Zhang et al APL 2017

Fully printed flexible carbon nanotube photodetectors Suoming Zhang, Le Cai, Tongyu Wang, Jinshui Miao, Nelson Sepúlveda...

0 downloads 63 Views 1MB Size
Fully printed flexible carbon nanotube photodetectors Suoming Zhang, Le Cai, Tongyu Wang, Jinshui Miao, Nelson Sepúlveda, and Chuan Wang

Citation: Appl. Phys. Lett. 110, 123105 (2017); doi: 10.1063/1.4978935 View online: http://dx.doi.org/10.1063/1.4978935 View Table of Contents: http://aip.scitation.org/toc/apl/110/12 Published by the American Institute of Physics

APPLIED PHYSICS LETTERS 110, 123105 (2017)

Fully printed flexible carbon nanotube photodetectors lveda, Suoming Zhang, Le Cai, Tongyu Wang, Jinshui Miao, Nelson Sepu and Chuan Wanga) Department of Electrical & Computer Engineering, Michigan State University, East Lansing, Michigan 48824, USA

(Received 9 December 2016; accepted 8 March 2017; published online 20 March 2017) Here, we report fully printed flexible photodetectors based on single-wall carbon nanotubes and the study of their electrical characteristics under laser illumination. Due to the photothermal effect and the use of high purity semiconducting carbon nanotubes, the devices exhibit gate-voltage-dependent photoresponse with the positive photocurrent or semiconductor-like behavior (conductivity increases at elevated temperatures) under positive gate biases and the negative photocurrent or metal-like behavior (conductivity decreases at elevated temperatures) under negative gate biases. Mechanism for such photoresponse is attributed to the different temperature dependencies of carrier concentration and carrier mobility, which are two competing factors that ultimately determine the photothermal effect-based photoresponse. The photodetectors built on the polyimide substrate also exhibit superior mechanical compliance and stable photoresponse after thousands of bending cycles down to a curvature radius as small as 3 mm. Furthermore, due to the low thermal conductivity of the plastic substrate, the devices show up to 6.5 fold improvement in responsivity compared to the devices built on the silicon substrate. The results presented here provide a viable path to low cost and high performance flexible photodetectors fabricated entirely by the printing process. Published by AIP Publishing. [http://dx.doi.org/10.1063/1.4978935]

Photodetectors are essential for applications in highresolution imaging, light wave communications, and optical interconnects.1–3 One particular type of photodetector is bolometer and it is extensively used in thermal imaging, astronomy, and particle physics.4 The bolometer converts the absorbed radiation into a rise of temperature, which subsequently results in a change in electrical resistance that can be measured using electronic circuits. Recent progress on nanomaterials provides a promising path for achieving bolometers with significantly improved responsivity and switching speed. Among those nanomaterials, semiconducting single walled carbon nanotubes (SWNTs) feature high carrier mobility and direct bandgap, and therefore, have been widely studied as the channel material for high-performance electronic and optoelectronic devices.5–10 Furthermore, owing to its inherently small diameters, networks composed of SWNTs exhibit excellent mechanical flexibility, and they can be processed using the solution-based room temperature process. The above advantages make the SWNT ideal for flexible electronics applications, such as flexible display, integrated circuits, and sensors.11–13 In terms of fabrication, printing processes such as roll-to-roll printing, screen printing, and ink-jet printing have been proposed as viable solutions for realizing flexible electronic systems over large areas at extremely low cost.13–18 The printing processes can also be used for fabricating high-performance carbon nanotube transistors or photodetectors. Several groups including ourselves have reported bolometers or photothermal-effectbased photodetectors using carbon nanotubes.19–27 Despite the success, all previously reported results had to rely on a)

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


conventional microfabrication processes such as metal deposition, plasma etching, or electron-beam lithography. These approaches either require high vacuum or are not suitable or economical for large area flexible electronics applications. Herein, we report fully printed thin-film transistors using solution-processed SWNTs inks. Unlike the conventional lithography- and etching-based fabrication processes, our printing method is maskless and only applies the materials into the active region, thus saving the materials. Such transistors can be printed directly onto a polyimide substrate and are thus flexible. Additionally, the transistors also work effectively as photodetectors due to the photothermal effect. Fig. 1(a) shows the schematic diagram of a fully printed back-gated thin film transistor used in this study. The fabrication process begins with spin coating of an ultrathin polyimide (HD MicroSystems, PI-2525) film with a thickness of 10 lm on top of a silicon wafer. The polyimide film serves as the substrate, and its ultra-small thickness is essential for achieving devices with good mechanical flexibility. Next, the back-gate electrode was printed using the silver nanoparticle ink (PG-007 AA, Paru Corporation, South Korea) by a GIX Microplotter (Sonoplot Inc.), followed by annealing at 180  C for 10 min to improve the conductivity of the silver electrode. Subsequently, an ink composed of BaTiO3 nanoparticles and PMMA (PD-100, Paru Corporation, South Korea) was printed as the gate dielectric layer. The substrate was placed on a hot plate at 60  C while printing in order to avoid the coffee ring effect and to achieve uniform printing results. The dielectric layer was printed for a total of three times in order to eliminate the possible gate leakage. The resulting dielectric film has a thickness of around 3.5 lm. Poly-L-lysine solution (0.1% w/v in water; Sigma Aldrich) was then printed on top of the gate dielectric layer followed

110, 123105-1

Published by AIP Publishing.


Zhang et al.

Appl. Phys. Lett. 110, 123105 (2017)

FIG. 1. (a) Schematic diagram of a fully printed flexible photodetector using SWNTs. (b) Optical microscope image of the printed transistor. Inset: SEM image of the printed SWNT network on top of the PMMA/BaTiO3 gate dielectric layer.

by rinsing with deionized water and blowing dry with nitrogen. This step functionalizes the dielectric surface with amine-termination, which leads to better adhesion for carbon nanotubes. With the surface functionalization, high purity semiconducting carbon nanotube ink (0.01 mg/ml, 99% semiconducting, NanoIntegris Inc.) could be printed into the transistor channel region with great uniformity. The carbon nanotube ink was printed for multiple times (typically 10–20 times) until desired nanotube network density and optimal transistor performance were achieved. Finally, source/drain (S/D) electrodes were printed following the same method as the printing of the gate (G) electrode and using the same silver ink. Once the fabrication process was completed, the polyimide substrate was peeled off from the silicon handling wafer, resulting in an extremely flexible thin-film transistor. Fig. 1(b) presents the optical micrograph of the printed transistor with the well-defined channel region and gate, source, and drain electrodes. In addition, the scanning electron microscopy (SEM) image is provided to show the uniformity of the printed SWNT network in the channel. The fully printed SWNT transistor was used as a photodetector, and its photoresponse was characterized and presented in Fig. 2. Figs. 2(a) and 2(b) show the transfer (IDS-VGS) and output (IDS-VDS) characteristics of the transistor with (dashed lines) and without (solid lines) laser illumination on the channel region. A red laser diode with a wavelength (k) of 650 nm and an intensity of 4 W/cm2 was used. The beam spot of the laser has a diameter of 600 lm so that it roughly covers the entire channel of the transistor. The transfer characteristics were measured with VGS varying from 40 V to 50 V under different drain voltages of 10 V,

20 V, 30 V, and 40 V. The output characteristics were measured by sweeping VDS from 0 V to 40 V under different VGS varying from 40 V to 35 V in 15 V steps. The applied gate and drain biases were fairly high due to the relatively large thickness (3.5 lm) of the printed gate dielectric layer. The transistors exhibit conventional field-effect transistor characteristics with on/off ratio of 500. With the laser turned on, both transfer and output characteristics underwent drastic changes. Specifically, Fig. 2(a) shows that the light illumination leads to increased drain current for positive gate biases (transistor off) and decreased drain current for more negative gate biases (transistor on). Similar phenomenon was observed and reported in our previously published work for devices built on rigid silicon substrates.28 The observed photoresponse can be attributed to the photothermal effect caused by the temperature rise in the carbon nanotubes network upon laser illumination. As reported in our previous work,28 due to the strong light absorption coefficient, low heat capacity of the carbon nanotubes network, and low thermal conductivity of the polyimide substrate, the temperature of the transistor channel region would increase dramatically upon laser irradiation. It is worth mentioning that while similar photothermal effectbased carbon nanotube photodetectors have been reported in the literature,23,24,26 all those devices are two-terminal devices, in which the carbon nanotube film works as a conductor whose resistivity only increases upon light illumination. Compared to a previous work, our work uses high-purity semiconducting carbon nanotube as the channel material and adopts a three-terminal device configuration with the addition of a gate electrode, thereby allowing us to directly

FIG. 2. (a) Transfer characteristics of the fully printed transistor with and without laser illumination. (b) Output characteristics of the fully printed transistor with and without laser illumination. The inset shows a zoomed in view for IDS from 0 to 0.2 lA and VDS from 40 to 35 V. (c) The relative change of channel resistance as a function of the laser intensity under different gate biases.


Zhang et al.

Appl. Phys. Lett. 110, 123105 (2017)

probe the effect of carrier concentration and carrier mobility on the photoresposnsivity. Generally speaking, the temperature rise of a semiconductor could result in more thermally excited electron-hole pairs and an increase in intrinsic carrier concentration (ni) as well as a decrease in the carrier mobility (l) due to stronger phonon scattering. These two competing factors have different temperature dependencies Eg (n / eð kT Þ for carrier concentration and l / T 3=2 for cari

rier mobility due to phonon scattering) and would determine the ultimate conductivity-temperature relationship of the semiconductor and consequently the photothermal effectbased photoresponse. Due to the use of high-purity semiconducting carbon nanotubes, we were able to observe gate voltage-dependent photoresponse in our devices and the transition from metal-like behavior (conductivity decreases at elevated temperatures) to semiconductor-like behavior (conductivity increases at elevated temperatures). More specifically for our devices, the channel is fully depleted under very positive gate biases (VG > 20 V in Fig. 2(a)), resulting in a very low carrier concentration. Consequently, the increased concentration of thermally excited carriers at higher temperatures upon laser illumination would be the prominent factor in affecting the conductivity, which leads to an increase in the drain current. In contrast, under more negative gate biases (VG < 0 V in Fig. 2(a)), the device is turned on and has rather high carrier concentration so that the conductivity would be mostly influenced by the reduced carrier mobility at elevated temperatures, resulting in a decreased output current. One could further conclude that a transition gate voltage should exist, under which the two factors discussed above would contribute equally in affecting the channel conductivity of the device, leading to zero photoresponse. Such transition gate voltages are indicated by the grey dashed line in Fig. 2(a). We have also extracted the relative change in channel resistance DR/R0 for various VG, which is plotted as a function of the incident laser intensity as illustrated in Fig. 2(c). Higher laser intensity generally leads to stronger photoresponse (larger DR/R0), which could be attributed to the higher temperature rise caused by more absorbed photons under high intensity laser illumination. Depending on the applied gate voltage, the increase in laser intensity could lead to either resistance decrease (for VG ¼ 40 or 25 V) or increase (for VG ¼ 10, 5, 20, or 35 V). When the gate voltage is somewhere in between 10 V and 25 V, the factor of increased carrier concentration and the factor of reduced mobility might contribute equally in affecting the conductivity of the SWNT network, under the condition where the DR/R0 could become more or less independent of the light intensity. For photodetection applications, this is the gate voltage that

should be avoided. In order to achieve good responsivity in such photothermal-effect-based photodetectors, more negative gate voltages are desired. The highest responsivity in our device is 1.1  103 A/W achieved at VG ¼ 35 V and VD ¼ 40 V. It is also worth mentioning that compared to our previously reported devices built on the silicon substrate whose responsivity was 1.7  104 A/W under the same laser wavelength (650 nm),28 the fully-printed flexible photodetectors in the work offer 6.5 times improvement in responsivity. The improvement in responsivity can be attributed to the low thermal conductivity of the plastic substrate used, which leads to the stronger photothermal effect. Table I summarizes and compares the key figures-of-merit for photothermal effect-base carbon nanotube photodetectors. Due to the use of the plastic substrate, whose low thermal conductivity leads to a stronger photothermal effect and thus larger photoresponse, the responsivity and DR/R0 under similar laser intensities are both orders of magnitude better than devices reported in the literature. The photo-switching behavior of the printed photodetector was also measured using the same 650 nm laser and the results are presented in Fig. 3. Fig. 3(a) shows the dynamic response of the device measured under different gate biases with the laser intensity fixed at 5.2 W/cm2. Similar to the results presented in Fig. 2, the device exhibits VG-dependent photoresponse, which means opposite responses are achieved between the on- (with negative VG applied) and off- (with positive VG applied) states of the transistor. With a gate bias of 20 V (the cyan curve), the device exhibit almost no photoresponse, due to the reasons explained above. The rise and fall times are extracted to be 2 ms and 4 ms, respectively. Compared to the results in the literature, our device offers one of the fastest response time, which could be attributed to the high quality and high purity semiconducting carbon nanotube used in the channel of the photodetectors. The high thermal conductivity and small heat capacity of the monolayer carbon nanotube network lead to the observed fast photothermal response. Additionally, as shown in Figs. 3(b) and 3(c), the device exhibits larger response under higher intensity laser illumination for both positive and negative VG biases, which is in agreement with the results in Fig. 2(c). Finally, because the mechanism of the photoresponse in our devices is mainly due to the photothermal effect, the magnitude of the photoresponse is mostly determined by the light absorption coefficient, thermal conductivity, and the heat capacity of the carbon nanotube film and plastic substrate, which are expected to remain stable over time. As a result, the magnitude of the photoresponse in these devices is also stable and remains almost unchanged after around 4 weeks as shown in Fig. 3(d).

TABLE I. Comparison of the photothermal effect-based carbon nanotube photodetector performance reported in the literature. References 23 24 26 Current work

CNT type


DR/R0 and the corresponding laser intensity

VD or ID bias

Response time


N/A N/A  30 V/W at 300 K (or 2.75  106 A/W) 1.1  103 A/W

1.2% at 3.5 mW/mm2 0.3% at 3.5 mW/mm2 0.7% at 0.12 lW (laser intensity N/A) 390% at 52 mW/mm2 81% at 5.3 mW/mm2

100 lA 1 lA N/A 40 V

1–2.5 ms 40–50 ms 50 ms 2–4 ms


Zhang et al.

Appl. Phys. Lett. 110, 123105 (2017)

FIG. 3. Dynamic response of the fully printed SWNT-based photodetector (a) under various gate biases with the laser intensity fixed at 5.2 W/cm2; (b) under various laser intensities with the gate bias fixed at 40 V; (c) under various laser intensities with the gate bias fixed at 40 V; and (d) measured on different dates to show the stability of the photoresponse.

FIG. 4. Bending test results of the printed photodetector showing the photoresponse under (a) different curvature radii and (b) different bending cycles.

Thanks to the ultrathin polyimide substrate as well as the remarkable mechanical flexibility of the SWNT network; the printed photodetector is highly flexible. Results about bending tests conducted on the printed transistors are illustrated in Fig. 4. The laser intensity was fixed at around 4 W/cm2, and the VGS and VDS used are 35 V and 40 V, respectively. The photoresponse of the transistor under laser illuminating exhibits only minimal variations even when the curvature radius goes down to 3 mm (Fig. 4(a)) and almost remains constant throughout a process of up to 1000 bending cycles (Fig. 4(b)) with a curvature radius of 5 mm. The inset of Fig. 4(a) shows the device wrapped onto a glass tube for the bending test. The stability and reliability of the devices under different curvature radii and bending conditions demonstrate its potential for applications in flexible electronics. To conclude, we have reported methods for fabricating SWNT-based thin-film transistors using a fully printed process and systematically studied their photoresponse. The photothermal effect was observed in such devices upon laser

illumination, which leads to gate-voltage-dependent photoresponse characteristics. Additionally, the photodetectors exhibit good reliability under bending conditions because of the ultrathin polyimide substrate used and the superior mechanical flexibility of the carbon nanotube network. Such photodetectors would be useful for the development of ubiquitous, low-cost, and large area flexible electronic systems. This work was funded by Michigan State University and the National Science Foundation under Grant No. ECCS-1549888. 1

R. Kapadia, Z. Fan, and A. Javey, Appl. Phys. Lett. 96(10), 103116 (2010). 2 S.-F. Leung, Q. Zhang, F. Xiu, D. Yu, J. C. Ho, D. Li, and Z. Fan, J. Phys. Chem. Lett. 5(8), 1479 (2014). 3 X. Liu, L. Gu, Q. Zhang, J. Wu, Y. Long, and Z. Fan, Nature Commun. 5, 4007 (2014). 4 P. Richards, J. Appl. Phys. 76(1), 1 (1994). 5 C. Wang, J. Zhang, and C. Zhou, ACS Nano 4(12), 7123 (2010).

123105-5 6

Zhang et al.

C. Wang, J. Zhang, K. Ryu, A. Badmaev, L. G. De Arco, and C. Zhou, Nano Lett. 9(12), 4285 (2009). 7 C. Wang, K. Takei, T. Takahashi, and A. Javey, Chem. Soc. Rev. 42(7), 2592 (2013). 8 C. Wang, D. Hwang, Z. Yu, K. Takei, J. Park, T. Chen, B. Ma, and A. Javey, Nature Mater. 12(10), 899 (2013). 9 P. H. Lau, K. Takei, C. Wang, Y. Ju, J. Kim, Z. Yu, T. Takahashi, G. Cho, and A. Javey, Nano Lett. 13(8), 3864 (2013). 10 X. He, N. Fujimura, J. M. Lloyd, K. J. Erickson, A. A. Talin, Q. Zhang, W. Gao, Q. Jiang, Y. Kawano, and R. H. Hauge, Nano Lett. 14(7), 3953 (2014). 11 C. Wang, J.-C. Chien, K. Takei, T. Takahashi, J. Nah, A. M. Niknejad, and A. Javey, Nano Lett. 12(3), 1527 (2012). 12 T. Takahashi, K. Takei, A. G. Gillies, R. S. Fearing, and A. Javey, Nano Lett. 11(12), 5408 (2011). 13 L. Cai, S. Zhang, J. Miao, Z. Yu, and C. Wang, Adv. Funct. Mater. 25(35), 5698 (2015). 14 J. A. Rogers, Z. Bao, K. Baldwin, A. Dodabalapur, B. Crone, V. Raju, V. Kuck, H. Katz, K. Amundson, and J. Ewing, Proc. Natl. Acad. Sci. U. S. A. 98(9), 4835 (2001). 15 Y. Liu and T. Cui, Macromol. Rapid Commun. 26(4), 289 (2005). 16 Z. Fan, J. C. Ho, T. Takahashi, R. Yerushalmi, K. Takei, A. C. Ford, Y. L. Chueh, and A. Javey, Adv. Mater. 21(37), 3730 (2009).

Appl. Phys. Lett. 110, 123105 (2017) 17

B. Chen, T. Cui, Y. Liu, and K. Varahramyan, Solid-State Electron. 47(5), 841 (2003). Z. Bao, Y. Feng, A. Dodabalapur, V. Raju, and A. J. Lovinger, Chem. Mater. 9(6), 1299 (1997). 19 G. Vera-Reveles, T. J. Simmons, M. Bravo-Sanchez, M. Vidal, H. Navarro-Contreras, and F. J. Gonzalez, ACS Appl. Mater. Interfaces 3(8), 3200 (2011). 20 M. Tarasov, J. Svensson, J. Weis, L. Kuzmin, and E. Campbell, JETP Lett. 84(5), 267 (2006). 21 M. Tarasov, J. Svensson, L. Kuzmin, and E. E. Campbell, Appl. Phys. Lett. 90(16), 163503 (2007). 22 M. Mahjouri-Samani, Y. Zhou, X. He, W. Xiong, P. Hilger, and Y. Lu, Nanotechnology 24(3), 035502 (2013). 23 R. Lu, J. J. Shi, F. J. Baca, and J. Z. Wu, J. Appl. Phys. 108(8), 084305 (2010). 24 R. Lu, Z. Li, G. Xu, and J. Z. Wu, Appl. Phys. Lett. 94(16), 163110 (2009). 25 R. Lu, R. Kamal, and J. Z. Wu, Nanotechnology 22(26), 265503 (2011). 26 M. E. Itkis, F. Borondics, A. Yu, and R. C. Haddon, Science 312(5772), 413 (2006). 27 A. E. Aliev, Infrared Phys. Technol. 51(6), 541 (2008). 28 S. Zhang, L. Cai, T. Wang, R. Shi, J. Miao, L. Wei, Y. Chen, N. Sep ulveda, and C. Wang, Sci. Rep. 5, 17883 (2015). 18