37][38]. Photographs of GO suspension in water are shown in the inset of Figure 3(b). GO was homogeneously dispersed to generate a deep yellowbrown suspension. Figure 3(b) shows the TEM image of GO, exhibiting a typical 2-D nanosheet morphology with a particle size of approximately 2 µm. The grid can be clearly observed because of the superior transparency of the prepared GO. The XRD patterns of the GF and GO are shown in Figure 3(c). The pattern of the GF shows a typical graphite peak of (002) at 26.8°, corresponding to an interlayer d-spacing of 0.330 nm. In the case of GO, a new typical peak of (001) at 10.5°corresponding to a d-spacing of 0.84 nm appeared, indicating that a large degree of exfoliation occurred in the graphitic layers. This result confirmed that the stacked graphitic layers were completely exfoliated and converted into GO, which is in good agreement with the previous results of a typical structure of GO. We employed Raman spectroscopy to observe the microstructural characteristics of the GF and GO. Basically, the Raman spectra of carbon nanomaterials present two peaks corresponding to the G-and D-bands at approximately 1,590 and 1,350 cm −1 due to the presence of sp 2 crystalline and sp 3 amorphous hybridized carbons, respectively [32,39]. As expected, two characteristic peaks in the GFs and  GO were observed at 1,584 and 1,352 cm −1 , as shown in Figure 4(a). To investigate the degree of amorphous and crystalline phases, the integrated intensity ratio ( / I I D G ) from the D-and G-bands was calculated. The / I I D G increased from 0.45 (GF) to 1.08 (GO), indicating a decrease in the average size of the sp 2 graphene domains in GO compared to the GF.
XPS analysis was conducted to determine the surface chemical compositions and functional groups of the GO. In Figure 4(b), the two predominant characteristic peaks observed at 285.0 and 531.0 eV can be attributed to C1s and O1s core-level emissions, respectively. The carbon/ oxygen (C/O) ratios were also calculated. The C/O ratio of the GFs was 6.14, which is mostly composed of carbon in the pristine GF, while the GO showed a much lower value of C/O ratio (1.13). To further confirm the surface functional groups of the GFs and GO, their high-resolution singlescan C1s spectra were deconvoluted. In Figure 4(c and d), the peaks of the GFs observed at 284.4 and 286.6 eV correspond to the C-C and C-O groups, respectively, indicating the presence of majorly conjugated sp 2 carbons and little oxygen. For the GO spectra, a broad peak was observed at 284.4 eV owing to an increase of full width at half maximum, resulting from the introduction of oxygenfunctional groups. Moreover, the prominent peak at 289.2 eV corresponding to the carboxylic group (-COOH) was observed, indicating successful oxidation.

Interfacial properties of EP/GO nanocomposites
Surface free energy (γ) can strongly affect the wettability, adhesion, adsorption, and morphology of the components of composites. The London dispersion force, specific component force, and surface tension of the liquids are given in Table S1. The contact angles of the composites are shown in Table S2. Using the data provided in Tables S1 and S2, we calculated the surface free energies of the EP/GO, as plotted in Figure 5  with an increase of the GO content ( Figure 5(b)). Here, all the EP/GO nanocomposites exhibit good linearity for the relationship between the surface free energies and work adhesion, which clearly demonstrates enhanced interfacial adhesion [40,41]. We studied the wetting behavior of distilled water (carried out at 27°C for 10 min) acting on the nanocomposites in more detail. It can be seen from Figure 5(c) that the wetting behavior of the EP/GO0.0 nanocomposites was maintained at roughly 60.4°exhibited no obvious changes. By contrast, the wetting behavior of the EP/GO0.8 nanocomposite is seen to decrease rapidly and become saturated at around 38.6°. This decrease is due to the large number of carboxylic groups on the GO surface, from which water droplets can rapidly diffuse, thus providing an enhanced wettability for the nanocomposites.

Mechanical properties of EP/GO nanocomposites
The bending strength of the EP/GO nanocomposites increased up to 0.6 wt% by weight and then decreased at a higher GO content (Figure 6(a)). The bending strength of the epoxy nanocomposites reached 55.7 MPa in EP/GO0.6 nanocomposites, which is twice that of EP/GO0.0 nanocomposites (28.5 MPa). In addition, the Young's modulus ( Figure 6(b)) showed a similar trend to the bending strength.
Normally, filler dispersion, interfacial interactions between the filler and polymer matrix, and the intrinsic characteristics of the filler are known to be three dominant factors affecting the performance of nanocomposites. In this work, since all the specimens were prepared through the ball-milling process, thereby ensuring good dispersion of the filler in the polymer matrix, the dispersion factor alone cannot account for enhanced mechanical properties. Meanwhile, the EP/GO nanocomposites exhibited superior interfacial interactions because abundant oxygen-functional groups were introduced onto the GO surface, as shown in Figure 5, which have strong affinity with the epoxy matrix. Thus, these results revealed that the perfect interfacial interaction between GO and the epoxy occurred, resulting in enhanced mechanical properties of the nanocomposites [42,43]. To further demonstrate the interfacial properties between the filler and the matrix, the flexural strength (σ) and flexural modulus (E b ) of the EP/GO nanocomposites with different GO loading amounts were also calculated, using the following equations (8) and (9) [44,45]: where P is the applied peak load, b is the sample width, d is the specimen thickness, L is the support span, P Δ is the variation in force at the linear portion, and m Δ is the relative deflection variation. Figure 7(a and b) shows the flexural strength and flexural modulus of EP/GO nanocomposites. With increasing GO loading fractions, the flexural strength gradually increased from 107 to 140 MPa. When the GO concentration exceeded 0.6% by weight, the flexural strength  decreased slightly. When the GO concentration exceeded 0.6% by weight, the flexural strength decreased slightly. The fracture toughness values were determined using the three-point bending tests, as shown in Figure 7(c and d).
The results demonstrate that GO enhances both the fracture toughness (K IC ) and fracture energy (G IC ) of the prepared nanocomposites. The K IC value of EP/GO0.6 nanocomposites increased by 109.3%; this enhancement can be attributed to the good dispersion of GO and thus the enhanced interfacial interactions within the epoxy matrix. Meanwhile, the EP/GO nanocomposites at higher fraction of GO (0.8% by weight) exhibited decreased K IC value, probably due to the formation of aggregates. Correspondingly, the EP/GO nanocomposites exhibited the highest G IC value in the EP/GO0.6 nanocomposites as well, demonstrating the best mechanical performances among the nanocomposites studied in the work.
It is well known that failure behaviors of nanocomposites are related to their structural integrity at both microscopic and macroscopic levels under external forces [44,46]. To reveal the failure behavior of the EP/GO nanocomposites, we observed their fracture surfaces using HR-SEM; the results are shown in Figure 8(a-d). With increasing GO loading fractions, the number of fatigue cracks increased and the surfaces became rougher. The equally distributed fatigue cracks on the fracture surfaces demonstrated homogeneously dispersed GO in an epoxy matrix. In EP/GO0.8 nanocomposites, the cracks were coarsened and pulled out, as shown in Figure 8(d), possibly due to the aggregation of the GO sheets. For the nanocomposites reinforced by a rigid additive, the plastic yielding of the epoxy matrix around the particles, subsequent void formations, and the interference of the rigid particles during crack propagation, including crack pinning and/or deflection, are considered as the main  toughening mechanisms. In case of the EP/GO nanocomposites, crack deflection could act as the main toughening mechanism due to the increased number of cracks in the EP/GO nanocomposites [47].
In addition, the improvements in K IC achieved by the addition of GO derived from pitch-based GFs in this study are compared with previously reported values for epoxy nanocomposites containing GO, as shown in Figure 9. It can be demonstrated that the prepared GO derived from pitch-based GFs plays an effective reinforcement for the K IC improvements, resulting from enhanced interfacial interaction between the GO and epoxy matrix [48][49][50][51][52][53][54][55][56][57][58][59].

Thermophysical properties of EP/GO nanocomposites
DMA is used not only to determine the amount of elastic energy stored in the prepared composites and the energy dissipated during strain, but also to examine the glass transition temperature (T g ), which is the temperature at which sufficient vibration energy was accumulated to rearrange the crosslinked polymer chains inside the polymeric matrix. T g is also known to be related to relaxation and is sensitive to structural transformations. The corresponding data of the temperature-dependent storage modulus and loss factor of the EP/GO nanocomposites are listed in Table 1. These data indicate that the homogenous dispersion of GO led to improved interfacial interaction between the GO sheets and the epoxy matrix, thereby substantially restricting the segmental movement of the epoxy chains and resulting in an enhanced storage modulus. The EP/GO nanocomposites exhibited a similar level of the dynamic storage moduli at the rubbery plateau area at the same temperature. We speculate that the presence of GO could play an important role in enhancing the dynamic storage modulus of the nanocomposites owing to the interfacial interaction between GO and the epoxy matrix. Moreover, the EP/GO nanocomposites exhibited higher T g values than that of EP nanocomposites. The GO sheets served as anchor points among polymer chains, which restricted their mobility and increased the T g . The T g values increased with increasing GO loading fractions, demonstrating the enhanced interfacial interaction. The increase in the T g value was not significant because the addition of a filler may reduce the cross-linking density of the epoxy, thus decreasing T g [60].

Conclusion
In this study, GO was successfully synthesized using GFs as a precursor. The EP/GO nanocomposites at loadings less than 1% by weight were fabricated via ball-milling process. Presented detailed investigation of the mechanical and thermophysical properties of the EP/GO nanocomposites revealed that the values depend on the loading level of GO. In particular, when the GO loading was 0.6 wt%, the fracture toughness of the EP/GO nanocomposites exhibited a significant improvement of ∼82% compared to that of EP nanocomposites. The fracture energies as well as the flexural strength and flexural modulus showed similar trends to the fracture toughness, demonstrating the effective toughness of GO. Moreover, the presence of GO improved the thermophysical properties of the epoxy matrix. These results can be attributed to the enhanced interfacial adhesion between the GO and the epoxy matrix, affecting the surface free energy and work adhesion. Collectively, these enhancements can be attributed to the good dispersion of GO, thereby enhancing the interfacial interactions between the GO and the epoxy matrix. Thus, the authors believe that the GF-derived GO in this work can be a promising candidate as a reinforcing material for improving the mechanical properties of nanocomposites in a wide range of applications. Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.

Conflict of interest:
The authors state no conflict of interest. Data availability statement: The datasets generated during and/or analyzed during the current study are available in the Supplementary material on the journal website.