Electrospun carbon nanofibrous mats surface-decorated with Pd nanoparticles via the supercritical CO₂ method for sensing of H₂

Abstract

Carbon nanofibers (in the form of an overlaid mat) were prepared by electrospinning of polyacrylonitrile followed by stabilization in air and carbonization in argon. Pd nanoparticles were then deposited on the surface of nanofibers by using Pd(acac)₂ as the precursor via the supercritical CO₂ method followed by pyrolysis at 600 °C. Structural characterizations indicated that the Pd nanoparticles were composed of single-crystalline clusters with sizes of a few nanometers. The hydrogen sensing properties of electrospun carbon nanofibrous mats surface decorated with Pd nanoparticles were investigated. The results indicated that the resistivity of Pd-decorated nanofibrous mats decreased linearly with the increase of H₂ volume fraction from 0 to 0.7 in H₂/N₂ mixture gas, and the response showed excellent reversibility.

---

Introduction

Carbonaceous nanomaterials (e.g., carbon nanotubes/nanofibers and graphenes) have been investigated extensively in the recent years. The superior properties of these materials could lead to a variety of applications such as reinforcement of composites and sensing of chemical/biological species.^1–6 Presently, the carbon nanofibers prepared via the technique of electrospinning followed by the heat treatments of stabilization and carbonization have attracted growing interest.^7,8 Unlike carbon nanotubes/nanofibers that are prepared by bottom-up synthetic methods (e.g., chemical vapour deposition), electrospun carbon nanofibers are produced through top-down nano-manufacturing processes. Hence, these carbon nanofibers are cost-effective; and the production can be readily scaled up.

The nanomaterial-processing technique of electrospinning utilizes the electric force to drive the spinning process and to produce the fibers with diameters typically ranging from tens to hundreds of nanometers (known as “electrospun nanofibers”), and the nanofibers are usually collected in the form of an overlaid mat.^9–11 Electrospun polyacrylonitrile (PAN) nanofibers are the commonly selected precursor for preparing carbon nanofibers. Electrospun carbon nanofibers have been studied to fabricate nanocomposites and/or hybrid multi-scaled composites;^7,8 they have also been studied for anode materials in batteries, for the support of catalysts in electrochemical/fuel cells, and for the counter electrode in dye-sensitized solar cells.^8,12 However, there have been few studies on the electrical properties of electrospun carbon nanofibers and their applications as the electronic and/or sensing materials.^13–15

Physical/mechanical properties of PAN-based carbonaceous materials are strongly affected by carbonization temperature. We have recently studied the temperature-resistivity dependence of electrospun PAN-based carbon nanofibers prepared at different carbonization temperatures.¹³ The results indicated that the electrospun carbon nanofibers behaved as typical semiconducting materials, and the activation energy decreased with the increase of carbonization temperature. In this study, these materials were investigated for making resistor-type H₂ sensors. The results suggested that the overlaid mats consisting of electrospun carbon nanofibers would be low-cost, flexible, and highly sensitive nanomaterials for sensing of H₂.

Carbonaceous nanomaterials such as carbon nanotubes have been investigated for sensing of H₂. To improve the sensitivity, these nanomaterials have to be surface-modified to introduce specific interactions with H₂; for example, the sensitivity of carbon nanotubes to H₂ can be substantially improved upon surface-deposition of Pd nanoparticles.^16–20 In general, Pd nanoparticles can be deposited on carbonaceous nanomaterials through physical vapor deposition¹⁶ or electrochemical deposition.^17,18 The physical vapor deposition produces the Pd nanoparticles with wide distribution in size and requires high-vacuum evaporation equipment, resulting in increased usage of expensive source materials and high fabrication cost. The electrochemical methods are simple and cost-effective;¹⁸ however, the methods involve chemical reactions that may introduce impurities and/or have the adverse effects on electrical properties.¹⁷ The supercritical CO₂ (SCCO2) method is well-known for the deposition of metallic nanostructures under mild conditions.^21,22 The method has many advantages including the controllable solubility, high surface diffusivity of reactants, low critical temperature, and effective generation of metal nanoparticles onto nanostructures.

In our previously reported research,²³ overlaid mats consisting of electrospun PAN nanofibers were first treated with NH₂OH to introduce amidoxime [–C(NH₂)=N(OH)] groups for chelating Pd²⁺ ions; these ions were then reduced by NH₂NH₂ into Pd nanoparticles. Finally, the amidoxime-functionalized PAN mat with Pd nanoparticles attached on fiber surfaces was stabilized in air and carbonized in argon. Since some nitrile (–C≡N) groups in the PAN precursor nanofibers (particularly on the fiber surfaces) were converted into amidoxime groups that could no longer undergo the cyclization reaction during stabilization, the resulting carbon nanofibers (particularly the fiber surfaces) possessed chaotic structures that had adverse effects on electrical properties. In this study, the overlaid mats consisting of electrospun PAN-based carbon nanofibers were prepared prior to surface-deposition of Pd nanoparticles via the SCCO2 method. Thereafter, the prepared electrospun carbon nanofibrous mats surface-decorated with Pd nanoparticles were investigated as the innovative nanomaterials for sensing of H₂.

Experimental

Materials

The polyacrylonitrile (PAN) used in this study was the Special Acrylic Fibers (SAF 3K) provided by Courtaulds, Ltd. (Nottingham, UK). The SAF 3K fibers were in the form of bundles with 3000 individual microfibers of a PAN copolymer. The copolymer was synthesized from acrylonitrile together with 1.2 wt% of itaconic acid and 6.0 wt% of methyl acrylate. Acetone, N,N-dimethylformamide (DMF), and palladium(II) acetylacetonate (Pd(acac)₂) were purchased from Sigma-Aldrich Chemical Co. (St. Louis, MO) and used without further purification.

Preparation of electrospun carbon nanofibrous mats

The preparation and characterization of electrospun carbon nanofibrous mats have been reported in a previous publication.¹³ In brief, a solution of 15 wt% PAN in DMF was placed in a syringe with a stainless-steel needle. During the electrospinning process, a positive voltage of 25 kV was applied to the needle, and the electrospun PAN nanofibrous mat was collected on the electrically grounded aluminum foil which covered a laboratory-produced roller with diameter of 25 cm. The distance between the tip of the needle and the collector was ∼25 cm. The collected PAN nanofiber mat had a thickness of ∼200 μm and a mass per unit area of ∼50 g m⁻². Subsequently, the PAN nanofibrous mat was stabilized in air at 280 °C for 6 h followed by being carbonized in argon at 700 °C or 900 °C for 1 h. The electrospun carbon nanofibrous mats were examined with a Zeiss Supra 40 VP field-emission scanning electron microscope (SEM) equipped with a PGT energy-dispersive X-ray spectrometer (EDS).

Deposition of Pd nanoparticles via the SCCO2 method

Prior to deposition of Pd nanoparticles, an electrospun carbon nanofibrous mat was first treated in 3.6 M HNO₃ at 90 °C for 1 h; the treated mat was then rinsed with de-ionized water for several times and dried at 110 °C for 1 h. The treatment would introduce functional groups (e.g., carbonyl and carboxyl) on the surface of carbon nanofibers,^24,25 which could facilitate the attachment of Pd(acac)₂ during the following SCCO2 process. Subsequently, the treated mat (0.1 g) was placed in a stainless steel basket in the upper part of a high-pressure vessel. The Pd precursor of Pd(acac)₂ (0.05 g) and 10 ml acetone were placed in a copper boat at the bottom of vessel. Hence, the carbon nanofibrous mat was not in direct contact with the precursor. Liquid CO₂ was then pumped into the vessel with high pressure up to 32 MPa; thereafter, the vessel was kept at 32 MPa and 150 °C for 24 h. During this process, the Pd precursor was dissolved in SCCO2/acetone and was carried onto the surfaces of electrospun carbon nanofibers. Finally, the carbon nanofibrous mat surface-attached with Pd(acac)₂ was heated in argon to 600 °C at 5 °C min⁻¹ and then kept at 600 °C for 1 h to convert Pd(acac)₂ into Pd. The electrospun carbon nanofibrous mats surface-decorated with Pd nanoparticles were characterized using a Zeiss Supra 40 VP SEM equipped with EDS and a JEOL JEM-2100 LaB6 transmission electron microscope (TEM) equipped with EDS.

Evaluation of hydrogen sensing properties

The resistivity and the hydrogen sensing characteristics (such as gas sensitivity, response time, etc.) were evaluated at room temperature (24 ± 2 °C) and ambient pressure in a test chamber equipped with electrical feed-through. Four different nanofibrous mats were studied: two nanofibrous mats carbonized at 700 °C and 900 °C, respectively, and the corresponding mats surface-decorated with Pd nanoparticles. Each nanofibrous mat with length, width, and thickness being 10, 7, and 0.18 mm was loaded into a sealed chamber with gas inlet and outlet. The set-up for sensing test is shown in Fig. S1 (ESI†). Electrical contacts were made of silver paste, and the resistivity values of nanofiber mats were acquired from the Keithley 2612 source-meter. The flow of H₂ or H₂/N₂ into the gas chamber was controlled by MKS flow meters using the Labview software. To test the sensitivity of nanofiber mats toward H₂, the resistivity of the mats was monitored as a function of time when the sample chamber was exposed alternatively between N₂ and H₂. Additionally, the change of resistivity of the nanofibrous mats under different concentrations of H₂ using N₂ as a carrier gas was investigated. The response was defined as the resistance ratio of a mat in a target gas versus in the pure N₂; i.e.,
where ρ₀ is the resistivity of a mat in N₂, and ρ is the resistivity when the mat is exposed to H₂/N₂ mixture.

Results and discussion

Structural characterization

SEM images in Fig. 1 show the representative morphologies of an electrospun carbon nanofibrous mat before and after surface-decoration with Pd nanoparticles. The mats carbonized at 700 °C and 900 °C did not have any appreciable difference in morphology (see Fig. S2, ESI†). The carbon nanofibers in the mat were reasonably uniform with diameters of ∼300 nm. After decoration with Pd nanoparticles, the overall morphology of nanofibers/mat was well-retained. As shown in Fig. 1(b), the Pd nanoparticles were evenly distributed on the surface of nanofibers. Under the adopted conditions, it was estimated that the density of Pd nanoparticles was in the range of 20–30 particles per micrometer-length of nanofiber. These Pd nanoparticles were well-separated, and no continuous Pd layer could be identified on the nanofibers.

Fig. 1 (a) SEM image showing a typical electrospun carbon nanofibrous mat. (b) SEM image showing a typical electrospun carbon nanofibrous mat after surface-decoration with Pd nanoparticles. (c) TEM image of a carbon nanofiber surface-decorated with a representative Pd nanoparticle. (d) High-resolution TEM image showing the lattice fringes of several Pd nano-clusters in the Pd nanoparticle; inset: fast-Fourier transform image from a Pd nano-cluster. (e) Energy-dispersion spectrum collected from the SEM measurement. (f) Energy-dispersion spectrum collected from the TEM measurement. The carbonization temperature was 700 °C.

The TEM image in Fig. 1(c) shows that the Pd nanoparticles on the surface of electrospun carbon nanofibers had sizes in tens of nanometers, and they were composed of smaller Pd clusters. The high-resolution TEM image in Fig. 1(d) showed the lattice fringes of several Pd nano-clusters, indicating that the Pd clusters were single crystalline with sizes of a few nanometers. The d-spacing of Pd (111) crystallographic planes, acquired from the fast-Fourier transform image (the inset in Fig. 1(d)), was 0.22 nm, confirming that the palladium in nano-clusters/nanoparticles was in elemental form. The composition of carbon nanofibrous mats decorated with Pd nanoparticles was further confirmed by the EDS results of the sample. The EDS from SEM in Fig. 1(e) showed the strong peaks of carbon (from the fiber) and Pd (from the Pd nanoparticles) only. This was consistent with the EDS results of the sample collected from TEM in Fig. 1(f), which indicated that the elements in the nanofibers were carbon and palladium. The copper peak in Fig. 1(f) was from the TEM grid. It is noteworthy that the nanofibers carbonized at 700 °C had a high content of amorphous carbon; with the increase of the carbonization temperature to 900 °C, the graphitic/crystalline content increased.¹³

Hydrogen sensing properties

The resistivity of an electrospun carbon nanofibrous mat would be lower when the mat was carbonized at higher temperature. Table 1 shows the resistivity values of electrospun nanofibrous mats carbonized at different temperatures. The resistivity value of the mat carbonized at 900 °C was 1.072 Ω cm, whereas the resistivity of the mat carbonized at 700 °C was 5403 Ω cm. The decrease of resistivity with increase of carbonization temperature was because the transformation of disordered/amorphous carbon to a more ordered graphitic structure would occur in the temperature range from 700 to 900 °C.^13,14 For the mat carbonized at 700 °C, the resistivity value after decoration of Pd nanoparticles was 253.6 Ω cm, more than 20 times lower than that of the carbon nanofibrous mat without Pd nanoparticles. The resistivity value of the mat carbonized at 900 °C only decreased approximately 3-fold after decoration of Pd nanoparticles. It is noteworthy that the electric current had to go through carbon nanofibers during the resistivity measurement, since the decorated Pd nanoparticles were well-isolated and/or non-continuous on the surface of these nanofibers.

Table 1 Electrical resistivity (ρ) values of different carbon nanofibrous mats

Carbonization temperatures	ρ values of the carbon nanofibrous mats (Ω cm)	ρ values of the carbon nanofibrous mats with Pd (Ω cm)
700 °C	5403	253.6
900 °C	1.072	0.345

---

The conduction mechanism is still not clear for electrospun carbon nanofibrous mats. Our previous studies have indicated that these mats are semiconducting in nature, and the activation energy decreases exponentially with the increase of carbonization temperature.¹³ It is important to note that the conduction mechanism of electrospun carbon nanofibrous mats is different from that of carbon nanotubes. While carbon nanotubes are p-type semiconductors, the conduction of structurally disordered electrospun carbon nanofibers (and nanofibrous mats) may be in the strong localization regime of both electrons and holes.²⁶ The temperature dependence of electrical conductivity of the carbon nanofibers prepared at 900 °C indicated that the fibers would be highly disordered and the conduction might follow the variable-range hopping mechanism.^15,26 The deposition of Pd nanoparticles on the surface of carbon nanofibers facilitated the hopping between the localized states, and thus the conductivity of the carbon nanofibers increased. The effect was more significant for the nanofibers carbonized at 700 °C, likely due to the large reduction of activation energy by the additional conduction paths provided by Pd nanoparticles.¹³

Prior to studying the hydrogen sensing properties, the effects of relative humidity on resistivity of Pd-decorated carbon nanofibrous mats were investigated; and the results are shown in Table 2. Current–voltage (IV) curves of the Pd-decorated mats carbonized at 700 °C exposed to different gases are shown in Fig. 2(a). It was evident that the carrier gas of either dry air or N₂ had no appreciable difference on resistivity of the Pd-decorated nanofibrous mats carbonized at 700 or 900 °C. In air with a relative humidity of 50%, the resistivity decreased slightly. For the Pd-decorated nanofibrous mats carbonized at 700 °C, the resistivity value was ∼3% lower in air with a relative humidity of 50% compared with that in dry air. Such a change was much smaller than the change of resistivity when the mat was exposed to H₂ (as shown in Fig. 2(b)). Similar results were observed for the Pd-decorated nanofibrous mats carbonized at 900 °C.

Fig. 2 (a) IV curves of the Pd-decorated nanofibrous mat (carbonized at 700 °C) exposed to N₂, dry air, and air with 50% humidity. (b) and (c) Sensor responses of the different carbon nanofibrous mats to the flow of pure H₂ being 500 sccm at room temperature (24 ± 2 °C), where the mats were carbonized at 700 °C (b) and 900 °C (c), respectively. The dashed (green) line represents the mats without Pd nanoparticles, and the solid (red) line represents the corresponding mats with Pd nanoparticles.

Table 2 Electrical resistivity (ρ) values of the electrospun PAN-based carbon nanofibrous mats with Pd nanoparticles upon exposure to dry air or the air with relative humidity of 50% at room temperature (24 ± 2 °C)

Carbonation temperatures	ρ values of carbon nanofibrous mats with Pd nanoparticles (Ω cm)
	In dry air	In air with relative humidity of 50%
700 °C	253.6	246.1
900 °C	0.3450	0.3415

---

Fig. 2(b) and 2(c) showed responses of the nanofibrous mats carbonized at 700 and 900 °C, respectively, when the flow of pure H₂ was turned on and off for 4 cycles at room temperature (24 ± 2 °C). For both mats, the resistivity decreased upon the exposure to H₂, and the resistivity was restored when H₂ was turned off. The change of resistivity was ∼4% and ∼0.1% for the mats carbonized at 700 and 900 °C, respectively, after the flow of H₂ for 10 min. The response of resistivity was sharply enhanced when the Pd-decorated carbon nanofibrous mats were used. For the 700 °C-carbonized nanofibrous mat with Pd nanoparticles, the resistivity decreased ∼18% after exposure to H₂ for 10 min, as shown in Fig. 2(b). Although the resistivity continually decreased thereafter, the sharp drop of resistivity occurred within the initial 2 min of exposure to H₂. The recovery of resistivity was fast upon removal of H₂ in the testing chamber; within 30 s, the resistivity was recovered by 90%. Therefore, the response of the mats to H₂ was reversible, indicating that the materials could be used for hydrogen sensing applications at room temperature. The response of the Pd-decorated nanofibrous mat carbonized at 900 °C was similar to that of the Pd-decorated nanofibrous mat carbonized at 700 °C, but the changes were merely ∼3.5% upon exposure to H₂ for 10 min. Slight baseline drift was observed in Fig. 2(b) and 2(c), and further investigations would be carried out to understand the reason for this drift.

The calibration curves for electrospun carbon nanofibrous mats surface-decorated with Pd nanoparticles are shown in Fig. 3, in which the change of resistivity after exposure to the mixed gases of H₂/N₂ for 10 min was plotted as a function of the volume fraction of H₂ in H₂/N₂ mixture. For both nanofibrous mats (carbonized at 700 and 900 °C), the resistivity change increased almost linearly with increasing the volume fraction of H₂ initially, and the change was saturated when the H₂ volume fraction was ∼70%. For the mat carbonized at 700 °C, the change of resistivity increased from 4% to 17% when the volume fraction of H₂ increased from 0.1 to 1.0, whereas the resistivity change increased from 0.7% to 3.6% for the mat carbonized at 900 °C over the same range of volume fraction. The sensitivity (the slope of the curve in Fig. 3) was higher for the mat carbonized at 700 °C than for the one carbonized at 900 °C. These results indicated that the sensitivity of the materials to H₂ could be tailored by adjusting the carbonization temperature.

Fig. 3 The sensing responses of electrospun carbon nanofibrous mats decorated with Pd nanoparticles after 10 min exposure to the mixture gases of H₂/N₂ with different volume fractions (concentrations) of H₂ at room temperature (24 ± 2 °C).

Many hydrogen sensors based on the change of resistance require Pd or Pt as the sensing element. Upon different sensing materials/mechanisms, the resistance can increase or decrease after exposure to H₂. In the hydrogen sensors based on Pd nanowires, the resistivity decreases when the nanowires are exposed to H₂.²⁷ In this case, the molecules of H₂ would be adsorbed by Pd, resulting in the swelling of Pd nanowires and leading to better contacts and conduction pathways between the Pd nanowires. In the hydrogen sensors based on metal oxide semiconductors or carbon nanotubes, the Pd or Pt nanoparticles are often deposited on the surface of these sensing materials. It is known that Pd is effective in the catalytic dissociation of H₂ molecule via formation of palladium hydride, and this is the mechanism of most Pd-based H₂ sensors.^{16,17,20,28–31} For the p-type carbon nanotubes, the resistivity increases upon exposure to H₂. In this case, the formation of palladium hydride decreases the work function of Pd, leading to the electron transfer from Pd nanoclusters to carbon nanotubes and the reduction of the concentration of the hole carriers. Consequently, the resistivity increases.^16–18 For the n-type metal oxides (e.g., ZnO), the formation of palladium hydride upon exposure to H₂ increases the number of electrons, and hence decreases the resistivity of the metal oxides.^28,29 The sensing mechanism in electrospun carbon nanofibers (and/or nanofibrous mats) may be quite different from that in metal oxides and/or carbon nanotubes. A recent study on the sensors made of carbon nanofibers toward NO and CO gases has revealed that the resistivity decreases for both gases, although NO is an oxidizing gas and CO is a reducing gas.³² We have observed that the resistivity of electrospun carbon nanofibrous mats would decrease upon exposure to H₂. However, as the conduction mechanism in these materials has not been fully understood, the exact mechanism for sensing of H₂ needs further investigation. Since both electrons and holes contribute to the conduction of carbon nanofibers,²⁶ we speculate that the formation of palladium hydride would lower work function of the Pd nanoparticles and transfer electrons from the Pd nanoparticles to the carbon nanofibers. As a result, the concentration of electrons increases, and the resistivity of the nanofibrous mats decreases.

Conclusions

In summary, the carbon nanofibrous mats were prepared via stabilization in air and then carbonization in argon of electrospun PAN nanofibrous mats. Pd nanoparticles were deposited on the surface of carbon nanofibers using Pd(acac)₂ as the precursor by the SCCO2 method followed by pyrolysis at 600 °C. Detailed structural characterizations indicated that the Pd nanoparticles were composed of single-crystalline clusters with sizes of a few nanometers. The resistivity of electrospun carbon nanofibrous mats surface-decorated with Pd nanoparticles was sensitive to H₂, and the changes were reversible at room temperature. The results suggested that the decoration of Pd nanoparticles via SCCO2 method to the surface of electrospun PAN-based carbon nanofibrous mats increases the sensitivity of the materials toward hydrogen gas significantly. The SCCO2 method could be utilized as a general approach for the decoration of metal nanoparticles on the surface of electrospun carbon nanofibers.

Acknowledgements

This work was supported by the National Aeronautics and Space Administration (Grant No.: NNX10AN34A), the National Science Foundation/EPSCoR (Grant No.: 0903804), and the State of South Dakota.

References

1. R. Baughman, A. Zakhidov and W. De Heer, Science, 2002, 297, 787.
2. V. Popov, Mater. Sci. Eng., R, 2004, 43, 61.
3. T. Zhang, S. Mubeen, N. Myung and M. Deshusses, Nanotechnology, 2008, 19, 332001.
4. A. Geim and K. Novoselov, Nat. Mater., 2007, 6, 183.
5. Y. Sun, Q. Wu and G. Shi, Energy Environ. Sci., 2011, 4, 1113.
6. M. Pumera, Mater. Today, 2011, 14, 308.
7. J. Liu, Z. Yue and H. Fong, Small, 2009, 5, 536.
8. M. Inagaki, Y. Yang and F. Kang, Adv. Mater., 2012, 24, 2547.
9. D. Reneker and I. Chun, Nanotechnology, 1996, 7, 216.
10. Y. Dzenis, Science, 2004, 304, 1917.
11. A. Greiner and J. H. Wendorff, Angew. Chem., Int. Ed., 2007, 46, 5670.
12. P. Joshi, L. Zhang, Q. Chen, D. Galipeau, H. Fong and Q. Qiao, ACS Appl. Mater. Interfaces, 2010, 2, 3572.
13. N. Hedin, V. Sobolev, L. Zhang, Z. Zhu and H. Fong, J. Mater. Sci., 2011, 46, 6453.
14. Y. Wang, S. Serrano and J. Santiago-Aviles, Synth. Met., 2003, 138, 423.
15. A. Wang, A. J. Hsieh and G. C. Rutledge, Polymer, 2005, 46, 3407.
16. J. Kong, M. Chapline and H. Dai, Adv. Mater., 2001, 13, 1384.
17. S. Mubeen, T. Zhang, B. Yoo, M. Deshusses and N. Myung, J. Phys. Chem. C, 2007, 111, 6321.
18. Y. Sun and H. Wang, Appl. Phys. Lett., 2007, 90, 213107.
19. J. Suehiro, S. Hidaka, S. Yamane and K. Imasaka, Sens. Actuators, B, 2007, 127, 505.
20. I. Sayago, E. Terrado, M. Aleixandre, M. Horrillo, M. Fernández, J. Lozano, E. Lafuente, W. Maser, A. Benito and M. Martinez, Sens. Actuators, B, 2007, 122, 75.
21. H. Wakayama, H. Itahara, N. Tatsuda, S. Inagaki and Y. Fukushima, Chem. Mater., 2001, 13, 2392.
22. A. Cooper, Adv. Mater., 2003, 15, 1049.
23. L. Zhang, X. Wang, Y. Zhao, Z. Zhu and H. Fong, Mater. Lett., 2012, 68, 133.
24. A. El-Hendawy, Carbon, 2003, 41, 713.
25. J. Li, M. Vergne, E. Mowles, W. Zhong, D. Hercules and C. Lukehart, Carbon, 2005, 43, 2883.
26. Y. Wang and J. Santiago-Avilés, J. Appl. Phys., 2003, 94, 1721.
27. F. Favier, E. Walter, M. Zach, T. Benter and R. Penner, Science, 2001, 293, 2227.
28. J. Wright, W. Lim, D. Norton, S. Pearton, F. Ren, J. Johnson and A. Ural, Semicond. Sci. Technol., 2010, 25, 024002.
29. H. Wang, B. Kang, F. Ren, L. Tien, P. Sadik, D. Norton, S. Pearton and J. Lin, Appl. Phys. Lett., 2005, 86, 243503.
30. J. Kim, F. Ren, B. Gila, C. Abernathy and S. Pearton, Appl. Phys. Lett., 2003, 82, 739.
31. K. Tsukada, T. Kiwa, T. Yamaguchi, S. Migitaka, Y. Goto and K. Yokosawa, Sens. Actuators, B, 2006, 114, 158.
32. J. Im, S. Kang, S. Lee and Y. Lee, Carbon, 2010, 48, 2573.

---

Footnote

† Electronic Supplementary Information (ESI) available. See DOI: 10.1039/c2ra21338a

---