Thermal conductivity and mechanical properties of a flake graphite/Cu composite with a silicon nano-layer on a graphite surface

Abstract

The interface between graphite flakes (G_f) and Cu in composites significantly affects the interfacial thermal diffusion and mechanical property characteristics. Silicon (Si) coating has been introduced to the surface of the G_f to investigate the thermal conductivity and flexural strength of G_f/Cu composites. Microstructural analysis demonstrates that (i) the high thermal conductivity of G_f/Cu composites is attributed to the homogeneous dispersion and well-controlled alignment of G_f in the composite matrix and (ii) silicon coating on the G_f slightly decreases the thermal conductivity of the composites, but greatly improves the bending strength. Compared with the raw G_f/Cu composites, the thermal conductivity of the Si-coated G_f/Cu composites along the plane parallel to the graphite laminate decreases from 676 to 610 W (m⁻¹ K⁻¹) when the volume fraction of G_f reaches 70 vol%. The bending strength of Si-coated graphite/Cu composites is significantly enhanced and the maximum bending strength is 110 MPa when the volume fraction of graphite is 40%. Additionally, the experimentally determined thermal conductivity is compared with the theoretically calculated value in this study.

---

1. Introduction

Efficient thermal management has become a critical necessity for guaranteeing performance and reliability in next-generation electronic devices. In this application, several factors must be taken into account: good machinability, a high thermal conductivity, a tailored coefficient of thermal expansion (CTE),^1,2 fabrication procedures and equipment. Among the different classes of materials nowadays being considered in electronics, graphite/metal composites have one main competitor: diamond/metal (Al, Cu or Ag) composites which exhibit excellent thermal properties with thermal conductivity approaching 600 W (m⁻¹ K⁻¹),^3–5 which have drawn much more attention from both scientific and industrial communities. However, the main factor that precludes these composites from a wide range of technological applications is their processability.⁶ In this case, special attention has been paid to G_f dispersed Cu composites in recent years.

G_f are attractive reinforcements due to its high thermal properties, low price and ease of processing. Combining G_f with copper will allow developing composites with high thermal conductivity, low CTE, good machinability, and reasonable cost.^7–10 However, this research is rarely done at present.¹¹ An essential problem for manufacturing G_f/Cu composites is that packed G_f tend to form a laminated structure and copper is a non-carbide forming material, which makes metal infiltration an insurmountable difficulty. In this case, G_f/Cu composites is prepared via vacuum hot pressing process. Since G_f is excellent heat conductors in only two dimension, the formation of well arrayed G_f in the composites can greatly improve the in-plane thermal conductivity, and the control of preferred orientation of G_f becomes important.^12,13 Several groups show that the efforts to improve the uniform dispersion of G_f in Cu matrix, while the enhancement of the thermal conductivity is not apparent in the composites.¹⁴ Boden et al. used planetary ball milling¹⁵ to align graphite nano platelets within a copper matrix. Q. Liu et al.¹⁶ manufactures 71 vol% G_f/Cu composites after the process of electro-less copper coating and improve thermal conductivity to 565 W (m⁻¹ K⁻¹). Ueno et al. control the orientation of G_f by a blade coating method, and the thermal conductivity of G_f/Cu composites reaches 632 W (m⁻¹ K⁻¹). Although a level of 600–700 W (m⁻¹ K⁻¹) has been attained for the G_f/Cu composites, the strength of graphite is too low, and the flexural strength for well-aligned graphite blocks perpendicular to the graphite layers is less than 35 MPa,¹⁷ which will greatly restrict the potential use of G_f/Cu composites. And the large difference of the density between graphite and Cu is also an obstacle to their uniform dispersion,¹⁸ producing much voids in the G_f/Cu composites with high volume fraction G_f, resulting the decrease of flexural strength. Up to now, there is little report of these G_f/Cu composites with super high thermal conductivity and good mechanical properties.

In this work, efforts are invested to improve the thermal conductivity and mechanical properties of composites, and silicon (Si), a carbide-forming element, is coated on the surface of G_f using the salt bath method to improve the interface bonding strength between G_f and Cu, followed by a simple vacuum hot pressing process to overcome those difficulty. Compared with the earlier studies, this study represents a simplification of the process based on high speed mixing and vibration to maintain the alignment and keep the uniform distribution of G_f in the composites. The highly oriented structure and interfacial configuration of Si coated G_f/Cu composites are demonstrated and discussed. In addition, the thermal properties and bending strength of the composites are also investigated.

2. Experimental procedures

2.1. Raw materials

Commercial graphite flakes (G_f) with lateral size ranging of 500 μm and thickness of around 40 μm, were purchased from Qingdao tianshengda Graphite Co. Ltd (China). The silicon powder with average particle size of 1 μm and a purity of >99.1% was supplied by Shanghai ST-Nanoscience and Technology Co., Ltd (China). The oxygen-free copper powder (99.9% in purity) with particle size in the range of 70–90 μm was used in the present study as the composite matrix. All other chemicals were of analytic reagent grade and used without further purification.

2.2. Silicon coating grown on the surface of G_f

The growth of silicon coating on the whole surface of the G_f is carried out by a simple heat-treatment process, in which a mixture of G_f, silicon powder and CaCl₂ powder is used as the source material. In the typical experiments, 100 g G_f, 100 g CaCl₂, and 20 g silicon powder (the mass ratio is about 5 : 5 : 1) are mixed using a DAC 150.1 FVZ-K SpeedMixer (FlackTek, Ltd. Germany) at a speed of 3000 rpm for 1 min. Then, the mixed powders are set into a graphite crucible and placed in the vacuum induction furnace to heat from 298 K to 1473 K in 150 min. The heating time is maintained for 120 min so that the melted CaCl₂ can transfer the silicon powder to the surface of the surface of the G_f. After the heat treated G_f being cooled, the extra silicon powders were simply sieved out, and then the G_f was cleaned in the distilled water for eliminating the CaCl₂, dried at 373 K and kept in desiccators.

2.3. Preparation of Si-coated G_f/Cu composites

Cu matrix composites containing different G_f contents (from 40 to 70 vol%) are prepared by the following procedures. A desired amount (40, 50, 60 and 70 vol%) of G_f is mixed with Cu powder using a SpeedMixer at a speed of 1500 rpm for 3 min. Then, the mixed powders is put into a graphite crucible and placed on a reciprocating vibration table for 1 h with the frequency of 150 Hz, contributing to the directional arrangement of lamella graphite. Finally, the graphite die with mixing of G_f and Cu particles packed perform is heated at 1333 K, and a pressure of 40 MPa is applied on the mixture perform for 30 min. Its synthetic route and the preparation sketch of the composites are shown in Scheme 1.

Scheme 1 Synthetic process for the G_f/Cu composites.

2.4. Microstructural characterization

The X-ray diffraction (XRD) patterns of the samples were recorded on a D8 DISCOVER with GADDS (BRUKER Ltd. Germany) with Cu Kα radiation. Morphology of the specimens were observed by field emission scanning electron microscopy (FE-SEM) on a JEOL JSM-6610 instrument at an accelerating voltage of 20 kV, and the fracture surface were deposited with sputter-gold to improve the conductivity. Element distribution across interface of the composite were characterized by energy disperse spectroscopy (EDS). The inter-facial micro-structures of Si-coated G_f/Cu composites obtained by transmission electron microscopy (TEM) using a JEOL 2100 instrument, operated at 200 kV. Thermal diffusivity (α) of the composites were measured with laser flash apparatus (LFA457, Netzsch) at room temperature. The samples were prepared in square-shaped forms, with a length of 10 mm and a thickness of about 2.5 mm. The specific heat (C_p), capacity of the composites was calculated by the linear rule of mixtures (ROM) using the specific heat capacity values of the components (i.e. G_f and Cu) and the bulk density (ρ) was calculated by the Archimedes method. The thermal conductivity is calculated from multiplication of thermal diffusivity, specific heat, and density. The flexural strength of the G_f/Cu composites were measured both parallel and perpendicular to the pressing direction at room temperature by an Instron-5569 universal test machine. The samples (10 mm × 10 mm × 50 mm) were tested by the three-point bending method.

2.5. Modeling approaches

Layers-in-parallel and layers-in-series models are used to calculate longitudinal (i.e. parallel to the X–Y plane) and transversal (i.e. perpendicular to the X–Y plane) thermal conductivity of the oriented G_f/Cu composites, respectively. The layers-in-parallel model considers heat-transfer characteristics parallel to the major heat flow direction of layered composites, while the layers-in-series model considers those perpendiculars to the major heat flow direction. The models are described as follows:¹⁹

Layers-in-parallel model:

K^L_C = fK^L + (1 − f)K_m (1)

Layers-in-series model:

where K_C, K and K_m represents the thermal conductivity of composites, G_f and metal, respectively; the subscripts L, T denotes the longitudinal and transversal directions and f is the volume fraction of G_f.

3. Results and discussion

3.1. Microstructure characterization of coated G_f

The typical SEM image of the Si-coated G_f is shown in Fig. 1. It can be found that the coating layer are densely and evenly covered on the surface of G_f after the coating process (Fig. 1(a) and (b)), in which the coating is patterned like a stack with layers. The edges of the G_f particle, usually the most difficult parts to be coated, are also well covered (Fig. 1(c) and (d)). Energy dispersive spectroscopy (EDS) element mapping (Fig. 2) of the coated sample shows a high silicon and carbon content accompanying with little carbon content. The continuous distribution of the silicon element further proves the continuity of the coating, and the little amount of oxygen detected is considered to come from surface oxygen absorption.

Fig. 1 SEM image of coating on the surface of G_f (a and b), coating on the edge of the coated G_f (c and d).

Fig. 2 EDS element mapping at the continuous layer of G_f surface.

To determine the composition of the coated G_f and coated G_f/Cu composites sample, the XRD patterns of the samples are shown in Fig. 3. As shown in Fig. 3(a), the diffraction peak centered at 2θ = 54.7° can be attributed to the (004) crystal planes of hexagonal graphite. Moreover, the two diffraction peaks at the 2θ angles of 36.7° and 60.1°, in accordance with the planes of SiC (102) and (110), respectively, confirms the formation of SiC. This XRD results indicates the existence of SiC in the products. The XRD analysis is further proved by Raman spectra of the coated G_f, as shown in Fig. S1.† Fig. 3(b) reveals the XRD profiles of the coated G_f/Cu composites in the range of 30–70° with a scan speed of 5° min⁻¹, two diffractions at about 2θ = 43.3° and 50.4, corresponding to the (111) and (200) crystal planes of Cu. A weak diffraction at about 2θ = 36.7° represents the existence of SiC.

Fig. 3 XRD patterns of coated G_f (a), coated G_f/Cu composites (b).

3.2. Microstructure characterization of coated G_f/Cu composites

Fig. 4 shows the typical microstructures of the Si-coated G_f/Cu composites. The elongated dark regions represent the aligned G_f while the light ones mean the Cu matrix. Almost all G_f are well aligned parallel to each other and oriented perpendicular to the pressing direction, which suggests the vibration of composite powder, followed by vacuum hot pressing process, can achieve the preferred orientation of the G_f/Cu composites. Besides, the observed interface is quite continuous and no obvious gaps have been observed, which means that the coating on the G_f improve the bonding strength between the G_f and Cu at the interface. Additionally, Fig. 4(b) shows EDS line-scan analysis of the major element distribution across the interface between the G_f and Cu. It can be found that the coating thickness of SiC is about 300 nm, and a copper and silicon elements miscible region is formed on the interface. The SEM analysis is further proved by TEM image of the composites, as shown in Fig. S2.† The interpenetrated interface structures consists of SiC coating and the miscible region, which is expected to reinforce the interface and improve the mechanical properties of the composites.

Fig. 4 SEM image of Si-coated G_f/Cu composite (a), EDS line-scan analysis across the interface of G_f and copper (b).

3.3. Thermal and mechanical properties of G_f/Cu composites

3.3.1. Thermal properties and modeling. Table 1 depicts the thermo-physical properties in parallel and perpendicular to the pressing direction of the G_f/Cu composites with different contents of G_f. On the whole, the thermal and mechanical properties of composites exhibit obviously anisotropic characteristics. One can see that thermal conductivity along X–Y and Z direction show a converse tendency as the volume fraction of G_f increases with different graphite fillers (raw G_f and Si-coated G_f). With increasing volume fraction of G_f from 40% to 70%, the thermal conductivity of the G_f/Cu composites in X–Y plane direction approach 676 W (m⁻¹ K⁻¹). It is well known that the increase of the volume fraction of G_f can increase the heat transfer paths through highly thermal conductive G_f, thus improving the thermal conductivity of composites in the X–Y plane direction. Obviously, due to the well orientation of G_f, the thermal conductivity in the Z-axis direction is far less than that in the X–Y plane direction. Meanwhile, the thermal conductivity and thermal diffusivity in Z-axis direction reduce with the increase of volume fraction of G_f.

Table 1 Thermal and mechanical properties of graphite flakes/Cu composites

Gf	Vf (%)	ρ (g cm^−3)	α (mm^2 s^−1)		TC (W (m^−1 K^−1))		Flexural strength (MPa)
			X–Y	Z	X–Y	Z	⊥^a	∥^b
a Perpendicular to X–Y plane, which is parallel to the loading direction.b Parallel to X–Y plane, which is perpendicular to the loading direction.
Raw	40	6.20	183	28.4	474	74	85	94
	50	5.50	215	27.6	518	66	62	78
	60	4.91	257	23.3	584	53	50	67
	70	4.23	323	19.2	676	40	41	52
Si coated	40	6.25	176	29.6	460	77	95	110
	50	5.58	201	27.2	491	66	79	89
	60	4.87	245	25.2	553	57	66	81
	70	4.16	296	17.5	610	36	47	75

---

In addition, it should be noted that the interface between graphite and Cu matrix plays an important role in determining the thermal conductivity of the composites, even when well-aligned G_f have been uniformly embedded in the Cu matrix. For Si-coated G_f/Cu composites, the introduction of Si element decreases the thermal conductivity in the base plane direction, whereas the thermal conductivity in the Z-axis direction shows no significant change. The Table 2 material parameters for theoretical calculation thermal conductivity of G_f/Cu composites in X–Y plane direction decreases from 676 to 610 W (m⁻¹ K⁻¹), when the raw G_f is coated with silicon. This phenomenon of reduction in thermal conductivity with G_f/Cu composites is different in comparison with the other carbon/metal composites.^20,21 In comparison with the previous works reported in the literature, the thermal conductivity of carbon/metal composites increase as the carbide element (Si) is added to the interface to create a strong chemical bond. However, in the G_f/Cu composites, the introductions of silicon element decrease the thermal conductivity of composites in X–Y plane direction. It might be attributed to slide of G_f and deformation of Cu at high temperature under pressure, resulting a good interface contact between G_f and Cu. And the interface thermal resistance coming from phonons coupling losses of Cu and graphite increase, as the formation of the solid solution at the interface would lead to an increase of phonon scattering and decrease the thermal conductivity of composites.^22–24

Table 2 Material parameters for theoretical calculation

Material	Density (kg m^−3)	Specific heat (J kg^−1 K^−1)	Phonon velocity (m s^−1)	TC (W (m^−1 K^−1))
a Parallel to the graphite layers.b Perpendicular to the graphite layers.
Graphite	2260	710	14 800 (ref. 19)	880^a (ref. 26)
				38^b (ref. 9)
SiC	3210	290	11 600 (ref. 27)	243 (ref. 9)
Cooper	8960	370	2500 (ref. 16)	400

---

Theoretical predictions and experimental data of the longitudinal and transversal thermal conductivity of Si-coated G_f/Cu composites, by means of the layers-in-parallel and layers-in-series models, are plotted versus in Fig. 5(a) and (b), respectively. From Fig. 5(a) one can see that the layers-in-parallel model overestimates the thermal conductivity of the coated G_f/Cu with increasing volume fraction of G_f. In order to take into account the effect of SiC coating layer, it can be solved by replacing the inclusion with a non-ideal interface by an “effective” inclusion having an effective thermal conductivity, K_eff, and can be expressed as

where h is the interface thermal conductance (i.e. the reciprocal of the interfacial thermal resistance), t and D denote the thickness and diameter of the inclusion. The value of h can be calculated by the acoustic mismatch model (AMM),²⁵ h is calculated to be:where C is the specific heat, ρ is the density and ν is the phonon velocity. Subscripts “in” and “tran’’ refer to incident side of phonon and transmission side of phonon, respectively. The velocities of transversal waves were used for the reason that these contribute the most to heat transfer. The material parameters for the calculation are given in Table 2.

Fig. 5 Comparison of experimental data with theoretical predictions: (a) the longitudinal thermal conductivity and (b) the transversal thermal conductivity of the composites with different G_f volume fraction composites.

In addition, for the Si-coated G_f/Cu composites, the SiC layer was referred to as the transmission side for the cooper/carbide layer and as the incident side for the carbide/graphite layer. The interface thermal conductance of the graphite/SiC/Cu can be calculated using eqn (5). We obtain h_Cu/SiC = 8.7 × 10⁷ W m⁻² K⁻¹ and h_SiC/graphite = 8.3 × 10⁸ W m⁻² K⁻¹ from the material parameters given in Table 2. The total interfacial thermal conductance (h_t) of the G_f/Cu composite with Si coating can be calculated by using the concept of interfacial thermal resistance:

where d_SiC is the thickness of coating layer and about 300 nm. K_SiC is thermal conductivity of SiC.

The calculation values of the interface thermal conductance of the Si-coated G_f/Cu composites h_t is 7.19 × 10⁷ W m⁻² K⁻¹ using eqn (6). The effective thermal conductivity of the G_f parallel and perpendicular to the X–Y plane are calculated from eqn (3) and (4) to be 838 and 37 W (m⁻¹ K⁻¹), respectively. Hence, we replace K in layers-in-parallel model and layers-in-series model by K_eff. The modified layers-in-parallel model and modified layers-in-series model are expressed as:

K^L_C = fK^L_eff + (1 − f)K_m (7)

The longitudinal thermal conductivity data predicted by the modified layers-in-parallel model are given in Fig. 5(a). Modified model is more suitable for predicting the thermal conductivity of the experimental data compared to the original one. But it still has deviations with the experimental data, and this might be attributed to three reasons in our case. First, although most of the G_f are arranged parallel to each other, but also a small amount in the state of disorder. Next, we all know that the mechanical strength of G_f only about 20 MPa, hot pressing process it easily bend, which could deflect flow of heat. At last, although graphite coating is thin, but the thermal conductivity of the coating is much lower than the G_f. Which increases the interfacial thermal resistance of the composite material. For these reasons, the experimental data and theoretical predictions some error is understandable.

Comparison of experimental data with theoretical predictions the transversal thermal conductivity of the composites with different G_f volume fractions given in Fig. 5(b). We found that the layers-in-series model can better predict the thermal conductivity in the longitudinal direction. In the longitudinal direction due to the thermal conductivity of the graphite layer structure height direction is less than the copper matrix, thus with increasing the volume fraction of graphite composite material to reduce thermal conductivity.

3.3.2. Mechanical properties and microstructures of G_f/Cu composites. The flexural strength of samples sintered under different pressures which are parallel and perpendicular to the loading direction are also presented in Table 1 respectively. Obviously, the flexural strength in the two direction decrease as the volume fraction of G_f increased, and the experimental measurements of bending strength for graphite–Si–Cu composites show higher bending strength than that of graphite–Cu composites due to the surface functionalization of G_f. Owing to the arrangement of G_f perpendicular the pressing direction, it is easy for G_f to slide under compression in X–Y direction which can weaken the composites. For 40 vol% G_f/Cu composites, the flexural strengths in direction perpendicular and parallel to X–Y plane are 85 MPa and 94 MPa, respectively. When the G_f are coated with silicon, the flake graphite is hard to slide, the flexural strength in direction perpendicular and parallel to X–Y increase to 95 MPa and 110 MPa, respectively.

Besides, the values of flexural strength in direction parallel to X–Y plane are larger than those in direction perpendicular to X–Y plane, which can be attributed to the contact area between the flake graphite and copper. When the samples are loaded in direction parallel to X–Y plane, the deformation of the graphite/Cu composites would perform like the deformation of “long fiber like filler” enhanced Cu composites, and “the long fiber like filler” is hard to be broken, which will increase the flexural strength of the graphite/Cu composites. For example, when the volume fraction of the silicon coated graphite increase from 40% to 70%, the flexural strength in direction parallel to X–Y plane decrease from 110 MPa to 75 Mpa, and the values in direction perpendicular to X–Y plane decrease from 95 MPa to 47 MPa.

Fig. 6 shows the microstructure of fracture surfaces of composites based on the Si-coated G_f and raw powder. In the typical case of composite based on the raw G_f, as illustrated in Fig. 6(a), the G_f can be directly broken into piece and the fracture surfaces are relatively flat, resulting in the disconnection and fracture of the tested samples. It can be observed from Fig. 6(b) that some of the G_f are being curved, and the bending deformation appears on the G_f. It means that the damage accumulation within the Si-coated G_f/Cu composites prior to rupture happens. And the cracks in the striped concave and convex areas need to consume more energy during the propagation process because of longer and more tortuous paths. Therefore the composites fail in the shear yielding inside the G_f and the fracture of G_f, which means the coating on the G_f has changed the fracture mode of nature G_f.

Fig. 6 SEM images of the fracture surface of G_f/Cu composites with (a) raw G_f, (b) Si-coated G_f; (c) schematic diagram of crack initiation and propagation along Z direction.

As stated above, the variation trend of the flexural strength along the X–Y and Z direction of the composites is the same, and it is easy to observe the microstructures transformation laws in the composites when the loadings are along the Z direction. Herein, we mainly focus on the different deformation mechanism along the Z direction in the composites with different G_f, including the raw and Si-coated G_f. On the whole, we assume that coating on the edge of G_f play an important roles on the fracture model of the G_f/Cu composites.²⁸ Fig. 6(c) gives the schematic of cracks initiation and propagation along parallel to sintering pressure direction, corresponding to Fig. 6(a) and (b). When the composites fabricated with raw G_f is loaded, the initial cracks will appear inside the layer of G_f, and then these cracks expand in the X–Y plane to the gap of Cu/G_f with increase of loading. For the weak bonding strength between the G_f and Cu, the cracks expand to the next G_f layer along the loading direction with some deflection, and low force is needed for the crack propagation until the samples fracture. When the G_f are coated with silicon, the strength of individual coated G_f might increase, means that it needs more energy for cracks initiation. And coating on the edge of the G_f is an obstacle for the slip of graphite, therefore the fracture morphology will be zigzag shaped for the enhanced bonding strength at the interface, which can effectively improve the fracture toughness of the composites. Compared with the composites fabricated with raw G_f, the crack propagation paths of fracture surface of coated G_f/Cu composites are more tortuous, more energy is used to keep the crack propagation. So the flexural strength of composites with coated G_f significantly increase due to the distinct fracture mode.

4. Conclusions

In view of our findings, a simple approach has been proposed to obtain the uniform dispersion and arrangement of G_f in a Cu matrix to enhance the thermal conductivity of composites, and a coating process is introduced onto the G_f to increase the bending strength of the composites. Microstructural characterization reveals that there is a distinguished interfacial layer homogeneously distributed along the G_f/Cu interface, which contains the elements C, Si, and Cu. The coating on the G_f decrease the thermal conductivity of coated G_f/Cu composite, improve the bending strength. With increasing volume fraction of G_f from 40% to 70%, the thermal conductivity in plane direction perpendicular to the pressing direction increase from 474 to 676 W (m⁻¹ K⁻¹), while the bending strength decrease from 85 to 41 MPa. However, for Si coated G_f, the thermal conductivity in the plane parallel to the graphite layers decreased from 676 to 610 W (m⁻¹ K⁻¹) as the volume fraction of G_f reach up to 70 vol%, while the flexural strength in direction perpendicular to the graphite layers increased from 94 to 110 MPa when the volume fraction of G_f in composites is about 40 vol%.

Acknowledgements

This work was financially supported by the National Natural Science Foundation of China (Grant No. 51501209) and Ningbo Natural Science Foundation (Grant No. 2015A610090).

References

1. K. Rajkumar and S. Aravindan, J. Mater. Process. Technol., 2009, 209, 5601–5605.
2. K. Rajkumar, K. Kundu, S. Aravindan and M. S. Kulkarni, Mater. Des., 2011, 32, 3029–3035.
3. P. W. Ruch, O. Beffort, S. Kleiner, L. Weber and P. J. Uggowitzer, Compos. Sci. Technol., 2006, 66, 2677–2685.
4. Z. Tan, Z. Li, G. Fan, Q. Guo, X. Kai, G. Ji, L. Zhang and D. Zhang, Mater. Des., 2013, 47, 160–166.
5. O. Beffort, F. A. Khalid, L. Weber, P. Ruch, U. E. Klotz, S. Meier and S. Kleiner, Diamond Relat. Mater., 2006, 15, 1250–1260.
6. R. Prieto, J. M. Molina, J. Narciso and E. Louis, Scr. Mater., 2008, 59, 11–14.
7. G. Yuan, X. Li, Z. Dong, A. Westwood, Z. Cui, Y. Cong, H. Du and F. Kang, Carbon, 2012, 50, 175–182.
8. Z. Liu, Q. Guo, J. Shi, G. Zhai and L. Liu, Carbon, 2008, 46, 414–421.
9. R. Prieto, J. M. Molina, J. Narciso and E. Louis, Composites, Part A, 2011, 42, 1970–1977.
10. I. Firkowska, A. Boden, B. Boerner and S. Reich, Nano Lett., 2015, 15, 4745–4751.
11. C. Zhang, X. He, Q. Liu, X. Ren and X. Qu, J. Compos. Mater., 2015, 49, 3323.
12. T. Ueno and T. Yoshioka, US Pat., 8501048 B2, 2013.
13. W. Li, Y. Liu and G. Wu, Carbon, 2015, 95, 545–551.
14. H. Fukushima, US Pat., 7851055B2, 2010.
15. A. Boden, B. Boerner, P. Kusch, I. Firkowska and S. Reich, Nano Lett., 2014, 14, 3640–3644.
16. Q. Liu, X. He, S. Ren and X. Qu, J. Alloys Compd., 2014, 587, 255–259.
17. T. Maruyama, M. Eto and T. Oku, Carbon, 1987, 25, 723–726.
18. S. Ren, Q. Hong, J. Chen, X. He and X. Qu, J. Alloys Compd., 2015, 652, 351–357.
19. C. Zhou, G. Ji, Z. Chen, M. Wang, A. Addad, D. Schryvers and H. Wang, Mater. Des., 2014, 63, 719–728.
20. L. Weber and R. Tavangar, Scr. Mater., 2007, 57, 988–991.
21. A. Veillère, J. M. Heintz, N. Chandra, J. Douin, M. Lahaye, G. Lalet, C. Vincent and J. F. Silvain, Mater. Res. Bull., 2012, 47, 375–380.
22. M. Battabyal, O. Beffort, S. Kleiner, S. Vaucher and L. Rohr, Diamond Relat. Mater., 2008, 17, 1438–1442.
23. A. Majumdar and P. Reddy, Appl. Phys. Lett., 2004, 84, 4768.
24. A. J. Schmidt, K. C. Collins, A. J. Minnich and G. Chen, J. Appl. Phys., 2010, 107, 104907.
25. E. T. Swartz and R. O. Pohl, Rev. Mod. Phys., 1989, 61, 605–668.
26. J. Barcena, J. Maudes, J. Coleto, J. L. Baldonedo and J. M. Gomez de Salazar, Compos. Sci. Technol., 2008, 68, 1384–1391.
27. K. Chu, C. Jia, W. Tian, X. Liang, H. Chen and H. Guo, Composites, Part A, 2010, 41, 161–167.
28. S. J. Guo, Q. S. Yang, X. Q. He and K. M. Liew, Composites, Part B, 2014, 58, 586–592.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra17804a

---