Investigation of the humidity-dependent conductance ...

Investigation of the humidity-dependent conductance of single-walled carbon nanotube networks Yunfeng Ling, Guiru Gu, Runyu Liu, Xuejun Lu, Vijaya Kayastha, Carissa S. Jones, Wu-Sheng Shih, and Daniel C. Janzen Citation: Journal of Applied Physics 113, 024312 (2013); doi: 10.1063/1.4774075 View online: http://dx.doi.org/10.1063/1.4774075 View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/113/2?ver=pdfcov Published by the AIP Publishing

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ] IP: 129.63.134.106 On: Tue, 21 Jan 2014 15:00:34

JOURNAL OF APPLIED PHYSICS 113, 024312 (2013)

Investigation of the humidity-dependent conductance of single-walled carbon nanotube networks Yunfeng Ling,1 Guiru Gu,1 Runyu Liu,1 Xuejun Lu,1,a) Vijaya Kayastha,2 Carissa S. Jones,2 Wu-Sheng Shih,2 and Daniel C. Janzen2

1 Department of Electrical and Computer Engineering, University of Massachusetts Lowell, One University Avenue, Lowell, Massachusetts 01854, USA 2 Brewer Science, Inc., 2401 Brewer Drive, Rolla, Missouri 65401, USA

(Received 4 September 2012; accepted 17 December 2012; published online 10 January 2013) In this paper, we investigate the conductance of single walled carbon nanotube (SWCNT) networks at different humidity levels and various device temperatures. The carrier transport processes are analyzed by performing a temperature-dependent conductance study. It is found that the conductance of the SWCNT networks is dominated by the thermal activation carrier hopping over the barriers between CNTs. The average separation between the SWCNTs is found to vary linearly with the humidity levels. The humidity-dependent conductance of the SWCNT network is modeled and compared with the experimental data. The model agrees well with the experimental C 2013 American Institute of Physics. [http://dx.doi.org/10.1063/1.4774075] data. V I. INTRODUCTION

II. EXPERIMENTAL

Due to their high surface area to volume ratio, nanostructured materials, such as nanowires,1 carbon nanotubes (CNTs),2–4 nanoparticles, and nanopillars5,6 have been extensively investigated to enhance the sensitivity of humidity sensors. Compared with other nanomaterials,1,5,6 CNTs also show excellent electronic and mechanical properties, including mechanical flexibility7 and high electron mobility.8,9 These properties make CNTs promising materials for various applications ranging from nanoelectronics10–12 to chemical13 and humidity sensors.14–16 Compared with a single CNT, CNT networks contain much denser CNTs, and thus, can offer much higher sensitivity for sensors. In addition, CNT networks can be fabricated using solution-cast or direct-printing methods on flexible substrates or conformal surfaces.17,18 This would not only enable the development of chemical sensors on any desired flexible surfaces but also allow them to be printed together with other flexible electronic devices, such as flexible thinfilm transistors17–19 and RF antennas, thereby making it promising for the development of integrated flexible sensors and electronics circuits with integrated sensing, signal processing, and signal transmission and receiving functionalities. Despite the aforementioned advantages, however, knowledge of the single walled carbon nanotube (SWCNT) network based humidity sensors is limited. In this paper, we investigate humidity-dependent conductance of SWCNT networks. The carrier transport processes is analyzed by performing a temperature-dependent conductance study. Humidity-dependent conductance is modeled and compared with the experimental data. Such knowledge about the humidity effect on the electrical properties of CNT networks will not only benefit reliability studies of the CNT-based flexible electronics but also facilitate the development of CNT-based humidity sensors with high sensitivity.

The humidity sensor is a SWCNT network based resistor (referred to as “the device” henceforth). Fig. 1(a) shows the microscopic picture of the device. It consists of a pair of gold (Au) electrodes and a thin layer SWCNT film coated on the Au electrodes. The device was built on a p-type silicon substrate with a 500 nm-thick SiO2 layer. The thickness of the Au electrodes is 200 nm. The spacing between two Au electrodes, L (i.e., the length of the CNT channel), is 20 lm. The width of the device, W, is 500 lm. Fig. 1(b) shows the atomic force microscopy (AFM) image of the SWCNT

a)

Author to whom correspondence should be addressed. Electronic addresses: [email protected].

0021-8979/2013/113(2)/024312/5/$30.00

FIG. 1. (a) Microscopic picture of the device with two Au electrodes and SWCNT networks on Si/SiO2 substrate. The length L and the width of the device W are 20 lm and 500 lm, respectively; (b) AFM image of the CNT networks.

113, 024312-1

C 2013 American Institute of Physics V


024312-2

Ling et al.

J. Appl. Phys. 113, 024312 (2013)

network film. The SWCNT network film was formed by dispensing a drop (approximately 10 ll) of ultrapure SWNT aqueous dispersion (Brewer Science, Inc. (BSI)) onto the Au electrodes on the surface of the silicon substrate. After the solution casting, the device was air-dried and put on a hot plate at 90 C under atmosphere for 30 min to improve its stability. A SWCNT network film was finally formed across the Au electrodes. The CNT film thickness was estimated to be less than 1 lm.17 The CNT dispersion is pure waterbased, surfactant-free and contains freely suspended SWCNTs. The material is purified to remove non-nanotube carbon impurities and catalyst metal particle impurities. The metal content in the water solution is less than 500 ppb. The tubes are functionalized with OH-groups to keep them disperse in the solution. The device substrate (p-type silicon) is attached to an Au-coated copper plate on the top of a four-stage thermoelectric (TE) module. The TE module was used to control the temperature of the device by a temperature controller (ILX-Lightwave, LDT-5412). The copper plate and the TE module were grounded in the measurement. Since the device is on a p-type silicon substrate that is attached to the grounded copper-plate, the gate of the device is effective grounded. The electrical characteristics of the device were measured using a Keithley 2602 dual source meter in a controlled humidity chamber, which is a closed chamber with a precise temperature tuning range up to 55 C and humidity control resolution of 0.5% relative humidity (RH). The environment temperature is fixed at 304.3 K. III. RESULTS AND DISCUSSION

We first determined the contact resistance between the electrode and the SWCNT network. The total resistance of the SWCNT network based resistor R can be written as20 R ¼ 2Rc þ Rsh

L ; W

(1)

where Rsh is the sheet resistance of the CNT networks, Rc is the contact resistance between the CNT network and metal electrode, and L is the spacing between the electrodes. The sheet resistance Rsh and the contact resistance Rc can be obtained by the transmission line method (TLM),21,22 which need to measure the total resistance R at different channel lengths. Fig. 2(a) depicts the picture of a 4-channel device for Rsh and Rc measurement. The channel lengths L are 5, 10, 20, and 40 lm. The width of channel W is 600 lm. Fig. 2(b) plots the measured resistance R at different channel lengths (R-L curve) at the humidity of 4.28 g/m3. The R-L curve agrees well with Eq. (1). The intercept and the slope of the R-L curve give 2Rc and Rsh/W, respectively. The Rc and Rsh are calculated to be 83.4 X and 26.8 kX/sq, respectively. Furthermore, Eq. (1) can also be written in terms of the characteristic length of the SWCNT networks Lt as R ¼ 2Rsh

Lt L þ Rsh ; W W

where Rc and Lt are related by

(2)

FIG. 2. (a) Picture of the 4-channel device for Rsh and Rc measurement. The spacings of the channels are 5, 10, 20, and 40 lm. The width of channel is 600 lm and width of the Au electrodes is 60 lm. (b) Device resistance as a function of spacing. The intercept and the slope of the R-L curve give 2Rc and Rsh/W, respectively.

Rc ¼ Rsh

Lt : W

(3)

From Eq. (3), the characteristic length of the SWCNT networks Lt can thus be obtained to be 1.87 lm. It was also verified that the Lt is quite stable (300 K. 1 Therefore, the r is primarily proportional to exp T T ; i.e., T1 : (8) r rth ¼ r0 exp T This indicates that the conductance contribution is mainly from the thermally activated carrier hoping and the conductance contribution from the carrier tunneling through the CNT barriers is low. The conductance of the whole CNT network film can be thus be expressed as T1 : (9) GN / exp T As shown in Eq. (5), the T1 values depend on the separation w between the CNTs. We now determine the change of the separation w at different humidity levels. Fig. 6 shows T1 values as a function of the humidity levels, qw. The diamonds are the experimental data, and the solid line is the linear fitting curve. A good linearity between T1 and the humidity, qw, is obtained with a correlation coefficient of 0.9949. The slope and the interception of the linear fitting are 13.58 K/g/cm3 and 767.53 K, respectively. The T1 can thus be written as T1 ðqw Þ ¼ 13:58qw þ 767:53:

(10)

Eq. (10) indicates that the average separation between the SWCNTs varies linearly with the humidity levels.

FIG. 6. T1 values as a function of the humidity levels, qw. The diamonds are the experimental data, and the solid line is the linear fitting curve. A good linearity between T1 and the humidity, qw, was obtained with a correlation coefficient of 0.9949. The slope and the interception of the linear fitting curves are 13.58 K/g/cm3 and 767.53 K, respectively.

Combining Eqs. (4) and (10), one obtains the linear relationship of the average separation between the CNTs, w, with the humidity, qw w¼

8pkB ð13:58qw þ 767:53Þ: SE20

(11)

The linear relationship of the separation, w, with the humidity, qw, suggests a linear dilution effect, i.e., the CNT separation increases evenly with the humidity levels, qw. Combining Eqs. (9) and (10), one gets 13:58qw þ 767:53 ; (12) GN ¼ A exp T where A is a temperature-independent pre-exponential factor. The factor A can be obtained by comparing the experimental data of the CNT networks conductance, GN, with the calculated values by using Eq. (9). To further investigate the dependence of the factor A on humidity, we plot the factor A in logarithmic scale and at different humidity levels in Fig. 7. The dots are the experimental data, and the solid line is the linear curve fitting. A linear relation was obtained between LnðAÞ and qw with a linearity of 0.9906. The factor Aðqw Þ can be thus expressed as Aðqw Þ ¼ expð0:0326qw 5:95Þ:

(13)

Very similar results were obtained for other temperatures as well, indicating that the factor A is truly temperatureindependent. The physics of the humidity dependence of the pre-exponential factor A is not clear and is subject to further research. Finally, combining Eqs. (12) and (13), one can express the model of the SWCNT networks conductance at different humidity levels and device temperatures by 13:58qw þ 767:53 ; GN ðT; qw Þ ¼ A0 exp 0:0326qw T (14) where A0 is a temperature- and humidity-independent constant. A0 ¼ 0:0026 S for this device.

FIG. 7. Relationship between the factor A in logarithm scale and different humidity levels at device temperature 312.9 K. The dots are the experimental data and the solid line is the linear curve fitting. A linear relation is obtained between LnðAÞ and qw with a linearity of 0.9906.


024312-5

Ling et al.

J. Appl. Phys. 113, 024312 (2013) 3

FIG. 8. Measured conductance of SWCNT networks compared with the calculated values using the model in Eq. (14) at different device temperatures and humidity levels. The model agrees well with the experimental data.

Fig. 8 illustrates the measured conductance of SWCNT networks compared with the calculated values using the model in Eq. (14) at different device temperatures and humidity levels. The model agrees well with the experimental data. In conclusion, we investigated the conductance of the SWCNT networks at different humidity levels and device temperatures. It is found that the conductance of the SWCNT networks is dominated by the thermal activation carrier hopping over the barriers between the SWCNTs. The average separation between the SWCNTs increases linearly with the humidity levels. The humidity-dependent conductance of the SWCNT network is modeled and compared with the experimental data. The model agrees well with the experimental data. ACKNOWLEDGMENTS

The authors would like to thank Dr. Xingwei Wang’s group at University of Massachusetts Lowell for providing the humidity test chamber in this research. This work was supported in part by AFOSR under Award No. FA9550-081-0070, and NSF STTR Phase II (Contract No. 924563). 1

Q. Kuang et al., “High-sensitivity humidity sensor based on a single SnO2 nanowire,” J. Am. Chem. Soc. 129(19), 6070–6071 (2007). 2 S. Iijima, “Helical micro-tubules of graphitic carbon,” Nature 345, 56–58 (1991).

J. T. W. Yeow and J. P. M. She, “Carbon nanotube-enhanced capillary condensation for a capacitive humidity sensor,” Nanotechnology 17(21), 5441 (2006). 4 J. Li et al., “Carbon nanotube sensors for gas and organic vapor detection,” Nano Lett. 3(7), 929–933 (2003). 5 M. M. Ahmad et al., “Dielectric behavior and ac conductivity study of NiO/Al2O3 nanocomposites in humid atmosphere,” J. Appl. Phys. 100(9), 094323 (2006). 6 S. J. Kim et al., “Humidity sensors using porous silicon layer with mesa structure,” J. Phys. D: Appl. Phys. 33(15), 1781 (2000). 7 J. D. W. Madden, “Stiffer than steel,” Science 323(5921), 1571–1572 (2009). 8 T. D€ urkop et al., “Extraordinary mobility in semiconducting carbon nanotubes,” Nano Lett. 4(1), 35–39 (2004). 9 K. Bradley et al., “Flexible nanotube electronics,” Nano Lett. 3(10), 1353– 1355 (2003). 10 Q. Cao et al., “Medium-scale carbon nanotube thin-film integrated circuits on flexible plastic substrates,” Nature 454(7203), 495–500 (2008). 11 J. Vaillancourt et al., “All ink-jet-printed carbon nanotube thin-film transistor on a polyimide substrate with an ultrahigh operating frequency of over 5 GHz,” Appl. Phys. Lett. 93(24), 243301 (2008). 12 G. Gu et al., “All-printed thin-film transistor based on purified singlewalled carbon nanotubes with linear response,” J. Nanotechnol. 2011, 823680 (2011). 13 T. Zhang et al., “Recent progress in carbon nanotube-based gas sensors,” Nanotechnology 19(33), 332001 (2008). 14 L. Liu et al., “Humidity sensitivity of carbon nanotube and poly (dimethyldiallylammonium chloride) composite films,” IEEE Sens. J. 9(10), 1308– 1314 (2009). 15 R. Pati et al., “Effect of H2O adsorption on electron transport in a carbon nanotube,” Appl. Phys. Lett. 81(14), 2638–2640 (2002). 16 Q. Y. Tang et al., “Fast response resistive humidity sensitivity of polyimide/multiwall carbon nanotube composite films,” Sens. Actuators B 152(1), 99–106 (2011). 17 J. Vaillancourt et al., “High-speed thin-film transistor on flexible substrate fabricated at room temperature,” Electron. Lett. 42(23), 1365–1367 (2006). 18 X. Han et al., “Printable high-speed thin-film transistor on flexible substrate using carbon nanotube solution,” Micro Nano Lett. 2(4) 96–98 (2007). 19 G. Gu et al., “All-printed CNT transistors with high on-off ratio and biasinvariant transconductance,” Proc. SPIE 8101, 8101B (2011). 20 R. Jackson and G. Samuel, “Specific contact resistance at metal/carbon nanotube interfaces,” Appl. Phys. Lett. 94, 012109 (2009). 21 W. M. Loh et al., “Modeling and measurement of contact resistances,” IEEE Trans. Electron Devices 34(3), 512–524 (1987). 22 H. H. Berger, “Contact resistance and contact resistivity,” J. Electrochem. Soc. 119(4), 507–514 (1972). 23 A. B. Kaiser, “Electronic transport properties of conducting polymers and carbon nanotubes,” Rep. Prog. Phys. 64(1), 1 (2001). 24 P. Sheng, “Fluctuation-induced tunneling conduction in disordered materials,” Phys. Rev. B 21, (6), 2180 (1980).
