pubs.acs.org/NanoLett
Suspended Graphene Sensors with Improved Signal and Reduced Noise Zengguang Cheng, Qiang Li, Zhongjun Li, Qiaoyu Zhou, and Ying Fang* National Center for Nanoscience and Technology, 11 Beiyitiao Street, Zhongguancun, Beijing 100190, People’s Republic of China ABSTRACT We report enhanced performance of suspended graphene field effect transistors (Gra-FETs) as sensors in aqueous solutions. Etching of the silicon oxide (SiO2) substrate underneath graphene was carried out in situ during electrical measurements of devices, which enabled systematic comparison of transport properties for same devices before and after suspension. Significantly, the transconductance of Gra-FETs in the linear operating modes increases 1.5 and 2 times when the power of low-frequency noise concomitantly decreases 12 and 6 times for hole and electron carriers, respectively, after suspension of graphene in solution from the SiO2 substrate. Suspended graphene devices were further demonstrated as direct and real-time pH sensors, and complementary pH sensing with the same nanodevice working as either a p-type or n-type transistor was experimentally realized by offsetting the electrolyte gate potential in solution. Our results highlight the importance to quantify fundamental parameters that define resolution of graphene-based bioelectronics and demonstrate that suspended nanodevices represent attractive platforms for chemical and biological sensors. KEYWORDS Graphene, noise, suspended, pH sensing
G
raphene is attracting tremendous interests due to its nearly perfect crystal quality and superb physical properties,1-5 and emerging applications are being developed based on this novel building block.6-9 In particular, owing to its exceptional carrier mobility and one-atomic thickness, graphene field effect transistors (Gra-FETs) have been proposed to hold great potentials for sensitive and label-free detection of chemical/biological species.9 To date, studies have been focused exclusively on graphene sensors supported on silicon oxide (SiO2) substrates,10-13 although charge traps at the interface and in the oxide have been shown to act as external scattering centers and degrade transport properties in single-layer graphene whose atoms are all exposed directly to fluctuations of extrinsic impurities.14-18 We have undertaken and report herein studies on performance improvement of graphene devices by suspending them in aqueous solution through a novel in situ etching technique. Our results show that, owing to concomitantly increased transconductance and decreased noise level by removal of the oxide, the signal-to-noise ratios of suspended graphene nanodevices were improved by 14 dB in lowfrequency regime (below 1 kHz) for both hole and electron carriers compared with those supported on SiO2 substrates. As an example, suspended devices were demonstrated as real-time and sensitive pH sensors, and complementary detection with either holes or electrons as charge carriers in the same graphene device was achieved, opening up new opportunities for suspended graphene as flexible candidates for bioelectronics.
Figure 1 illustrates the overall design of our experiments. First, Gra-FETs were fabricated from mechanically exfoliated single-layer graphene supported on top of a silicon substrate with 300 nm SiO2.5,19 Briefly, source-drain contacts to graphene were defined by e-beam lithography and subsequently metallization with 5 nm Cr/70 nm Au. Nanodevices in our study were shown to have smooth graphene surface with height of ∼1.0 nm by atomic force microscopy (AFM) (see Supporting Information, Figure S1a). Typical 2-probe mobilities of a 5 µm long Gra-FET gated through 300 nm SiO2 in air were calculated as 4600 and 4100 cm2/V-sec for hole and electron carriers (see Supporting Information, Figure S1b), respectively, which are consistent with reported values for pristine graphene at room temperature.5
FIGURE 1. Graphene FETs in the electrolyte solution. (a) Schematic representation of our experimental setup where a single-layer graphene is supported in solution by Cr/Au contacts to bridge a trench in the oxide. (b) In situ etching of SiO2 underneath graphene. The conductance of graphene starts to drop gradually after buffered HF was added to the PDMS chamber. Single-layer devices usually stabilize within 50 to 100 s, indicating the complete suspension of graphene in solution. Arrows indicate the time when solution was switched in the PDMS chamber. The inset shows a scanning electron microscope (SEM) image taken of a suspended graphene device after solution measurements. Scale bar is 0.5 µm.
* To whom correspondence should be addressed. E-mail:
[email protected]. Received for review: 02/22/2010 Published on Web: 04/07/2010 © 2010 American Chemical Society
1864
DOI: 10.1021/nl100633g | Nano Lett. 2010, 10, 1864–1868
conductance of the device is large at negative electrolyte gate, and it starts to decrease with increased gate voltage due to the depletion of hole carriers in the p-type graphene until reaches its minimum at the Dirac point, then the conductance starts to increase again with the accumulation of electron carriers in the currently n-type graphene. Interestingly, the Dirac point of the graphene device on SiO2 settles at 0.1 V when the chemical potential of the neutral solution (pH 7) is at 0 V. We observed a large range of Dirac point potentials in our Gra-FETs on SiO2, (0.1 ( 0.2) V, which was also reported by similar studies on graphene with naked surfaces.12,13,23 This variation of device properties in GraFETs on SiO2 is apparently disadvantageous in practical applications where reproducible characteristics are critical for devices as, for example, reliable clinical diagnostic sensors.24 After in situ etching of SiO2, the response of the Gra-FET’s conductance to the electrolyte gate potential was characterized again in the neutral electrolyte solution as shown by the red curve in Figure 2. Valuable information can be immediately deduced by comparing transport properties of the same device before and after suspension. (1) The Dirac point of graphene moved to ∼0 V after etching of oxide, which was reproduced by other devices in our study, indicating that the Fermi level of suspended graphene is controlled solely by the solution. (2) The operating principle of electrical sensors is based on their conductance changes responding to local potential changes, so their signal sensitivity can be described as dG/dVgate, that is, the transconductance of devices.10,24 Significantly, the transconductance, or the signal sensitivity, of the graphene device in the linear operating regions increases 1.5 and 2 times for hole and electron carriers, respectively, after suspension. (3) The hole-electron asymmetry was found for both electrolytegate and back-gate geometry of the Gra-FET on SiO2 (black curve in Figure 2 and Supporting Information Figure S3b). Asymmetric hole and electron branches were previously reported in both nonsuspended and suspended Gra-FETs with channel length below 1 µm in vacuum, although the origin of the asymmetry was not understood then.14 Surprisingly, the asymmetry observed in our short-channel devices decreases greatly after suspension of graphene in electrolyte solutions (red curve in Figure 2), indicating effects from dielectric screening and reduced scattering may be responsible for the improved symmetry in solution,25,26 although a detailed study is needed in future to reveal the underlying mechanism. The level of signal that can be processed, or the resolution, of a sensor is ultimately determined by its signal-tonoise ratio. Thus the noise spectra of graphene devices in electrolyte solutions were further examined under our experimental condition, and we will concentrate our discussion on low-frequency (below 1kHz) noise which has been reported to be dominant in limiting performance of nanodevices.15 Figure 3a and Supporting Information Figure S4b
FIGURE 2. Transport characterization of a Gra-FET in solution. Black curve is conductance vs the electrolyte gate voltage of the graphene device supported on SiO2 substrate, and red curve is of the same device after suspension of graphene in solution. The black and red diamonds highlight electrolyte gate potentials at which the comparison of transconductance and noise spectra was performed in the text and Figure 3.
A polydimethylsiloxane (PDMS) chamber was then incorporated over the Gra-FET chip to confine the electrolytic solution of 100 mM potassium chloride and 10 mM phosphate (pH7) (Figure 1a). A nonleak Ag/AgCl reference electrode was placed in the center of the reservoir as the electrolyte gate. The conductance of graphene sensors was monitored through the source and drain gold electrodes using lock-in detection by applying a 30 mV sinusoidal bias voltage (Vsd) with a frequency of 109 Hz. To quantify exclusively effects from the underlying oxide substrate, we developed an in situ etching technique by carefully keeping devices immersed in liquid during switching of solutions and measuring of Gra-FETs’ conductance. We eliminated the drying steps of suspended graphene devices reported by former studies14,16 because the cycle of drying and wetting can easily generate stress to deform graphene and their metal contacts, and alter the electrical properties of devices unintentionally (see Supporting Information, Figure S2). Etching of SiO2 underneath graphene was monitored in real-time by tracking conductance signals through devices (Figure 1b), and we found that incubation in buffered HF results in ca. 100 nm/min etching rate underneath graphene. We note that etching of devices in the following discussion was carried out for 1 min.14,20,21 The electrical measurements of a typical Gra-FET (0.5 µm long and 0.6 µm wide) before and after suspension of graphene in solution were summarized in Figure 2 and Supporting Information Figure S3. The black curve in Figure 2 shows the conductance (dI/dVsd or G) of the device plotted against the electrolyte gate voltage (Vgate) with graphene supported on SiO2 substrate. The transconductance of holes (dG/dVgate) in the nanodevice reaches as high as 1mS/V in the linear operating mode because the electrolyte gate was able to modulate effectively the Fermi level in single-layer graphene through the double layer of ions.22 Note that the © 2010 American Chemical Society
1865
DOI: 10.1021/nl100633g | Nano Lett. 2010, 10, 1864-–1868
FIGURE 3. Noise characterization of graphene device in electrolyte solution. (a) Normalized current power spectral density is shown as a function of frequency of the graphene device on SiO2 at different electrolyte gate potentials. The power spectra were measured at a DC mode with a bias voltage of 30 mV, and sharp peaks (50 Hz and its harmonics) were caused by extrinsic noise from the power source. (b) Comparison of graphene’s noise power spectra in the linear operating modes with holes as carriers before (black, Vgate ) -0.10 V) and after suspension (red, Vgate ) -0.15 V). (c) Comparison of graphene’s noise power spectra with electrons as carriers before (black, Vgate ) 0.20 V) and after suspension (red, Vgate ) 0.15 V).
and SI has a unit of A2/Hz. Thus the power of the current noise, Pnoise, for a given frequency band [ f1, f2] in the lowfrequency region can be calculated as
summarize the current noise power spectral density normalized by the mean square of current, S ) (SI)/(I2), as a function of frequency, f, of the same graphene device in Figure 2 before suspension. The normalized power spectra overlap in the linear operating modes and around the Dirac point of the Gra-FET, indicating that noise density in graphene gated by the electrolyte gate is independent of the gate potential, although gate-dependent noise spectra have been reported in graphene nanobelts gated through SiO2 dielectric.15 Further, the power spectral density of the graphene device was inversely proportional to the frequency, representing the 1/f flicker noise characteristic ubiquitous in nanodevices. Thus the noise spectral density in graphene devices supported on SiO2 can be expressed as27
SI )
Pnoise )
2
1
2
1
f2 AI2 df ) AI2 ln f f1
and Pnoise is with a unit of A2. After suspension of graphene in solution, the normalized power spectral density of the device dropped 12 and 6 times for hole and electron carriers, respectively, compared with that on SiO2 substrate (Figure 3b,c). It has been found that trapped charges at the interface and in the oxide degrade transport characteristics of single-layer graphene.14,17 Former study of graphene devices on SiO2 substrate with low 2-probe carrier mobility in vacuum (∼100 cm2/V-sec) were reported to show ∼20 times improved performance by ion screening of external impurity charges in high ionic solutions.25 Our Gra-FETs, on the other hand, have high mobility (4000 to 5000 cm2/V-sec) on SiO2 in the back-gate geometry
AI2 f
in the low-frequency region for the electrolyte gate geometry in solution, where A ∼ 10-6 is the amplitude of the 1/f noise, © 2010 American Chemical Society
∫f f SI df ) ∫f f
1866
DOI: 10.1021/nl100633g | Nano Lett. 2010, 10, 1864-–1868
in air. The further improved performance in our suspended graphene shows that device degradation associated with oxide substrates is still dominant in our nonsuspended devices, despite any possible ionic screening effect in 100 mM solutions. It is important to note that low ionic strength that offers long Debye length is desirable for detection of charged biomolecules with nano-FETs,24,28 thus the elimination of the oxide by suspension can be increasingly critical to the reduction of noise for single-layer graphene in solutions with low ionic strength. Overall, the signal-to-noise ratio defined as29
SNR(dB) ) 20 log
Asignal Psignal ) 10 log Anoise Pnoise
was increased by 14 dB in low-frequency regime for both hole and electron carriers in graphene devices after suspension. Thus our results explicitly prove that for single-layer graphene FET based sensors, a platform with suspended nanodevices is advantageous to achieve low detection limit in solution. To demonstrate the viability of suspended graphene device as direct chemical sensors, we applied our nanodevices to pH sensing. Solutions from pH 6 to pH 9 were delivered to the surface of a graphene device sequentially, and for each solution, the electrolyte gate response of the device’s conductance was measured as shown in Figure 4a. We found that the Dirac point of graphene shifts positively from pH 6 to pH 9 (see Supporting Information, Figure S5), which is consistent with the increased chemical potential of solution and confirms the capability of suspended graphene for chemical sensing in solution.30-32 Compared to former unipolar silicon nanowire (SiNW) or carbon nanotube (CNT) based p-type sensors,33,34 suspended graphene shows nearly symmetric transport around the Dirac point in neutral solution (Figure 4a), so it can work as either a p-type or n-type sensor when a small gate potential, without introducing any redox reaction of molecules in solution, is applied to switch its carrier characteristic. We investigated the time-dependent response of the Gra-FET’s conductance to different pH solutions when the electrolyte gate is biased at -0.05 and 0.05 V, respectively. The conductance of the nanosensor was monitored in real-time as solutions with pH 6 to pH 9 were sequentially delivered to the surface of the suspended graphene. As shown in Figure 4b, the graphene works as a p-type material when -0.05 V gate potential is applied. Similar to former p-type devices of SiNWs or CNTs,33,35 conductance of the graphene device increases with increased pH values. On the contrary, when the gate potential is switched to +0.05 V, the signal flips its sign and the conductance of the graphene device decreases with increased pH. The polarity flip of signals verifies that the sensing mechanism in graphene nanodevices is due to a field-effect. The unique capability to © 2010 American Chemical Society
FIGURE 4. Graphene as pH sensor. (a) Gra-FET’s conductance as a function of the electrolyte gate voltage in pH 6, 7, 8, and 9 solutions. (b) Real-time detection of solution’s pH by conductance change of the same graphene device with the electrolyte gate potential biased at -0.05 (left branch) and +0.05 V (right branch).
tune easily Gra-FETs between p-type and n-type characteristics presents powerful confirmation on the nature of sensing signals. We further note that the pH detection of our graphene devices is reversible. Thus complementary and reversible measurements in different operating regions of graphene demonstrate the flexibility of our suspended graphene sensors. In summary, quantitative studies on the improved performance of suspended graphene devices in solution were reported. Our results show that, in low-frequency regime, the signal-to-noise ratio of graphene devices increased 14 dB for both hole and electron carriers as a result of concomitantly increased mobility and decreased noise level by suspension in solution. Direct and real-time detection of solution’s pH by suspended graphene was described, and complementary signals from the same graphene device as a p-type or n-type transistor were presented. The enhanced electrical characteristics of suspended graphene devices in aqueous solution show that they are advantageous for chemical and biological detection than their counterparts supported on substrates, and our results are expected to 1867
DOI: 10.1021/nl100633g | Nano Lett. 2010, 10, 1864-–1868
(15) Lin, Y. M.; Avouris, P. Nano Lett. 2008, 8 (8), 2119–2125. (16) Bolotin, K. I.; Sikes, K. J.; Jiang, Z.; Klima, M.; Fudenberg, G.; Hone, J.; Kim, P.; Stormer, H. L. Solid State Comm. 2008, 146 (910), 351–355. (17) Liu, G.; Stillman, W.; Rumyantsev, S.; Shao, Q.; Shur, M.; Balandin, A. A. Appl. Phys. Lett. 2009, 95 (3), 033103. (18) Cho, S.; Fuhrer, M. S. Phys. Rev. B 2008, 77 (8), 084102. (19) Li, Q.; Li, Z. J.; Chen, M. R.; Fang, Y. Nano Lett. 2009, 9 (5), 2129– 2132. (20) Long etching time was found to increase the contact resistance of Gra-FETs due to, for example, the deformation of metal contacts. (21) Akarvardar, K.; Eggimann, C.; Tsamados, D.; Chauhan, Y. S.; Wan, G. C.; Lonescu, A. M.; Howe, R. T.; Wong, H. S. P. IEEE Trans. Electron Devices 2008, 55 (1), 48–59. (22) Rosenblatt, S.; Yaish, Y.; Park, J.; Gore, J.; Sazonova, V.; McEuen, P. L. Nano Lett. 2002, 2 (8), 869–872. (23) Chen, F.; Qing, Q.; Xia, J. L.; Li, J. H.; Tao, N. J. J. Am. Chem. Soc. 2009, 131 (29), 9908–9909. (24) Zheng, G. F.; Patolsky, F.; Cui, Y.; Wang, W. U.; Lieber, C. M. Nat. Biotechnol. 2005, 23 (10), 1294–1301. (25) Chen, F.; Xia, J. L.; Ferry, D. K.; Tao, N. J. Nano Lett. 2009, 9 (7), 2571–2574. (26) Jang, C.; Adam, S.; Chen, J. H.; Williams, D.; Das Sarma, S.; Fuhrer, M. S. Phys. Rev. Lett. 2008, 101 (14), 146805. (27) Lin, Y. M.; Appenzeller, J.; Knoch, J.; Chen, Z. H.; Avouris, P. Nano Lett. 2006, 6, 930–936. (28) Stern, E.; Klemic, J. F.; Routenberg, D. A.; Wyrembak, P. N.; Turner-Evans, D. B.; Hamilton, A. D.; LaVan, D. A.; Fahmy, T. M.; Reed, M. A. Nature 2007, 445 (7127), 519–522. (29) Mitchell, F. H. Introduction to electronics design, 2nd ed.; PrenticeHall, Inc: Upper Saddle River, NJ, 1992. (30) Larrimore, L.; Nad, S.; Zhou, X. J.; Abruna, H.; McEuen, P. L. Nano Lett. 2006, 6 (7), 1329–1333. (31) Eisele, I.; Doll, T.; Burgmair, M. Sens. Actuators, B 2001, 78 (13), 19–25. (32) Oprea, A.; Simon, E.; Fleischer, M.; Frerichs, H. P.; Wilbertz, C.; Lehmann, M.; Weimar, U. Sens. Actuators, B 2005, 111, 582–586. (33) Cui, Y.; Wei, Q. Q.; Park, H. K.; Lieber, C. M. Science 2001, 293 (5533), 1289–1292. (34) Li, C.; Curreli, M.; Lin, H.; Lei, B.; Ishikawa, F. N.; Datar, R.; Cote, R. J.; Thompson, M. E.; Zhou, C. W. J. Am. Chem. Soc. 2005, 127 (36), 12484–12485. (35) Lee, K.; Kwon, J. H.; Moon, S. I.; Cho, W. S.; Ju, B. K.; Lee, Y. H. Mater. Lett. 2007, 61 (14-15), 3201–3203.
bring attentions to suspended nanodevices in general as promising platforms for bioelectronics. Acknowledgment. Y.F. acknowledges support of this work by Special Presidential Foundation of the Chinese Academy of Sciences, China (08172911ZX), “973” Fund (2009CB930200), and the National Natural Science Foundation of China (20973045). Supporting Information Available. Methods and Figures S1-S5. This material is available free of charge via the Internet at http://pubs.acs.org. REFERENCES AND NOTES (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14)
Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, V. V.; Morozov, S. V.; Geim, A. K. Proc. Natl. Acad. Sci. U.S.A. 2005, 102 (30), 10451–10453. Zhang, Y. B.; Tan, Y. W.; Stormer, H. L.; Kim, P. Nature 2005, 438 (7065), 201–204. Geim, A. K.; Novoselov, K. S. Nat. Mater. 2007, 6 (3), 183–191. Novoselov, K. S.; Jiang, Z.; Zhang, Y.; Morozov, S. V.; Stormer, H. L.; Zeitler, U.; Maan, J. C.; Boebinger, G. S.; Kim, P.; Geim, A. K. Science 2007, 315 (5817), 1379–1379. Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Science 2004, 306 (5296), 666–669. Meyer, J. C.; Girit, C. O.; Crommie, M. F.; Zettl, A. Nature 2008, 454 (7202), 319–322. Wang, X.; Zhi, L. J.; Mullen, K. Nano Lett. 2008, 8 (1), 323–327. Liu, Z.; Robinson, J. T.; Sun, X. M.; Dai, H. J. J. Am. Chem. Soc. 2008, 130 (33), 10876. Geim, A. K. Science 2009, 324 (5934), 1530–1534. Schedin, F.; Geim, A. K.; Morozov, S. V.; Hill, E. W.; Blake, P.; Katsnelson, M. I.; Novoselov, K. S. Nat. Mater. 2007, 6 (9), 652– 655. Mohanty, N.; Berry, V. Nano Lett. 2008, 8 (12), 4469–4476. Ohno, Y.; Maehashi, K.; Yamashiro, Y.; Matsumoto, K. Nano Lett. 2009, 9 (9), 3318–3322. Ang, P. K.; Chen, W.; Wee, A. T. S.; Loh, K. P. J. Am. Chem. Soc. 2008, 130 (44), 14392–14393. Du, X.; Skachko, I.; Barker, A.; Andrei, E. Y. Nat. Nanotechnol. 2008, 3 (8), 491–495.
© 2010 American Chemical Society
1868
DOI: 10.1021/nl100633g | Nano Lett. 2010, 10, 1864-–1868
