Microfluidic Devices for

Md. Azahar Ali , Saurabh Srivastava , Pratima R. Solanki , Venu Reddy , Ved V. Agrawal , CheolGi Kim , Renu John , Bansi D. Malhotra. Scientific Repor...
29 downloads 0 Views 322KB Size
Download PDF
13406

2005, 109, 13406-13408 Published on Web 06/28/2005

Integrated Single-Walled Carbon Nanotube/Microfluidic Devices for the Study of the Sensing Mechanism of Nanotube Sensors Qiang Fu and Jie Liu* Gross Chemistry Laboratory, Duke UniVersity, Durham, North Carolina 27708 ReceiVed: May 16, 2005; In Final Form: June 13, 2005

A method to fabricate integrated single-walled carbon nanotube/microfluidic devices was developed. This simple process could be used to directly prepare nanotube thin film transistors within the microfluidic channel and to register SWNT devices with the microfludic channel without the need of an additional alignment step. The microfluidic device was designed to have several inlets that deliver multiple liquid flows to a single main channel. The location and width of each flow in the main channel could be controlled by the relative flow rates. This capability enabled us to study the effect of the location and the coverage area of the liquid flow that contained charged molecules on the conduction of the nanotube devices, providing important information on the sensing mechanism of carbon nanotube sensors. The results showed that in a sensor based on a nanotube thin film field effect transistor, the sensing signal came from target molecules absorbed on or around the nanotubes. The effect from adsorption on metal electrodes was weak.

A single-walled carbon nanotube (SWNT) is a molecule wire formed by rolling a single graphene sheet into a seamless cylinder. A unique property of SWNTs is that all the carbon atoms of a SWNT are on the surface and exposed to the environment. The conduction of current through a semiconducting SWNT is greatly affected by the environment, a feature that has been exploited to develop chemical sensors both in the gas phase and in physiological solution.1-4 For instance, the binding of proteins to SWNTs or to receptors immobilized on SWNTs can be detected by monitoring the conductance of nanotube field effect transistors (FETs) prepared by connecting two metal electrodes with a nanotube.2-4 The combination of nanotube FETs with microfluidic channels offers unique advantages for sensor applications in a liquid. The microfluidic channels provide a convenient and controlled way to deliver liquid to the sensor system.5,6 To fabricate such integrated devices, the nanotube FET needs to be registered with the microfluidic channel in order to precisely deliver liquid to the FET. However, registration of the microfluidic channel with the nanotube FET is difficult, which currently relies upon careful alignments under an optical microscope.7 Here, we describe a simple process to directly fabricate SWNT FETs inside a microfluidic channel via the method of Kenis and co-workers.8 No registration between the nanotube FET and the microfluidic channel is required and the integrated device can be used as a chemical sensor without being disassembled. More importantly, this device enables the manipulation of the width and location of the liquid flow within the channel, providing important information on how molecules affect the conductance of a nanotube FET. The devices were fabricated on a dense SWNT network that was grown on a SiO2 substrate by chemical vapor deposition.9 * Corresponding author. E-mail: [email protected].

10.1021/jp0525686 CCC: $30.25

First, metal electrode patterns of 2 nm of chromium and 40 nm of gold were fabricated on top of the nanotube network using lithographic methods. The pattern is a continuous thin metal line connected to two connecting pads (Figure 1b). To eliminate the excessive conduction channel, SWNTs not covered by the pattern were removed with O2 plasma at 200 mTorr for 5 min (PDC-32G, Harrick Plasma), during which the electrode pattern acts as a mask to protect nanotubes under the gold pattern. Then, a PDMS membrane embossed with a microfluidic channel was put on top of the substrate to form a sealed channel. The electrodes form a right angle with the channel (Figure 1b). The microfluidic channel has a three-flow inlet; an etching solution of 2 mM iron nitrate and 3 mM thiourea flowed through the center inlet, and DI water flows through the two side inlets (Figure 1a). The chemical etching formed a gap on the electrode pattern (Figure 1c). In the gap, SWNT networks originally underneath the gold film acted as a conduction channel and electrically connected the separated electrodes in the channel (Figure 1d). The integrated devices fabricated this way can be directly used as nanotube sensors by making electrical connections to the two separated electrodes and introducing liquid to the system using the microfluidic channels. One unique advantage of using an integrated nanotube/microfluidic device for sensing is that the location and width of the liquid stream containing the molecules of interest can be controlled within the channel when multiple liquid streams are introduced into the system. Because of the small dimensions of the channel, the liquid streams in the channel are laminar and can maintain their shape without developing turbulence.8 When three liquid streams are introduced into the main channel, the width of each stream depends on the relative flow rates of the three streams. Hence, we can easily control the width and location of the center flow, delivering the etching solution or the liquid flow containing the molecule of interest in our experiments. © 2005 American Chemical Society

Letters

J. Phys. Chem. B, Vol. 109, No. 28, 2005 13407

Figure 1. (a) Schematic illustration of a three-stream flow in a microfluidic channel. (b) Optical graph of SWNT devices in the channel prior to etching a gap on the Au line (scale bar, 400 µm). (c) Optical graph of SWNT devices in the channel after removing sections of Au through chemical etching. (d) Scanning electron microscope image of the SWNT network in the black square in part c.

Although SWNTs have been used as sensors for biological systems, the sensing mechanism is still unclear. For example, it was reported that the adsorption of proteins on metal electrodes could be the source of the sensing signal.4 In this reported experiment, it was shown that when the electrodes were protected from protein adsorptions, the sensing signal disappeared. There is a need to understand how the biological binding events can be translated into detectable signals and where these events take place. However, most of the previous investigations were performed without the control of the exact location of the liquid, making it difficult to extract such information. Previously, we have shown that the SWNT network FETs behave as typical p-type semiconductor FETs. The presence of cationic surfactants such as cetyltrimethylammonium bromide (CTAB) can reduce the drain-source current (IDS) value of the devices by neutralizing the negative charges on the SiO2 surface.10 However, previous experiments cannot rule out the effect of molecules absorbed on the electrodes. The devices fabricated within microfluidic channels have similar performances, but the liquid inside a microfluidic channel can be conveniently manipulated (Supporting Information). This capability enables us to study how the conductance of SWNT devices is affected by the location of adsorbed molecules, essential to develop SWNT sensors. In this study, we use the model system in our earlier studies, in which the adsorption of CTAB reduces the IDS value of the device. The device contains an etched gap of 130 µm between the two metal electrodes and an overall width of the microfluidic channel of 400 µm (Figure 2). A bias voltage of 200 mV was applied in all of our experiments. Under such experimental conditions, the conduction through the bulk liquid can be negligible.10 CTAB solution from the center inlet and water from

the side inlets were introduced. By altering the relative flow rate, the width and location of the CTAB flow within the channel can be controlled precisely. A significant IDS decrease occurs when the CTAB solution fills the whole channel and contacts both electrodes and SWNTs, similar to previous experiments.10 No obvious change of IDS is observed when the stream of CTAB solution only flows over the metal electrodes, suggesting that the decrease of IDS is induced by CTAB molecules adsorbed around SWNTs. IDS decreases significantly when CTAB solution flows over the SWNTs, and the decrease of IDS is strongly affected by the width of the CTAB stream (Figure 2, state I). When the CTAB stream width increases past the SWNT network and over the gold electrodes, IDS becomes a constant value and is independent of the stream width (Figure 2, state II). Such a relationship between IDS and the width of the CTAB stream suggests that CTAB affects IDS only by adsorbing on or around the SWNTs in the devices. The relationship between the CTAB stream width and IDS can be calculated using a simple model considering only the resistances of SWNTs covered by CTAB solution and by pure water in serial connection. The overall IDS value can be written as

IDS ) VDS/(RC(VC/VT) + RU(1 - VC/VT))

(1)

Here, VDS is the applied bias, RC is the resistance when the CTAB stream covers the whole area of the SWNTs in the microfluidic channel, RU is the resistance when CTAB is absent in the channel, VC is the flow rate of CTAB solution with different flow widths, and VT is the CTAB flow rate required to just cover the entire SWNT network, which can be measured experimentally. The calculated values of IDS are close to the

13408 J. Phys. Chem. B, Vol. 109, No. 28, 2005

Letters

Figure 2. (right) Drain-source current (IDS) versus the width of a 1 mM cetyltrimethylammonium bromide (CTAB) stream (channel width between drain and source, 130 µm; bias, 200 mV; Vg ) 0 V). The calculated values are calculated using eq 1 in the text. (left) Schematic illustration showing two different states in which CTAB solutions contact only SWNTs (state I) or both SWNT and electrodes (state II).

values measured experimentally (Figure 2), further verifying that the change of IDS is induced by the CTAB absorbed around SWNTs. Such an effect is different from the observed effect of protein absorptions on SWNT devices, in which proteins adsorbed on the electrodes are responsible for the change of conductance.4 We believe that such a discrepancy may be caused by the difference in length of the conduction channel and the relative significance of the contact resistance versus nanotube resistance. The conduction channel in our devices (130 µm) is wider than that used in the study of protein adsorptions (5 µm). This means that the resistance in SWNT networks determines the overall device conductance instead of the resistance from SWNT-electrode contacts. A recent study also demonstrated that the contact resistance is small compared to the total resistance in nanotube thin film FETs.11 In summary, we have developed a simple process to integrate SWNT devices with microfluidic devices without the need for optical alignment and registration of the two different devices. This integration makes it possible to control the width and location of the liquid stream with the molecule of interest. The conductance of the devices is sensitive to the adsorption of cationic surfactants in the microfluidic channel. Simultaneous conductance measurements show that the conductance of devices changes only when the charged molecules contact SWNTs. Molecules absorbed on electrodes contribute little to the conductance change, suggesting that sensors with selectivity can be constructed by modifying the SWNT area in the channel. Such an integrated system of SWNTs and a microfluidic channel can be used to study the effects of other molecules on SWNT devices, such as proteins or DNA. In biological/chemical detection, such devices can be integrated into a microanalysis system, combining the high sensitivity of SWNT sensors and the versatility of microfluidic devices in liquid handling.

Acknowledgment. The project was supported in part by NSF NIRT Program (NSF CCR-0326157) and a 2002 young professor award from DuPont to J.L. The authors are indebted to Dr. Bo Zheng at University of Chicago for advices on microfluidic devices and Soojin Oh and Michael Staderman at University of North Carolina at Chapel Hill for help in photolithography. Supporting Information Available: A detailed experimental description, more data on the time dependent monitoring of sensing signal, and a movie showing the laminar liquid flow within the microfludic channel as a function of the relative flow rates of the flows. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) Kong, J.; Franklin, N. R.; Zhou, C. W.; Chapline, M. G.; Peng, S.; Cho, K. J.; Dai, H. J. Science 2000, 287, 622-625. (2) Besteman, K.; Lee, J. O.; Wiertz, F. G. M.; Heering, H. A.; Dekker, C. Nano Lett. 2003, 3, 727-730. (3) Chen, R. J.; Bangsaruntip, S.; Drouvalakis, K. A.; Kam, N. W. S.; Shim, M.; Li, Y. M.; Kim, W.; Utz, P. J.; Dai, H. J. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 4984-4989. (4) Chen, R. J.; Choi, H. C.; Bangsaruntip, S.; Yenilmez, E.; Tang, X. W.; Wang, Q.; Chang, Y. L.; Dai, H. J. J. Am. Chem. Soc. 2004, 126, 1563-1568. (5) Cui, Y.; Wei, Q. Q.; Park, H. K.; Lieber, C. M. Science 2001, 293, 1289-1292. (6) Duffy, D. C.; Schueller, O. J. A.; Brittain, S. T.; Whitesides, G. M. J. Micromech. Microeng. 1999, 9, 211-217. (7) Wu, H.; Brittain, S.; Anderson, J.; Grzybowski, B.; Whitesides, S.; Whitesides, G. M. J. Am. Chem. Soc. 2000, 122, 12691-12699. (8) Kenis, P. J. A.; Ismagilov, R. F.; Whitesides, G. M. Science 1999, 285, 83-85. (9) Fu, Q.; Huang, S. M.; Liu, J. J. Phys. Chem. B 2004, 108, 61246129. (10) Fu, Q.; Liu, J. Langmuir 2005, 21, 1162-1165. (11) Zhou, Y. X.; Gaur, A.; Hur, S. H.; Kocabas, C.; Meitl, M. A.; Shim, M.; Rogers, J. A. Nano Lett. 2004, 4, 2031-2035.