Copper/graphene composites: a review

The composite powders can be prepared by simple mixing techniques including mechanical stirring, magnetic stirring, sonication and vortex mixing [28, 29, 32, 47, 49, 51, 52, 55, 57, 60, 63, 65, 66, 68‐70, 73, 103, 104]. However, high-energy processes such as ball milling (BM) or mechanical alloying (MA) have been also employed [12, 27, 28, 36, 38, 40, 53, 60, 61, 64, 101, 102]. Mechanical alloying is the solid-state processing of powder materials which is often used to produce alloys and composites that are difficult to obtain from conventional melting and casting techniques [1]. The process of MA starts with mixing graphene with the metallic powders in the desired proportion and then loading the powder mix into a mill (shaker mill, planetary mill or attritor) along with the grinding medium (generally steel balls) [116]. The mix is milled for the desired length of time, usually in a protective atmosphere to prevent Cu oxidation. During mixing, the impacted powders undergo repeating fracture, deformation and welding processes, which leads to the intimate mixing of the constituent powder particles on an atomic scale [1]. The total milling energy can be tailored by varying the charge ratio (the ratio of the weight of balls to the powder), ball mill design, milling atmosphere, time, speed and temperature. In certain cases, a process control agent (PCA), such as stearic acid or petroleum ether, is added to the powder mixtures to prevent excessive sticking and agglomeration of Cu powders during milling [12, 28, 36, 60]. The PCA adsorbs on the surface of the powder particles and minimises cold welding between impacted particles, thereby preventing agglomeration [1]. Moreover, mixing techniques such as mechanical stirring, magnetic stirring and sonication and, occasionally, BM are performed in certain organic solvents (e.g. ethanol, acetone, etc.), which hinders the agglomeration of graphene into clusters. The solvents must be then evaporated to obtain dry composite powders before compaction and/or consolidation. For this purpose, vacuum-drying, air-drying and rotary evaporation are commonly used, although other less common techniques such freeze–drying or vacuum infiltration have been also employed.

Mechanical alloying can produce composites with finer microstructures and a better distribution of graphene in the Cu matrix [1]. However, the processing steps must be handled with care in order to retain the structural integrity of graphene. Yue et al. [60] reported that with increase in BM time the size of the composite powders decreases and the dispersion of graphene improves, but the damage of the graphene intrinsic structure inevitably increases. Figure 1 shows SEM micrographs of Cu-0.5 wt% GO powders after BM for different times varying from 1 h to 7 h. It can be seen that the shape of the Cu/GO powders undergoes a change from flake-like to more granular morphology with increase in BM time due to the shearing effect of the balls (Fig. 1) [60].

Figure 2 displays Raman spectra of the composite powders after BM for different times [60]. These spectra show the typical D and G of the GO nanosheets located at around 1349 and 1595 cm⁻¹, respectively. It can be seen that the ratio of I_D/I_G increases from 0.84 to 1.42 with increase in BM time, indicating that the degree of damage of the GO increases with increase in BM time. Additionally, Cui et al. [27] found, for CMCs reinforced with GNPs, that the higher the milling speed, the higher the degree of exfoliation of GNPs. Nevertheless, the I_D/I_G values increased with increase in the milling speed, indicating that the degree of structural damage of graphene also increases with increase in the speed of BM. Thus, the BM conditions need to be balanced to obtain a uniform dispersion of fine graphene particles in Cu matrix while reducing structural damage to the graphene [60].

Realising that the most critical issues in processing graphene-reinforced MMCs are the dispersion of graphene and the interfacial bond strength between the graphene and the matrix, many researchers have adopted modified steps in their approach [2]. Gao et al. [47] coated Cu powders with hexadecyl trimethyl ammonium bromide (CTAB), a cationic surface agent, to obtain a positive surface charge. The results showed that GO, with a negative charge, is adsorbed on the surface of CTAB coated Cu powder, realising the homogeneous dispersion of graphene in the CMCs [47]. A schematic of the fabrication process of Cu/graphene composites following this approach is given in Fig. 3.

Most researchers have used sintering to consolidate the composite powders. In a few works, green compacts, generally prepared using a press or a testing machine, were sintered in a conventional [38, 52, 57, 63, 65, 73, 104] or microwave furnace [57]. The major advantage of the microwave sintering over conventional sintering is that it provides rapid heating, resulting in much finer grain sizes. A larger number of researchers have used hot pressing (HP) consolidation of powders or compacts [12, 36, 47, 52, 60, 63, 64, 102, 103]. This is a high-pressure consolidation technique working at a temperature high enough to induce sintering. It is conducted by placing either the composite powders or the composite compacts into a suitable die, typically graphite, and applying uniaxial pressure, while the entire system is held at an elevated temperature. So, by hot pressing, consolidation is achieved by the simultaneous application of heat and pressure. Spark plasma sintering (SPS), a comparatively new sintering technique, has also been explored [27, 29, 32, 40, 49, 51, 53, 55, 66, 68‐70, 101]. In this process, a pulsed direct current is passed through a graphite die where the powder mixtures or compacts are pressed uniaxially [1]. When a spark discharge appears at the contact point between the particles of a material, a local high-temperature condition is created, resulting in rapid heating and hence increasing the sintering rate [1], so that grain growth, graphene agglomeration and thermal decomposition of graphene can be minimised during consolidation [23]. Efficient densification can be achieved by applying a combination of spark impact pressure, joule heating and electrical field diffusion [32, 117].

Kim et al. [28] rolled composite powders to achieve a better density and distribution of graphene in a Cu matrix; the powders were balled milled, followed by encapsulation in a pure Cu tube and degasification, and then subjected to equal speed rolling (ESR) or conventional rolling and to high-ratio differential speed rolling (HRDSR). All the ESR- and HRDSR-processed Cu and Cu composites showed high densities between 98.8 and 99.4%, indicating that almost full densification was obtained after rolling.

The electrodeposition technique is an easy, cost-effective and scalable method to fabricate Cu/graphene composite coatings [72, 112‐114] and foils or films [19‐21, 24, 31, 33, 48]. In addition, electrodeposition being a low temperature process preserves the properties of graphene during the preparation of the composites, unlike in the conventional sintering processes, which may damage graphene because they may involve temperatures higher than its decomposition temperature (> 600 °C) [31]. Electrodeposition takes place from a dispersion of graphene in an electrolytic bath consisting of copper sulphate as a source of Cu²⁺ ions, the graphene content in the Cu/graphene composites depending on the amount of dispersed graphene in the bath. To disperse graphene sheets uniformly into the electrolyte is one of the main challenges to synthesise graphene enhanced nanocomposites by electrodeposition [72]. Stirring [24, 31, 33, 48, 114] can be used to keep graphene in suspension during electrodeposition. Additions of anionic or polymeric surfactants have also been used to improve the wettability of the substrate to be coated and to prevent agglomeration [31, 113, 114]. These additions may, however, introduce heterogeneous impurities, that weaken the interfacial bonding of graphene sheets and matrix, adversely affecting the mechanical and physical properties of the composite coatings. As an alternative, Mai et al. [72] proposed a surfactant-free colloidal solution comprising copper (II)-ethylene diamine tetra acetic acid ([Cu^IIEDTA]²⁻) complexes and GO sheets to prepare Cu/RGO composites. The anionic complexes stably coexist with negatively charged GO sheets due to the electrostatic repulsion between them, facilitating the electrochemical reduction and the uniform dispersion of RGO sheets into the Cu matrix.

Both direct and pulse reverse current have been used by Pavithra et al. [31, 48] for electrodeposition of Cu/graphene nanocomposite films (Fig. 4). Pulse reverse (PR) is advantageous over direct current (DC) electrodeposition because it allows the optimisation of several key processing parameters including applied current, pulse duration and duty cycle that enables a smooth, highly dense, uniform deposit, while minimising hydrogen embrittlement. This in turn improves the properties of the deposited material. The forward pulse restricts the mass transfer and hence controls the grain size, whereas the reverse pulse minimises the dendritic morphology and helps in the removal of extended graphene and loosely adsorbed Cu or graphene, in addition to removal of entrapped hydrogen during each pulse. Furthermore, PR electrodeposition facilitates a uniform distribution of graphene sheets into the Cu matrix, where they spread around the grain boundaries to achieve an improved interface with the Cu throughout the composite.

In the case of DC electrodeposition, the deposition is rapid at the most active nucleation sites and, due to the continuous application of current, the continuous incorporation of graphene along with the Cu deposition results in a rough surface with graphene clusters in the matrix.

An electroless plating process consisting in situ chemical or thermal reduction has been used to manufacture graphene-metal nanoparticles (MNPs) hybrids [29, 32, 35, 50, 55, 64, 66, 105‐107, 110, 111] or sandwich-like 2D Cu/RGO nanocomposites composed of continuous Cu layers on both sides of the central RGO [108]. Copper-nanoparticle/graphene composite powders fabricated by this technique were further consolidated by SPS to obtain bulk Cu/graphene composites [35] or used as such for different applications [105‐107, 110, 111]. Graphene decorated with other metallic nanoparticles such as Ag or Ni was also fabricated and afterwards successfully introduced as fillers into Cu matrices by processing techniques such as PM routes or molecular level mixing (MLM) in order decrease the contact angle of Cu on graphene and thus to improve the wettability between graphene and the Cu matrix [29, 32, 50, 55, 64, 66]. The fabrication of graphene-MNPs hybrids (Fig. 5) usually consists of the in situ nucleation of MNPs on the graphene sheets by reducing a mixture of GO and metallic ions. Metal ions prefer to nucleate at the sites of functional groups. For this reason, when GNPs are used as precursor materials, they are sensitised and activated before being decorated with the metallic particles [50, 55, 66].

Another simple, but usually multi-step electroless plating technique, molecular level mixing (MLM), has been used to fabricate Cu/graphene composite powders which are subsequently consolidated by SPS [23, 34, 39, 43, 50, 54, 67, 74, 109]. A schematic diagram of a fabrication process of Cu/RGO nanocomposites by MLM is given in Fig. 6 [23]. Firstly, GO and Cu ions are homogeneously mixed in deionised water. Chemical bonds are then formed between the functional groups of GO and the Cu ions. Finally, Cu/GO nanocomposites are thermally reduced in H₂, the as-reduced Cu/RGO composite powders being subsequently consolidated by SPS. However, additional steps involving the generation of copper oxides (CuO and Cu₂O) as intermediate products are usually required. Graphene-MNPs hybrids or GNPs can be also used as raw material [34, 43, 50, 54, 67]. However, since formation of Cu–O–C chemical bonds (whose origin is in the reaction between the carboxyl or hydroxyl groups and Cu) plays a major role in the adsorption of graphene on the Cu surface, GNPs are usually sensitised and activated by a hydrochloric acid solution of SnCl₂ and PdCl₂, respectively, beforehand.

Figure 7 displays the evolution of the nanocomposite powders during the MLM and SPS process as proposed by Hwang et al. [23]. Figure 7a shows an atomic force microscopy (AFM) image of GO fabricated by the Hummers method. Figure 7b shows that after mixing the GO and Cu salts, the GO layer was not agglomerated and was homogeneously mixed with the Cu ions. After oxidation, GO particles were fully covered with ellipsoidal CuO particles of about 500 nm in size (Fig. 7c). The Cu/RGO nanocomposite powders obtained by H₂ thermal treatment of Cu/CuO powders are shown in Fig. 7d. CuO particles that were formed on GO were reduced to form islands with average size of 30 nm, while CuO particles that were formed without GO were reduced to form large Cu particles and connected to each other during the thermal treatment. The fine size of Cu particles on the RGO originated from the difficulty of Cu diffusion on the surface of RGOs. After consolidation by SPS, the RGO layers were dispersed homogeneously in the Cu matrix without further agglomeration (Fig. 7e). The Raman spectra in Fig. 7f illustrate the evolution of defects in GO and RGO. The I_D/I_G ratio increased from 0.78 for GO to 0.81 for Cu²⁺/GO, indicating an increase in defects in the GO structure after mixing with Cu ions. Graphene oxide with Cu ions could be more defective because the interaction of the Cu ions with the GO surface could damage the sp² bonding network of the graphene further. The continuous, conformal coating of CuO on the GO flakes immediately after the oxidation process blocked the characteristic Raman signals of GO (i.e. D and G bands) from the CuO/GO samples. The I_D/I_G ratio of the Cu/RGO nanocomposite powders was markedly lower (i.e. 0.40) because the reduction process removed functional groups and partially recovered the graphene structure.

It is important to note that the reaction products, as well as their morphology, in a MLM process strongly depend on the reaction conditions, with the pH and the reaction temperature being the most critical factors. Yang et al. [71] carried out experiments at different pH values of 5.9, 6.6, 12.8 and 13.6, respectively. XRD patterns of the as-collected samples are given in Fig. 8a. It can be seen that the phase constitution of the composite powders is pH-sensitive. When the pH value is lower than 6.6, the diffraction peaks are assigned to the crystal planes of Cu₂(OH)₃Ac. However, once the pH value is increased to 12.8, the major diffraction peaks match well with Cu(OH)₂ and CuO. Figure 8b, c shows the morphology change of the composite powders at carious pH values as indicated by SEM analysis. As shown in Fig. 8b, Cu₂(OH)₃Ac is in the form of sheets of about 5 μm in size. When pH value is increased to 12.8 and 13.6, the Cu₂(OH)₃Ac sheets transform into Cu(OH)₂ and CuO nanofibers (Fig. 8c). The nanofibers are uniformly dispersed on GO sheets with diameters of a few tens of nanometres. Most importantly, the edges of GO sheets can be easily observed and are distributed almost parallel to each other, which can be considered the ideal framework of micro-layered composites with nacre-inspired architecture [74]. Nacre is a natural inorganic/organic composite material that gains its toughness from a microstructure that consists of sheets of calcium carbonate separated by layers of elastic biopolymers.

A modified MLM process, comprising the self-assembly, reduction and consolidation of CuO/GO/CuO or 2D Cu/CuO sandwich-like nanosheets, has been employed to fabricate multilayer Cu/graphene composites with a nacre-inspired architecture [45, 71, 74]. This process leads simultaneously to a uniform dispersion and high alignment of graphene in the metal matrices. An example of such processing technique is shown schematically in Fig. 9 [45]. First, GO is synthesised from natural graphite flakes by a modified Hummers method. In order to assist the dispersion of GO in aqueous media and direct the deposition of CuO on the surface of GO, surfactant sodium dodecyl sulphate (SDS) was chosen to adsorb electrostatically and self-assemble onto the surface of the GO. Cu cations were bound to the surfactant assembled onto the GO, forming CuO/GO/CuO sandwich-like nanosheets in alkaline solution with the decomposition of the added urea in at elevated temperature. CuO was deposited on both sides of the GO (Fig. 9a) and prevented them from restacking. Subsequently, the bottom-up assembly of CuO/GO/CuO sandwich-like nanosheets was carried out by vacuum filtering the parent solution (Fig. 9b). Afterwards, by reducing the assembled CuO/GO/CuO films (Fig. 9c), Cu/RGO/Cu films were achieved. Finally, these films were stacked and consolidated by HP to produce bulk nano-laminated composites (Fig. 9d).

This novel technique was employed by Xiong et al. [41] to fabricate a nature-inspired CMC, where RGO was chosen as “brick” because of its inherent 2D geometry and good mechanical properties and Cu was used as mortar. The entire process consisted of three steps: replication of the ordered porous structure of fir wood with Cu, absorption of RGO into the porous Cu preform and HP compaction. Fir wood has a highly ordered layer porous structure, in which pores are in rectangular shape with an average size of ~ 20 × 30 μm and a wall thickness of 1.5 μm. The authors applied a chemical route comprising copper oxide replication and subsequent reduction to replicate the porous structure of fir wood with Cu.

Cold spraying (CS) is a relatively new technique in which the composite powders are accelerated to very high velocities (~ 500–1200 ms⁻¹) at low temperature and impacted on a substrate [1, 2]. During the process, powders are accelerated by injection into a stream of a gas in a converging diverging de-Laval type nozzle. The gas is heated, without using combustion, only to increase the gas and particle velocity [123]. The particles are in solid state when they impact the surface, where they undergo severe plastic deformation. The high kinetic energy upon impact ensures good adhesion of the particles on the substrate. Since the temperature of the process is below the melting point, oxidation and phase transformations can be avoided. Cold spraying in conjunction with ball milling has been shown to be successful in the fabrication of Cu coatings, where non-agglomerated and uniformly distributed GNPs were embedded [115].

Accumulative roll bonding (ARB) is a severe plastic deformation technique consisting of multiple cycles of cutting, stacking and roll bonding [124]. Hence, large strains can be accumulated in the material and significant structural refinement can be achieved [125]. As a result, good mechanical properties at low and high temperatures have been observed for different metals and alloys over their coarse-grained counterparts [125]. In addition, several researchers have also used ARB for the successful fabrication of particle reinforced MMCs [126‐129]. So, by manually distributing SiC or Al₂O₃ particles between the two metallic strips prior to each roll-bonding steps, Al or Cu matrix composites with excellent distributions of reinforcing particles, good interfacial bonding and no porosity were obtained after several ARB cycles. More recently, Liu et al. [37] adopted a similar approach to fabricate GNPs reinforced CMCs. In this way, pure Cu was ARB up to eight cycles at RT, a GNPs dispersion being sprayed on the surface of the Cu strips before each rolling step.

The simplest model that can be used to predict the mechanical properties of MMCs is the rule of mixtures (ROM), in which the desired property can be estimated from the weighted average of the individual components as follows [7, 15]:where β is the property of interest (Young’s modulus or yield strength), V is the volume fraction, and the subscripts c, m and r refer to the composite, matrix and reinforcement, respectively. Limitations to the ROM have resulted in models which take into account additional phase parameters other than the content.

The Cox shear-lag model, for example, assumes a perfectly bonded interface, so that the applied stress is transferred from the matrix to the fibre through interfacial shear stress [1, 2, 7]. According to the modified shear-lag model [131], the Young’s modulus or yield stress of a randomly distributed graphene composite is expressed as [32, 54]:where p is the aspect ratio of the graphene reinforcement.

The Halpin–Tsai model, however, considers not only the reinforcement aspect ratio, but also its spatial distribution [132]. Considering a random or a unidirectional distribution of graphene into the matrix, the Halpin–Tsai model is expressed by the following empirical equations [12]:where the subscripts random and || refer to the composites with randomly oriented and unidirectionally distributed graphene, respectively, and η_L and η_T are parameters defined by:

The incorporation of graphene into Cu can lead to grain refinement of the matrix phase [12, 22, 26, 35, 37, 43, 55, 61, 67, 69, 73]. The dependency of yield stress (σ_y) on grain size (D) generally follows the Hall–Petch relationship [133, 134]:where σ₀ is the friction stress and K is the Hall–Petch slope, which is associated with a measure of the resistance to dislocation motion caused by the presence of grain boundaries.

Grain refinement in Cu/graphene composites has been ascribed to an acceleration of the BM process by graphene and oxide particles, obtaining much smaller particles [36, 38, 61], and to the pinning effect of graphene or carbides on the grain boundaries during the consolidation processes [12, 35, 37, 38, 41, 43, 53, 55, 58, 61, 69, 73].

Graphene itself can also impede dislocation motion during mechanical tests. Assuming that graphene is not sheared by dislocations, the flow stress would be then controlled by the stress required to bend graphene particles and subsequent form loops around them, as proposed by Orowan [135]. The following expression could be used to calculate the Orowan increment of the yield stress [136]:where G is the shear modulus of the matrix, b is the magnitude of the Burgers vector, λ is the effective planar interparticle spacing, ν is the Poisson’s ratio of the matrix, d_p is the mean planar diameter of the particles, and r₀ is the core radius of the dislocations in the matrix.

Generally, the Orowan looping mechanism is more pronounced in MMCs reinforced with particles of low aspect ratio [1]. So, in principle, its contribution to the strengthening of MMCs reinforced with graphene it is expected to be little. Moreover, to play an important role in Orowan strengthening, graphene should be finely dispersed within the grains, because particles in grain boundaries are not expected to effectively impede the movement of dislocations in grain interiors [28]. This is quite challenging because, in MMCs, graphene has a tendency to distribute along the grain boundaries in most of the fabrication routes, [22, 47, 52, 60, 103]. For example, in a composite produced by MLM followed by SPS a small amount of spherical-shape GNPs, RGO or Ni-plated GNPs were observed within the Cu grain interiors [50, 54]. However, only after the combination of BM and HRDSR, nanosized graphene particles were densely and uniformly dispersed in the grain interiors, attributable to the large shear stress introduced during the rolling process [28].

When uniform dispersions of fine graphene particles are not achieved, graphene can also act, due its two-dimensional geometry, as an effective obstacle for dislocation motion [23, 29, 31, 32, 37, 44, 49, 51, 56, 65, 69, 103]. As a result, dislocations are at the grain boundaries, piled up at the interface region under loading causing an enhancement of the yield stress. This has been proved to be the main strengthening mechanism in nanolayered composites consisting of alternating layers of Cu and monolayer graphene, where graphene acts as an efficient barrier to dislocation propagation and provides a strengthening effect which can reach far beyond the simple ROM prediction [25]. As a result of the gliding dislocations being blocked by the metal–graphene interface, ultra-high flow stresses were observed for the Cu/graphene multilayers, these increasing systematically with a reduction in metal layer spacing (Fig. 17a). The flow stresses at 5% plastic strain for the Cu/graphene nanopillars were extracted and plotted against the corresponding metal layer spacing (Fig. 17b). The slope of the log–log is − 0.402, which is in close agreement with the Hall–Petch exponent, σ ∝ h^−1/2, where h is the repeat layer thickness. This finding suggests multiple dislocation pile-up at the interface, consistent with the studies on metal nanolayered composites that demonstrate Hall–Petch-like behaviour.

Since immobilised dislocations also hinder the movement other dislocations, the enhancement of the yield strength in MMCs has been also related to an increase in dislocation density in the matrix (Δρ_dis) originated mainly from different thermal contractions [43, 54, 55, 60, 67, 103], but also from the additional plastic deformation induced by the presence of reinforcing phases during processing [43]. The dependence of the yield stress of composites upon of dislocation density in the matrix can be expressed as [137]:where a is a constant.

The significant mismatch of coefficient of thermal expansion (CTE) between the Cu matrix (~ 24 × 10⁻⁶ K⁻¹ at RT) and graphene (− 6 × 10⁻⁶ K⁻¹ in-plane at RT) causes a residual plastic strain during processing, thereby generating dislocations at the interface whose density is given by [138]:where A is a geometrical constant, ΔC is the value of CTE mismatch between graphene and the Cu matrix, and ΔT is the temperature change.

The accumulation of dislocations at the interfaces after yielding, caused by the blocking effect of graphene, leads to an enhancement of the work hardening and thus of the maximum strength and hardness compared with the unreinforced materials [74]. This is because the dislocations pin each other and form tangles so that an increase in stress is required to continue plastic deformation. The presence of geometrical constraints also generates additional dislocations in the matrix during the mechanical tests [51]. Hence, during deformation of a ductile matrix containing a dispersion of hard particles, continued plastic flow necessitates the formation of dislocations in order to avoid the void formation [15]. The density of geometrically necessary dislocations (GNDs) is given by [139]:where γ is the shear strain, λ is the geometric slip distance, and b is the Burgers vector.

The strengthening efficiency R, defined as the ratio of the amount of yield strength increase in the composite to that of the matrix by the addition of reinforcing materials, can be expressed as [29, 39, 54, 60]:where σ_y,c and σ_y,m are the yield stress of the composite and the matrix, respectively.

The reinforcing phase content of graphene in the available works is presented either in volume fraction or weight fraction. The relation between the volume fraction (V_r) and weight fraction (W_r) of the reinforcement of a composite is given by:where ρ_r and ρ_m are the density of the graphene reinforcement (2.2 g/cm³) and Cu matrix (8.96 g/cm³), respectively.

By using Eq. (12), the strengthening effect of various processing routes for Cu/graphene composites with different graphene derivatives can be compared. Moreover, the level of reinforcement imparted by graphene on pure Cu is usually calculated as the percentage increase in the yield stress compared to the matrix. However, in terms of the enhancement of mechanical properties upon the addition of graphene, it is also instructive to examine the relationship between the experimental data and the theoretical predictions. It is found that the experimental data generally lie close to the expectations only at very low graphene volume fractions (< 0.1%) and then fall away, especially above volume fractions of 1% as is found for both aluminium-matrix [140] and polymer-matrix systems [141]. There are a number of possible reasons why this might be the case:

Due to the excellent mechanical properties of graphene together with the changes induced in the matrix microstructure, it is expected that both strength and stiffening increase with increase in the graphene content. However, although the available results show that graphene is generally a good reinforcement for Cu, there exists an optimal loading with the mechanical performance deteriorating above this loading. This degradation of the mechanical properties has been mainly related to a poor stress transfer, mainly caused by aggregative trend of graphene and interface debonding. For example, Chu and Jia [12] prepared Cu/GNPs composites by a combination of BM and HP processing. Compared to the unreinforced Cu, the Cu/GNPs composites showed a remarkable increase in the yield stress and Young’s modulus up to 114% and 37% at 8 vol% GNPs content, respectively (Fig. 18a). This extraordinary reinforcement was attributed to the homogeneous dispersion attained by BM for 0–8 vol% GNPs contents and to grain refinement, the average grain size decreases from ~ 10 μm for the Cu matrix to ~ 4 μm for the Cu-8 vol% GNPs composite. However, as seen in Fig. 18a, with further increasing GNPs content up to 12 vol%, the increments of the yield strength and Young’s modulus dramatically decrease to 46% and 24%, respectively. The less effective enhancement in Cu-12 vol% GNPs composites arises mainly from the GNPs aggregations in the BMed powders. Moreover, the mechanical improvement of the Cu/GNPs composites was still below the theoretical value. So, as shown in Fig. 18b, it is obvious that the Young’s modulus measurements are lower than the predictions made by the Halpin–Tsai model, implying that there is still some room for further enhancement in the mechanical performance of the Cu/GNPs composites. The gap between the predictions and the experimental results was attributed by the authors to following three reason: the loss of intrinsic properties of GNPs due to the introduction of structural defects during the BM process, an insufficient interfacial bonding due to no-wetting of the GNPs and Cu and a reduction in the strengthening efficiency of the GNPs due to their random orientation.

Cu/GNPs composites were also prepared via MLM and SPS [43]. With the exception of ductility, the mechanical performance of Cu was improved evidently by the graphene addition. However, the strengthening effect was first enhanced and then deteriorated by increasing graphene content (Fig. 19). So, for pure Cu, the average yield strength is 142 MPa and fracture elongation is about 30%. With the increase in graphene content, the fracture elongation decreases from 30 to 3.5% (Fig. 19a). In contrast, the yield strength is first increased to 310 MPa, corresponding to an enhancement of 118%, at the graphene content of 0.6 vol% and then drops to 200 MPa when the graphene content further increases to 4.0 vol%. Similarly, both the elastic modulus and hardness of the composites increase to their maximum values before decrease beginning at the graphene content around 0.6–0.8 vol% (Fig. 19b). The highest elastic modulus and hardness obtained were 147 and 1.75 GPa, the increment compared with pure Cu (E ≈ 89 GPa, H ≈ 1.01 GPa) being 65% and 75%, respectively. The strengthening effect of graphene was attributed to a high dislocation density generated in the vicinity of GNPs due to the large thermal expansion mismatch between GNPs and Cu and to grain refinement. However, interface debonding was observed to take place in the composites during loading. So, the decline of mechanical performance for GNPs contents greater than 0.6–0.8 vol% could be attributed to poor interfacial bonding.

Chen et al. [44] fabricated Cu/graphene composites through in situ growth of graphene on flaky Cu powders and HP. Figure 20a shows the stress–strain curves of the composites with different graphene contents and of pure Cu. It is obvious that there is a marked improvement on the mechanical properties of the Cu/graphene composites, the strengthening effect of in situ grown graphene attributable to load transfer and the role of graphene as an obstacle to the propagation of dislocations during deformation. A yield strength of 144 MPa and a tensile strength of 274 MPa are achieved by the composite with 0.95 wt% graphene (graphene/Cu-2 composite), which are, respectively, a 177% and 27% enhancement over pure Cu. However, for a higher graphene content (graphene/Cu-3 composite), the mechanical properties fall to a lower level. The poor enhancement of graphene/Cu-3 was mainly attributed to a worse bonding between the Cu matrix and graphene. However, the Raman spectra of the composites showed an increased I_D/I_G ratio for such composite (Fig. 20b). So, its poor enhancement in mechanical properties could be also attributed to an increase in defects in the in situ grown graphene obtained.

GNPs and Cu powders were mixed by BM to produce composite powders using different milling speeds and times. Then, Cu-2.4 vol% GNPs bulk composites were fabricated by SPS [27]. The compressive yield strength and maximum compressive strength of the composites are shown in Fig. 21a. The yield strength of the composite fabricated at 100 rpm for 4 h is 376 MPa. However, the yield strength of the composites decreased from 376 MPa to 337 MPa as the milling speed increased from 100 rpm to 300 rpm. Moreover, for a rotating speed of 300 rpm, the yield strength decreased from 337 MPa to 325 MPa by increasing the milling time from 4 h to 8 h. This was attributed to the increase in the defect concentration, showed by the decrease of the I_D/I_G ratios in the Raman spectra (Fig. 21b) with increase in the milling speed and time.

The combination of BM followed by equal speed rolling (ESR) or high-rate differential speed rolling (HRDSR) was applied to fabricate 0.5 and 1 vol% Cu/GNPs composites [28]. Following the same procedures, two pure Cu sheets were also fabricated using BMed powders. All the materials exhibited similar grain sizes and grain boundary misorientations, indicating that neither the consolidation process nor the additions of GNPs contributed to grain size reduction. However, the results indicate that HRDSR process increases the efficiency of the addition of GNPs for strengthening and strain hardening.

Cu/GNP composites, as well as pure Cu, were fabricated by ARB at room temperature up to 8 cycles [37]. It was observed that the dispersion of GNPs and the interface bonding improves and the matrix grain size decreases with increase in the number of ARB cycles. In agreement, the tensile strength of the Cu/GNPs composites increased with increase in the number of cycles. After 6 ARB cycles, the tensile strength of the Cu/GNPs composites reached 496 MPa, which is higher than that of the annealed Cu by 275 MPa.

Cu metal matrix composites reinforced with varying amounts (0.9, 1.8, 2.7 and 3.6 vol%) of graphene particles were fabricated through powder metallurgy route by employing conventional and microwave sintering processes [57]. In both cases, it was found that with the addition of graphene, hardness increases as compared to pure Cu sample due to the superior mechanical properties of graphene. However, for the same graphene content, microwave sintered samples exhibited higher hardness compared to conventional counterparts. The highest hardness value of 89 ± 2.4 HV₁₀₀ was observed for the microwaved sintered Cu-3.6 vol% graphene and 82 ± 2.2 HV₁₀₀ for the conventional sintered one. This difference was attributed to the more rapid heating during microwave sintering resulting in a more refined and homogeneous microstructure.

An Ag-RGO hybrid was employed as reinforcement to prepare Cu-0.15 wt% graphene composites via BM followed by vacuum hot pressing sintering at 800 °C using a pressure of 30, 40, 50 or 60 MPa [64]. For comparison, pure Cu specimens were also fabricated at 60 MPa. Due to the good bonding interface between RGO and Cu (Fig. 22a), promoted by the Ag NPs, the micro-hardness of the Cu/Ag-RGO was higher than that of pure Cu and increased with increase in sintering pressure (Fig. 22b). In addition to micro-hardness measurements, nanoindentation [150, 151] can also be employed to follow the effect of the addition of graphene in, for example, metal foils [31].

Cu-2 wt% GNPs composites were fabricated by Ponraj et al. [104] through mixing the pure Cu and the GNPs powders using a mechanical stirrer, followed by compaction and conventional sintering. Two different Cu powders were used: spherical powders with an average size of 45 μm and dendritic powders with an average size of 70 μm. Moreover, they were subjected to BM for different times before being mixed with GNPs. It was found that, for the same starting Cu powder morphology, the hardness of the composites increases with increase in the milling time. This was attributed to a higher reduction in the Cu powder size and to a better dispersion of GNPs into the Cu matrix. It was also apparent that, for the same milling time, the hardness of the composites fabricated with dendritic Cu powders was higher than that of the composites fabricated with spherical Cu powders, suggesting that the morphology of the starting Cu powders also has an effect on the mechanical properties of the Cu/GNPs composites.

Cu matrix composites with a homogeneous dispersion of RGO sheets were successfully fabricated by a MLM method [109]. The composite powders were reduced in H₂ at 350, 450 and 550 °C and then consolidated by SPS. It was found that both the compressive yield stress and hardness increase with decrease in the reduction temperature. This has been related to an increase in the interface strength between graphene and copper caused by a lower degree of reduction in the functional groups in the graphene surface, which are required to form the oxygen-mediated bonding between Cu and graphene.

The effect of graphene structural defects on the mechanical behaviour of CMCs was also investigated by Li et al. [40]. Different amounts of RGO or high quality graphene (HQG) were mixed with copper powders by BM followed by SPS. The HQG was obtained from regular RGO by a hot pressing treatment. The hardness of both the Cu/RGO and Cu/HQG composites was higher than that of pure Cu. However, due to the absence of defects on the surface of the HQG, the hardness of the Cu/HQG composites is generally higher than that of the Cu/RGO composites.

Zhang and Zhan [54] used two kinds of graphene derivatives, namely GNPs and RGO, to fabricate Cu matrix composites through a MLM process followed by SPS. Neither Cu/GNPs nor Cu/RGO composites showed obvious grain refinement compared to pure Cu, and both GNPs and RGO were well bonded with the Cu matrix after sintering. RGO was more defective than GNPs. However, GNPs showed an obvious aggregative trend when the volume fraction was above 0.5%. Consequently, GNPs showed better strengthening efficiency at content below 0.5 vol%, while RGO performed better when the content increased from 0.5 to 1 vol% (Fig. 23).

In agreement with these results, the same authors observed in a different work [50] that the tensile strength of a Cu-0.5 vol% RGO composite increased by 22 MPa compared with a Cu-0.5 vol% GNPs one, both fabricated by MLM and SPS (Table 1). However, in this case they attributed the difference to the fact that the interface bonding between RGO and Cu was stronger than that of GNPs with Cu. In fact, a combination of mechanical and metallurgical bonding was observed between GNPs and the Cu matrix, while the interfacial adhesion between RGO and the copper matrix was oxygen-mediated chemical bonding.

Ni decorated graphene nanoplatelets (Ni–GNPs), consisting of well-dispersed Ni NPs strongly attached on GNPs, were synthesised by chemically reducing Ni ions on the surface of GNPs, which were then added to a Cu matrix to synthesise a Cu-0.8 vol% Ni/GNPs composite by sonication in ethyl alcohol and SPS [29]. For comparison, pure Cu specimens and a Cu/GNPs composite with 0.8 vol% GNPs were also fabricated under the same processing conditions. The Cu/Ni–GNPs composite exhibited a significant improvement in ultimate tensile strength (UTS), being 42% higher than that of the monolithic Cu (Fig. 24a). In contrast, the UTS of the Cu/GNPs composite was lower than that of the monolithic Cu (Fig. 24a). The significant strength enhancement of the first composite was attributed to the unique role of Ni NPs, which generate a good dispersion and strong Cu–GNPs bonding. Hence, more stress can be transferred to the GNPs during deformation. Figure 24b, c shows the representative microstructure of the Cu/GNPs and Cu/Ni–GNP composites. According to the SEM image of the Cu/GNP composite (Fig. 24b), the GNPs appear to be poorly dispersed in the Cu matrix forming aggregates in the surface. The authors claim [29] that the micrograph of the Cu/Ni–GNP composite (Fig. 24c) does not show any GNP aggregates although without any elemental mapping it is not possible to draw clear conclusions.

Zhang and Zhan [55] fabricated Cu-0.5 vol% GNPs by a PM route. Electroless Cu and Ni plating were firstly performed on the surface of the GNPs before mixing with Cu powders in order to improve their wettability. The yield strength of the composites is higher than that of pure Cu, which was attributed to grain refinement, an increase in dislocation density and better load transfer. The yield strength of the Cu/GNPs–Cu and Cu/GNPs–Ni composites is higher than that of the Cu/GNPs composites, attributable to more uniform dispersion of the GNPs in the Cu matrix and to a better interfacial bonding. However, since the bonding of graphene is higher to Ni than to Cu [90, 91, 93, 94, 97], the wettability of the GNPs–Ni is better than that of the GNPs–Cu. Therefore, the Cu/GNPs–Ni composite is stronger than the Cu/GNPs–Cu composite.

Copper matrix composites reinforced with carbide-coated GNPs were investigated in order to understand the role of the carbide interlayers on different properties of the Cu/GNPs composites [67]. Figure 25 presents the stress–strain curves of pure Cu and the Cu/0.5 wt% GNPs and Cu/0.5 wt% GNPs–TiC composites. The tensile strength of the two composites is higher than that of pure Cu, which was attributed to grain refinement, dislocation strengthening and load transfer. However, for the Cu/GNPs–TiC composite, the tensile strength was increased to 470 ± 7 MPa, which is 40% higher than that of the Cu/GNPs composite. This was attributed to an improvement of the interfacial properties and thus to a more efficient load transfer when reinforcing with Ti-coated GNPs than with bare GNPs.

The effect of graphene content upon electrical conductivity for W₇₀Cu₃₀/graphene composites fabricated by BM and LPS is shown in Fig. 27a [61]. It can be seen from the picture that, unlike hardness, which increases with increase in graphene content, the electrical conductivity tends to increase gradually first and decrease sharply then with increase in the graphene content. The electrical conductivity of the W₇₀Cu₃₀ alloy was 42% IACS, while that of the W₇₀Cu/graphene composites was ~ 46% IACS when the doping amount of graphene was 0.5 wt%, which is attributed to the high electrical conductivity of graphene. However, the electrical conductivity of 1.0 wt% bulk composites was only 38.3% IACS, which was reduced by 10% compared with W₇₀Cu₃₀ alloys without any additive. For W₇₀Cu₃₀ composites with 1.0 wt% graphene addition, tungsten carbides (WC and W₂C phases) were formed. Figure 27b, c presents the XRD pattern of the W₇₀Cu₃₀ alloy doped with different graphene contents both after BM (Fig. 27b) and sintering by LPS (Fig. 27c) [61]. In Fig. 27b, it is clearly seen that all the samples have only the major W and Cu peaks after BM. However, it is notable that new peaks are still observed after sintering (Fig. 27c). In particular, for 1 wt% graphene, tungsten carbides, whose formation may be promoted by the presence of defects in graphene (Fig. 14), can be identified. It is not clear from this study, however, why carbide formation only takes place for 1 wt% graphene loading. For the electrical conductivity, the presence of carbide is equivalent to adding obstacles to the electron mobility in the WCu alloys. Furthermore, the electrical conductivity of tungsten carbides is much lower than that of pure W and Cu as well as they inevitably increase the number of interphases and hence electron scattering. As a result, the W₇₀Cu₃₀ alloys with 1.0 wt% graphene addition have lower conductivity in comparison with that of W₇₀Cu₃₀ alloys without any addition.

Cu/graphene composites have been prepared by electroless plating of graphene with Ni particles [66], and it is found that at a content of 0.13 wt%, the electrical conductivity of the composite is comparable to that of pure Cu. However, a dramatic drop of the electrical conductivity occurs at higher EPG contents. This decline has been ascribed to a weakening of the interface bonding with increase in the content of EPG.

Figure 28a shows the electrical conductivity of Cu/GNPs composites synthesised by BM and HP with varying graphene content [102]. Upon increasing the graphene content, the electrical conductivity first decreases and then increases with the minimum electrical conductivity reached at 3–4 wt% graphene. At low graphene content, the movement of free electrons, and thus the electrical conductivity, decreases with increase in graphene content. However, compactness is another important factor for electrical conductivity. The density and relative density of the composites is shown in Fig. 28b. It is observed that the relative density of the composites increases with increase in the graphene weight fraction, that could promote an enhancement of the electrical conductivity for graphene contents above 3–4 wt%. From Fig. 28a, it can be also seen that the composites are anisotropic. So, the electrical conductivity perpendicular (P⊥) to the direction of sintering pressure is higher than the electrical conductivity parallel (P//) to the direction of sintering pressure. This can be explained because graphene usually aligns in the direction perpendicular to the consolidation force in the Cu matrix and the electrical conductivity of graphene is much higher in the in-plane than the through-plane direction. It appears, however, that the alignment of the graphene changes with graphene loading.

The effect of two different post-processing treatments (cold uniaxial repressing annealing and hot isostatic pressing) on the electrical conductivity of pure Cu and Cu/GNPs composites fabricated by wet mixing and sintering is shown in Fig. 29a [73]. It can be seen that in all the cases the electrical conductivity increases after post-processing, especially after hot isostatic pressing, which can be mainly attributed to a decrease of the residual porosity. Figure 29b shows the comparison of the theoretical and measured densities of Cu as a function of graphene content. In agreement with the relative electrical properties, it is evident that the densities of pure Cu and Cu composites increased when using post-processing techniques, especially with hot isostatic pressing.

The electrical conductivity of Cu-2.4 vol% RGO composites fabricated by MLM using 350, 450 or 550 °C as the H₂ reduction temperature and then consolidated by SPS is shown in Fig. 30 [109]. It can be seen that the conductivity of the Cu/RGO-450 is the highest among those of the three composites. In order to understand the change of conductivity, the measured results of density show that the relative densities of the Cu/RGO composites reduced at temperatures of 350, 450 and 550 °C are 87.3, 93.5 and 88.9%, respectively. It is worth noting that the variation trend of the conductivity is similar to that of the density of the composites, suggesting again that porosity is the key factor for the change of electrical conductivity. It can be also seen that the conductivities of the composites treated by hot rolling after SPS are higher than those of the composites before hot rolling. So, hot rolling can improve the electrical conductivity of the composites. This has been mainly attributed to a reduction in porosity.

Li et al. [40] reported better electrical conductivity for the CMCs containing HQG than for those containing regular RGO. The higher electrical conductivity of the Cu/HQG composites for graphene contents lower than 5 wt% was attributed to the much higher electrical conductivity in HQG than in RGO. It was also shown that, when the HQG content is lower than 1 wt%, the electrical conductivity increased gradually with increase in the graphene content. However, when the HQG content was higher than 1 wt%, the electrical conductivity began to decrease. SEM examination revealed that with increase in the HQG content, the cavities in the composite gradually increased in quantity and size because of the poor wettability between graphene and Cu. This is the reason why the electrical conductivity of the composites was reduced over the optimum HQG content.

The electrical conductivity of CMC coatings filled with different contents of GO, chemically reduced GO (RGO) and thermally reduced GO (TRGO) is presented in Fig. 31 [114]. It can be observed that the Cu/GO composite coatings show very low conductivity compared to the Cu/RGO and the Cu/TRGO composite coatings, which is due to the insulating nature of GO caused by the presence of a high amount of oxygen. In contrast, the better electrical conductivity of the Cu/RGO and the Cu/TRGO coatings was attributed to the reduction in major oxygen-containing functional groups during the reduction process, especially during thermal reduction.

Transition metal carbide (TiC and VC) coatings were synthesised on GNPs to improve the interfacial properties of Cu/GNPs composites fabricated by MLM [67]. However, this had no effect on the electrical conductivity of the Cu/GNPs composites, which probably due to the presence of pores, is lower than that of pure Cu (Fig. 32). On the contrary, the good electrical conductivity of the Cu-0.15 wt% Ag/RGO composite, which is 15% higher than that of pure Cu synthetised by the same route, could be attributed to the good bonding interface of Ag–Cu and Ag-RGO (Fig. 22a) [64].

Chen et al. [43] fabricated Cu/GNPs composites via MLM and SPS. Since the orientation of graphene was affected by content, the thermal diffusivity of the composites was tested vertical (α_vertical) and horizontal (α_horizontal) to the direction of the consolidation force. The thermal diffusivity (α) is expressed as:where k is the thermal conductivity, ρ is the density, and C_p is the specific heat capacity.

It was found that the thermal performance of Cu deteriorated upon the addition of graphene. Both α_vertical and α_horizontal decreased significantly with the increase in graphene concentration, especially when the graphene content was over 0.8 vol%. The decrease of thermal diffusivity induced by graphene additions was attributed to the decrease of the mean-free path of heat carriers, the interfacial thermal resistance and the voids formed during sintering, that serve as insulating barriers to the heat flow. By comparing α_vertical and α_horizontal of each composite, it was found that the difference between the values is varied with the graphene content. For the composites with 0.2 and 0.8 vol% graphene, α_horizontal was almost equivalent to α_vertical. However, for the composites with 2 and 4 vol% graphene, α_horizontal was considerably higher than α_vertical. This was attributed to the difference of graphene alignment in the composites. As the GNPs were oriented randomly in the composite with 0.2 and 0.8 vol% GNPs, there was no difference between the thermal diffusivity in the two directions. For the composites with 2 and 4 vol% graphene, as the GNPs are aligned along the direction perpendicular to the consolidation force, α_horizontal and α_vertical are the thermal diffusivity along the in-plane and the through-plane direction of graphene, respectively. It is well known that the thermal diffusivity of graphene at the in-plane direction is much higher than that at the through-plane direction. This explains the large difference between α_horizontal and α_vertical for the highest loadings.

Gao et al. [47] mixed GO with cationic surface agent coated Cu powders to obtain Cu/GO powders by electrostatic self-assembly. Afterwards, the composite powders were consolidated by HP. Figure 34 shows the thermal conductivity of the synthesised Cu/GO composites as a function of the graphene content. It can be seen that the addition of graphene into the Cu matrix can improve the thermal conductivity. When the content of graphene is small, the thermal conductivity gradually increases with increase in the graphene content, reaching its maximum at 0.3 wt% GO. However, for higher graphene contents, the thermal conductivity significantly decreases. The reason was the formation of agglomerates, resulting in the loss of associativity among the Cu grains. Moreover, pores and defects at the interfaces could promote the presence of interfacial thermal resistance contacts in the composites, acting as sites for phonon scattering.

The thermal conductivity at different temperatures of pure Cu and Cu/GNPs composites fabricated by wet mixing, sintering and hot isostatic pressing has also been investigated [73]. The samples containing 1 vol% GNPs present an improved thermal conductivity (up to 17%) with respect to pure Cu. However, when 2 vol% were added, only a slight increase was achieved with respect to pure Cu, indicating that there is a critical graphene content for the attainment of the maximum thermal conductivity. For higher graphene contents, the thermal conductivity of the composites is lower than that of pure Cu. This could be mainly attributed to the tendency of graphene to form agglomerates with increase in content. It was found that increasing the graphene content from 4 to 8 vol% increases the presence of clusters in the Cu matrix [73].

Cu-35 vol% GNPs composites were prepared by vacuum filtering and SPS from mixed powders obtained by either vortex mixing or ball milling [70]. As shown in Fig. 35a, the in-plane thermal conductivity (TC) of the composite derived from the ball-milled powders was 18.5% lower than that of the composite derived from the vortex-mixed powders. Raman spectra shown in Fig. 35b clearly revealed that the D band intensity of ball-milled powders was apparently higher than that of vortex-mixed powders, demonstrating that graphene structural defects were introduced during the BM process. These extra defects can impair the intrinsic TC of GNPs by acting as obstacles for strong phonon scattering, resulting in a diminished in-plane TC. Therefore, vortex mixing is superior to BM for the V–GNP/Cu composites in terms of achieving a high in-plane TC, because there are almost no extra graphene defects introduced during the vortex-mixing process.

Cu-35 vol% GNPs composites were prepared by vortex mixing and vacuum filtering followed by SPS from either large-sized (25 μm in average lateral size) or small-sized (~ 3 μm in average lateral size) [70]. Figure 36a shows that the in-plane thermal conductivity (TC) of the composites exhibits an astonishing drop from 525 W/mK to 275 W/mK when the GNP lateral size changes from 25 to 3 μm, suggesting that the GNP lateral size plays a paramount role in dictating the in-plane TC. This surprisingly low in-plane TC of the composites with small-sized GNPs was ascribed to two factors. First, reducing GNP lateral size creates more Cu–GNP interfaces, especially GNP edge-Cu interfaces, which cause a large thermal resistance. Second, as shown in Fig. 36b, the small-sized GNPs tend to be randomly distributed in the Cu matrix rather than aligned in one direction, as the large-sized GNPs are, regardless of using vacuum filtration, which hinders the effective TC contribution of GNPs.

Regarding the graphene thickness, it was found to have a key role on the improvement in the thermal conductivity of the Cu/graphene composites [19, 21, 53]. In particular, the lower thermal conductivity of smaller graphene particles is considered to be a limiting factor [19, 21]. On the contrary, thicker graphene platelets with more than three atomic layers are expected to possess higher thermal conductivity in the presence of matrix so that the quenching of phonons that carry the heat in graphene is confined only to the outer layers [21]. It is worth-mentioning that, similarly, single-layer graphene is not expected to give rise to as large an electrical conductivity improvement as graphene platelets or graphene with few atomic layers as the interface carrier scattering is more significant in single-layer graphene. So, as for the thermal conductivity, multilayer graphene is expected to provide better electrical conductivity improvement as the inner atomic layers are free from interface scattering.

Figure 37 shows the thermal properties of pure Cu and Cu-0.5 wt% GNPs, Cu-0.5 wt% GNPs–TiC and Cu-0.5 wt% GNPs–VC composites fabricated by MLM and SPS [67]. The thermal diffusivity of the composites is lower than that of pure Cu. However, the thermal diffusivity of the Cu/GNPs–TiC and Cu/GNPs–VC composites, where the interface is continuous and tightly bonded is slightly higher than that of the Cu/GNPs composite. The TiC and VC interlayers, with a mixture of metallic and covalent bonding, can reduce the energy of electron–phonon coupling and by this decreasing the interfacial thermal resistance. Similarly, the enhancement of the thermal conductivity of Cu-0.5 wt% Ag/RGO composites (Table 3) was attributed to a good interfacial bonding [64]. This is in agreement with some investigations on phonon transmission across the graphene/Cu interfaces using different simulation methods, showing the critical importance of interfacial properties of graphene-metal systems, in applications of graphene in integrated electronics, as thermal materials, and in electromechanical devices [93, 94, 153, 154].