Enhanced cross-plane thermal conductivity and high resilience of three-dimensional hierarchical carbon nanocoil–graphite nanocomposites

Abstract

Three-dimensional hierarchical carbon nanocoil–graphite (CNC–GT) nanocomposite blocks were prepared by the growth of CNCs at the interlayer of expanded GT using chemical vapor deposition followed by hot-pressing. The distribution and density of the CNCs were tuned by vacuum impregnation for catalyst loading and growth time, respectively. Helical CNCs with spring-like structures were observed by scanning electron microscopy and transmission electron microscopy. The CNC–GT blocks showed a higher density and lower porosity than GT due to the intercalation of CNC fillers. The thermal conductivities of the CNC–GT blocks in the cross-plane (λ_⊥) and in-plane (λ_‖) directions were controlled by the consolidating pressure and growth time of the CNCs. The remarkable increase in λ_⊥ and the resilience of the CNC–GT blocks were further optimized using microstructures of CNCs at the interface. The maximum λ_⊥ of the CNC–GT blocks (∅ 3 cm × 2 mm) of up to 23.6 W m⁻¹ K⁻¹ was about five-fold higher than that of GT at 4.9 W m⁻¹ K⁻¹. This feature arose from improved phonon transfer in the cross-plane through intercalated CNCs at the interlayer. Moreover, a high resilience ratio of 84.1% and a low compressibility (17%) were also obtained for the CNC–GT blocks due to the excellent elasticity of CNCs. The CNC–GT blocks with high λ_⊥, good resilience properties and dimensional stability could be developed to be highly thermally conductive and resilient interface materials for heat sealing.

---

1. Introduction

Highly thermally conductive materials have received much interest for heat absorption, transfer, exchange and dissipation. As a typical heat sink material, graphite (GT) with a layered structure shows a high thermal conductivity of up to 2000 W m⁻¹ K⁻¹ in the in-plane direction due to its perfect crystalline structure.^1–3 However, GT-based materials exhibit a distinct anisotropy in their thermal conductivity.^4–8 Specifically, when GT is consolidated into blocks under a high compressive pressure, the thermal conductivity in the cross-plane direction is only in the range of 5–10 W m⁻¹ K⁻¹ because of low connectivity between adjacent layers and porous structures at the interlayer,^4,5 which greatly inhibit phonon mobility across the GT layers.^6–8 Furthermore, GT blocks show very low mechanical properties and almost no resilience due to weak van der Waals interactions between layers.

One of the important methods of improving the thermal conductivity in the cross-plane direction and the mechanical properties of GT-based materials is to design and prepare unique microstructures.

Carbon nanomaterials (CNs), such as nanotubes, nanofibers, and nanorods, with superior thermal and mechanical properties have been regarded as ideal one-dimensional (1D) candidates for the intercalation of GT-based materials with high-performance thermal management.^1–20 Recently, a three-dimensional (3D) hierarchical nanostructure built with a 1D nanopillar on GT exhibited enhanced phonon mobility in the cross-plane direction and excellent mechanical properties.^19,21–24 Lu et al. demonstrated that GT blocks with carbon nanotubes (CNTs) grown at the interlayer showed an increase in thermal conductivity in the cross-plane direction of 10.3% compared with pure GT blocks.⁶

Carbon nanocoils (CNCs), including helical nanotubes and nanofibers, are excellent 1D nanofillers for the design of 3D hierarchical blocks due to their unique spring-like shape, super-elasticity and good heat conduction along the axis of the helix. According to previous studies, CNCs not only show comparable thermal conductivity to that of carbon nanofibers,²⁵ but also exhibit an elongation strain up to 42–60% and a compressive strain up to 35% under an external force due to their spiral structures,^26,27 and they can regain their original shape when the load is released. Good elasticity and excellent thermal properties enable CNCs to be 1D nanofillers for the fabrication of a 3D hierarchical composite with high thermal conductivity and good resilience by optimizing the nanostructure.

In this paper, 3D hierarchical CNC–GT nanocomposite blocks were prepared by the growth of CNCs at the interlayer followed by hot-pressing. The catalyst loading at the interface of expanded GT (EG) was controlled by vacuum impregnation. Cross-plane thermal conductivity and resilience were improved by the intercalation of CNC into the interlayer of GT. The morphologies of the helical CNCs distributed on the surface and interlayer of EG were observed by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). The density and open porosity of the CNC–GT blocks were also studied. Thermal conductivities of the CNC–GT blocks in the cross-plane (λ_⊥) and in-plane (λ_‖) directions were controlled by the growth time of the CNCs and the consolidating pressure. The increase in λ_⊥ was due to enhanced phonon transfer in the cross-plane of GT. The mechanical properties of the CNC–GT blocks were also studied to confirm the low compressibility and high resilience.

2. Experimental

2.1 Materials

GT intercalation compounds were purchased from Qingdao Wanxing Graphite Technology Co., Ltd. Coal pitch was purchased from Wuhan Kexing Chemical Industry Co., Ltd. Chemical reagents purchased from Tianjin Jiangtian Chemical Company were used without further treatment.

2.2. Growth of CNCs at the interlayer of EG

EG (expansion rate 300%) was prepared by heating GT intercalation compounds at 1000 °C for 10 s. EG (1 g) was immersed in a 50 mL solution of ferric nitrate (0.05 mol L⁻¹) and nickel nitrate (0.05 mol L⁻¹), and the mixture was ultrasonicated for 30 min. The distribution of the catalysts on EG was controlled by vacuum impregnation. The details of the conditions for catalyst loading and the compositions of the blocks and composites are shown in Table 1. The resulting catalyst-treated EG-1 and EG-2 were obtained after drying at 80 °C for 12 h. The growth of CNCs on the interlayers of EG-1 and EG-2 by chemical vapor deposition (CVD) to form CNC–EG-1 and CNC–EG-2 respectively is shown below. Catalyst-treated EG in a quartz boat was placed in the middle of a horizontal quartz tube and was heated to 500 °C under argon at a rate of 10 °C min⁻¹. Hydrogen was then fed into the tube at 500 °C for the reduction of ferric nitrate and nickel nitrate to metal nanoparticles. The furnace was heated to 810 °C at a rate of 10 °C min⁻¹ under a mixture of argon and hydrogen with flow rates of 300 and 200 sccm respectively. The growth of CNCs was carried out for 15–150 min using acetylene (carbon source), hydrogen and argon at flow rates of 65, 200 and 75 sccm, respectively. The CNC–EG-1 and CNC–EG-2 composites were acquired when the furnace was cooled to room temperature under argon.

Table 1 The details of the conditions for catalyst loading (between rinsing and drying) and the compositions of the blocks and composites

Block	Composite	Material	Conditions for catalyst loading
			Pressure (kPa)	Time (h)
GT	—	EG	—	—
CNC–GT	CNC–EG-1	EG-1	1.013 × 10^2	2
—	CNC–EG-2	EG-2	<1.013	2

---

2.3. Preparation of CNC–GT blocks

CNC–EG-1 powders were compacted in a GT-based mould, and then pressed uniaxially with a hot-pressing device (ZRY45) to obtain consolidated CNC–GT blocks. As indicated in Table 1, GT blocks were also prepared using EG powers by hot-pressing under the same conditions for a comparison of the thermal conductivity and mechanical properties. The hot-pressing was carried out at 1600 °C for 30 min, and the pressure was varied from 10 MPa to 40 MPa. A schematic of the preparation of a 3D hierarchical CNC–GT block is shown in Scheme 1.

Scheme 1 Schematic illustration of the preparation of a 3D hierarchical CNC–GT block.

2.4. Characterization

The morphologies of CNCs grown at the interlayer of EG and the cross-sections of CNC–GT blocks were observed by SEM (Hitachi S-4800) and TEM (FEI-Tecnai G2 F20). Open porosities were measured using a mercury porosimeter instrument (Autopore 9000). The crystalline structures were studied by X-ray diffraction (XRD, D8 Advanced) analysis using Cu Kα radiation (wavelength = 0.15406 nm). Two kinds of cylinders cut from the consolidated CNC–GT blocks (Scheme 2) were used to measure λ_‖ and λ_⊥ using a DRX-II-RL apparatus and the ASTM E1530 guarded heat flow meter method. Thermal conductivities were calculated according to eqn (1):where λ is the thermal conductivity (W m⁻¹ K⁻¹), Q is the measured heat flow through the sample (W m⁻²), d is the sample thickness (m), A is the surface area of the sample (m²) and ΔT is the temperature difference between the two sides of the sample (K). The compression and resilience properties of the CNC–GT blocks were investigated using an electronic universal testing machine (CSS-44001) under a load of 18 MPa. Compressibility and resilience ratios were calculated using (t₀ − t₁)/t₀ and (t₂ − t₁)/(t₀ − t₁) respectively, where t₀, t₁ and t₂ are the primary, loading and unloading heights, respectively. The coefficient of thermal expansion (CTE) of the CNC–GT blocks was also tested using a thermo-mechanical analyzer (TMA/SS 7000) at 50 °C to 400 °C with a heating rate of 10 °C min⁻¹ and a force of 0.1 N.

Scheme 2 Schematic illustration of the two kinds of cylinders cut from the CNC–GT blocks.

3. Results and discussion

The morphologies of the EG, CNC–EG-1 and CNC–EG-2 composites were observed by SEM. EG (Fig. 1a and S1†) shows a honeycomb-shaped structure with many micro-pores among the layers for catalyst loading. Fig. S2† indicates that the interplanar distances between two adjacent layers of EG range from 2 to 10 μm with an average distance of 4.7 μm, which provides enough free space for the growth of CNCs. In the case of CNC–EG-1 (Fig. 1b), the CNCs are mainly distributed on the surface of EG with no growth at the interlayer. This feature indicates that the catalysts are mostly loaded onto the surface of EG rather than spaces at the interlayer due to strong capillary action. The vacuum impregnation controlled distribution of grown CNCs at the interlayer of EG is shown in Fig. 1c. A 3D hierarchical structure with a high-density of CNCs throughout the surface and interlayer of EG is acquired for the CNC–EG-2 composite due to good distribution of the catalysts by vacuum impregnation. The results reveal that vacuum impregnation is critical for catalyst loading at the interlayer of EG, which is used for the growth of CNCs.

Fig. 1 SEM images of (a) EG, (b) CNC–EG-1, and (c) CNC–EG-2 composites.

The density of CNCs on EG was controlled by the growth time. Fig. 2a–d show SEM images of CNC–EG-2 composites with growth times of 15, 60, 120 and 150 min, respectively. It can be seen that an increase in the growth time leads to an increase in the density and length of CNCs throughout EG. At short growth times (15 and 60 min), the low-density and short CNCs at the interlayer are not long enough to connect adjacent layers of GT. When the growth time is 120 min, the high-density and long CNCs at the surface and interlayer of EG are desirable for the fabrication of 3D hierarchical CNC–GT blocks.^8,28 The approximate average length of the CNCs is larger than 6 μm. CNCs are overloaded on EG when the growth time is 150 min. Fig. 2d shows many entangled CNC bundles and carbonaceous impurities on EG, which are due to excessive decomposition of the carbon sources.

Fig. 2 SEM images of CNC–EG-2 composites prepared at growth times of (a) 15, (b) 60, (c) 120 and (d) 150 min, and (e) high-resolution image of CNCs. (f) TEM image of an as-grown CNC where the inset is at high resolution.

It can also be seen that, compared with the average distance of EG, CNCs (>5–6 μm long at growth times of 120 and 150 min) are long enough after hot-pressing to connect adjacent layers within a limited space despite the entanglement and independent growth. Yields of CNCs at various growth times are shown in Fig. S3.† The microstructures of CNCs were also observed by high-resolution SEM (Fig. 2e) and TEM (Fig. 2f). A helical multiwalled carbon nanotube with a diameter of about 60 nm is shown.²⁹ The helical conformation is important for the super-elasticity of CNCs and resilience properties of CNC–GT blocks. This 3D hierarchical nanostructure offers great potential for efficient thermal conduction in the cross-plane direction and good resilience of CNC–GT blocks.

CNC–GT and GT blocks were prepared by hot-pressing using CNC–EG-1 and EG respectively. Fig. 3 shows SEM images of the cross-section morphologies of CNC–GT and GT blocks. The arrow indicates the direction of hot-pressing. Both GT and CNC–GT blocks show a typical lamellar structure of the GT sheets due to the preferred orientation of the crystal plane during hot-pressing.⁸ CNC–GT blocks are more dense and compact than GT because the space at the interlayer is filled with CNCs at a growth time of 120 min.³⁰ The oriented crystal planes of the CNC–GT blocks in the direction perpendicular to the hot-pressing direction is confirmed by XRD (Fig. S4†). The rough surface of the CNC–GT blocks is also observed (Fig. S5†). Compared with GT, the alignment of the CNC–GT blocks is slightly decreased due to randomly disordered CNCs at the interlayer.^31,32

Fig. 3 SEM images of the cross-sections of (a) GT and (b) CNC–GT blocks (the growth time of CNCs was 120 min) prepared by hot-pressing at a temperature of 1873 K and a pressure of 40 MPa. The arrow indicates the direction of hot-pressing.

The hot-pressing pressure controlled densities and porosities of the CNC–GT and GT blocks are shown in Fig. 4a and b. Both blocks show an increased density and a decreased open porosity when the hot-pressing pressure increases due to more compact stacking of GT sheets. Furthermore, the CNC–GT blocks exhibit a higher density and a lower porosity than GT at the same pressure because the space at the interlayer is filled with dense CNCs. The density of the CNC–GT blocks prepared at a pressure of 40 MPa reaches a maximum of 1.2 g cm⁻³. Meanwhile, the density of GT varies from 0.8 to 0.9 g cm⁻³ by consolidating EG under a pressure of 30 to 40 MPa, which is comparable to the densities of previously reported compacted GT plates.^8,33,34 Thus, CNC–GT blocks with higher density and lower porosity than GT show great potential for efficient heat conduction due to the CNC connections between adjacent layers.

Fig. 4 (a) Densities, (b) open porosities, (c) λ_⊥ and (d) λ_‖ of (1) CNC–GT and (2) GT blocks controlled by the hot-pressing pressure.

Fig. 4c and d display λ_⊥ and λ_‖ of the CNC–GT and GT blocks. Both GT and CNC–GT blocks show a clear anisotropy, in that λ_‖ is almost one order of magnitude higher than λ_⊥ because of efficient phonon mobility through the oriented crystal planes.^8,19 Furthermore, an increase in both λ_⊥ and λ_‖ of the CNC–GT and GT blocks is observed with increasing hot-pressing pressure, which is attributed to the higher density and the lower open porosity (Fig. 4a and b). The λ_⊥ of CNC–GT is much higher than that of GT blocks due to the growth of CNCs at the interface. The maximum λ_⊥ of CNC–GT is 23.6 W m⁻¹ K⁻¹, which is 5 times higher than that of GT blocks (about 4.9 W m⁻¹ K⁻¹). The increase in λ_⊥ of the CNC–GT blocks is attributed to enhanced phonon transfer across the GT layers by intercalated CNCs based on the 3D hierarchical structure.^19,20,22 CNC–GT blocks also show slightly lower λ_‖ than GT, indicating that CNCs might affect the orientation and size of the crystalline structures during hot-pressing (confirmed by the less intense diffraction peaks in the XRD patterns in Fig. S4† and the rough surface shown in Fig. S5†),⁸ thus resulting in a decrease in λ_‖.

The density, porosity and thermal conductivity of CNC–GT blocks were also controlled by the growth time of CNCs. Fig. 5a and b show that an increase in the CNC growth time leads to an increased density and a decreased open porosity of the CNC–GT blocks (prepared using a hot-pressing pressure of 40 MPa) because thousands of long and dense CNCs effectively fill the open spaces in GT. Interestingly, as shown in Fig. 5c and d, λ_⊥ and λ_‖ display different trends with changes in the growth time of CNCs. A remarkable continuous increase in λ_⊥ and a slight decrease in λ_‖ are observed for CNC–GT blocks with increasing growth time. The maximum λ_⊥ of CNC–EG blocks of 23.6 W m⁻¹ K⁻¹ observed after a growth time of 120 min is about 2.5 fold higher than that observed after a growth time of 15 min (9.6 W m⁻¹ K⁻¹). Moreover, a slight decrease in λ_⊥ (23.3 W m⁻¹ K⁻¹) is observed when the growth time is 150 min. The decrease in λ_⊥ might arise from randomly entangled CNC bundles and the formation of carbonaceous impurities (Fig. 2d), which affect phonon mobility in the cross-plane direction.²⁸ This feature indicates that λ_⊥ is affected by the microstructures of CNCs at the interface of GT after hot-pressing.³⁴ Meanwhile, for the same growth time, λ_‖ (284.6 W m⁻¹ K⁻¹) only shows a slight decrease of 4.56% compared with that for 15 min (298.2 W m⁻¹ K⁻¹) due to the slightly reduced size and larger number of defects of the crystalline structures. The high λ_⊥ of the CNC–EG blocks is comparable to or much higher than those of GT-based blocks observed in previous studies.^4–6,8,18 In addition, the variation of λ_⊥ with the growth time of CNCs indicates that although the CNCs are long (>5–6 μm) and stiff enough to connect adjacent layers of GT after hot-pressing, the random disorder and entanglement of unaligned CNCs inevitably affect the λ_⊥ of the CNC–GT blocks. Thus, the alignment of well-distributed and stiff CNCs of a suitable length at the interface of GT needs to be optimized in the future.

Fig. 5 (a) Densities, (b) open porosities, (c) λ_⊥ and (d) λ_‖ of (1) CNC–GT and (2) GT blocks controlled by the CNC growth time.

The mechanical properties of CNC–GT and GT blocks are also important for thermally conductive interface materials. The resilience ratio and compressibility of CNC–GT blocks were controlled by the growth time (Fig. 6). GT blocks show a much lower resilience ratio of 73.9% and a higher compressibility of 20.1% than those of all CNC–GT blocks. This data indicates poor resilience properties of traditional GT-based blocks.

Fig. 6 (a) Resilience ratios and (b) compressibilities in the cross-plane direction of (1) CNC–GT and (2) GT blocks controlled by the CNC growth time.

As shown in Fig. 6b, the compressibility of CNC–GT blocks decreases with increasing growth time. CNC–GT blocks show the lowest compressibility of 17% at a growth time of 90 min, which is 3.1% lower than that of GT blocks (20.1%). The decrease in compressibility is confirmed by the increasing resilience ratio (Fig. 6a) when the load is released. CNC–GT blocks exhibit a maximum resilience ratio of 84.1% at a growth time of 90 min, which is 10.2% higher than that of GT blocks (73.9%). The high resilience of CNC–GT blocks is attributed to the storage of mechanical energy in super-elastic helical CNCs at the interface of GT. CNC–GT blocks also exhibit a decrease in resilience and an increase in compressibility when the growth time of CNCs is longer than 90 min. This feature might arise from the randomly entangled CNCs in the limited space at the interface, which is consistent with the growth time controlled results for λ_⊥. When a load is applied, the distribution of mechanical energy is affected by disordered and entangled CNC bundles. The resilience of compressed CNCs is limited by intertwined long CNCs in a narrow space at the interface. Thus, CNC–GT blocks show a decreased resilience at a long CNC growth time. CNC–GT shows almost a one-fold increase in resilience and a 50% decrease in compressibility compared with GT-based materials with similar densities,³⁵ which is favorable for thermally conductive interface materials.

The excellent resilience of CNC–GT blocks is attributed to the elasticity of tough CNC nanofillers at the interlayers of GT sheets. As shown in Scheme 1, the intercalation of CNCs forms a “spring-sheet” 3D hierarchical structure of GT, in which CNCs act as the “spring” that facilitates the configuration recovery.^29,36 The open interspace between the GT layers is filled with CNCs. The CNCs show compression deformation during hot-pressing and also enable configuration recovery when the pressure is unloaded, resulting in an increase in resilience. However, this recovery can be partially limited by the disordered and entangled CNCs. Thus, the resilience of CNC–GT blocks shows a slight decrease when the growth time is longer than 90 min. The results indicate that the compressibility and resilience of CNC–GT can be improved by controlling the microstructures (alignment, order, density and length) of helical CNCs with super-elasticity. Excellent dimensional stability of the CNC–GT blocks at high temperature is also confirmed by a low coefficient thermal expansion (CTE) in the cross-plane direction (Fig. S6†). CNC–GT blocks with excellent resilience, high λ_⊥ and λ_‖, and good dimensional stability can be developed for highly thermally conductive interface materials for heat sealing.

4. Conclusions

3D hierarchical CNC–GT nanocomposite blocks were prepared by the growth of CNCs at the interlayer of GT followed by hot-pressing. The well-distributed growth of CNCs at the interface of EG relied on the catalyst loading controlled by vacuum impregnation. The intercalation of helical CNCs not only enhanced phonon mobility in the cross-plane direction, but also resulted in super-elasticity and good resilience. Compared with GT, CNC–GT blocks exhibited a remarkable increase in λ_⊥ with a higher density and lower porosity due to the presence of CNC nanofillers at the interlayer. λ_⊥ of CNC–GT blocks was further improved by increasing the growth time and the hot-pressing pressure. When the growth time of CNCs was 120 min, the maximum λ_⊥ of the CNC–GT blocks (∅ 3 cm × 2 mm) of up to 23.6 W m⁻¹ K⁻¹ was about five-fold higher than that of GT of 4.9 W m⁻¹ K⁻¹. The resilience ratio in the cross-plane direction of the CNC–GT blocks (84.1%) was 10.2% higher than that of the GT blocks due to the good elasticity of CNCs in 3D hierarchical structures. These properties were much better than those of various GT-based blocks observed previously. The results indicated that the macrostructures of stiff CNCs with suitable length and density at the interface of GT are crucial for the improvement of λ_⊥ and high resilience ratios of 3D hierarchical GT–CNC blocks. A low CTE in the cross-plane direction of the CNC–GT blocks also indicated good dimensional stability. CNC–GT blocks combining high λ_⊥, good mechanical resilience and dimensional stability are important candidates for highly thermally conductive and resilient interface materials for use in heat sealing.

Acknowledgements

This work was financially supported by the National Basic Research Program of China (Grant no. 2012CB626800 and 2010CB934700), the National Natural Science Foundation of China (Grant no. 51073115, 51173127, 51273144 and 51373116), the Research Fund for the Doctoral Program of Higher Education of China (Grant no. 20110032110067) and the Program for New Century Excellent Talents in University (NCET-13-0403).

Notes and references

1. J. Norley, Elecron. Cooling., 2001, 7, 50–51.
2. A. A. Balandin, Nat. Mater., 2011, 10, 569–581.
3. P. G. Klemens, J. Wide Bandgap Mater., 2000, 7, 332–339.
4. W. S. Kuo, T. H. Ko and H. F. Lu, Micro Nano Lett., 2010, 5, 219–221.
5. Z. H. Feng, T. Q. Li, Z. J. Hu, G. W. Zhao, J. S. Wang and B. Y. Huang, Carbon, 2012, 50, 3947–3948.
6. H. F. Lu, W. S. Kuo and T. H. Ko, Nanoscale Microscale Thermophys. Eng., 2011, 15, 209–219.
7. A. V. Bershtein, V. M. Gun'ko, L. M. Egorova, Z. W. Wang, M. Illsley, E. F. Voronin, G. P. Prikhod'ko, P. N. Yakushev, R. Leboda, J. Skubiszewska-Zieba and S. V. Mikhalovsky, RSC Adv., 2012, 2, 1424–1431.
8. X. J. Tian, M. E. Itkis, E. B. Bekyarova and R. C. Haddon, Sci. Rep., 2013, 3, 1710.
9. M. S. Dresselhaus, G. Dresselhaus, J. C. Charlier and E. Hernández, Philos. Trans. R. Soc. London, Ser. A, 2004, 362, 2065–2098.
10. R. S. Ruoff and D. C. Lorents, Carbon, 1995, 33, 925–930.
11. Z. L. Wang, D. W. Tang, X. H. Zheng, W. G. Zhang and Y. T. Zhu, Nanotechnology, 2007, 18, 475714.
12. Y. K. Choi, K. I. Sugimoto, S. M. Song and M. Endo, Mater. Lett., 2005, 59, 3514–3520.
13. A. A. Balandin, S. Ghosh, W. Z. Bao, I. Calizo, D. Teweldebrhan, M. Feng and C. N. Lau, Nano Lett., 2008, 8, 902–907.
14. C. Lee, X. D. Wei, J. W. Kysar and J. Hone, Science, 2008, 321, 385–388.
15. T. Tong, Y. Zhao, L. Delzeit, A. Kashani, M. Meyyappan and A. Majumdar, IEEE Trans. Compon. Packag. Technol., 2007, 30, 92–98.
16. M. Bonnissel, L. Luo and D. Tondeurb, Carbon, 2001, 39, 2151–2161.
17. M. T. Pettes, H. Ji, R. S. Ruoff and L. Shi, Nano Lett., 2012, 12, 2959–2965.
18. Q. Z. Liang, X. X. Yao, W. Wang, Y. Liu and C. P. Wong, ACS Nano, 2011, 5, 2392–2401.
19. B. Li, H. Y. Jung, H. Wang, Y. L. Kim, T. Kim, M. G. Hahm, A. Busnaina, M. Upmanyu and Y. J. Jung, Adv. Funct. Mater., 2011, 21, 1810–1815.
20. M. G. Hahm, H. Wang, H. Y. Jung, S. Hong, S. G. Lee, S. R. Kim, M. Upmanyu and Y. J. Jung, Nanoscale, 2012, 4, 3584–3590.
21. L. Q. Xu, N. Wei, Y. P. Zheng, Z. Y. Fan, H. Q. Wang and J. C. Zheng, J. Mater. Chem., 2012, 22, 1435–1444.
22. S. S. J. Aravind and S. Ramaprabhu, RSC Adv., 2013, 3, 4199–4206.
23. A. Aitkaliyeva, D. Chen and L. Shao, Sci. Rep., 2013, 3, 2774.
24. V. Varshney, S. S. Patnaik, A. K. Roy, G. Froudakis and B. L. Farmer, ACS Nano, 2010, 4, 1153–1161.
25. H. Ma, L. J. Pan, Q. Zhao, Z. B. Zhao and J. S. Qiu, Carbon, 2012, 50, 778–783.
26. X. Q. Chen, S. L. Zhang, D. A. Dikin, W. Q. Ding, R. S. Ruoff, L. J. Pan and Y. Nakayama, Nano Lett., 2003, 3, 1299–1304.
27. L. Z. Liu, H. L. Gao, J. J. Zhao and J. P. Lu, Nanoscale Res. Lett., 2010, 5, 478–483.
28. A. K. Roy, B. L. Farmer, V. Varshney, S. W. Sihn, J. H. Lee and S. Ganguli, ACS Appl. Mater. Interfaces, 2012, 4, 545–563.
29. T. Y. Tsou, C. Y. Lee and H. T. Chiu, ACS Appl. Mater. Interfaces, 2012, 4, 6505–6511.
30. Y. C. Wang, T. K. Huang, S. H. Tung, T. M. Wu and J. J. Lin, Sci. Rep., 2013, 3, 2621.
31. G. M. Yuan, X. K. Li, Z. J. Dong, A. Westwood, Z. W. Cui, Y. Cong, H. D. Du and F. Y. Kang, Carbon, 2012, 50, 175–182.
32. Y. Zhao, J. L.Shi, H. Q. Wang, Z. C. Tao, Z. J. Liu, Q. G. Guo and L. Liu, Carbon, 2013, 51, 427–434.
33. L. W. Wang, Z. Tamainot-Telto, S. J. Metcalf, R. E. Critoph and R. Z. Wang, Appl. Therm. Eng., 2010, 30, 1805–1811.
34. L. W. Wang, S. J. Metcalf, R. E. Critoph, R. Thorpe and Z. Tamainot-Telto, Carbon, 2011, 49, 4812–4819.
35. Y. Leng, J. L. Gu, W. Q. Cao and T. Y. Zhang, Carbon, 1998, 36, 875–881.
36. T. Wei, K. Wang, Z. J. Fan, J. Yan, W. Z. Qian and F. Wei, Carbon, 2010, 48, 305–312.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3ra45647a

---