Tailoring nanocomposite interfaces with graphene to achieve high strength and toughness

See allHide authors and affiliations

Science Advances  14 Oct 2020:
Vol. 6, no. 42, eaba7016
DOI: 10.1126/sciadv.aba7016


The nanofiller reinforcing effect in nanocomposites is often far below the theoretically predicted values, largely because of the poor interfacial interaction between the nanofillers and matrix. Here, we report that graphene-wrapped B4C nanowires (B4C-NWs@graphene) empowered exceptional dispersion of nanowires in matrix and superlative nanowire-matrix bonding. The 0.2 volume % B4C-NWs@graphene reinforced epoxy composite exhibited simultaneous enhancements in strength (144.2 MPa), elastic modulus (3.5 GPa), and ductility (15%). Tailoring the composite interfaces with graphene enabled effective utilization of the nanofillers, resulting in two times increase in load transfer efficiency. Molecular dynamics simulations unlocked the shear mixing graphene/nanowire self-assembly mechanism. This low-cost yet effective technique presents unprecedented opportunities for improving nanocomposite interfaces, enabling high load transfer efficiency, and opens up a new path for developing strong and tough nanocomposites.


Nanofillers such as nanowires and nanoparticles, which have much larger specific surface areas than microfillers, are theoretically predicted to be ideal reinforcements to enable exceptional joint enhancements in strength and toughness. However, nanocomposites have not fulfilled this promise, largely because of the poor interfacial bonding between the fillers and matrix. As one of the third hardest materials known in nature, boron carbide (B4C) is often prized for its eminent physical and mechanical properties, including low density (2.5 g/cm3), extreme hardness (27.4 to 37.7 GPa), and high elastic modulus (460 GPa) (14). However, when used as nanoreinforcements in nanocomposites, B4C nanowires (B4C-NWs) did not show their entire reinforcing effect because of the poor dispersion of B4C-NWs in matrix and the weak interfacial bonding between B4C-NWs and matrix (5). Thus, engineering nanocomposite interfaces is key to realize the full potential of nanofillers in their composites.

Many approaches have been explored to improve nanofiller dispersion and filler-matrix interfacial interaction. Interfacial engineering techniques include the attachment of small-molecule surfactants, such as silane coupling agents (6, 7), and grafted polymer chains, like polyacrylamide and polystyrene (8, 9). These surface treatments, to a certain extent, mitigated the interfacial problems; the trade-off was the reduction of the nanofiller intrinsic properties and extra treatment cost (10). In this context, we have a pressing need for seeking a new kind of interface modifier that can simultaneously achieve the homogenous dispersion of nanofillers and improve interfacial bonding in nanocomposites. Graphene, on account of its exceptional high crystallinity and mechanical prowess (1113), has stimulated widespread scientific interest. The recently developed mechanical shear mixing enables at-scale production of graphene at low cost, promoting its practical applications, especially in polymer-based composites (1418). The high specific area makes graphene and its derivatives the excellent interface agent to enhance the bonding between nanofillers and polymer matrix through π-π interaction, hydrogen bonding, van der Waals force, electrostatic interaction, and chemical bonding (19, 20). However, graphene interface engineering has not been explored.

Here, we report a graphene interface engineering technique that glued B4C-NWs with graphene, exceptionally enabling enhancements in both strength and toughness. Specifically, high-density B4C-NWs were obtained via a vapor-liquid-solid (VLS) process (21). High-quality graphene sheets were converted directly from graphite and simultaneously wrapped onto the B4C-NWs by shear mixing. The as-obtained graphene-wrapped B4C-NWs (B4C-NWs@graphene) exhibited excellent dispersion in water and epoxy. The 0.2 volume % (vol %) B4C-NWs@graphene reinforced epoxy composite exhibited joint enhancements in strength (144.2 MPa), elastic modulus (3.5 GPa), and fracture strain (15.0%).


First, the B4C-NWs were uniformly grown on the surface of carbon fiber cloth (fig. S1) via a typical VLS method (21) where cotton served as carbon source, amorphous boron powders served as boron source, and Ni(NO3)2 6H2O served as catalyst (1, 22). The B4C-NWs of 20 to 300 nm in diameter and around 5 μm in length were then separated from the substrate by ultrasonic vibration. The chemical bonding states in the B4C-NWs were studied by x-ray photoelectron spectroscopy (XPS). The B 1s peaks revealed the existence of B─B (187.9 eV), B─C (188.7 eV), and B─O (193.4 eV) bonds (fig. S1C). In the C 1s spectra, two peaks centered at 284.5 and 282.2 eV were observed, corresponding to C─C and C─B bonds, respectively (fig. S1D) (23). The atomic ratio of B to C was measured to be 3.56 ± 0.68, which is in a reasonable range of B4C stoichiometry, confirming the production of high-quality B4C-NWs.

The self-assembled B4C-NWs@graphene were directly synthesized by shear mixing the mixture of graphite powders and B4C-NWs (Fig. 1). Transmission electron microscopy (TEM) inspection (Fig. 2C) showed that graphite was successfully exfoliated to graphene, while B4C-NWs remained intact in the shear mixing. Most B4C-NWs were fully wrapped with graphene, and some redundant graphene was also observed having been self-assembled (fig. S2). The diameter of B4C-NWs@graphene was increased by 30 to 200% compared with pristine B4C-NWs. Following the same procedure, pristine graphene alone was fabricated using the shear mixing method and served as the control sample (Fig. 2B). To investigate the quality of B4C-NWs@graphene, the suspension of B4C-NWs@graphene in dilute water was monitored immediately after shear mixing without any further treatments (Fig. 2D). By using the same processing, the suspensions of B4C-NWs and pristine graphene were also prepared for comparison. The above-prepared B4C-NWs@graphene, B4C-NWs, and graphene were separately dispersed in water, and then, the coarse powders that were not thoroughly shear mixed gradually precipitated down to the bottom. At the early stage, the settling rates of the sediments in both graphene and B4C-NW suspensions were much larger than that in B4C-NWs@graphene, indicating that graphene and B4C-NWs tended to agglomerate in their individual suspensions, whereas the B4C-NWs@graphene exhibited superlative dispersion. After 6 hours of sedimentation, both graphene and B4C-NW suspensions exhibited aggregation and deposition, and after 12 hours, they were separated completely into the clean supernatants and the solid residues. In contrast, the B4C-NWs@graphene remained well dispersed, with little precipitates. The productivities of the graphene derived from pristine graphite and the B4C-NWs@graphene were calculated after 6 hours of sedimentation to be 9.1 ± 1.5 and 78.2 ± 3.0%, respectively. Adding B4C-NWs in the suspension effectively facilitated the exfoliation of graphite and the dispersion of as-synthesized graphene. In conclusion, graphene sheets were fabricated from graphite together with B4C-NWs by shear mixing in which graphene sheets were simultaneously self-assembled onto the B4C-NW surface.

Fig. 1 Schematic illustration of the synthesis process steps of B4C-NWs@graphene.

Fig. 2 Synthesis of nanofillers in dilute water by shear mixing.

TEM images of (A) B4C-NWs, (B) multilayered graphene, and (C) B4C-NWs@graphene. (D) Chronological digital photos of the suspensions of B4C-NWs, graphene, and B4C-NWs@graphene. Photo credit: Ningning Song, University of Virginia.

High-resolution TEM (HRTEM) inspection and the corresponding fast Fourier transform (FFT) pattern revealed that graphene sheets were crumpled and self-assembled onto the B4C-NWs (Fig. 3A and fig. S2, E to H). The graphene on B4C-NWs showed highly complex wrinkled/crumpled texture. X-ray diffraction (XRD) inspection (Fig. 3B) revealed three typical diffraction peaks of B4C, which can be indexed to the (110), (114), and (021) diffractions of rhombohedral boron carbide [the Joint Committee on Powder Diffraction Standards (JCPDS) no. 6-0555]. The sharp XRD peak at 26.0° is ascribed to graphene, and other peaks resulted from nickel boride, the catalyst for B4C-NW growth. The electron energy-loss spectroscopy (EELS) spectrum showed the B(1s) K-edge and the C(1s) K-edge (fig. S3A). The Raman spectra obtained from the B4C-NWs@graphene structure (Fig. 3C) exhibited the typical peaks of B4C and graphene. Because of the coexistence of two components and their heterogeneous distribution, the Raman spectra varied at different acquisition positions. In some areas of B4C-NWs@graphene, the coexistence of B4C and self-assembled graphene induced Raman peak shift and broadening (fig. S3B), due to the amorphous phase inclusions in B4C (24) and the crumpled and stacked graphene sheets. B4C crystal lattice has a rhombohedral arrangement consisting of 12-atom icosahedra and 3-atom linear chains (R3¯m space group, a = 5.16 Å, and α = 65.7°) (1, 25). Close-up HRTEM observation (Fig. 3D) and the FFT pattern with a zone axis [012¯] (Fig. 3E) jointly verified that in the B4C-NWs@graphene structure, the B4C-NWs are of perfect rhombohedral crystalline structure. The measured interplanar spacing of 0.256 nm pointed toward the axial growth plane (121). The representative B4C-dominated Raman spectrum (Fig. 3F) displayed the peaks that can be ascribed to the intraicosahedral and intericosahedral modes (188, 720, 813, 978, and 1068 cm−1) and the vibrations of chain structures linking icosahedra (377, 484, and 531 cm−1) (26). The HRTEM image and corresponding FFT pattern of graphene validated that the graphene sheets on B4C-NWs are of high quality, with monolayered and multilayered features (Fig. 3, G and H). The presence of monolayered graphene in the B4C-NWs@graphene structure was confirmed by Raman spectroscopy (Fig. 3I), displaying a symmetrical two-dimensional (2D) band with a full width at half maximum of 38.1 cm−1 and 2D/G intensity ratio of 1.65.

Fig. 3 Characterization of B4C-NWs@graphene.

(A) TEM image, (B) XRD pattern, and (C) background-corrected Raman spectrum of B4C-NWs@graphene. (D) HRTEM image, (E) the corresponding FFT, and (F) background-corrected Raman spectrum of the B4C-NWs in B4C-NWs@graphene. (G) HRTEM image, (H) the corresponding FFT, and (I) background-corrected Raman spectrum of the monolayered graphene in B4C-NWs@graphene. a.u., arbitrary units.

The B4C-NWs@graphene were dispersed into epoxy resin to fabricate epoxy nanocomposites. Three-point bending tests were carried out on the B4C-NWs@graphene composites and epoxy specimens [five beam specimens with the dimension of 40 mm by 7 mm by 5 mm (length by width by height) were tested for each material]. The typical flexural stress-strain curves (Fig. 4A) demonstrate a general trend that the flexural strength and elastic modulus increased with increasing nanofillers. The pure epoxy sample exhibited a linear elastic stress-strain relationship without having plastic deformation, whereas the B4C-NWs@graphene nanocomposites underwent a large portion of plastic deformation before fracture. The fractographic analysis was performed to investigate the dispersion quality of nanofillers, with the goal to understand B4C-NWs@graphene strengthening and toughening mechanisms. The pure epoxy control sample showed catastrophic failure with a rather smooth fracture surface (fig. S4A), whereas the B4C-NWs@graphene composite exhibited a much rougher fracture surface with “sea-island”–like morphology (Fig. 4B), indicating crack pinning and/or deflection when encountering with the B4C-NWs@graphene. The frequently observed nanowire pullout sites suggest that yielding of the matrix around the fillers was first generated, followed by plastic void formation and growth (Fig. 4C), and the primary crack deflected when encountering with B4C-NWs@graphene (fig. S5). Graphene, as an interfacial agent, largely strengthened the bonding between B4C-NWs and epoxy matrix. As a result, the pulled-out B4C-NWs@graphene nanofillers exhibited larger diameters of ~500 nm because of the attached epoxy coatings (Fig. 4C and fig. S4G). The debonding of B4C-NWs@graphene from epoxy consumed more energy. Therefore, crack pinning, deflection, debonding of B4C-NWs@graphene from matrix, void formation around the nanofillers, and nanofiller pullout jointly contribute to the enhanced toughness of the B4C-NWs@graphene composites. For comparison, pure graphene and B4C-NW reinforced epoxy composites were respectively fabricated and characterized following the same procedure. The graphene composites and B4C-NW composites exhibited a plethora of large agglomerations on their fracture surfaces (fig. S4, C and E) that induced microcrack coalescence, promoting the primary crack propagation (fig. S4, B and D). The pulled-out B4C-NWs in the B4C-NW composites showed smooth surface, suggesting the poor bonding between the B4C-NWs and matrix (fig. S4F). For the B4C-NWs@graphene composites, the pulled-out B4C-NWs, which were not fully wrapped by graphene, were observed with smooth surface (fig. S4, H and I). In sum, graphene rendered nanofillers better dispersion ability and improved load transfer, leading to joint amplifications in strength and toughness.

Fig. 4 Reinforcing effects of B4C-NWs@graphene.

(A) Flexural stress-strain curves of epoxy and B4C-NWs@graphene (0.1, 0.2, and 0.3 vol %) reinforced composites. (B and C) Scanning electron microscopy images of the fracture surface of 0.2 vol % B4C-NWs@graphene reinforced composite. Comparison of experimentally measured (scatter plot) and theoretically predicted elastic modulus values of (D) B4C-NWs@graphene composites, (E) B4C-NW composites, and (F) graphene composites.

To evaluate the dispersion quality of B4C-NWs@graphene, theoretical elastic moduli of the composites were calculated using the Voigt approximation (upper bound), the Reuss approximation (lower bound), and the Halpin-Tsai model (empirical model) as follows (27, 28)Voigt approximation: Ec=νEf+(1ν)Em(1)Reuss approximation: 1Ec=νEf+1νEm(2)Halpin-Tsai model: Ec=Em[381+η1(2lf/df)ν1η1ν+581+2η2ν1η2ν]η1=Ef/Em1Ef/Em+2lf/df, η2=Ef/Em1Ef/Em+2(3)where Ec, Ef, and Em are the elastic moduli of composite, fillers, and matrix, respectively. ν, lf, and df represent the volume fraction, length, and diameter of the nanofillers. The elastic moduli of B4C-NW (24), graphene (29), and epoxy were given by 435, 250, and 2.7 GPa, respectively. The elastic modulus of B4C-NWs@graphene was considered to be approximately equal to that of B4C-NW. As shown in Fig. 4D, the elastic moduli of B4C-NWs@graphene composites are much higher than the empirical values and very close to the upper limit. For comparison, graphene sheets and B4C-NWs were separately dispersed into the epoxy resin, and the graphene epoxy and B4C-NW epoxy composites were respectively characterized by three-point bending (fig. S6). With a low volume fraction of nanofillers, both graphene and B4C-NW composites exhibited enhanced strength and toughness. However, the resultant strength and elastic modulus of the B4C-NW composites were lower than those of the B4C-NWs@graphene composites. All B4C-NWs@graphene (0.1, 0.2, and 0.3 vol %) reinforced composites exhibited large plastic deformations before failure. However, the plastic deformation portions in the stress-strain diagrams of both graphene and B4C-NW composites gradually reduced with increasing nanofiller content and disappeared completely for 0.3 vol % composites, indicating that graphene and B4C-NWs tended to agglomerate at high volume fractions of reinforcement. Accordingly, because of the deficient load transfer, the elastic moduli of B4C-NW composites and graphene composites are much lower than their upper limit (Fig. 4, E and F).

The 0.1, 0.2, and 0.3 vol % B4C-NWs@graphene composites presented the flexural strengths of 127.4, 144.2, and 156.2 MPa, representing respectively 13.9, 28.9, and 39.6% amplification over the pure epoxy specimen (111.9 MPa); the elastic moduli of 3.0, 3.5, and 3.7 GPa; and 11.1, 29.6, and 37% enhancement compared with the epoxy control sample (2.7 GPa). The fracture strains of the B4C-NWs@graphene nanocomposites are approximately increased by 173%, indicating that the toughness of B4C-NWs@graphene composites is largely enhanced (Fig. 5C). For comparison, Fig. 5 (A and B) summarizes the strength, toughness, elastic modulus, and fracture strain values of nanofiller reinforced composites. The composites’ properties were normalized by being divided by the corresponding properties of the matrix. After being tailored with graphene at the interface, the B4C-NW composites exhibited outstanding enhancement in strength, toughness, elastic modulus, and fracture strain by 65.6, 1083.2, 15.2, and 378.4%, respectively. B4C-NWs@graphene composites have an exceptional combination of strength and toughness compared with other composites reported in literature (3044).

Fig. 5 Mechanical performance of B4C-NWs@graphene composites.

(A and B) Comparison of mechanical properties of 0.3 vol % B4C-NWs@graphene composites with other typical nanofiller reinforced composites [derived from (3044)]. (C) Comparison of flexural strength, elastic modulus, and fracture strain for pure epoxy and B4C-NWs@graphene reinforced composites. (D) Load transfer efficiency versus density chart showing that the B4C-NWs@graphene composite had exceptional interface properties [mechanical properties of 1D nanofiller reinforced composites were derived from (3093)]. CNT, carbon nanotube.

1D nanofiller reinforced composites have, for decades, been explored, including polymer, natural material, ceramic, and metal-reinforced composites (Fig. 5D) (3093). However, poor load transfer efficiency, to a certain extent, worsened their mechanical properties. To evaluate load transfer efficiencies, the Cox-Krenchel model (94) was applied as followsEcomposite=ηeffνEf+(1ν)Em(4)where Ecomposite is the elastic modulus of composite measured by the three-point bending and ηeff is the effective load transfer efficiency factor, involving the filler orientation factor. It turned out that 0.2 vol % composites achieved their respective highest effective efficiencies, which were calculated to be 39.7, 31.0, and 92.5% for B4C-NW, graphene, and B4C-NWs@graphene composites, respectively. Impressively, tailoring the composite interfaces with graphene enabled effective utilization of the nanofillers, resulting in two times increase in load transfer efficiency, specifically from 39.7% (blue star mark in Fig. 5D) to 92.5% (red star mark in Fig. 5D).

Molecular dynamics (MD) simulations were carried out to unveil how graphene sheets edited the B4C-NW surface, how graphene facilitated the dispersion of B4C-NWs, and how graphene enhanced the load transfer in the composites. The initial atomic configuration consists of an individual B4C-NW with a diameter of 10 Å and three monolayered graphene sheets with the dimension of 50 Å by 50 Å (fig. S7). The interaction between B4C and graphene is only described by van der Waals forces [Lennard-Jones (LJ) potential]. The MD simulation results unveil that the B4C-NW remains stable while being blended and wrapped by graphene sheets. Similarly, self-assembled wrapping process was observed with the initial atomic structure consisting of an individual B4C-NW with a diameter of 100 Å and three monolayered graphene sheets with the dimension of 400 Å by 400 Å (fig. S8). After the nanowire was fully wrapped up with graphene, the excess graphene sheets were absorbed in an edge-to-edge mode and/or partially folded into bilayer graphene with closed edges, generating a hybrid structure in equilibrium (movies S1 and S2). Without B4C-NWs, graphene sheets alone aggregate and form multilayered graphene, which is energetically favorable (fig. S7).

The interaction energy profiles between the nanofillers (B4C-NWs, multilayered graphene sheets, and B4C-NWs@graphene) in the aggregation process are calculated throughϕinteraction(d)=[ϕxx(d)2ϕx]/n(5)where ϕinteraction is the normalized interaction energy between two nanofillers, ϕx is the total potential energy of an individual nanofiller (graphene, B4C-NW, or B4C-NWs@graphene), ϕxx is the total potential energy of a system with two nanofillers (graphene/graphene, B4C-NW/B4C-NW, or B4C-NWs@graphene/B4C-NWs@graphene), d is the separation distance between the center of mass of two nanofillers, and n is the total atom number in the system. The liquid/solid interface friction is described (Fig. 6A) as (95)F=μSv(6)where μ is the friction coefficient and can be determined from the liquid-graphene interaction (96). S represents the surface area of the nanofiller. v is the velocity of liquid flow at the interface, which was calculated according to the aggregation rate of nanofillers. Considering different liquids, various friction coefficients were selected in the range from ~103 to ~105 Ns/m3 (9698), and the corresponding friction per atom was in the interval of 1.06 × 10−19 to 1.06 × 10−17 eV/Å. As shown in Fig. 6B, with decreasing the distance between two nanofillers, no obvious energy barriers are observed in both the interaction energy profiles of B4C-NW and graphene, indicating that both graphene sheets and B4C-NWs tend to agglomerate. In the B4C-NWs@graphene system, with decreasing the distance between two B4C-NWs@graphene fillers, the interaction energy suddenly increases at a certain position and gradually reaches the maximum of 85.78 ± 2.53 eV/atom. The maximum interaction energy between two B4C-NWs@graphene fillers presented little changes with different friction coefficients (table S1) because the frictions are negligibly small, approximately 15 orders of magnitude lower than the average force exerted on each atom by its neighbors (fig. S9). The high energy barrier is enabled by the hybrid structure with the coexistence of graphene sheets and B4C-NWs and by the severe deformation of the graphene sheets in the aggregation process (fig. S10). In the process of finding the equilibrium position, graphene sheets and B4C-NWs move or deform as a unit, and simultaneously interlock each other. Armed with experimental observations, the MD simulations uncover that graphene-tailored B4C-NWs notably enhance the interaction energy barrier, making aggregation difficult and thus largely improving the dispersion performance.

Fig. 6 MD simulations of the nanofiller interactions.

(A) MD snapshots of the initial structure (B4C-NWs@graphene/B4C-NWs@graphene) for calculating the interaction energy. (B) Interaction energy profiles between two nanofillers of the same type (graphene/graphene, B4C-NW/B4C-NW, and B4C-NWs@graphene/B4C-NWs@graphene).

MD simulations were performed to investigate the pullout process of the nanofillers from the epoxy matrix (fig. S11), and the interaction energy was calculated to evaluate the adhesive strength between the nanofillers and matrix. The normalized interaction energy (per atom) had a maximum value of 0.71 kcal/mol when the B4C-NW was fully embedded in the epoxy matrix and gradually decreased to zero at 70 Å displacement when completely pulled out. After being tailored with graphene, the maximum interaction energy per atom between B4C-NWs@graphene and epoxy was substantially increased to 1.86 kcal/mol, 162.0% higher than that of B4C-NW (fig. S11C). Accordingly, pullout force was also monitored during the entire process. To enable the nanofiller sliding at early stage, the pullout force had a rapid increase in the peak value with an approximate linear relationship with the increase in displacement (fig. S11D) and gradually decreased until full separation. The peak force per atom over B4C-NWs@graphene was calculated to be 121.54 pN, 18.5% higher than that over B4C-NW (102.56 pN). In summary, B4C-NWs@graphene had higher interaction energy with epoxy and larger pullout peak force because graphene rendered the nanofiller with higher surface area, larger number of interacting atoms, and complex geometries and thereby enhanced the interfacial strength and load transfer efficiency of the composites.

In summary, graphene sheets were used to tailor the interface between B4C-NW and epoxy. The B4C-NWs@graphene were directly synthesized by shear mixing the mixture of graphite powders and B4C-NWs in dilute water. The as-obtained B4C-NWs@graphene suspension exhibited homogeneous dispersion in both water and epoxy and enhanced load transfer efficiency from the matrix to reinforcements, leading to the overall improved mechanical performance of the composites. In addition, B4C-NWs@graphene enabled hybrid toughening effects in the epoxy matrix via crack pinning and deflection, debonding of B4C-NWs@graphene from matrix, void formation around the nanofillers, and nanofiller pullout. The 0.2 vol % B4C-NWs@graphene composite exhibited an exceptional combination of mechanical properties in terms of flexural strength (144.2 MPa), elastic modulus (3.5 GPa), and fracture strain (15.0%). This low-cost yet effective technique presents unprecedented opportunities for improving nanocomposite interfaces, enabling high load transfer efficiency, and opens up a new path for developing strong and tough nanocomposites. The graphene wrapping technique may find applications in medicine such as pharmacology and drug delivery in which graphene can be wrapped onto nanoparticles to compromise efflux pumps and overcome drug resistance.


Synthesis of B4C-NWs

All chemicals were purchased from Sigma-Aldrich without further treatment. Cotton was used to provide a carbon source to synthesize B4C-NWs on the surface of carbon fibers via the VLS method. A Ni-B emulsion was prepared by mixing 7 g of Ni(NO3)2 6H2O (catalyst) and 4 g of amorphous boron powders (boron source) into 10 ml of ethanol under ultrasonic vibration. The carbon fiber cloths were cut into small pieces, immersed in the Ni-B emulsion, and dried at 70°C in a preheated oven for 3 hours. The B4C-NWs were synthesized in a horizontal alumina tube furnace (diameter, 60 mm; length, 790 mm). The boron-loaded carbon fiber cloth, sandwiched by two pieces of cotton textile, was placed in the middle of the tube furnace and heated up to 1160°C and held for 2 hours with 300–standard cubic centimeter continuous flow of argon. The as-synthesized samples were simply treated by ultrasonication to peel off B4C-NWs.

Synthesis of B4C-NWs@graphene

B4C-NWs@graphene were simply synthesized by shear mixing. B4C-NW powders (0.5 g) and graphite powders (1.0 g; 99.9% purity) were mixed into 200 ml of H2O and shear mixed at 3000 rpm at room temperature for 3 hours with a L5M-A shear mixer. The as-obtained suspension of B4C-NWs@graphene in dilute water was kept standing for 6 hours. After removing the solid residues at the bottom, the suspension was further treated by freeze drying for composite fabrication. To quantify the conversion efficiency of the nanofillers, the productivity was calculated by the weight of the nanofillers remaining in the upper liquid. The nanofiller powders were achieved by drying 10 ml of the upper suspension and weighed by a high-accuracy scale. The total weight in the 200-ml suspension was calculated by multiplying the one in 10-ml suspension by 20. For comparison, graphene alone was fabricated using the shear mixing method following the same procedure. The graphene and B4C-NWs were separately used for fabricating graphene epoxy composites and B4C-NW epoxy composites as control samples.

Epoxy composite fabrication

The as-obtained fillers (graphene, B4C-NWs, and B4C-NWs@graphene) were dispersed into epoxy resin (EPO-TEK, 302-3M, Epoxy Technology Inc.) to fabricate their epoxy composites, with the reinforcement content varying from 0.1 to 0.3 vol %. The blend was sonicated for 5 min to degas and then cured for 24 hours at 60°C. Pure epoxy resin, as a control sample, was prepared following the same procedure. The pure epoxy and composites were cut and polished into three-point bending specimens of 40 mm by 7 mm by 5 mm (length by width by height). Three-point bending tests were carried out using an ADMET eXpert 2600 tensile universal testing machine with a deflection speed of 5 mm/min.

Materials characterization

The morphology, structure, and composition of the as-synthesized samples were characterized by scanning electron microscopy (Quanta 650), HRTEM (FEI Titan 80, equipped with EELS), XRD [PANalytical X’Pert Pro Multipurpose Diffractometer equipped with Cu Kα radiation (λ = 0.15406 nm)], XPS (ULVAC-PHI Inc.), and inVia Raman microscopy (Raman, Renishaw; at the wavelength of 514 nm).

MD simulations

MD simulations were carried out using the Large-scale Atomic/Molecular Massively Parallel Simulator (99). B4C has a rhombohedral unit cell (R3¯m space group, a = 5.16 Å, and α = 65.7°) consisting of 12-atom icosahedra and 3-atom linear chains (24, 28). The primary unit cell of graphene is extracted as the monolayer of graphite with hexagonal crystallography structure (P63mc space group and a = b = 2.47 Å). The initial atomic configurations were replicated, truncated, and combined to serve as the initial input structure (B4C rod with a diameter of 10 Å and monolayered graphene sheet with the dimension of 50 Å by 50 Å) for MD simulations. The dangling bonds at the graphene edge were decorated with hydrogen atoms to avoid covalent bonding between two graphene sheets (100). The atomic configuration with larger dimensions (B4C rod with a diameter of 100 Å and monolayered graphene sheet with the dimension of 400 Å by 400 Å) was built to investigate the size effect of the initial input structure on the wrapping process of graphene.

Periodic boundary conditions were applied in all directions. Boron carbide interactions were described by the reactive force field (101), and graphene was modeled by the adaptive intermolecular reactive empirical bond order potential (102). The nonbond interactions were described by the LJ potential. According to the LJ parameters for graphite interlayer binding (103) and graphene/nanoparticle interactions (104), the LJ parameters in this study were selected as ε = 2 meV and σ = 3 Å, which are in the appropriate range for these typical materials. The canonical ensemble (N, V, and T) was applied to relax the structure with a time step of 0.25 fs, and the constant temperature was controlled by a Nose-Hoover thermostat method. Simulations temperature was set to be 10 K to ignore the thermal effect. After 500,000 MD steps for reaching equilibrium, the as-obtained structures were replicated into two structures (graphene/graphene, B4C-NW/B4C-NW, or B4C-NWs@graphene/B4C-NWs@graphene). The initial distance between the two structures was set to be 100 Å, and the angle was set to be 60°. By decreasing the distance, the interaction energy between two structures was calculated accordingly. In the pullout simulations, the epoxy system was constructed with EPON 826 (epoxy resin) and diethyltoluenediamine (DETDA, curing agent) monomer polymer consistent force field (105).


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgment: We thank the staff members at the University of Virginia NMCF for electron microscopy technical support. Funding: Financial support for this study was provided by the NSF (CMMI-1537021). Author contributions: N.S. and X.L. conceived the research. N.S. contributed to this work in the experimental planning, experimental measurements, MD simulations, data analysis, and manuscript preparation. Z.G. assisted the fabrication of graphene. X.L. contributed to the experimental planning, data analysis, and manuscript preparation. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article