Bioinspired, graphene-enabled Ni composites with high strength and toughness

See allHide authors and affiliations

Science Advances  31 May 2019:
Vol. 5, no. 5, eaav5577
DOI: 10.1126/sciadv.aav5577


Nature’s wisdom resides in achieving a joint enhancement of strength and toughness by constructing intelligent, hierarchical architectures from extremely limited resources. A representative example is nacre, in which a brick-and-mortar structure enables a confluence of toughening mechanisms on multiple length scales. The result is an outstanding combination of strength and toughness which is hardly achieved by engineering materials. Here, a bioinspired Ni/Ni3C composite with nacre-like, brick-and-mortar structure was constructed from Ni powders and graphene sheets. This composite achieved a 73% increase in strength with only a 28% compromise on ductility, leading to a notable improvement in toughness. The graphene-derived Ni-Ti-Al/Ni3C composite retained high hardness up to 1000°C. The present study unveiled a method to smartly use 2D materials to fabricate high-performance metal matrix composites with brick-and-mortar structure through interfacial reactions and, furthermore, created an opportunity of developing advanced Ni-C–based alloys for high-temperature environments.


A joint enhancement of strength and toughness is a vital requirement for next-generation structural materials. Unfortunately, this pursuit often ends with a compromise between hardness and ductility (1). Such a dilemma originates from the fact that the size of the plastic deformation zone in front of the crack tip, which works to dissipate local stress, is inversely proportional to the yield strength (2). Moreover, in most engineering materials, once fracture is initiated, cracks propagate rapidly without any shielding behind the crack tip (3). The wisdom of nature addresses this conflict by constructing materials with a hierarchical architecture, all the while using only limited materials and nontoxic processes (4). A common example is nacre, or mother-of-pearl. As one of the most well-known natural armors, nacre is endowed by a brick-and-mortar structure composed of aragonite (a mineral form of CaCO3) platelets and biopolymer (5). Acting as the major load bearers, aragonite platelets (the bricks) that are 5 to 10 μm in length and 0.5 μm in thickness are constructed by nanocrystals (5). The biopolymer (the mortar), which is only several nanometers in thickness, closely binds the aragonite platelets together (6). Such a complex structure enables multiple extrinsic toughening mechanisms on different length scales, leading to an outstanding combination of strength and toughness that is hardly seen in engineered materials. The layer-by-layer architecture redirects the crack growth into a tortuous path, effectively consuming fracture energy via extension of the crack length and reduction in stress concentration (7). In addition, mineral bridges shield the crack opening (8), while biopolymer layers dissipate fracture energy (9). In the meantime, surface nanoasperities interlock the aragonite platelets, preventing large-scale delamination (10). While entirely mimicking the reinforcing multitude of these scale levels is difficult, the hierarchical architecture may hold the key to suppressing the dilemma between strength and toughness. Therefore many researchers have gained inspiration from these biodesigns for high-performance composite materials. Similar to nacre, ceramic composites with additional soft phases such as polymers or metals have shown that brittle ceramic materials can be converted into tough materials via architecture design (1118). Bioinspired polymer composites with added hard ceramic platelets exhibited high strength, outperforming most engineering polymers, while retaining ductility (1923). These studies are nothing short of remarkable; the toughness of such composites was magnitudes higher than the simple mixture of constituents. However, the intrinsically low ductility of ceramics and the low strength of polymers limit the overall potential mechanical performance. Moreover, weak bonding between hard phases and soft phases may also lead to interface delamination. Therefore, it can be expected that cloning nacre’s architecture with stronger constituents such as metals in engineered composites is a more promising, as well as a more challenging, task.

Previously, ceramics and intermetallic compounds have been used as hard phases in constructing metal-based composites with brick-and-mortar structure, which exhibited notable mechanical properties (2427). Graphene, a single layer of carbon atoms with sp2 bonds, is considered an ideal reinforcing agent for metal matrix composites because of its two-dimensional (2D) morphology, ultrahigh elastic modulus of 1 TPa, and high strength of 140 GPa (28). So far, graphene has been composited with metals such as Al, Cu, and Ni for the construction of laminated structures (2934). However, the agglomeration and degradation of graphene sheets, as well as the poor bonding between graphene and the metal matrix, or the occurance of unexpected reactions, resulted in a much lower than expected reinforcement efficiency. Therefore, unlocking the true potential of graphene in metal composites is still unresolved. An alternative approach is to use the unique morphological features of graphene, inducing interfacial reactions to form nacre-like, brick-and-mortar architecture in composites (29). Ni and Ni alloys are widely used in different applications, especially in high-temperature, extreme environments such as combustion engines or turbine blades because of their outstanding mechanical performance and stability (35). Considering the substantial significance of developing advanced Ni alloys with superior properties, several studies have attempted to composite graphene with Ni, but the high solubility of carbon in Ni and the tendency of forming coarse Ni3C particles (36) required electrochemical deposition (32), spark plasma sintering (33), and/or laser sintering (34) as the only methods to sinter Ni/graphene composites. The obtained composites, although exhibiting superior hardness, are unlikely to be mass-produced. Therefore, the question remains: Can we fabricate graphene-enabled, high-performance Ni matrix composites with nacre-like, brick-and-mortar structure via feasible and scalable procedures?

Here, a graphene-derived Ni/Ni3C composite with a characteristic bioinspired, brick-and-mortar architecture was fabricated by conventional powder metallurgy. Ni powders were homogeneously coated with graphene by shear mixing and freeze drying. At high temperature, carbon dissolved into Ni, facilitating the sintering process. Subsequently, part of the carbon atoms reacted with Ni, forming Ni3C second-phase particles along grain boundaries. The Ni3C second-phase particles were deformable and aligned into thin, long stripes, forming a brick-and-mortar structure via cold rolling. Another portion of carbon remained in the Ni matrix as interstitial solid solution atoms. The Ni3C platelets served as major load bearers and strengthened the composite, while the Ni matrix ensured ductility. Because of the confluence of strengthening and toughening mechanisms, the fabricated composite exhibited a 73% improvement in strength and only a 28% reduction in ductility, leading to a notable enhancement of toughness. By adding 2 weight % (wt %) of Ti and Al, the graphene-derived Ni-Ti-Al/Ni3C composite exhibited a high hardness up to 1000°C, indicative of possible advanced Ni-based superalloys. The 2D material–enabled powder processing can be applied to different material combinations, creating unlimited possibilities for new metal matrix composites.


Fabrication of graphene/Ni powders

Shear mixing was used to produce few-layered graphene and coat the graphene sheets on Ni particles simultaneously, largely improving the production efficiency. Specifically, 1.5 g of graphite was dispersed in 200 ml of dilute water and shear-mixed at 3000 rpm for 1 hour. The relationships between the shear mixing rate, duration, and the amount of graphene production were thoroughly studied by Paton et al. (37). After shear mixing, the coarse, undefoliated graphite powders were subsided and removed, leaving few-layered graphene in the liquid. Under the current conditions, the weight of the fabricated graphene decreased from 1.5-g graphite to about 0.09 to 0.12 g. The obtained graphene sheets exhibited thin, flexible morphology (fig. S1A) with an almost intact crystal structure (fig. S1, B and C). Without further treatment, the graphene-containing suspension was shear-mixed with 5-g Ni powders with irregular shape for 2 hours, corresponding to about 2 wt % and about 10 atomic % (at %) of carbon. Subsequently, the Ni/graphene powders were freeze-dried for 6 hours. After these procedures, scanning electron microscopy (SEM) observation showed that the graphene sheets were closely coated over the Ni powders without noticeable aggregation (Fig. 1A). Transmission electron microscopy (TEM) inspection unveiled that the graphene had closely coated the Ni powders without free space (Fig. 1B). The thickness of the graphene coating layer ranged from 10 to 15 nm, equaling to 20 to 30 atomic layers. An in situ heating observation revealed that the graphene sheets gradually dissolved into the Ni particle even without any compression (Fig. 1C), which is indicative of a very intimate bonding between graphene and Ni. Ni powders with 4 and 6 wt % graphene were also prepared as references. Both shear mixing and freeze drying are critical for homogeneously coating graphene on Ni powders. Without shear mixing, graphene sheets were loosely attached on the Ni powders and could not permeate into the small gaps or crevices on the particle surface (fig. S2A). Without freeze drying, graphene sheets agglomerated into coarse graphite particles (fig. S2B). The method can be feasibly adapted to other materials and may find more applications in various fields.

Fig. 1 Ni/graphene powders after shear mixing and freeze drying.

(A) SEM image of Ni/graphene powders, showing no noticeable aggregation of graphene sheets. (B) TEM image of the surface of a Ni/graphene powder, showing that few-layered graphene closely coated around the Ni particle. (C) In situ heating observation of a Ni/graphene powder. Graphene gradually dissolved into Ni with increasing temperature.

Microstructure and mechanical performance of graphene-enabled Ni/Ni3C composite

The Ni particles with 2 wt % graphene produced by shear mixing and freeze drying were compressed in a round mold that was 25 mm in diameter to a pressure of 80 MPa and then sintered at 1450°C under the protection of argon, slightly lower than the melting point of Ni (1455°C). After sintering, the Ni/graphene powders were melted together, showing no evidence of pulverization or fracture (fig. S3A, inset). After chemical etching, the Ni/graphene powder–derived ingot exhibited clean, well-defined grain boundaries, indicative of the formation of grain boundary precipitates (fig. S3A). Ni powders without graphene and with 4 and 6 wt % of graphene were also compressed and sintered under the same condition as references. The Ni powder–derived ingot displayed no obvious grain boundaries after chemical etching (fig. S3B). Sintering enabled the Ni powders with 4 wt % graphene to form a coin, which was broken into pieces after minor deformation, indicative of weak bonding between particles (fig. S4A). The Ni powders with 6 wt % of graphene could not be sintered together (fig. S4B). Close-up inspection of the sintered sample with 4 wt % graphene showed a discontinuous microstructure with isolated particles (fig. S4C). Energy-dispersive x-ray spectroscopy (EDS) carbon map uncovered that carbon segregated on the particle surfaces (fig. S4D). According to the Ni-C phase diagram, the highest solubility of carbon in nickel is 2.7 at %, and the eutectic reaction point is located at 10 at % of carbon (36), which corresponds to about 2 wt %. When 4 wt % graphene (or 20 at %) was composited with Ni, some carbon remained on the Ni particle surfaces even though part of the Ni powders melted. The excessive carbon left on the Ni particle surfaces prohibited the sintering of Ni powders.

Cold rolling with a 40% deformation reduction in thickness was first applied to the sintered coins. The grain boundary precipitates were broken into long, thin stripes, which gradually aligned along the rolling direction during deformation. After the deformation reduction in thickness was increased to 80%, the boundary precipitates were aligned in a parallel fashion, forming a brick-and-mortar architecture (Fig. 2A) [RD (rolling direction), TD (transverse direction), and ND (normal direction)]. The fracture surface also exhibited a laminated feature with elongated dimples (Fig. 2B), which had the same shape as the second-phase particles in Fig. 2A. TEM inspection unveiled large second-phase particles embedded within the matrix (Fig. 2C). The dislocation density within the second-phase particles was lower than that of the Ni matrix. In addition, the grain size and shape of the Ni matrix were different near and beyond the second-phase particles. Close-up inspection of the Ni matrix exhibited stripe-like grains with a thickness ranging from 100 to 300 nm, a typical cold-rolled microstructure (Fig. 2D). Two-beam dark-field imaging revealed a high concentration of dislocations, and their migration was prohibited (Fig. 2E), indicative of the existence of precipitates or solid solution atoms. In contrast, Ni grains near the large second-phase particle were smaller in size with an equiaxed morphology. The difference in grain size and shape may derive from higher deformation energy, which may stimulate dynamic recrystallization. High-resolution TEM (HRTEM) imaging of the boundary showed no noticeable defects such as voids or cracks (Fig. 2G), indicating that the cold deformation did not break the bonding between the precipitates and the matrix. Instead, there was an amorphous transition zone between the Ni matrix and the second-phase particles, suggesting that the second-phase particles precipitated out from the Ni matrix.

Fig. 2 Microstructure of the graphene-enabled Ni/Ni3C composite.

(A) SEM image of cold-rolled Ni/Ni3C composite, showing brick-and-mortar structures. (B) Fracture surface of Ni/Ni3C composite, showing laminated structure constructed by elongated dimples. (C) Low-magnification TEM image, showing a large second-phase particle embedded in the Ni matrix. (D) After cold rolling, Ni grains were deformed into long stripes with the thickness ranging from 100 to 300 nm. (E) Two-beam diffraction dark-field image of the Ni matrix, showing a high concentration of dislocations. (F) Close-up observation of the Ni/Ni3C boundary. (G) HRTEM image of the interface between Ni and a second-phase particle, revealing a transition zone. (H) Ni3C crystal on [−110] plane. (I) HRTEM image of the [−110] plane of Ni3C particle, showing identical atomic arrangement as in the Fig. 2H. (J) Schematic illustration of the formation of Ni/Ni3C composite with a brick-and-mortar structure.

A critical question then arises: What is the second-phase particle? The x-ray diffraction (XRD) spectrum of the Ni/graphene-derived composite exhibited a weak peak at about 43° (fig. S5A), which was similar to the Ni3C (002) peak illustrated in (36). To determine the composition of the second-phase particles, EDS analysis and HRTEM observation were used. EDS spectra showed the existence of Ni and carbon with a Ni/C ratio of about 3:1 (fig. S5B). Therefore, we can likely assume that the second-phase particle is Ni3C. Ni3C is a stable nickel/carbon compound below 300°C. It has a hexagonal close-packed structure with lattice parameters of a = 0.26 nm and c = 0.43 nm. The unit cell of Ni3C plotted by Materials Studios is shown in fig. S6A, in which Ni and C atoms are arranged layer by layer. According to the crystal structure, on the [−110] plane of the Ni3C particle, Ni atoms should exhibit visual distances of 0.23 and 0.26 nm (fig. S6B and Fig. 2H) under HRTEM. The HRTEM image (Fig. 2I) and corresponding fast Fourier transform (FFT) pattern (fig. S6C) exactly matched the calculation. Therefore, by coupling the results from XRD, EDS, and HRTEM, the large second-phase particles are conclusively determined to be Ni3C. A close [01–1]//[−110] orientation relation between Ni and Ni3C can be derived from Fig. 2 (G and I), which may lead to the weak XRD peaks. Armed with solid experimental results, we can conclude the formation mechanism of the bioinspired Ni/Ni3C composite with brick-and-mortar structure. As illustrated in Fig. 2J, Ni powders were closely wrapped by graphene after shear mixing and freeze drying. The graphene sheets dissolved into the Ni matrix and precipitated out as Ni3C in the sintering and cooling. The Ni3C second-phase particles were broken and elongated into long strips during cold deformation, forming the brick-and-mortar structure.

Encouragingly, this graphene-derived Ni/Ni3C composite with brick-and-mortar structure exhibited outstanding mechanical performance. The tensile specimens were fabricated into a dog bone shape with a sample size demonstrated in Fig. 3A. Tensile tests were carried out at a strain rate of 0.06 mm/min. The strain was measured by Correlated Solutions’s Vic-2D digital image correlation system with one photo every 2 s. Three tests were conducted on each group of samples. In a typical tensile test, the Ni/Ni3C composite showed a yield strength of 780 GPa and an ultimate tensile strength of 1095 GPa (Fig. 3A). This strength is comparable to the strongest Ni-based alloys (38). Although the strength had an obvious improvement, the ductility only exhibited a minor decrease. The Ni/Ni3C composite has a joint enhancement of mechanical properties: A 73% improvement in strength, a 6% increase in Young’s modulus, and an 82.3% enhancement of hardness with only a 28% reduction in ductility (Fig. 3B). Such outstanding combination between strength and ductility resulted in a 44% increase in toughness (the area under the strength/strain curve) than the pure Ni reference sample, indicating that the bioinspired, brick-and-mortar architecture effectively mitigated the conflict between strength and toughness. So far, a large number of bioinspired, nacre-like composites including ceramic-based (diamond marks in Fig. 3C) (1118), polymer-based (round marks in Fig. 3C) (1923), and metal-based (square marks in Fig. 3C) (2427, 2934) composites have been developed. Freeze casting has been widely used to construct laminated ceramic composites (1113); polymer cross-linking, on the other hand, is the primary route to fabricate layer-by-layer, polymer-based composites (20, 21); compressing sintering, powder processing, electrochemical deposition, and laser sintering have been used to synthesize nacre-like, metal-based composites (2934). The soft phases (polymers and metals) with a volume fraction ranging from 5 to 40% in the nacre-like, ceramic composites effectively improved the ductility of the composites, with a trade-off of reduction in the characteristically high strength of ceramics (diamond marks in Fig. 3C). The hard phase (mainly ceramic flakes) in nacre-like, polymer-based composites prohibited the decoiling of polymer chains, which notably enhanced the strength of the polymer matrices (round marks in Fig. 3C). In the nacre-like, metal-based composites, the volume fraction of metal matrix, which is normally considered as the soft phase, is often more than 70%. Hard phases with platelet-like morphology, such as graphene, ceramics, and intermetallic compounds, were homogeneously dispersed in the metal matrices, facilitating a joint enhancement of strength and toughness (square marks in Fig. 3C). Because of the intrinsically high strength of Ni and the constructed brick-and-mortar architecture, the graphene-enabled Ni/Ni3C composite outperformed most other nacre-like composites in terms of the combination of yield strength and ductility (star mark in Fig. 3C).

Fig. 3 Mechanical properties of graphene-enabled Ni/Ni3C composite with a brick-and-mortar structure.

(A) Tensile stress-strain curves of Ni, Ni produced by powder metallurgy, and Ni/Ni3C composite (inset shows the size of tensile specimen). (B) Comparative bar chart of mechanical properties of Ni and Ni/Ni3C composite. (C) Elongation versus yield strength plot showing that the as-fabricated Ni/Ni3C composite had an outstanding combination of strength and ductility (mechanical properties of nacre-like composites were derived from (1127, 2934).

Stiffening, strengthening, and toughening mechanisms

It is important to understand the stiffening, strengthening, and toughening mechanisms of the graphene-derived Ni/Ni3C composite. The Young’s modulus of the composite was slightly higher than that of pure Ni (Fig. 3B). Apparently, the enhancement of Young’s modulus derives from the Ni3C platelets. A volume fraction of 13.3% of the Ni3C platelets was calculated from five low-magnification SEM images on ND/TD and ND/RD planes, respectively. Low-load nanoindentation tests were carried out to identify the mechanical properties of the Ni3C platelets. As shown in a typical nanoindentation displacement-load curve, the Ni3C platelets exhibited higher hardness and reduced modulus (Fig. 4A). On average, the Ni3C platelet has a hardness of 6.5 GPa (3.4 GPa higher than that of Ni matrix) and an elastic modulus of 364 GPa (154 GPa higher than Ni). The hardness (Fig. 4B) and reduced modulus (Fig. 4C) maps derived from nanoindentations exhibited an alternating hard-soft-hard structure with the hard part as the Ni3C platelets. Because of the linear nature of elastic deformation and the intimate bonding between Ni matrix and Ni3C platelets, the enhancement of Young’s modulus can be estimated by the rule of mixturesE=E1ν1+E2ν2(1)where E is the modulus of the composite, E1 is the modulus of Ni matrix (210 GPa), ν1 is the volume fraction of Ni (86.7%), E2 is the modulus of Ni3C platelets (364 GPa), and ν2 is the volume fraction of Ni3C platelets (13.3%). The theoretical Young’s modulus of the composite was calculated to be 230.5 GPa, which is close to the average value of Young’s modulus obtained from tensile tests (222 GPa).

Fig. 4 Strengthening and toughening mechanisms of graphene-derived Ni/Ni3C composite with brick-and-mortar structure.

(A) Nanoindentation load-displacement curves of Ni and Ni3C platelet. (B) Hardness map derived from nanoindentation tests. (C) Reduced modulus map derived from nanoindentation tests. (D) Finite element simulation of the Ni/Ni3C composite under tension. (E) APT map of Ni and C atom distribution. (F) APT map of C atom distribution. (G) In situ tensile test with strain map. (H) In situ three-point bending test under SEM.

Subsequently, we attempted to unveil the strengthening mechanisms of the graphene-enabled Ni/Ni3C composite. According to the tensile stress-strain curves in Fig. 3A, the yield strength of Ni/Ni3C composite had a 330-MPa improvement comparing with that of the Ni sample produced by powder metallurgy. Apparently, the Ni3C platelets contributed to the improvement of yield strength. On the basis of the experimental results, a finite element model (FEM) was constructed (Fig. 4D and fig. S7A). When a tensile stress of 600 MPa was exerted along the x axis, the FEM simulation showed a high concentration of stress on the platelets, indicating that the platelets acted as the load bearers and effectively strengthened the composite. The strengthening effect from Ni3C platelets can be validated by the rule of mixtures because we can assume that the Ni3C platelets deformed synchronously with the Ni matrix during elastic deformation. On the basis of the tensile test results, the Ni/Ni3C composite started yielding at a strain of about 0.3%. The improved yield strength was then calculated by the equationΔσsec=εE2ν2(2)where ε is the strain at yield point (0.3%). The improvement of yield strength from Ni3C platelets was calculated to be 145.2 MPa.

The hardness derived from nanoindentation on the Ni sample produced by powder metallurgy was about 2.2 GPa, which was 0.85 GPa lower than that of the Ni matrix in the Ni/Ni3C composite (fig. S7B). Therefore, the Ni matrix should be strengthened by other mechanisms. The second source of strengthening may derive from the grain boundaries, i.e., the grain size. The grain boundaries act as pinning points to impede dislocation propagation. The relationship between grain size and strength can be demonstrated by the Hall-Patch equation (39, 40)Δσy=σ0+kyd(3)where σy is the yield strength, σ0 is a materials constant for the starting stress for dislocation movement, ky is the strengthening coefficient, and d is the grain size. Therefore, the change of the yield strength due to the reduction of grain size should beΔσgb=kyd1kyd2(4)

On the basis of TEM inspections (Fig. 2D and fig. S7C), the average grain size along the ND of the Ni/Ni3C composite was 198 nm (d1), and that of the Ni sample produced by powder metallurgy was 543 nm (d2). The finer grain size after adding graphene may originate from the formation of second-phase particles, which can prohibit recrystallization and grain growth. The ky was measured to be 4.9 MPa mm1/2 (40). Thus, the strength enhancement derived from grain boundaries was 138.3 MPa.

The overall enhancement from the Ni3C platelets and grain boundaries was 283.5 MPa, at least 66.5 MPa less than the experimental results. Because the deformation was the same for all the samples, the remaining mechanisms were precipitate strengthening and solution atom strengthening. A rational hypothesis is that carbon dissolved into the Ni matrix and then precipitated out as atom clusters and/or second-phase particles, which pinned the migration of dislocations and strengthened the Ni matrix. A direct evidence of the precipitates should be the weak secondary patterns that appear in selected-area electron diffraction (SAED) patterns. However, the SAED pattern of the Ni matrix along both [011] and [112] directions showed no other patterns except for the Ni (fig. S8), eliminating the possibility of nanosized precipitates or atom clusters. Thus, solid-solution strengthening had the highest likelihood for improving the yield strength. Figure 4 (E and F) shows atom probe tomography (APT) maps, showing homogeneously dispersed carbon atoms in the Ni matrix. The atomic percentage of carbon in Ni matrix was about 1 at %. The strengthening effect of interstitial solution atoms originates from the pinning of dislocation due to lattice distortion. The strength contribution can be expressed as (2)Δσss=kc(5)where k is a parameter related to shear modulus and lattice distortion and c is the concentration of interstitial solution atoms. For body-centered cubic metals, such as Fe, carbon atoms can generate a nonsymmetrical stress field, which strongly interacts with dislocations, leading to a large k of 5G (G is the shear modulus). However, the smaller lattice distortion and symmetric stress field stemmed from carbon interstitial atoms in face-centered cubic Ni have much weaker pinning effects on dislocations, making the k only G/10 (2). The shear modulus of Ni is 72 GPa, and the concentration of carbon was 1 at % based on the APT result. The theoretical yield strength enhancement contributed by carbon interstitial atoms was calculated to be 72 MPa. Therefore, the Ni/Ni3C composite is triply strengthened by Ni3C platelets, grain boundaries, and carbon interstitial solution atoms. The theoretical improvement of yield strength was calculated as followsΔσ=Δσsec+Δσgb+Δσss=355.5 MPa(6)which is close to the experimentally derived improvement of yield strength.

In engineering alloys, the increase of strength derived from interstitial atoms and second-phase particles usually trades off with a reduction in ductility. Especially large, brittle carbides often introduce defects, which may become the source of cracks. However, the graphene-derived Ni/Ni3C composite exhibited an obvious plastic deformation stage and a higher static toughness than the pure Ni (Fig. 3, A and B). An in situ tensile test was carried out under SEM to determine the influence of Ni3C platelets on crack formation and propagation during deformation. A speckled coating created by a mix of conductive silver glue and carbon black was coated on the sample to trace the evolution of strain via digital image correlation. A small notch was also made on the sample edge. As shown in Fig. 4G, no apparent strain concentration was found at a low displacement of 0.05 mm. With an increase of the exerted load, a crack started at the artificial notch. The crack propagated along 45o against the loading direction, a typical fracture of ductile metals. Therefore, the Ni3C platelets likely did not act as the fracture source to stimulate cracking. Close-up observation of the crack initiation near the artificial notch showed that multiple small cracks appeared before the major crack propagated (fig. S9), indicating that, in addition to the intrinsic toughening mechanisms, such as fine grain size and parallelly aligned grain orientation, the brick-and-mortar structure may introduce extrinsic toughening mechanisms to further improve the toughness. To reveal the possible extrinsic toughening mechanisms, an in situ three-point bending test was carried out (Fig. 4H). Although an artificial notch was made, the crack initiated at a large deflection, indicative of outstanding ductility. After the crack initiated, instead of propagating perpendicular toward another side of the three-point bending plate, the crack was gradually deviated to be parallel to the three-point bending sample length direction by the parallelly aligned Ni3C platelets (fig. S10), which shifted the crack mode, leading to lower effective stress around the crack tip and higher difficulty for crack opening. Moreover, the interlacing Ni3C platelets resulted in a zigzag morphology of crack edges and formation of small cracks near the primary crack both in-plane and out-of-plane, which inevitably elongated the crack length. The energy required to propagate the crack, Ws, is related to the crack length by (29)Ws=2abγ(7)where a is the crack length, b is the out-of-plane thickness of the solid material, and γ is the sum of surface energy (γs) and energy related to plastic deformation (γp). Apparently, the longer the crack length is, the higher the toughness. In addition to the crack deflection, metal bridges appeared behind the crack tip, and the layer-by-layer structure blunted the crack tip, which further prohibited the crack opening and propagating (fig. S10A). Thus, the Ni3C/Ni brick-and-mortar structure contributed to the improvement of toughness. Worth mentioning is that the inspection of the chemically etched sample showed that the Ni3C platelets in fact deformed with the Ni matrix (fig. S10B). This result demonstrated that the Ni3C platelets are truly ductile; they can deform with the matrix without inducing notable cracks. This answered two essential questions: (i) why the Ni3C grain boundary precipitates formed the brick-and-mortar structure after cold working without introducing large defects and cracks and (ii) why the coarse Ni3C platelets did not induce fracture during deformation.

Microstructure and high-temperature hardness of Ni-Ti-Al/Ni3C composite

Ni alloys are widely used in high-temperature environments because of their outstanding stability and creep resistance. To verify the impact of the Ni3C platelets and carbon interstitial solution atoms on the high-temperature performance, 2 wt % Ti and 2 wt % Al were added to the Ni/graphene powders and sintered together. The obtained Ni-Ti-Al/Ni3C composite also exhibited a brick-and-mortar structure (Fig. 5A) and stripe-like grains (Fig. 5B) after cold rolling. The atomic-resolution EDS analysis of Ni, Ti, and Al (Fig. 5, C to E) showed that Ti and Al had concentrated in some grains, forming a laminated structure at the submicrometer scale. The distribution of carbon was homogeneous (Fig. 5F), which was consistent with the APT map of the Ni/Ni3C composite (Fig. 4F). High-temperature Vickers hardness tests were carried on pure Ni, Ni/Ni3C composite, Ni-Ti-Al/Ni3C composite, and commercial HR-224 superalloy. At room temperature, the Ni/Ni3C composite and Ni-Ti-Al/Ni3C composite exhibited a high hardness of 3.7 and 4.6 GPa, respectively. Alloying the Ni/Ni3C with Ti and Al further increased the strength. The Ni/Ni3C composite maintained a high hardness from room temperature to 300°C. After 300°C, the hardness decreased rapidly. The rapid decrease of hardness may be mainly due to the failure of the pinning effect of interstitial solution carbon atoms on the dislocations. In comparison, the Ni sample showed a constant decrease in hardness with increasing temperature; the Ni-Ti-Al/Ni3C composite and HR-224 superalloy showed almost no hardness reduction up to 500°C. When the temperature further increased, the hardness of Ni-Ti-Al/Ni3C composite decreased gradually and remained 1.8 GPa at 1000°C, which is higher than that of HR-224 superalloy, which dropped down to 1.11 GPa. Close-up inspections of the centers of Vickers indentations on the Ni-Ti-Al/Ni3C composite at room temperature and 1000°C are shown in Fig. 5 (H and I, respectively). At room temperature, the indentation surface was relatively smooth (Fig. 5H). High-temperature (1000°C) indentation impression showed an oxidized surface with large, irregular particles (Fig. 5I). Therefore, although the alloy recipes and heat treatments require further studies, the proposed Ni/Ni3C composite is promising as a basis for next-generation superalloys.

Fig. 5 Microstructure of Ni-Ti-Al/Ni3C composite and high-temperature Vickers hardness of Ni, graphene-derived Ni/Ni3C composite, Ni-Ti-Al/Ni3C composite, and HR-224 superalloy.

(A) SEM image of Ni-Ti-Al/Ni3C composite after chemical etching. (B) High-angle annular dark-field (HAADF) image of the Ni-Ti-Al/Ni3C composite. (C to F) High-resolution EDS of Ni, Ti, Al, and C maps. (G) Hardness values from high-temperature Vickers hardness tests. (H) Room temperature Vickers hardness indentation impression on Ni-Ti-Al/Ni3C composite (the edge length of the inset image is 180 μm). (I) High-temperature (1000°C) Vickers hardness indentation impression on Ni-Ti-Al/Ni3C composite (the edge length of the inset image is 180 μm).


In summary, a prototypical graphene-derived Ni/Ni3C composite with bioinspired, brick-and-mortar structure was developed. Graphene closely wrapped the Ni powders via shear mixing and freeze drying. The Ni/graphene powders were compressed and sintered at 1450°C, forming Ni3C at the grain boundary area. The Ni3C platelets were deformable. They were rolled into long stripes during the cold deformation, leading to the formation of a brick-and-mortar structure. Additional carbon atoms were dissolved into the Ni matrix and existed as interstitial solution atoms. The Ni3C platelets not only acted as the load bearers but also redirected crack propagation. The small grains and interstitial solution atoms prohibited the dislocation propagation and enhanced the Ni matrix. In total, the confluence of multiple strengthening and toughening mechanisms enabled a 73% increase on strength and a 6% increase on Young’s modulus, with only 28% reduction in ductility, leading to a 44% improvement in toughness. The Ni-Ti-Al/Ni3C composite exhibited superior strength over commercial superalloys up to 1000°C. This strategy presents a new promise for the design and synthesis of advanced bioinspired materials to achieve exceptionally high mechanical robustness for applications in an extensive range of fields.


Graphite and Ni powders were purchased from Sigma-Aldrich Company without further purification. Graphene was prepared using the shear mixing method. Specifically, 1.5 g of graphite powders with purity of 99.9% was added into 200 ml of H2O and then shear-mixed at 3000 rpm at room temperature for 1 hour by a Silverson L5M-A shear mixer. Subsequently, the suspension stood for 2 hours, and the large particles deposited at the bottom were filtrated for reuse. The upper, transparent liquid contained about 0.09 to 0.12 g of graphene. Ni powders (5 g) with 99% purity were then shear-mixed with the graphene sheets for 2 hours. After shear mixing, the Ni/graphene powders were collected and lyophilized for 6 hours. The obtained powders were compressed in a round mold with a diameter of 25 mm at 80 MPa and then sintered at 1450°C for 1 hour with the protection of Ar gas. After sintering, the coin was rolled at room temperature for a reduction in thickness of 80%. For comparison, five reference samples were fabricated. For the first reference sample, a pure nickel plate (99% purity) was purchased from ESPI Metals and cold-rolled with the same thickness reduction as used for the Ni/Ni3C composite plates. For the second reference sample, Ni powders were shear-mixed, freeze-dried, compressed, and sintered without adding graphene. The sample was cold-rolled with the same thickness reduction as used for the Ni/Ni3C composite plates. For the third and fourth reference samples, Ni powders were composited with 4 and 6 wt % graphene, respectively, and hybrid powders were then compressed and sintered under the same conditions. The last control sample was a commercial HR-224 superalloy.

Tensile testing was carried out on an Admet eXpert 2600 tensile universal testing machine with an extension speed of 0.06 mm/min. Strain was measured using the Vic-2D digital image correlation system from Correlated Solutions with one photo every 2 s. Specimens for tensile tests were prepared in a dog bone shape, and the sample size is demonstrated in Fig. 3A. In situ tensile tests and three-point bending tests were carried out on an MTI Instruments SEMTester 1000 tensile stage. High-temperature hardness testing was performed using a Bruker UMT-3 tribometer with a peak indentation load of 30 N for 30-s holding. Nanoindentation tests were carried out using a MicroMaterials Vantage nanoindenter with a load of 200 and 10 mN. The nanoindenter was a diamond Berkovich tip whose shape function was carefully calibrated. The elastic modulus of the diamond tip was 1140 GPa, and the Poisson’s ratio was 0.07. XRD patterns were obtained using PANalytical X’Pert Pro MPD equipped with Cu Kα radiation (λ = 0.15406 nm). The microstructure of the specimens was characterized with a FEI Quanta 650 SEM with an EDS detector and a FEI Titan G2 aberration-corrected TEM. TEM specimens for cross-sectional imaging were cut using a Helios dual-beam focused ion beam. APT experiments were carried on a LEAP5000XS.

Finite element simulation was performed on ANSYS student edition. The model size was 25 μm by 25 μm by 125 μm (W × H × L). The constituent particle size was determined on the basis of the SEM images, ranging from 20 to 70 μm in length and approximately 2 to 3 μm in thickness, with a standard width of 5 μm. The volume fraction of the constituent particles in the model was 10%. Boundary conditions were assigned as fixed support at one end and distributed face load applied at the other. Loading was a ramped load uniformly over 30 s to maximum of 0.375 N (600 MPa) along a positive x direction. The unit cell of Ni3C (R3¯c space group, a=4.553 Å, and c=12.92 Å) was constructed via density functional theory using the Cambridge Serial Total Energy Package. The generalized gradient approximation–Perdew-Burke-Ernzerhof was selected to describe the exchange and correlation energy. The Broyden-Fletcher-Goldfarb-Shanno minimizer was used to perform cell optimization, and the convergence tolerance of total energy was set to be 1 × 10−6 eV/atom.


Supplementary material for this article is available at

Fig. S1. Graphene sheets produced by shear mixing.

Fig. S2. Microstructure of Ni/graphene particles without shear mixing or without freeze drying.

Fig. S3. SEM images of sintered Ni/graphene and Ni sample produced by powder metallurgy after chemical etching.

Fig. S4. Ni with 4 and 6 wt % graphene after sintering.

Fig. S5. XRD and EDS spectra of Ni/Ni3C composite and Ni produced by powder metallurgy.

Fig. S6. Crystal structure and FFT pattern of Ni3C particle.

Fig. S7. Strengthening of Ni/Ni3C composite.

Fig. S8. SAED patterns of the Ni matrix.

Fig. S9. Close-up observation of the crack initiation at the artificial notch.

Fig. S10. Close-up inspection of the crack propagation of the three-point bending Ni/Ni3C sample.

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.


Acknowledgments: We thank the staff members at the University of Virginia NMCF and North Carolina State University AIF for electron microscopy technical support. Funding: Financial support for this study was provided by the U.S. National Science Foundation (CMMI-1537021). Author contributions: Y.Z. and X.L. designed the experiment. Y.Z. fabricated the composite and carried out microstructural/mechanical tests. Y.Z. and F.M.H. conducted the in situ tensile and bending tests. F.M.H. performed digital image correlation analysis. J.L.B. constructed finite element analysis models. N.S. analyzed crystal structure of Ni3C. D.I. carried out APT tests. Y.Z. and X.L. wrote the article. All the authors proofread the manuscript. Competing interests: X.L., Y.Z., F.M.H., J.L.B., and N.S. are inventors on a U.S. provisional patent application related to this worked filed by the University of Virginia (no. 62/817,142, filed 12 March 2019). The authors declare no other 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