- Email: [email protected]
Fluctuation-induced tunneling dominated electrical transport in multi-layered single-walled carbon nanotube films
Thin Solid Films 519 (2011) 7987–7991
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Fluctuation-induced tunneling dominated electrical transport in multi-layered single-walled carbon nanotube films Yanli Zhao a,⁎, Wenzhi Li b a b
Wuhan National Laboratory for Optoelectronics, School of Optoelectronics Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, China Department of Physics, Florida International University, Miami, FL 33199, United States
a r t i c l e
i n f o
Article history: Received 14 July 2010 Received in revised form 16 May 2011 Accepted 21 May 2011 Available online 27 May 2011 Keywords: Single-walled carbon nanotubes Thin films Tunneling Temperature coefficient Bolometer Atomic force microscopy
a b s t r a c t Low temperature measurements may give some insight into the transport mechanism of single-walled carbon nanotube (SWCNT) films, which could lead to an optimal SWCNT film with designed photoelectric properties. Despite intense research efforts on the low temperature transport in SWCNT films, it is still an open question for the low temperature transport in multi-layered SWCNT films. In this work, the multi-layered SWCNT films were prepared with a layer by layer vacuum filtration. It suggests that the space between different layers of the multi-layered SWCNT can be ignored. For deposition of different-layered SWCNT films using the same total amount of SWCNT suspension, the increase of the layer numbers can reduce the density of the resulting films, which may account for the low temperature transport. The effect of thermal annealing and subsequent nitric acid treatment on the electrical properties of the SWCNT films has also been investigated. At the temperature range of 80–300 K, the transport of the multi-layered SWCNT films can be explained by a fluctuation-induced tunneling model. Our results could build a bridge connecting measured temperature coefficient of resistance and the microscopic tunneling barrier. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Single-walled carbon nanotubes (SWCNTs) are attractive materials from both fundamental and technological points of view. For fundamental research, individual SWCNTs have been attracted a lot of attention especially in the nanoelectronics domain, such as a field effect transistor, which outperforms the state of art silicon technologies in various figures of merit [1]. However, an as-prepared SWCNT can be either metallic or semiconducting depending on its chirality. In general, one third of the synthesized SWCNTs are metallic, while the other two thirds are semiconducting [2,3]. Even though the separation of metallic SWCNTs from a mixture of both metallic and semiconducting SWCNTs has recently become possible [4], the repeatable fabrication of two identical SWCNT devices is further than a near future work. SWCNT films draw a lot of attention in recent years, since the ensemble averaging over tubes makes that SWCNT films can be mass produced in a cost effectively manner with a series of repeatable properties. Moreover, from the point of technological application, SWCNT film can be used in SWCNT thin film transistors [5], a potential replacement of indium tin oxide (ITO) electrode, especially on flexible substrates [6,7], infrared bolometers [8], gas sensors [9], and optical modulators [10]. Q. Cao et al. presented a comprehensive review which covers relevant work in this field over the last 10 years or so [11].
⁎ Corresponding author. E-mail address: [email protected] (Y. Zhao). 0040-6090/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2011.05.059
Vacuum filtration has been found an effective approach for repeatable fabrication of SWCNT films with a designed thickness and film density [6]. In this work, an improved vacuum filtration method (hereafter, it was called a layer-by-layer vacuum filtration) was adopted for the fabrication of multi-layered SWCNT films. A layer-by-layer vacuum filtration for fabrication of SWCNT films has been proposed previously, aiming at a high quality of transparent and conductive electrode which can be used as an ITO replacement, especially on flexible substrates [12]. Low temperature measurements may give some insight into the transport mechanism of SWCNT films, which could lead to an optimal SWCNT film with designed electrical properties [13]. Despite intense research efforts on the low temperature transport in SWCNT films [14–16], the study on the low temperature transport in the multi-layered SWCNT films has not been found. Moreover, in addition to being a potential replacement for ITO as a conductive and transparent electrode material, SWCNT thin films have also attracted a lot of attention recently due to their large bolometric photoresponse and high temperature coefficient of resistance (TCR) [17]. High TCR SWCNT films are highly desired for its potential application in infrared (IR) sensor. The TCR for a multi-layered SWCNT film has not been reported. In this work, a layer-by-layer vacuum filtration was adopted for the fabrication of multi-layered SWCNT films. The low temperature transport in the single and multi-layered SWCNT films has been investigated comparatively. The effect of thermal annealing and nitric acid (HNO3)-treatment on the electrical properties of the multi-layered SWCNT films has also been discussed. Aiming to ultimately produce an SWCNT film with optimal electrical
7988
Y. Zhao, W. Li / Thin Solid Films 519 (2011) 7987–7991
and optical properties, our experiment provides insight into the design and selection of SWCNT films.
2. Experiments In this research, we used home-made purified SWCNTs for fabrication of SWCNT films. The SWCNT raw material was prepared by an ethanol chemical vapor deposition (CVD) using Co catalyst, as we reported before [3]. The as-synthesized SWCNT raw material (soot) was scratched from the Si substrate and purified by HNO3. To form a nanotube film, an improved layer-by-layer vacuum filtration method was adopted and the process for each layer of the SWCNT films was similar to that reported by Wu et al. [18]. The SWCNT soot was dispersed in aqueous solutions of 1% v/v surfactant (Triton-100) via 1 h ultrasonication and 0.5 h centrifugation. The concentration of the obtained SWCNT suspension is about 5 × 10 − 3 mg/ml. The obtained SWCNT suspensions were vacuum filtered onto 0.1 μm pore size mixed cellulose ester membrane (Millipore) and followed by washing with copious quantities of deionized water to remove the surfactant. The so-called layer-by-layer method is described as follows: after the first layer of SWCNTs was transferred to the glass substrate, a second vacuum filtrated layer of SWCNTs will be transferred on the first layer, and the third vacuum filtrated layer will be transferred on the second layer. All samples in this work can be written as sample(s) n–m, where n denotes the amount (ml) of SWCNT suspension used for the film preparation, while m denotes the layer number. The film thickness and the standard error associated with the thickness measurements are tabulated in Table 1. It is not difficult to know that the film density can be calculated with an equation ρ = M/S ⋅ d, where ρ denotes volume density, M denotes the mass, d is film thickness, and S denotes the film area, which keeps a same value for all SWCNT films prepared with the same apparatus for vacuum filtration. For simplicity, in this work, the so-called “relative density” is also defined as the ratio of a film density to the density of sample 1–1, as presented in Table 1. Table 2 tabulates the low temperature properties of some typical SWCNT thin films. According to above definition, sample 1–2 is a two-layered SWCNT film, and thereinto each layer was prepared using 1 ml SWCNT suspension. Sample 2–1 has only one layer prepared by vacuum filtration with 2 ml SWCNT suspension. Sample 2–3 is a three-layered SWCNT film, the suspension amount for each layer of this film is exactly the same as that of the sample 2–1. The as-prepared SWCNT films were thoroughly cleaned in acetone and then in ethanol for future electrical property measurements. Post-treated samples (samples 2–3a and 2–3aN) were prepared by annealing the as-prepared SWCNT film named sample 2–3 at 300 °C in an Ar gas environment for 12 h with and without subsequent HNO3-treatments, respectively. The HNO3treatments were carried out by submerging the annealed SWCNT film in 12-M HNO3 for 0.5 h. Table 1 The thickness and relative density of the SWCNT thin films. Sample name
Sample description
1–1
1 ml SWCNT suspension, 1 layer 6 ml SWCNT suspension, 1 layer 2 ml SWCNT suspension, 1 layer 2 ml SWCNT suspension, 2 layer 2 ml SWCNT suspension, 3 layers 1 ml SWCNT suspension, 2 layers
6–1 2–1 2–2 2–3 1–2
Thickness (nm)
Standard error
Relative density ρ ( n–m ρ1–1 )
22.0
1.7
1.0
83.0
3.8
1.59
35.0
2.5
1.26
70.0
3.7
1.26
105.0
5.3
1.26
45.0
3.5
0.98
/
Table 2 The low temperature properties of the SWCNT thin films. Sample name
Sample description
Tb (K)
Ts (K)
TCR (%/K) (at 280 K)
1–2 2–1 2–3 2–3a 2–3aN
1 ml SWCNT suspension, 2 layers 2 ml SWCNT suspension, 1 layer 2 ml SWCNT suspension, 3 layers Sample 2−3 after annealing at 300 °C, 12 h Sample 2–3a after HNO3-treatment
55.0 34.5 23.7 30.9 7.7
21.1 17.5 19.8 16.6 12.1
− 0.065 − 0.042 − 0.028 − 0.038 − 0.010
Atomic force microscopy (AFM) topography images of the SWCNT thin films were acquired in the tapping mode. AFM was also used to measure the thickness of the SWCNT films. The dependence of the resistivity on temperature has been measured using a dc standard fourprobe method. Before measurement, Ag metal film was deposited on the SWCNT films to get a good ohmic contact. The resistance between the electrodes with different spacing was then measured and used to determine the contact resistance of the silver/nanotube contact and the intrinsic resistance of the SWCNT films, as reported in references [6] and [19]. Our results show that compared with the intrinsic resistance, the contact resistance is small. The dimension of the SWCNT films under test was 2 mm × 20 mm, and the area of the circular electrodes is less than 2 mm in diameter. The constant current was flowing along the longer edge (20 mm) and the voltage drop between the inner two electrodes was recorded automatically with a program-controlled Keithley voltmeter during low temperature measurement. For measurement of temperature-dependent resistivity, SWCNT film fixed in a cryogenic system was first cooled down from room temperature. The measurements were then taken while SWCNT films were heating in the temperature range of 80–280 K. Since the maximum temperature is well below the T*dedop (350 K) suggested by Teresa M. Barnes et al. [16], it is not difficult to understand that no obvious hysteresis was observed in the temperature-dependent resistivity obtained from different cycles of heating or cooling. 3. Results and discussion Fig. 1a shows isolated SWCNTs grown on a 5-pulse Co catalytic substrate as we reported before [3], which allows us to derive the diameter of SWCNTs from AFM topography data, and the typical results are shown in Fig. 1b. For every nanotube, tens of positions were measured to calculate the value of the average heights. The diameter distribution was calculated with the value of average height from more than 50 CNTs. It is found that the nanotube diameters are in the range of 0.6 nm to 1.6 nm, with a mean diameter about 1.0 nm. Fig. 1c illustrates SWCNT bundles grown on 10-pulse Co catalytic substrate with a high ethanol pressure during CVD. The increase of pulse number and ethanol pressure increases the yield of SWCNTs. The as-synthesized SWCNT bundles shown in Fig. 1c was scratched from the Si substrate and purified by HNO3. After purification, the obtained soot contains at least 90% SWCNTs by weight, and was used as the raw material for SWCNT film fabrication. Fig. 1d shows a typical AFM image of the SWCNT films fabricated by vacuum filtration. The root mean square (rms) roughness for the SWCNT film deduced from AFM data is about 8.0 nm, which is nearly same as the previous report [20]. The multi-layered SWCNT films were prepared with the “improved” vacuum filtration. Since each layer can be thoroughly washed from both sides of the layer with copious amounts of water [12], it is thought that the advantage of the layer-by-layer vacuum filtration over “normal” vacuum filtration reported by Wu et al. [18] might be that the improved vacuum filtration produces a “clean” multi-layered SWCNT film if we neglect (or remove) the residues of the (cellulose ester) membrane left after every deposition. As suggested by Wu et al. [18], the density and thickness of SWCNT films can be controlled well with the “normal” vacuum filtration. With
Y. Zhao, W. Li / Thin Solid Films 519 (2011) 7987–7991
7989
Fig. 1. (a) An AFM image of SWNTs grown on a 5-pulse Co catalytic Si substrate; (b) diameter distributions of isolated SWNTs shown in (a); (c) an AFM image of SWNT bundles grown on a 10-pulse Co catalytic Si substrate; (d) a typical AFM image of the SWCNT film prepared with vacuum filtration.
the improved vacuum filtration, it is possible to prepare a series of films with same density but different thickness, simply by adding or reducing the number of the layers. For this purpose, the most important issue is that the amount of SWCNT suspension used for every layer deposition must be kept as an exact constant. The above predication has been proved to be true in this work, as tabulated in Table 1. Three series of SWCNT films were prepared and summarized as follows. The first series of samples are samples n–1 (n = 1, 2, and 6) prepared with the “normal” vacuum filtration. The thickness increases from 22 nm, 35 nm to 83 nm when the amount of SWCNT suspension used for film fabrication changes from 1 ml, 2 ml to 6 ml, while the relative density changes from 1.0, 1.26 to 1.59 for samples 1–1, 2–1, and 6–1, respectively. Here the so-called “relative density” denotes a normalized value, which is defined as the ratio of a film density to the density of sample 1–1. The second series of SWCNT films are named as samples 2–m (m = 1, 2, and 3). The thickness increases from 35 nm, 70 nm to 105 nm when the amount of SWCNT suspension used for film fabrication changes from 1 ml, 2 ml to 3 ml, while the relative density keeps as a constant (1.26). The thickness of samples 2–m (m = 1, 2, 3) is proportional to the layer number. It is also found from Table 1 that the thickness and density of the third series of samples named sample 1–m (m = 1, 2) also follows the same rule as the samples 2–m, which indicates that the measurement is reliable. The deviation in relative density of sample 1–2 from sample 1–1 can be neglected, which may originate from the experimental error in thickness measurement. The comparison of sample 6–1 with sample 2–3 reveals that the multi-layer technique can reduce the density of the resulting films when keeping the total amount of the used SWCNT suspension as a constant (also proved by sample 2–1 versus sample 1–2). Since it has been proved with AFM that the thickness of the multi-layered SWCNT films is indeed proportional to the layer number, it is thought that the effect of the possible interface formed between adjacent layers for the multi-layered film can be ignored. On basis of the above discussion, two points for the multi-layer technique should be addressed here. First, the technique can prepare a series of films with same density but different thickness, simply by adding or reducing the number of the layers (e.g. sample 2–1 versus
2–3, which are prepared with same amount of SWCNT suspension for each layer of sample 2–3 and the single layer of sample 2–1); secondly, the multi-layer technique can reduce the density of the resulting films (e.g. sample 2–1 versus sample 1–2). Keeping the total amount of the used SWCNT suspension as a constant, the increase of the layer number of multi-layered SWCNT film will reduce the film density. Hopefully, for a future work, simply by changing the amounts of SWCNT suspension for each layer during deposition of a multi-layered SWCNT film, the layer-by-layer technique can produce multi-layered SWCNT film with tunable electric or optical properties in the thickness direction and form a gradient material. The gradient material with tunable electric and optical properties may draw a lot of attention in different fields, such as being a gradient refractive index material [21]. In theory, the temperature dependence of resistivity for SWCNT films can be understood by two existing mechanisms, one is variable range hoping (VRH), and the other is fluctuation-induced tunneling (FIT) [14–16]. The expression for the FIT-dominated resistivity can be written as Tb R = A exp ; TS + T
ð1Þ
where A is a constant, kTb reflects the order of magnitude of the barrier energies (where k is the Boltzmann constant). TS is another important parameter since FIT model controls the temperature dependence of resistivity for temperatures that are well above TS. Behnam et al. advocated that there exists different transport regimes for temperature dependent resistivity, and suggested that VRH is valid in the extremely low temperature range, while FIT can account for the transport in SWCNT films at a moderate temperature range below room temperature [14]. In this work, we repeat the similar measurement on single and the multi-layered SWCNT films at the temperatures between 80 K and 300 K. The simulation of the experimental data with the VRH model or combination of VRH model with FIT model has been performed, even though a large range of parameters have been initialized, the deviation of theory from the measurement cannot be neglected. However, in Fig. 2, it is shown that
7990
Y. Zhao, W. Li / Thin Solid Films 519 (2011) 7987–7991
the measurement can be fit satisfactorily based on FIT model, which is consistent with the work from Behnam et al. [14]. Fig. 2 shows the log plot of resistivity versus temperature for all the samples tabulated in Table 2. The sheet resistance of all samples changes consecutively with temperature. The decrease of the film layer number leads to an increase in the sheet resistance, as shown by the plots for samples 2–3 and 2–1 in Fig. 2. Fig. 2 also shows the change of sheet resistance of the SWCNT films before and after thermal annealing (plots for sample 2–3 and 2–3a) and after further HNO3-treatment (plot for 2–3aN). It demonstrates that thermal annealing results in an increase of the sheet resistance, while further HNO3-treatment will lead to a decrease of the sheet resistance. The lines are fits to the experimental data based on FIT model. The values of Tb and TS for all samples are tabulated in Table 2. Our results suggest that the single and multilayered SWCNT films can only be understood by FIT model at the moderate temperature regime. Bolometric responsivity is defined by the resistance variation due to a heating. Fig. 3 depicts the variation in TCR for samples 1–2, 2–1 and 2–3 as a function of temperature, which is deduced from Fig. 2 with the formula TCR = (ΔR/ΔT)/R, where R is the sheet resistance at temperature T which is obtained from Fig. 2. Fig. 4 shows the typical TCR ~ T plots for sample 2–3 before and after annealing and acid treatments. After the thermal annealing, the TCR absolute values increase from 0.028% (sample 2–3) to 0.038% (sample 2–3a) at 280 K. It is also revealed in Fig. 4 that the HNO3-treatment to the annealed samples will result in a remarkable decrease in the TCR absolute value; see the plot for sample 2–3aN in Fig. 4. For building a theoretical relationship of TCR with Tb based on FIT model, from Eq. (1), we get, 1 dR 1 = TCR = R dT A exp T
S
Tb + T
h d A exp T
S
Tb + T
i
dT
ð2Þ
then, TCR =
1 dR 1 = = R dT A exp T
Tb + TS
A exp
Tb T + TS
−
Tb 1 dTb : + ðT + TS Þ dT ðT + TS Þ2
ð3Þ
Further, considering Tb is a constant for a specialized SWCNT film, dT the term b is zero. If we ignore the term TS, it is reasonable since TS is dT
Fig. 2. The log plot of resistivity versus temperature for all the samples tabulated in Table 2. The lines are fits to the experimental data based on FIT model.
Fig. 3. The variation of TCR for SWCNT films as a function of temperature.
about 20 K, which is small compared with T (280 K), a simple expression will be derived, TCR∝−
Tb T ≈− b2 : ðT + TS Þ2 T
ð4Þ
Fig. 5 shows a TCR ~ Tb plot for all SWCNT films tabulated in Table 2 and a fit to the experimental data based on Eq. (4). For clarity, error bars are placed on all the experimental data points. As we expected, basically, the theory and experiment are consistent with each other. The deviation of experimental data from theory may originate from the approximation during Eq. (4) deduction or from the measurement error. From Eq. (4), we know that TCR is proportional to Tb at a fixed temperature. By comparing the TCR vs. T plots for samples 1–2 and 2–1 in Fig. 4, one can see that the magnitude of the TCR absolute value of higher density sample 2–1 is smaller than that of lower density sample 1–2, which reveals that the increase of SWCNT film density results in a decrease in the TCR absolute value, and, intrinsically, may originate from the change in Tb for the SWCNT films. For SWCNT films with same density, the comparison of TCR for samples 2–1 and 2–3 reveals that the TCR absolute value decreases with the increase in thickness of the SWCNT films.
Fig. 4. TCR~T plots for as-prepared, annealed and acid-treated three-layered SWCNT films.
Y. Zhao, W. Li / Thin Solid Films 519 (2011) 7987–7991
7991
hints on the application of SWCNT film in bolometers. It suggests that the layer-by-layer technique can produce a multi-layered SWCNT film with tunable electric or optical properties in the thickness direction, and form a gradient material, which may draw a lot of attention in different fields. The further study on a new gradient material based on multi-layered SWCNT films is underway. Acknowledgment This work was supported by the National Hi-Tech Research and Development Program of China (No. 2008AA01Z207, 863), Natural Science Foundation of Hubei Province, China (Grant No. 2010CDB01606) and Scientific Research Foundation for the Returned Overseas Chinese Scholars. References Fig. 5. TCR ~ Tb plot for all SWCNT films tabulated in Table 2 and a fit to the experimental data based on Eq. (4).
As we know, the as-prepared SWCNT film is of p-type which may originate from the acid reflux-based purification process of SWCNTs and atmospheric impurities, such as oxygen gas. Thermal annealing can remove the p-type dopants and make the SWCNT more semiconducting, and therefore increases the resistivity of single SWCNT and also the barrier energies among SWCNTs, which leads to an increase in Tb. We think accordingly that the increase of Tb may account for the increase of TCR after annealing. On the other hand, acid treatment can be ascribed to the charge transfer doping in the SWCNT films [22]. The decrease of the TCR of the acid treated SWCNT film means that HNO3-treatment makes SWCNTs more metallic, and barrier energies among SWCNTs decrease, which leads to a decrease in Tb. 4. Conclusions In this work, the low temperature electrical transport properties of the multi-layered single-walled carbon nanotube (SWCNT) films have been investigated. Similar to the single layer SWCNT film, our results suggest that the multi-layered SWCNT films prepared with a layer-bylayer vacuum filtration can also be understood by FIT model at a moderate temperature regime. The temperature coefficient of resistance (TCR) of the multi-layered SWCNT films has also been studied. The decrease in layer number of the multi-layered SWCNT films, decrease of the film density and the thermal annealing treatment are shown to be effective approaches to obtain a SWCNT film with a large TCR absolute value. A bridge is built connecting the measured TCR with the tunneling barrier, which may provide more
[1] P. Avouris, Z. Chen, V. Perebeinos, Nat. Nanotechnol. 2 (2007) 605. [2] M.S. Dresselhaus, G. Dresselhaus, R. Saito, A. Jorio, Phys. Rep 409 (2005) 47. [3] Yanli Zhao, Haruhisa Nakano, Hirohiko Murakami, Toshiki Sugai, Hisanori Shinohara, Yahachi Saito, Appl. Phys. A 85 (2006) 103. [4] Alexander A. Green, Mark C. Hersam, Nano Lett. 8 (2008) 1417. [5] Matthew A. Meitl, Yangxin Zhou, Anshu Gaur, Seokwoo Jeon, Monica L. Usrey, Michael S. Strano, John A. Rogers, Nano Lett. 4 (2004) 1643. [6] L. Hu, D.S. Hecht, G. Grüner, Nano Lett. 4 (2004) 2513. [7] Fumiaki N. Ishikawa, Hsiao-kang Chang, Koungmin Ryu, Po-chiang Chen, Alexander Badmaev, Lewis Gomez De Arco, Guozhen Shen, Chongwu Zhou, ACS Nano 3 (2009) 73. [8] Mikhail E. Itkis, Ferenc Borondics, Aiping Yu, Robert C. Haddon, Science 312 (2006) 413. [9] Chang-Seung Woo, Chae-Hyun Lim a, Chi-Won Cho b, Bonghyun Park a, Heongkyu Ju a, Dong-Hun Min b, Cheol-Jin Lee c, Seung-Beck Lee, Microelectron. Eng. 84 (2007) 1610. [10] Moon-Ho Ham, Byung-Seon Kong, Woo-Jae Kim, Hee-Tae Jung, Michael S. Strano1, Phys. Rev. Lett. 102 (2009) 047402. [11] Q. Cao, J.A. Rogers, Adv. Mater. 21 (2008) 29. [12] Yu Wang, Chong-an Di, Yunqi Liu, Hisashi Kajiura, Shanghui Ye, Lingchao Cao, Dacheng Wei, Hongliang Zhang, Yongming Li, Kazuhiro Noda, Adv. Mater. 20 (2008) 4442. [13] Alexandru R. Biris, Dan Lupu, Zhongrui Li, Hom R. Kandel, Enkeleda Dervishi, Viney Saini, Alexandru S. Biris, Appl. Phys. Lett. 91 (2007) 053115. [14] Ashkan Behnam, Amlan Biswas, Gijs Bosman, Ant Ural, Phys. Rev. B. 81 (2010) 125407. [15] Roderick K. Jackson, Andrea Munro, Kenneth Nebesny, Neal Armstrong, Samuel Graham, ACS Nano 4 (2010) 1377. [16] Teresa M. Barnes, Jeffrey L. Blackburn, Jao van de Lagemaat, Timothy J. Coutts, Michael J. Heben, ACS Nano 2 (2008) 1968. [17] Rongtao Lu, Guowei Xu, Judy Z. Wu, Appl. Phys. Lett. 93 (2008) 213101. [18] Z.C. Wu, Z.H. Chen, X. Du, J.M. Logan, J. Sippel, M. Nikolou, K. Kamaras, J.R. Reynolds, D.B. Tanner, A.F. Hebard, A.G. Rinzler, Science 305 (2004) 1273. [19] Fiona M. Blighe, Yenny R. Hernandez, Werner J. Blau, Jonathan N. Coleman, Adv. Mater. 19 (2007) 4443. [20] Yangxin Zhou, Liangbing Hu, George Grünera, Appl. Phys. Lett. 88 (2006) 123109. [21] Hao Lv, Aimei Liu, Jufang Tong, Xunong Yi, Qianguang Li, J. Opt. Soc. Am. A: 26 (2009) 1085. [22] Bhupesh Chandra, Ali Afzali, Neeraj Khare, Mostafa M. El-Ashry, George S. Tulevski, Chem. Mater. 22 (2010) 5179.
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Fluctuation-induced tunneling dominated electrical transport in multi-layered single-walled carbon nanotube films Yanli Zhao a,⁎, Wenzhi Li b a b
Wuhan National Laboratory for Optoelectronics, School of Optoelectronics Science and Engineering, Huazhong University of Science and Technology, Wuhan 430074, China Department of Physics, Florida International University, Miami, FL 33199, United States
a r t i c l e
i n f o
Article history: Received 14 July 2010 Received in revised form 16 May 2011 Accepted 21 May 2011 Available online 27 May 2011 Keywords: Single-walled carbon nanotubes Thin films Tunneling Temperature coefficient Bolometer Atomic force microscopy
a b s t r a c t Low temperature measurements may give some insight into the transport mechanism of single-walled carbon nanotube (SWCNT) films, which could lead to an optimal SWCNT film with designed photoelectric properties. Despite intense research efforts on the low temperature transport in SWCNT films, it is still an open question for the low temperature transport in multi-layered SWCNT films. In this work, the multi-layered SWCNT films were prepared with a layer by layer vacuum filtration. It suggests that the space between different layers of the multi-layered SWCNT can be ignored. For deposition of different-layered SWCNT films using the same total amount of SWCNT suspension, the increase of the layer numbers can reduce the density of the resulting films, which may account for the low temperature transport. The effect of thermal annealing and subsequent nitric acid treatment on the electrical properties of the SWCNT films has also been investigated. At the temperature range of 80–300 K, the transport of the multi-layered SWCNT films can be explained by a fluctuation-induced tunneling model. Our results could build a bridge connecting measured temperature coefficient of resistance and the microscopic tunneling barrier. © 2011 Elsevier B.V. All rights reserved.
1. Introduction Single-walled carbon nanotubes (SWCNTs) are attractive materials from both fundamental and technological points of view. For fundamental research, individual SWCNTs have been attracted a lot of attention especially in the nanoelectronics domain, such as a field effect transistor, which outperforms the state of art silicon technologies in various figures of merit [1]. However, an as-prepared SWCNT can be either metallic or semiconducting depending on its chirality. In general, one third of the synthesized SWCNTs are metallic, while the other two thirds are semiconducting [2,3]. Even though the separation of metallic SWCNTs from a mixture of both metallic and semiconducting SWCNTs has recently become possible [4], the repeatable fabrication of two identical SWCNT devices is further than a near future work. SWCNT films draw a lot of attention in recent years, since the ensemble averaging over tubes makes that SWCNT films can be mass produced in a cost effectively manner with a series of repeatable properties. Moreover, from the point of technological application, SWCNT film can be used in SWCNT thin film transistors [5], a potential replacement of indium tin oxide (ITO) electrode, especially on flexible substrates [6,7], infrared bolometers [8], gas sensors [9], and optical modulators [10]. Q. Cao et al. presented a comprehensive review which covers relevant work in this field over the last 10 years or so [11].
⁎ Corresponding author. E-mail address: [email protected] (Y. Zhao). 0040-6090/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2011.05.059
Vacuum filtration has been found an effective approach for repeatable fabrication of SWCNT films with a designed thickness and film density [6]. In this work, an improved vacuum filtration method (hereafter, it was called a layer-by-layer vacuum filtration) was adopted for the fabrication of multi-layered SWCNT films. A layer-by-layer vacuum filtration for fabrication of SWCNT films has been proposed previously, aiming at a high quality of transparent and conductive electrode which can be used as an ITO replacement, especially on flexible substrates [12]. Low temperature measurements may give some insight into the transport mechanism of SWCNT films, which could lead to an optimal SWCNT film with designed electrical properties [13]. Despite intense research efforts on the low temperature transport in SWCNT films [14–16], the study on the low temperature transport in the multi-layered SWCNT films has not been found. Moreover, in addition to being a potential replacement for ITO as a conductive and transparent electrode material, SWCNT thin films have also attracted a lot of attention recently due to their large bolometric photoresponse and high temperature coefficient of resistance (TCR) [17]. High TCR SWCNT films are highly desired for its potential application in infrared (IR) sensor. The TCR for a multi-layered SWCNT film has not been reported. In this work, a layer-by-layer vacuum filtration was adopted for the fabrication of multi-layered SWCNT films. The low temperature transport in the single and multi-layered SWCNT films has been investigated comparatively. The effect of thermal annealing and nitric acid (HNO3)-treatment on the electrical properties of the multi-layered SWCNT films has also been discussed. Aiming to ultimately produce an SWCNT film with optimal electrical
7988
Y. Zhao, W. Li / Thin Solid Films 519 (2011) 7987–7991
and optical properties, our experiment provides insight into the design and selection of SWCNT films.
2. Experiments In this research, we used home-made purified SWCNTs for fabrication of SWCNT films. The SWCNT raw material was prepared by an ethanol chemical vapor deposition (CVD) using Co catalyst, as we reported before [3]. The as-synthesized SWCNT raw material (soot) was scratched from the Si substrate and purified by HNO3. To form a nanotube film, an improved layer-by-layer vacuum filtration method was adopted and the process for each layer of the SWCNT films was similar to that reported by Wu et al. [18]. The SWCNT soot was dispersed in aqueous solutions of 1% v/v surfactant (Triton-100) via 1 h ultrasonication and 0.5 h centrifugation. The concentration of the obtained SWCNT suspension is about 5 × 10 − 3 mg/ml. The obtained SWCNT suspensions were vacuum filtered onto 0.1 μm pore size mixed cellulose ester membrane (Millipore) and followed by washing with copious quantities of deionized water to remove the surfactant. The so-called layer-by-layer method is described as follows: after the first layer of SWCNTs was transferred to the glass substrate, a second vacuum filtrated layer of SWCNTs will be transferred on the first layer, and the third vacuum filtrated layer will be transferred on the second layer. All samples in this work can be written as sample(s) n–m, where n denotes the amount (ml) of SWCNT suspension used for the film preparation, while m denotes the layer number. The film thickness and the standard error associated with the thickness measurements are tabulated in Table 1. It is not difficult to know that the film density can be calculated with an equation ρ = M/S ⋅ d, where ρ denotes volume density, M denotes the mass, d is film thickness, and S denotes the film area, which keeps a same value for all SWCNT films prepared with the same apparatus for vacuum filtration. For simplicity, in this work, the so-called “relative density” is also defined as the ratio of a film density to the density of sample 1–1, as presented in Table 1. Table 2 tabulates the low temperature properties of some typical SWCNT thin films. According to above definition, sample 1–2 is a two-layered SWCNT film, and thereinto each layer was prepared using 1 ml SWCNT suspension. Sample 2–1 has only one layer prepared by vacuum filtration with 2 ml SWCNT suspension. Sample 2–3 is a three-layered SWCNT film, the suspension amount for each layer of this film is exactly the same as that of the sample 2–1. The as-prepared SWCNT films were thoroughly cleaned in acetone and then in ethanol for future electrical property measurements. Post-treated samples (samples 2–3a and 2–3aN) were prepared by annealing the as-prepared SWCNT film named sample 2–3 at 300 °C in an Ar gas environment for 12 h with and without subsequent HNO3-treatments, respectively. The HNO3treatments were carried out by submerging the annealed SWCNT film in 12-M HNO3 for 0.5 h. Table 1 The thickness and relative density of the SWCNT thin films. Sample name
Sample description
1–1
1 ml SWCNT suspension, 1 layer 6 ml SWCNT suspension, 1 layer 2 ml SWCNT suspension, 1 layer 2 ml SWCNT suspension, 2 layer 2 ml SWCNT suspension, 3 layers 1 ml SWCNT suspension, 2 layers
6–1 2–1 2–2 2–3 1–2
Thickness (nm)
Standard error
Relative density ρ ( n–m ρ1–1 )
22.0
1.7
1.0
83.0
3.8
1.59
35.0
2.5
1.26
70.0
3.7
1.26
105.0
5.3
1.26
45.0
3.5
0.98
/
Table 2 The low temperature properties of the SWCNT thin films. Sample name
Sample description
Tb (K)
Ts (K)
TCR (%/K) (at 280 K)
1–2 2–1 2–3 2–3a 2–3aN
1 ml SWCNT suspension, 2 layers 2 ml SWCNT suspension, 1 layer 2 ml SWCNT suspension, 3 layers Sample 2−3 after annealing at 300 °C, 12 h Sample 2–3a after HNO3-treatment
55.0 34.5 23.7 30.9 7.7
21.1 17.5 19.8 16.6 12.1
− 0.065 − 0.042 − 0.028 − 0.038 − 0.010
Atomic force microscopy (AFM) topography images of the SWCNT thin films were acquired in the tapping mode. AFM was also used to measure the thickness of the SWCNT films. The dependence of the resistivity on temperature has been measured using a dc standard fourprobe method. Before measurement, Ag metal film was deposited on the SWCNT films to get a good ohmic contact. The resistance between the electrodes with different spacing was then measured and used to determine the contact resistance of the silver/nanotube contact and the intrinsic resistance of the SWCNT films, as reported in references [6] and [19]. Our results show that compared with the intrinsic resistance, the contact resistance is small. The dimension of the SWCNT films under test was 2 mm × 20 mm, and the area of the circular electrodes is less than 2 mm in diameter. The constant current was flowing along the longer edge (20 mm) and the voltage drop between the inner two electrodes was recorded automatically with a program-controlled Keithley voltmeter during low temperature measurement. For measurement of temperature-dependent resistivity, SWCNT film fixed in a cryogenic system was first cooled down from room temperature. The measurements were then taken while SWCNT films were heating in the temperature range of 80–280 K. Since the maximum temperature is well below the T*dedop (350 K) suggested by Teresa M. Barnes et al. [16], it is not difficult to understand that no obvious hysteresis was observed in the temperature-dependent resistivity obtained from different cycles of heating or cooling. 3. Results and discussion Fig. 1a shows isolated SWCNTs grown on a 5-pulse Co catalytic substrate as we reported before [3], which allows us to derive the diameter of SWCNTs from AFM topography data, and the typical results are shown in Fig. 1b. For every nanotube, tens of positions were measured to calculate the value of the average heights. The diameter distribution was calculated with the value of average height from more than 50 CNTs. It is found that the nanotube diameters are in the range of 0.6 nm to 1.6 nm, with a mean diameter about 1.0 nm. Fig. 1c illustrates SWCNT bundles grown on 10-pulse Co catalytic substrate with a high ethanol pressure during CVD. The increase of pulse number and ethanol pressure increases the yield of SWCNTs. The as-synthesized SWCNT bundles shown in Fig. 1c was scratched from the Si substrate and purified by HNO3. After purification, the obtained soot contains at least 90% SWCNTs by weight, and was used as the raw material for SWCNT film fabrication. Fig. 1d shows a typical AFM image of the SWCNT films fabricated by vacuum filtration. The root mean square (rms) roughness for the SWCNT film deduced from AFM data is about 8.0 nm, which is nearly same as the previous report [20]. The multi-layered SWCNT films were prepared with the “improved” vacuum filtration. Since each layer can be thoroughly washed from both sides of the layer with copious amounts of water [12], it is thought that the advantage of the layer-by-layer vacuum filtration over “normal” vacuum filtration reported by Wu et al. [18] might be that the improved vacuum filtration produces a “clean” multi-layered SWCNT film if we neglect (or remove) the residues of the (cellulose ester) membrane left after every deposition. As suggested by Wu et al. [18], the density and thickness of SWCNT films can be controlled well with the “normal” vacuum filtration. With
Y. Zhao, W. Li / Thin Solid Films 519 (2011) 7987–7991
7989
Fig. 1. (a) An AFM image of SWNTs grown on a 5-pulse Co catalytic Si substrate; (b) diameter distributions of isolated SWNTs shown in (a); (c) an AFM image of SWNT bundles grown on a 10-pulse Co catalytic Si substrate; (d) a typical AFM image of the SWCNT film prepared with vacuum filtration.
the improved vacuum filtration, it is possible to prepare a series of films with same density but different thickness, simply by adding or reducing the number of the layers. For this purpose, the most important issue is that the amount of SWCNT suspension used for every layer deposition must be kept as an exact constant. The above predication has been proved to be true in this work, as tabulated in Table 1. Three series of SWCNT films were prepared and summarized as follows. The first series of samples are samples n–1 (n = 1, 2, and 6) prepared with the “normal” vacuum filtration. The thickness increases from 22 nm, 35 nm to 83 nm when the amount of SWCNT suspension used for film fabrication changes from 1 ml, 2 ml to 6 ml, while the relative density changes from 1.0, 1.26 to 1.59 for samples 1–1, 2–1, and 6–1, respectively. Here the so-called “relative density” denotes a normalized value, which is defined as the ratio of a film density to the density of sample 1–1. The second series of SWCNT films are named as samples 2–m (m = 1, 2, and 3). The thickness increases from 35 nm, 70 nm to 105 nm when the amount of SWCNT suspension used for film fabrication changes from 1 ml, 2 ml to 3 ml, while the relative density keeps as a constant (1.26). The thickness of samples 2–m (m = 1, 2, 3) is proportional to the layer number. It is also found from Table 1 that the thickness and density of the third series of samples named sample 1–m (m = 1, 2) also follows the same rule as the samples 2–m, which indicates that the measurement is reliable. The deviation in relative density of sample 1–2 from sample 1–1 can be neglected, which may originate from the experimental error in thickness measurement. The comparison of sample 6–1 with sample 2–3 reveals that the multi-layer technique can reduce the density of the resulting films when keeping the total amount of the used SWCNT suspension as a constant (also proved by sample 2–1 versus sample 1–2). Since it has been proved with AFM that the thickness of the multi-layered SWCNT films is indeed proportional to the layer number, it is thought that the effect of the possible interface formed between adjacent layers for the multi-layered film can be ignored. On basis of the above discussion, two points for the multi-layer technique should be addressed here. First, the technique can prepare a series of films with same density but different thickness, simply by adding or reducing the number of the layers (e.g. sample 2–1 versus
2–3, which are prepared with same amount of SWCNT suspension for each layer of sample 2–3 and the single layer of sample 2–1); secondly, the multi-layer technique can reduce the density of the resulting films (e.g. sample 2–1 versus sample 1–2). Keeping the total amount of the used SWCNT suspension as a constant, the increase of the layer number of multi-layered SWCNT film will reduce the film density. Hopefully, for a future work, simply by changing the amounts of SWCNT suspension for each layer during deposition of a multi-layered SWCNT film, the layer-by-layer technique can produce multi-layered SWCNT film with tunable electric or optical properties in the thickness direction and form a gradient material. The gradient material with tunable electric and optical properties may draw a lot of attention in different fields, such as being a gradient refractive index material [21]. In theory, the temperature dependence of resistivity for SWCNT films can be understood by two existing mechanisms, one is variable range hoping (VRH), and the other is fluctuation-induced tunneling (FIT) [14–16]. The expression for the FIT-dominated resistivity can be written as Tb R = A exp ; TS + T
ð1Þ
where A is a constant, kTb reflects the order of magnitude of the barrier energies (where k is the Boltzmann constant). TS is another important parameter since FIT model controls the temperature dependence of resistivity for temperatures that are well above TS. Behnam et al. advocated that there exists different transport regimes for temperature dependent resistivity, and suggested that VRH is valid in the extremely low temperature range, while FIT can account for the transport in SWCNT films at a moderate temperature range below room temperature [14]. In this work, we repeat the similar measurement on single and the multi-layered SWCNT films at the temperatures between 80 K and 300 K. The simulation of the experimental data with the VRH model or combination of VRH model with FIT model has been performed, even though a large range of parameters have been initialized, the deviation of theory from the measurement cannot be neglected. However, in Fig. 2, it is shown that
7990
Y. Zhao, W. Li / Thin Solid Films 519 (2011) 7987–7991
the measurement can be fit satisfactorily based on FIT model, which is consistent with the work from Behnam et al. [14]. Fig. 2 shows the log plot of resistivity versus temperature for all the samples tabulated in Table 2. The sheet resistance of all samples changes consecutively with temperature. The decrease of the film layer number leads to an increase in the sheet resistance, as shown by the plots for samples 2–3 and 2–1 in Fig. 2. Fig. 2 also shows the change of sheet resistance of the SWCNT films before and after thermal annealing (plots for sample 2–3 and 2–3a) and after further HNO3-treatment (plot for 2–3aN). It demonstrates that thermal annealing results in an increase of the sheet resistance, while further HNO3-treatment will lead to a decrease of the sheet resistance. The lines are fits to the experimental data based on FIT model. The values of Tb and TS for all samples are tabulated in Table 2. Our results suggest that the single and multilayered SWCNT films can only be understood by FIT model at the moderate temperature regime. Bolometric responsivity is defined by the resistance variation due to a heating. Fig. 3 depicts the variation in TCR for samples 1–2, 2–1 and 2–3 as a function of temperature, which is deduced from Fig. 2 with the formula TCR = (ΔR/ΔT)/R, where R is the sheet resistance at temperature T which is obtained from Fig. 2. Fig. 4 shows the typical TCR ~ T plots for sample 2–3 before and after annealing and acid treatments. After the thermal annealing, the TCR absolute values increase from 0.028% (sample 2–3) to 0.038% (sample 2–3a) at 280 K. It is also revealed in Fig. 4 that the HNO3-treatment to the annealed samples will result in a remarkable decrease in the TCR absolute value; see the plot for sample 2–3aN in Fig. 4. For building a theoretical relationship of TCR with Tb based on FIT model, from Eq. (1), we get, 1 dR 1 = TCR = R dT A exp T
S
Tb + T
h d A exp T
S
Tb + T
i
dT
ð2Þ
then, TCR =
1 dR 1 = = R dT A exp T
Tb + TS
A exp
Tb T + TS
−
Tb 1 dTb : + ðT + TS Þ dT ðT + TS Þ2
ð3Þ
Further, considering Tb is a constant for a specialized SWCNT film, dT the term b is zero. If we ignore the term TS, it is reasonable since TS is dT
Fig. 2. The log plot of resistivity versus temperature for all the samples tabulated in Table 2. The lines are fits to the experimental data based on FIT model.
Fig. 3. The variation of TCR for SWCNT films as a function of temperature.
about 20 K, which is small compared with T (280 K), a simple expression will be derived, TCR∝−
Tb T ≈− b2 : ðT + TS Þ2 T
ð4Þ
Fig. 5 shows a TCR ~ Tb plot for all SWCNT films tabulated in Table 2 and a fit to the experimental data based on Eq. (4). For clarity, error bars are placed on all the experimental data points. As we expected, basically, the theory and experiment are consistent with each other. The deviation of experimental data from theory may originate from the approximation during Eq. (4) deduction or from the measurement error. From Eq. (4), we know that TCR is proportional to Tb at a fixed temperature. By comparing the TCR vs. T plots for samples 1–2 and 2–1 in Fig. 4, one can see that the magnitude of the TCR absolute value of higher density sample 2–1 is smaller than that of lower density sample 1–2, which reveals that the increase of SWCNT film density results in a decrease in the TCR absolute value, and, intrinsically, may originate from the change in Tb for the SWCNT films. For SWCNT films with same density, the comparison of TCR for samples 2–1 and 2–3 reveals that the TCR absolute value decreases with the increase in thickness of the SWCNT films.
Fig. 4. TCR~T plots for as-prepared, annealed and acid-treated three-layered SWCNT films.
Y. Zhao, W. Li / Thin Solid Films 519 (2011) 7987–7991
7991
hints on the application of SWCNT film in bolometers. It suggests that the layer-by-layer technique can produce a multi-layered SWCNT film with tunable electric or optical properties in the thickness direction, and form a gradient material, which may draw a lot of attention in different fields. The further study on a new gradient material based on multi-layered SWCNT films is underway. Acknowledgment This work was supported by the National Hi-Tech Research and Development Program of China (No. 2008AA01Z207, 863), Natural Science Foundation of Hubei Province, China (Grant No. 2010CDB01606) and Scientific Research Foundation for the Returned Overseas Chinese Scholars. References Fig. 5. TCR ~ Tb plot for all SWCNT films tabulated in Table 2 and a fit to the experimental data based on Eq. (4).
As we know, the as-prepared SWCNT film is of p-type which may originate from the acid reflux-based purification process of SWCNTs and atmospheric impurities, such as oxygen gas. Thermal annealing can remove the p-type dopants and make the SWCNT more semiconducting, and therefore increases the resistivity of single SWCNT and also the barrier energies among SWCNTs, which leads to an increase in Tb. We think accordingly that the increase of Tb may account for the increase of TCR after annealing. On the other hand, acid treatment can be ascribed to the charge transfer doping in the SWCNT films [22]. The decrease of the TCR of the acid treated SWCNT film means that HNO3-treatment makes SWCNTs more metallic, and barrier energies among SWCNTs decrease, which leads to a decrease in Tb. 4. Conclusions In this work, the low temperature electrical transport properties of the multi-layered single-walled carbon nanotube (SWCNT) films have been investigated. Similar to the single layer SWCNT film, our results suggest that the multi-layered SWCNT films prepared with a layer-bylayer vacuum filtration can also be understood by FIT model at a moderate temperature regime. The temperature coefficient of resistance (TCR) of the multi-layered SWCNT films has also been studied. The decrease in layer number of the multi-layered SWCNT films, decrease of the film density and the thermal annealing treatment are shown to be effective approaches to obtain a SWCNT film with a large TCR absolute value. A bridge is built connecting the measured TCR with the tunneling barrier, which may provide more
[1] P. Avouris, Z. Chen, V. Perebeinos, Nat. Nanotechnol. 2 (2007) 605. [2] M.S. Dresselhaus, G. Dresselhaus, R. Saito, A. Jorio, Phys. Rep 409 (2005) 47. [3] Yanli Zhao, Haruhisa Nakano, Hirohiko Murakami, Toshiki Sugai, Hisanori Shinohara, Yahachi Saito, Appl. Phys. A 85 (2006) 103. [4] Alexander A. Green, Mark C. Hersam, Nano Lett. 8 (2008) 1417. [5] Matthew A. Meitl, Yangxin Zhou, Anshu Gaur, Seokwoo Jeon, Monica L. Usrey, Michael S. Strano, John A. Rogers, Nano Lett. 4 (2004) 1643. [6] L. Hu, D.S. Hecht, G. Grüner, Nano Lett. 4 (2004) 2513. [7] Fumiaki N. Ishikawa, Hsiao-kang Chang, Koungmin Ryu, Po-chiang Chen, Alexander Badmaev, Lewis Gomez De Arco, Guozhen Shen, Chongwu Zhou, ACS Nano 3 (2009) 73. [8] Mikhail E. Itkis, Ferenc Borondics, Aiping Yu, Robert C. Haddon, Science 312 (2006) 413. [9] Chang-Seung Woo, Chae-Hyun Lim a, Chi-Won Cho b, Bonghyun Park a, Heongkyu Ju a, Dong-Hun Min b, Cheol-Jin Lee c, Seung-Beck Lee, Microelectron. Eng. 84 (2007) 1610. [10] Moon-Ho Ham, Byung-Seon Kong, Woo-Jae Kim, Hee-Tae Jung, Michael S. Strano1, Phys. Rev. Lett. 102 (2009) 047402. [11] Q. Cao, J.A. Rogers, Adv. Mater. 21 (2008) 29. [12] Yu Wang, Chong-an Di, Yunqi Liu, Hisashi Kajiura, Shanghui Ye, Lingchao Cao, Dacheng Wei, Hongliang Zhang, Yongming Li, Kazuhiro Noda, Adv. Mater. 20 (2008) 4442. [13] Alexandru R. Biris, Dan Lupu, Zhongrui Li, Hom R. Kandel, Enkeleda Dervishi, Viney Saini, Alexandru S. Biris, Appl. Phys. Lett. 91 (2007) 053115. [14] Ashkan Behnam, Amlan Biswas, Gijs Bosman, Ant Ural, Phys. Rev. B. 81 (2010) 125407. [15] Roderick K. Jackson, Andrea Munro, Kenneth Nebesny, Neal Armstrong, Samuel Graham, ACS Nano 4 (2010) 1377. [16] Teresa M. Barnes, Jeffrey L. Blackburn, Jao van de Lagemaat, Timothy J. Coutts, Michael J. Heben, ACS Nano 2 (2008) 1968. [17] Rongtao Lu, Guowei Xu, Judy Z. Wu, Appl. Phys. Lett. 93 (2008) 213101. [18] Z.C. Wu, Z.H. Chen, X. Du, J.M. Logan, J. Sippel, M. Nikolou, K. Kamaras, J.R. Reynolds, D.B. Tanner, A.F. Hebard, A.G. Rinzler, Science 305 (2004) 1273. [19] Fiona M. Blighe, Yenny R. Hernandez, Werner J. Blau, Jonathan N. Coleman, Adv. Mater. 19 (2007) 4443. [20] Yangxin Zhou, Liangbing Hu, George Grünera, Appl. Phys. Lett. 88 (2006) 123109. [21] Hao Lv, Aimei Liu, Jufang Tong, Xunong Yi, Qianguang Li, J. Opt. Soc. Am. A: 26 (2009) 1085. [22] Bhupesh Chandra, Ali Afzali, Neeraj Khare, Mostafa M. El-Ashry, George S. Tulevski, Chem. Mater. 22 (2010) 5179.