Enhanced thermal conductivity for poly(vinylidene fluoride) composites with nano-carbon fillers

Abstract

Polymer composites with high thermal conductivity have recently attracted much attention, along with the rapid development of electronic devices toward higher speeds and performance. Here, we reported a facile method to prepare poly(vinylidene fluoride) (PVDF) composites with nano-carbon fillers including zero-dimensional superfullerene (SF), one-dimensional carbon nanotubes (CNT) and two-dimensional graphene sheets (GS) by simple solution blending and compression molding. The effects of these nano-carbon fillers on the thermal conductivity of PVDF composites were systematically investigated. It was found that PVDF composites exhibit a higher thermal conductivity than that of neat PVDF. Among them, the thermal conductivity of PVDF composites with 20 wt% two-dimensional GS reaches a maximum (2.06 W m⁻¹ K⁻¹), which is approximately 10-fold enhancement in comparison to that of the neat PVDF. Such highly thermal conductive PVDF composites may enable some prospective applications in advanced thermal management.

---

1. Introduction

Polymer composites with excellently thermal conductivity are playing an important role in replacing traditional metal and ceramic parts in several applications, such as heat exchangers, electronic devices, and generators, etc., due to the polymer advantages including low mass, corrosion resistance and good formability.^1–4 Among polymer composites, poly(vinylidene fluoride) (PVDF) is often chosen as the polymer matrix because of its good chemical resistance, mechanical strength, high permittivity, excellent piezoelectric and pyroelectric properties.^5–8 However, the thermal conductivity of PVDF is in the order of 0.1 W m⁻¹ K⁻¹, which can't meet the heat-dissipating demands of electric devices and electrical equipment.⁹ Therefore, it is crucial to dissipate the heat generated from devices in an effective manner and keep the operating temperature at a normal level.

Over the last decade, a large amount of investigations have been conducted in pursuit of improving the thermal conductivity of polymer composites. Adding fillers with high thermal conductivity to polymer matrix to fabricate composite is believed to be a simple and effective way, such as silicon carbide (SiC),^10,11 silica (SiO₂),^12,13 silicon nitride (Si₃N₄),^14,15 boron nitride (BN),^16,17 alumina (Al₂O₃),^18,19 aluminum nitride (AlN).^20,21 In these cases, a high content of fillers is often needed to attain percolation thresholds of thermal conductivity of polymer to form a successive thermal conductive network, which consequently brings about a high density and poor mechanical properties. Comparison with these ceramic materials, nano-carbon fillers exhibit significant advantages in enhancing the thermal conductivity of polymer composites at a low fraction while retaining the density and mechanical properties.^22,23 Graphene sheet (GS) has been considered a filler for promoting thermal conductivity of the polymer matrix due to its high intrinsic thermal conductivity (theoretical 5300 W m⁻¹ K⁻¹)²⁴ and large specific surface area. Meanwhile, carbon nanotube (CNT) was utilized to reinforce the heat dissipation in polymer composites as a result of their unique one dimensional structure, high aspect ratio and high thermal conductivity (about 3000 W m⁻¹ K⁻¹).^25,26 Besides, fullerene and its derivative were also used to enhance thermal stability of polymer.^27,28

In this study, we proposed a facile approach to fabricate PVDF composites where nano-carbon fillers are well dispersed in matrix by a simple solution blending and investigated the effect of different dimensional nano-carbon fillers (superfullerene, CNT, and GS) on the thermal conductivity of PVDF matrix. To the best of our knowledge, no one has reported that SF was chosen as filler for preparing PVDF composites before, especially for the research on the thermal conductive properties.

2. Experimental

2.1. Materials

Commercial graphene sheets (GS) with a mean particle size of <20 μm were produced by Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences (China). CNT with 1–25 μm in length and 40 nm in diameter are obtained from the CNT Co. Ltd, Incheon (Korea). Superfullerene (SF) was supplied by Southwest University of Science and Technology (China). PVDF powder was purchased from 3F Co. Ltd, Shanghai (China). N,N-Dimethylformamide (DMF) with purity of 99.5% was purchased from Sinopharm Chemical Reagent Co. Ltd, Shanghai (China) and used without further purification.

2.2. Preparation of PVDF/nano-carbon composites

A solution method was adopted to fabricate the PVDF composites. Firstly, nano-carbon filler (SF or CNT or GS) was dissolved in DMF (1 mg ml⁻¹) and dispersed under ultrasonication. Meanwhile, poly(vinylidene fluoride) (PVDF) was dissolved in DMF at 60 °C. Then the prepared suspension of fillers/DMF and the solution of PVDF/DMF were poured into a three-necked flask and the resulting mixture was stirred with an electric stirrer at a speed of 3000 rpm. At the same time, distillation process was carried out and the temperature was set at 175 °C. Finally, the residual after distillation was dried in a vacuum oven at 120 °C for 3 h to eliminate residual DMF, and dried mixture was compressed into films with a thickness ranging from to 1.2 to 2.0 um at 180 °C under a pressure of about 10 MPa. The detailed preparation process is graphically exhibited in Scheme 1.

Scheme 1 The preparation process of PVDF/nano-carbon composites.

2.3. Characterizations

The X-ray diffraction (XRD) patterns of the specimens were carried out using a D8 DISCOVER with GADDS (BRUKER Ltd. Germany) at a scan rate of 10 °C min⁻¹ with a 2θ ranging from 10° to 90° with Cu Kα radiation (λ = 1.5406 Å) at room temperature. The Raman spectra were recorded using a Reflex Raman system (RENISHAW plc, Wotton-under-Edge, UK) with a laser wave length of 532 nm. X-ray photoelectron spectroscopy (XPS) was carried out with an AXIS Ultra DLD spectrometer (Kratos, Japan). Nuclear Magnetic Resonance (NMR) was obtained from 400 MHz AVANCE III (Bruker, Switzerland). The carbon fillers and fractured surfaces of PVDF samples were investigated by a field emission scanning electron microscopy (FE-SEM, QUANTA FEG250, USA) and a field emission scanning electron microscopy (FE-SEM, HITACHI S4800, Japan) at an accelerating voltage of 20 kV, 8 kV, respectively, and the fractured surfaces were sputtered with a thin layer of gold to avoid the accumulation of charge. The optical micrograph (OM) image was captured by optical microscope (OM, Leica DM2500M, Germany). The microstructures of carbon fillers were observed from JEOL JEM-2100 (Electron Optics Laboratory CO., Ltd, Japan) instrument with an acceleration voltage of 200 kV, and samples for TEM measurements were fabricated by drop casting on carbon-coated copper grids followed by solvent evaporation in the air. Thermal conductivities of the composites were measured using LFA 447 Nanoflash (NETZSCH, Germany). Samples were prepared in cylindrical shape of 12.7 mm in diameter and 1.0–2.0 mm in thickness. Differential scanning calorimetry (DSC) experiments were carried out under nitrogen by a Pyris Diamond DSC (Perkin-Elmer, U.S.A). Thermogravimetric analysis (TGA) were taken by a TGA 209 F3 (NETZSCH, Germany). The samples were close to 10 mg and all the measurements were performed under nitrogen atmosphere. Dynamic mechanical analysis (DMA) were performed on a DMA Q800 dynamic mechanical analyzer (TA instruments, USA), operating in the tension mode at an oscillation frequency at 1 Hz. The IR-photos were captured by infrared camera (Fluke, Ti400, U.S.A.).

3. Results and discussion

3.1. Characterization of nano-carbon fillers

The morphologies of SF, CNT and GS were characterized using SEM and TEM, as shown in Fig. 1. Fig. 1(a) shows that there are many SF representing ball-like structures, which gather together in the carbon substrate. In order to further understanding of the structure of SF, TEM was used to observe SF as shown in Fig. 1(b). Shown in Fig. 1(b), the SF is sphere-like structure with a diameter of between 50 and 100 nm. Besides, SF was characterized by AVANCE III (Bruker, Switzerland) and its ¹³C and ¹H NMR is present in Fig. S1.† Fig. 1(c) clearly shows that CNT are likely to aggregate to form crowed CNT bundles because of van der Waal forces. The TEM image in the Fig. 1(d) further demonstrates the microstructure of CNT with entanglement behaviors. Besides, there are some black spots existing in the structure of CNT, indicating that CNT is impure. From Fig. 1(e), one can observe that GS possess an amount of platelets with wrinkles. GS also tend to fold and aggregate as the same with CNT and act like a bunch of flower with a size ranging from a few to approximately 20 um. The optical micrograph (OM) image of GS indicates the aggregation either, as shown in Fig. S2.† Fig. 1(f) presents a typical TEM image of GS. It reveals that GS are quite thin and possess transparent sheet structure which resembles silk appearance. In addition, GS exhibits overlapping in some district due to scroll with each other.

Fig. 1 FE-SEM images of (a) SF, (c) CNT, (e) GS and TEM image of (b) SF, (d) CNT, (f) GS.

XRD measurement is employed to study the structure of SF, CNT and GS, as shown in Fig. 2(a). It is clearly observed that SF does not have an obvious peak in comparison to GS and CNT, indicating that SF is not crystal. GS and CNT represent a sharp diffraction peak occurring at 26.32° and 25.96°, respectively, corresponding to (0 0 2) graphite plane made up of well-aligned graphene with an interlayer spacing of 0.335 nm.²⁴ The Raman spectra is an effective and clipping method tool to study crystalline, nanocrystalline, and the amorphous of graphitic base materials.^29,30 The D band at ∼1330 cm⁻¹ is usually associated with disordered sp³-hybridized carbon which becomes active at the edge.³¹ The G band at ∼1580 cm⁻¹ is usually representing in-plane bond stretching motion of sp² bonded carbon atoms.³² In addition, the intensity ratio from the D band to G band (I_D/I_G) demonstrates the structure quality of SF, CNT and GS. It can be seen from Fig. 2(b) that GS and CNT have three peaks at around 1348, 1582, 2719 cm⁻¹ and 1343, 1572, 2677 cm⁻¹, respectively, corresponding to the D band, the G band and the 2G band and the I_D/I_G ratio are 0.14, 0.59, respectively. Quite different from GS and CNT, SF does not have obvious D band and only possesses a weak and wide peak at 1585 cm⁻¹.

Fig. 2 (a) XRD pattern and (b) Raman spectra of SF, CNT and GS. (c, e and g) XPS survey scans of SF, CNT and GS. (d, f and h) The corresponding C1s XPS spectra of SF, CNT and GS.

XPS analysis was employed to clarify the surface composition and explore the functional groups and the results are shown in Fig. 2. From Fig. 2(c), one can see that there are three main peaks assigned to C1s, O1s and Cl2p and the proportions are 80.2%, 16.2% and 3.6%, respectively. Furthermore, it should note that the proportion of oxygen in SF is much more than GS and CNT, which means SF exits more functional groups containing oxygen atoms. Fig. 2(d) confirms that the deconvolution of C1s peaks occurs at 284.6, 286.0, and 288.2, originated from the C=C, C–O and C=O, respectively. Fig. 2(e) and (f) show that CNT chiefly contains carbon (98.2%) and oxygen (1.8%) and the specific C1s peaks emerge at 284.6, 285, 285.9, and 288.1 which corresponds to C=C, C–C, C–O and O–C=O, respectively. Fig. 2(g) presents the XPS survey spectra of GS in the range of 0–1200 eV and quantify the surface elements composition, namely: carbon (94.4%), oxygen (3.0%), and nitrogen (2.6%). Detailed analysis of the XPS spectra offers clear evidence that GS is made up of different functional groups and the C1s core level spectra are illustrated in Fig. 2(h). It revealed peaks appearing at 284.6, 285, 286.3, 287.3 and 286.9 which derives from the C=C, C–C, C–OH, C=O, and C–O–C, respectively.³³ Thus, these structures and functional groups are also consistent with Fig. 2(d) and (f).

3.2. Morphology of composites

In order to investigate the dispersion and compatibility of nano-carbon fillers in the PVDF matrix, the morphology of the fractured surfaces of neat PVDF and PVDF composites were observed by SEM, as shown in Fig. 3. It can be seen from Fig. 3(a) that the fracture surface of neat PVDF is quite smooth with a few cracks, which is the representative fracture way of polyolefin and the reason may be ascribed to the fact that surface cannot elastically release in the case of external force. For the PVDF composites, the rough and ridged structure characteristic on fracture were displayed in Fig. 3(b–d). Fig. 3(b) shows that the fractured surface of PVDF/SF composites owns a rougher fracture than neat PVDF and appears as ripples. It reveals interaction between the SF and PVDF matrix transferring the exterior force to filler. For PVDF composites containing CNT, the SEM image indicates that CNT exhibit an excellent compatibility in the PVDF matrix, as shown in Fig. 3(c). Besides, CNT show poor dispersion in PVDF matrix and form large agglomerates due to strong van der Waals forces. Compared to the PVDF/CNT composites, the fracture of the PVDF/GS composites shows that GS present good dispersion and homogeneity in the matrix and form cross-linked network, as shown in Fig. 3(d). More interesting, there are some pulled out GS appearing at fracture surface while the sample was broken into two parts, which may be attribute to strong interfacial interaction between matrix and GS filler.

Fig. 3 SEM images of the fractured surface of (a) neat PVDF and (b) PVDF/SF composite (c) PVDF/CNT composite and (d) PVDF/GS composite.

3.3. Thermal conductivity of composites

Thermal conductivity shows the performance of heat transfer and is affected by many factors, such as filler structure, dispersion, thermal interfacial resistance, and temperature measured. Fig. 4(a) presents thermal diffusivity and thermal conductivity of neat PVDF and PVDF composites with 20 wt% nano-carbon filler loading. It is evident that the thermal diffusivity of neat PVDF and PVDF composites containing SF, CNT, GS are 0.094, 0.096, 0.299 and 1.02 mm² s⁻¹, respectively, their corresponding thermal conductivity (0.189, 0.194, 0.604 and 2.06 W m⁻¹ K⁻¹). By comparing thermal conductivity, we can draw the conclusion that GS have a more remarkable thermal enhancement. Besides, the heating DSC thermograms of neat PVDF and PVDF composites was obtained by Pyris Diamond DSC (Perkin-Elmer, American) and shown in Fig. S3.† From Fig. S3,† we can clearly see the melting temperature of neat PVDF and PVDF composites and find that the change trend of melting temperature of PVDF composites is similar to the thermal conduction properties. Fig. 4(b) clarifies the thermal conductivity enhancement (TCE) of PVDF composites prepared with SF, CNT and GS in detail. Obviously, PVDF/GS composites exhibit maximum thermal conductivity enhancement and reach approximately a 10-fold enhancement in comparison with neat PVDF, whereas the thermal conductivity enhancement of PVDF composites containing SF and CNT only achieve 0.02 and 2, as shown in the inset of Fig. 4(b), which is dislike the GS filler for enhancing thermal conductivity of PVDF composites. For SF and CNT, even for a direct contact, the geometry of the junctions between two crossed nanoparticles or nanotubes results in a point contact with extremely small contact area. Thus the interaction between filler and filler is very weak and the wrapping the polymer provides the dominant contributed to the heat flow. The higher thermal conductivity PVDF/GS composites may be attributed to the following two factors: (1) large surface aspect of GS improve the interaction between GS and PVDF matrix.²⁶ (2) The strong interface interaction between GS and PVDF matrix provides good interface compatibility and reduces interfacial thermal resistance.³⁴

Fig. 4 (a) Thermal diffusivity and thermal conductivity of neat PVDF and PVDF composites. (b) Thermal conductivity enhancement (TCE) of PVDF composites compared to neat PVDF. (c) Infrared images of neat PVDF, PVDF/SF, PVDF/CNT, and PVDF/GS composites upon heating. The temperature gradient scale bar at left shows the highest and lowest temperatures of 100 °C and 29 °C, respectively. (d) Surface temperature variation of the neat PVDF, PVDF/SF, PVDF/CNT, and PVDF/GS composites with time upon heating and cooling event. Cooling started at the 556 seconds for each sample.

To visually verify the heat dissipation performance of neat PVDF and PVDF composites, we make use of an infrared camera. Before taking pictures, we made some necessary preparations. First of all, four samples are put on a heater in order of neat PVDF, PVDF/SF, PVDF/CNT, and PVDF/GS and each size of which is 12.7 mm (diameter) × 1.45 mm (thickness). Meantime, it should be noted that the system was kept at room temperature before heating. Then we started to heat the heater and utilized an infrared camera to take some photos to record thermal state of four samples, as shown in Fig. 4(c). With the temperature of heater increasing, it is easy to observe that the surface color of each sample is becoming more and more bright, especially, the color of PVDF composites is quite brighter in comparison to neat PVDF. Specifically, when the heater temperature rose to 130 °C within 45 s, the temperature of neat PVDF, PVDF/SF, PVDF/CNT, and PVDF/GS reached 71, 76, 91 and 100 °C, respectively. This set of data clearly presents that PVDF/GS composites exhibit a better heating dissipation behavior, which is consistent with thermal conductivity in Fig. 4(a). Besides, quantitative heating and cooling results are exactly measured by an equipment containing a hot plate, a thermocouple element and versatile voltmeter, as shown in Fig. 4(d). We can easily observe the heating and cooling trend of neat PVDF and PVDF composites. It is obvious that PVDF composites with SF, CNT, and GS reach their highest temperatures at 107, 121, and 142 °C, respectively, whereas neat PVDF just ascends to 106 °C by the end of 556 seconds, which is consistent with infrared images. Besides, Fig. S4† shows the TGA and DTG curves of neat PVDF and PVDF composites and indicates that PVDF composites have better thermal stability than neat PVDF. To further demonstrate the release performance of neat PVDF and PVDF composites, we deal with the cooling curve with differential treatment ranging from 540 to 650 seconds as shown in the inset of Fig. 4(d). Apparently, PVDF composites containing GS are the first down to room temperature, followed by PVDF/CNT, PVDF/SF, and neat PVDF. In summary, PVDF composites exhibit a better heating and cooling behaviors compared to neat PVDF. Among all of composites, PVDF/GS composites are illustrated to be most excellent, which means GS are the most promising filler to enhance the thermal conductivity of PVDF composite.

In order to further understand the thermal conductive mechanism of PVDF composites, a visualized model was proposed to present the thermal path of PVDF composites and is shown in Fig. 5. PVDF/SF composite does not exhibit a good thermal conductivity because of the following two factors: (1) SF was separated by the polymer layer which hinders the SF from contacting each other. (2) The phonon mismatch between SF particles and polymer matrix results in a large thermal interface resistance. With regard to phonon, it is believed that phonon is a quasi-particle inspired by crystalline structure and its chemical potential is zero.³⁵ It is a quantum mechanical description of an elementary vibrational motion in which a lattice of atoms or molecules uniformly oscillates at a single frequency.³⁶ Besides, phonon is boson carrying a quasi-momentum, which obeys the Bose–Einstein statistics. Phonon mismatch represents that phonon cannot be absorbed by crystalline structure in the process of transmission. Namely, the energy phonon carries is not susceptive to crystal lattice. The heat flow model of the PVDF/SF composite is first shown in Fig. 5. For PVDF/CNT composites, there is an obvious improvement in thermal conductivity in comparison to PVDF/SF composite, which may originate from the unique one dimensional structure, high intrinsic conductivity and high aspect ratio of CNT. On the other hand, the thermal conductivity enhancement in CNT filler polymers does not attain the theoretically predicted value because of large thermal interface resistance caused by phonon mismatch. The thermal conductive path can be seen in the middle of Fig. 5. It has been suggested that for CNT incorporated into a polymer matrix the formation of a percolation network for heat flow is suppressed due to the presence of a thin polymer layer, which separates the CNT in the network and interrupt the direct CNT–CNT phonon transfer. Even in case of a direct contact between CNT and CNT, the contact area is extremely small. Thus, the interaction between two entangled CNT is very weak and the wrapping polymer matrix provides the dominant contribution for heat conduction.^26,37 However, based on our research, one can see that GS present an extraordinary enhancement in thermal conductivity in PVDF matrix and the heat dissipating model is also illustrated in the right of Fig. 5. It is known to us that PVDF is a semi-crystalline polymer containing amorphous regions and semi-crystalline regions and the boundary between these two regions exits remarkable thermal interface resistance which was greatly hinder the heat flow in PVDF. However, when GS are added to PVDF and the loading of GS is low and does not achieve the percolation, GS can serve as a bridge between PVDF spherulites, resulting in heat transfer from spherulites to spherulites. When the amount of GS is beyond percolation, a direct contact will be taken place between GSs and thus leading to form an effective thermally conductive pathway. If we continue to add GS, a network thermal path will be formed because of the mutual connect between local chain and thermal conductive path. To sum up, two-dimensional GS exhibit an outstanding enhancement in thermal conductivity of polymer matrix and can be used as a promising filler applied for thermal interface materials in the future.

Fig. 5 The model of heat flow for the PVDF composites with SF, CNT, and GS.

4. Conclusions

In summary, PVDF/nano-carbon composites were fabricated by a simple solution blending and compression molding. The effect of different dimensional nano-carbon fillers (SF, CNT, and GS) on the thermal conductivity of PVDF composites was systematically investigated. It was found that the thermal conductivity of the PVDF/SF, PVDF/CNT, and PVDF/GS composites with 20 wt% are 0.194, 0.604 and 2.06 W m⁻¹ K⁻¹, which corresponds to 0.02, 2, and 10-fold enhancement in comparison to that of neat PVDF, respectively. Considered the high thermal conductivity, PVDF/GS composites are expected to be useful as interfacial materials for thermal management applications.

Acknowledgements

The authors are grateful for the financial support by the National Natural Science Foundation of China (51303034, 51573201), Natural Science Foundation of Ningbo (2014A610111), Public Welfare Project of Zhejiang Province (2016C31026) and International Science and Technology Cooperation Program of Ningbo (2015D10003) and Shanxi (201603D421024) for financial support. We also thank the Chinese Academy of Science for Hundred Talents Program, Chinese Central Government for Thousand Young Talents Program.

References

1. S. Deng, J. Wang, G. Zong, F. Chen, S. Chai and Q. Fu, RSC Adv., 2016, 6, 10185–10191.
2. Z. Han and A. Fina, Prog. Polym. Sci., 2011, 36, 914–944.
3. U. Kalsoom, A. Peristyy, P. N. Nesterenko and B. Paull, RSC Adv., 2016, 6, 38140–38147.
4. R. Qian, J. Yu, C. Wu, X. Zhai and P. Jiang, RSC Adv., 2013, 3, 17373–17379.
5. B. Ameduri, Chem. Rev., 2009, 109, 6632–6686.
6. S. B. Lang and S. Muensit, Appl. Phys. A, 2006, 85, 125–134.
7. A. J. Lovinger, Science, 1983, 220, 1115–1121.
8. L. Priya and J. P. Jog, J. Polym. Sci., Part B: Polym. Phys., 2002, 40, 1682–1689.
9. C. T'Joen, Y. Park, Q. Wang, A. Sommers, X. Han and A. Jacobi, Int. J. Refrig., 2009, 32, 763–779.
10. G.-W. Lee, M. Park, J. Kim, J. I. Lee and H. G. Yoon, Composites, Part A, 2006, 37, 727–734.
11. G. A. Slack, J. Appl. Phys., 1964, 35, 3460.
12. I. A. Rahman and V. Padavettan, J. Nanomater., 2012, 2012, 8.
13. H. Zou, S. Wu and J. Shen, Chem. Rev., 2008, 108, 3893–3957.
14. F. L. Riley, J. Am. Ceram. Soc., 2000, 83, 245–265.
15. W. Zhou, C. Wang, T. Ai, K. Wu, F. Zhao and H. Gu, Composites, Part A, 2009, 40, 830–836.
16. H. Ishida and S. Rimdusit, Thermochim. Acta, 1998, 320, 177–186.
17. C. Zhi, Y. Bando, C. Tang, H. Kuwahara and D. Golberg, Adv. Mater., 2009, 21, 2889–2893.
18. Y. Shimazaki, F. Hojo and Y. Takezawa, Appl. Phys. Lett., 2008, 92, 133309.
19. Y. Shimazaki, F. Hojo and Y. Takezawa, ACS Appl. Mater. Interfaces, 2009, 1, 225–227.
20. J.-W. Bae, W. Kim, S.-H. Cho and S.-H. Lee, J. Mater. Sci., 2000, 35, 5907–5913.
21. S. Kume, I. Yamada, K. Watari, I. Harada and K. Mitsuishi, J. Am. Ceram. Soc., 2009, 92, S153–S156.
22. M. Bozlar, D. He, J. Bai, Y. Chalopin, N. Mingo and S. Volz, Adv. Mater., 2010, 22, 1654–1658.
23. K. M. F. Shahil and A. A. Balandin, Nano Lett., 2012, 12, 861–867.
24. A. A. Balandin, S. Ghosh, W. Bao, I. Calizo, D. Teweldebrhan, F. Miao and C. N. Lau, Nano Lett., 2008, 8, 902–907.
25. P. Kim, L. Shi, A. Majumdar and P. L. McEuen, Phys. Rev. Lett., 2001, 87, 215502.
26. A. Yu, P. Ramesh, X. Sun, E. Bekyarova, M. E. Itkis and R. C. Haddon, Adv. Mater., 2008, 20, 4740–4744.
27. L. Derue, O. Dautel, A. Tournebize, M. Drees, H. Pan, S. Berthumeyrie, B. Pavageau, E. Cloutet, S. Chambon, L. Hirsch, A. Rivaton, P. Hudhomme, A. Facchetti and G. Wantz, Adv. Mater., 2014, 26, 5831–5838.
28. C. Lindqvist, J. Bergqvist, O. Backe, S. Gustafsson, E. Wang, E. Olsson, O. Inganas, M. R. Andersson and C. Muller, Appl. Phys. Lett., 2014, 104.
29. A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Piscanec, D. Jiang, K. S. Novoselov, S. Roth and A. K. Geim, Phys. Rev. Lett., 2006, 97, 187401.
30. A. C. Ferrari and J. Robertson, Phys. Rev. B: Condens. Matter Mater. Phys., 2000, 61, 14095.
31. A. C. Ferrari, Solid State Commun., 2007, 143, 47–57.
32. F. Tuinstra, J. Chem. Phys., 1970, 53, 1126.
33. Y. Wang, J. Yu, W. Dai, Y. Song, D. Wang, L. Zeng and N. Jiang, Polym. Compos., 2015, 36, 556–565.
34. J. Yu, X. Huang, L. Wang, P. Peng, C. Wu, X. Wu and P. Jiang, Polym. Chem., 2011, 2, 1380–1388.
35. J. J. Sakurai, Advanced Quantum Mechanics, Pearson Education, India, 1967.
36. S. H. Simon, The Oxford Solid State Basics, OUP, Oxford, 2013.
37. N. Shenogina, S. Shenogin, L. Xue and P. Keblinski, Appl. Phys. Lett., 2005, 87, 133106.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra11178e

---