APPLIED PHYSICS LETTERS VOLUME 83, NUMBER 20 17 NOVEMBER 2003 Patterned growth of single-walled carbon nanotube arrays from a vapor-deposited Fe catalyst H. B. Peng Division of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts 02138 T. G. Ristroph, G. M. Schurmann, G. M. King, and J. Yoona) Department of Physics, Harvard University, Cambridge, Massachusetts 02138 V. Narayanamurtib) and J. A. Golovchenkob),c) Division of Engineering and Applied Sciences, Harvard University, Cambridge, Massachusetts 02138 ͑Received 27 June 2003; accepted 10 September 2003͒ Single-walled carbon nanotubes have been grown on a variety of substrates by chemical vapor deposition using low-coverage vacuum-deposited iron as a catalyst. Ordered arrays of suspended nanotubes ranging from submicron to several micron lengths have been obtained on Si, SiO2 , Al2O3 , and Si3N4 substrates that were patterned on hundred nanometer length scales with a focused ion beam machine. Electric fields applied during nanotube growth allow the control of growth direction. Nanotube circuits have been constructed directly on contacting metal electrodes of Pt/Cr patterned with catalysts. Patterning with solid iron catalyst is compatible with modern semiconductor fabrication strategies and may contribute to the integration of nanotubes in complex device architectures. © 2003 American Institute of Physics. ͓DOI: 10.1063/1.1627935͔ Carbon nanotubes have attracted intense attention in recent years due to their interesting physical properties and potential applications.1,2 The controlled growth of carbon nanotubes is critical in practical applications and a number of recent reports have addressed this issue.3–9 For example, Dai’s group8 obtained suspended single-walled nanotubes ͑SWNTs͒ oriented parallel to the substrate surfaces on specific sites by a methane chemical vapor deposition ͑CVD͒ growth method. This method relies on liquid solutions containing catalyst nanoparticles, which are transferred to patterned substrate surfaces by contact printing. There are also reports of multiwalled carbon nanotube growth from vapor deposited solid catalyst films.4,5,9–11 However, little work has been done on directing the controlled growth of SWNTs on specific sites using solid catalyst films.12 Here, we show that very low coverage of vapor-deposited Fe serves as a reliable catalyst for the growth of SWNTs. By patterning the Fe catalyst, we obtain ordered arrays of suspended SWNTs on predefined locations using a methane CVD growth method. Furthermore, locally applied, ‘‘on chip’’ electric fields13 were utilized to successfully manipulate the growth directions of SWNTs. Finally, device-quality SWNTs were grown directly on catalyzed metal electrodes with good electrical contact. In our experiments, Fe was deposited by thermal evaporation in a vacuum of 10Ϫ6 Torr. The Fe coverage was typically 5ϫ1015 atoms/cm2 ͑corresponding to a few monolayers͒ as determined by Rutherford backscattering. Catalyst regions were patterned into arrays of small posts or parallel stripes of different dimensions by using a focused ion beam ͑FIB͒ machine ͑FEI Micrion 9500, 50 kV͒. Unless otherwise a͒Also at: Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139. b͒Also at: Department of Physics, Harvard University, Cambridge, MA 02138. c͒Electronic mail: golovchenko@physics.harvard.edu specified in the letter, nanotubes were then grown in a quartz tube oven at a temperature of 900 °C in the flow of methane ͑1000 sccm͒ for 4 –20 min.8 Heating up and cooling down were done in an argon flow. High-resolution scanning electron microscopy ͑SEM͒ and transmission electron microscopy ͑TEM͒ were used to characterize the carbon nanotubes. For TEM imaging, multiple gaps or holes were ion beam etched all the way through self-supporting 0.5-␮m-thick Si3N4 membranes.14 The substrate then formed a natural electron microscope grid upon which nanotubes could be grown and studied by TEM directly. Figures 1͑a͒–1͑d͒ show the growth behavior of nano- tubes on patterned Si3N4 membranes as observed with a SEM operated at 1 keV. Using such a low energy SEM al- lows direct observation of suspended nanotubes. The width of the Fe coated Si3N4 regions on different patterns ranged from 200 to 500 nm. These regions clearly serve as the ini- tiation and termination sites in the nanotube growth process. Figure 1͑a͒ shows the nanotubes grown across 500 nm wide gaps. The growth time was 10 min in the methane flow. Most crossing nanotubes were grown in directions approxi- mately perpendicular to the long gap direction. As the gap width is increased to 700 nm, 1.3, and 3 ␮m ͓Figs. 1͑b͒, 1͑c͒, 1͑d͒, respectively͔, the nanotube density decreases and the tubes that succeed in traversing the gap are more nearly aligned perpendicular to it. A likely explanation is that the Van der Waals’ interaction between the thermally vibrating nanotubes13 and the substrate successfully returns most nanotubes to the substrate segment they start from, unless their initial growth direction is nearly perpendicular to the gap, or a tube–tube collision results in growth in that direc- tion. Long suspended nanotubes can be easily obtained even for gaps exceeding 5 ␮m. Near the edges, two or more tubes usually join together to form a long tube across the gaps. This demonstrates that interactions between the floating 0003-6951/2003/83(20)/4238/3/$20.00 4238 © 2003 American Institute of Physics Downloaded 09 Dec 2005 to 128.103.60.225. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp Appl. Phys. Lett., Vol. 83, No. 20, 17 November 2003 Peng et al. 4239 FIG. 2. SEM images of SWNTs connecting stripes or posts patterned on ͑a͒ a SiO2 ͑5000 Å͒/Si substrate ͑grains due to etching are seen in the bottom of the trenches. Scale bar: 500 nm͒ and ͑b͒ an Al2O3 ͑1000 Å͒/Si substrate ͑Scale bar: 500 nm͒. ͑c͒ SEM images of SWNTs across gaps in Si3N4 membranes with 4 min growth time, under applied electric field perpendicu- lar to the gaps ͑0.5 V/␮m͒ and ͑d͒ at a 45° angle relative to the gaps ͑2.4 V/␮m͒ ͑Scale bar: 1 ␮m͒. FIG. 1. SEM ͑a͒–͑d͒ and TEM ͑e͒, ͑f͒ images of nanotubes across parallel gaps etched through free-standing Si3N4 membranes. ͑a͒ 500 nm wide gaps with remaining stripes 300 nm wide. Scale bar: 1 ␮m. ͑b͒ Gaps 700 nm wide. Scale bar: 500 nm. ͑c͒ Gaps 1.3 ␮m wide. Scale bar: 500 nm. ͑d͒ Gaps 3 ␮m wide. Scale bar: 1 ␮m. ͑e͒ TEM image of a suspended individual SWNT. Scale bar: 10 nm. ͑f͒ From bottom left to top right: the split of a nanotube bundle into two individual SWNTs. From top left to bottom right: an out-of-focus image of another nanotube. Scale bar: 5 nm. tubes are connecting arrays of post on Al2O3 coated Si substrate. So far, the directional alignment of nanotubes in our experiment is mainly determined by nanotube-substrate and nanotube–nanotube interactions and thermal vibrations, which make most of the nanotubes orient themselves roughly in a direction perpendicular to the gaps or trenches. To get nanotubes during the growth process are important in forming nanotube structures. TEM ͑200 kV͒ images were taken of suspended nano- better control in the growth of nanotubes we explored the utility of applying electric fields to align the nanotubes during the growth process.13,19 Here we show that locally ap- tubes grown across gaps on catalyst-coated Si3N4 membranes. In Fig. 1͑e͒, the wall structure of an individual SWNT is clearly seen. Figure 1͑f͒ shows the split of a two- tube bundle into two individual SWNTs 2 nm in diameter. plied electric fields are very effective in manipulating the growth direction of SWNTs from our Fe catalyzed surfaces. In our experimental setup, two 20 ␮m wide electrodes ͑Pt 40 nm/Ti 50 nm͒ spaced 10 ␮m apart end on, were patterned by Most of the nanotubes observed in our samples, even very optical lithography on self supported insulating Si3N4 memlong nanotubes, are isolated individual SWNTs, or bundles branes. The Fe catalyst was deposited in the space between made of only a few SWNTs. The diameter of most SWNTs the electrodes, and then that space was patterned into parallel ranges from 1 to 4 nm from TEM studies. It has been sug- gaps with the FIB. During the CVD growth process, a bias gested that nanometer-sized catalyst particles are required to voltage was applied between the electrodes to generate an seed nanotube growth.15–18 Gaseous carbon compounds pre- electric field in the gap areas. Figure 2͑c͒ shows the result for sumably decompose on the surface of catalytic metal drop- a sample under a bias voltage of 5 V and 4 min growth time. lets, and carbon nanotube growth occurs by diffusion-driven The electric field direction was perpendicular to the gaps. precipitation of carbon from the supersaturated catalyst Compared with the results without electric fields during nanoparticles. For low coverage Fe depositions, nanoscale growth, the nanotubes look straighter, and are more aligned droplets nucleate at the surface during the high temperature with the applied field. It has been previously suggested that treatment.10,11 From atomic force microscopy images taken the interaction between the external electric field and the on Fe deposited surfaces before and after annealing at induced dipole moment of a SWNT is the driving force for 900 °C with argon flow, we clearly observe the aggregation the alignment.13 To further demonstrate the effect of electric of nanoparticles ͑ranging from nanometers to hundreds of fields, we also prepared samples with gaps at a 45° angle nanometers͒ due to the heating process. Also, at both ends of relative to the applied field direction. As illustrated in Fig. the nanotubes observed by SEM, the size of nucleation sites 2͑d͒, most of the SWNTs are now aligned to the new electric is of order or less than the resolution of the SEM ͑ϳ10 nm͒. field direction, although there are still some SWNTs in the Therefore, they possibly grow from 10 nm or smaller Fe perpendicular direction, which is due to the competing driv- particles. ing forces as described previously ͑Van der Waals interac- We also studied the growth behavior of nanotubes on tions and thermal vibrations͒. It should be clear from these different substrates commonly encountered in semiconductor results that orientational growth can be programmed to ob- device fabrication, such as Si͑111͒, Si͑100͒, SiO2 , and tain structures that contain crossed nanotubes, which could Al2O3 . Figure 2͑a͒ shows nanotubes crossing trenches 700 be very important as device structures. nm wide on a Si substrate coated with insulating SiO2 layer Finally, we show that the growth strategy can be applied for a growth time of 20 min. In Fig. 2͑b͒, suspended nano- to obtain SWNTs directly on metal electrodes with good Downloaded 09 Dec 2005 to 128.103.60.225. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp 4240 Appl. Phys. Lett., Vol. 83, No. 20, 17 November 2003 Peng et al. FIG. 3. ͑a͒ A suspended SWNT connecting 2 ␮m wide metal electrodes ͑Pt 50 nm/Cr 50 nm͒, across a gap etched all the way through the Si3N4 membrane. Scale bar: 2 ␮m. ͑b͒ I vs V curves for a sample with suspended semiconducting SWNTs at temperatures of Tϭ293 K ͑triangles͒ and T ϭ77 K ͑squares͒. The current (I) values for Tϭ77 K are multiplied by a factor of 100. ͑c͒ I – V curve for a sample with metallic SWNTs at room temperature. electrical contact. Two 2 ␮m wide electrodes ͑Pt 50 nm/Cr 50 nm͒ with Fe catalyst on top, spaced apart end on with a ϳ2 ␮m gap between them, were lithographically patterned in the center of Si3N4 membrane. A FIB was used to mill the gaps between the electrodes all the way through the Si3N4 membranes. CVD nanotube growth was carried out from 2 to 5 min in a methane flow of 200 sccm at 900 °C, which usually yields only one or a few suspended SWNTs across the gaps between electrodes. Figure 3͑a͒ shows a typical sample with a SWNT bridging an electrode gap. With this growth condition, amorphous carbon deposition on the substrate makes no contribution to the measured nanotube electrical properties. All the electrical measurement were made on the as-grown samples, without any postgrowth contacting process. Figure 3͑b͒ shows the I – V curves for a semiconducting SWNT. At room temperature, the low-bias differential resistance is ϳ100 M⍀, and band gap-related nonlinearity is already visible. An asymmetry appears which is most likely due to contact difference at the two ends of the tube. When the temperature is reduced to 77 K, a gap is clearly observed in the I – V curve. Figure 3͑c͒ shows the room tem- perature I – V curve of a sample with metallic SWNTs. The linear resistance for this metallic sample is ϳ10 k⍀. In summary, ordered SWNT arrays were grown by ther- mal CVD from patterned vapor deposited Fe catalysts on different substrates. External electric fields were shown to be effective in manipulating the orientation of SWNTs during the growth process despite competing tendencies. SWNT cir- cuits with electrical contacts were constructed using the method, which may be useful in practical electronic device fabrication strategies. The controlled growth, and incorpora- tion, of nanotubes in different geometries is important for applications in electronic, mechanical, and chemical devices.2 For example, nanotubes across gaps in the free- standing membranes provide a good device geometry for chemical sensors. Crossed nanotubes bridged by appropriate molecules may provide true single molecule negative differential resistance device capabilities.14 We note that our growth strategy is highly compatible with, and adaptable to, standard semiconductor lithography processing techniques. It may therefore be important for the adoption of carbon nano- tubes in future large scale practical applications which will most likely require integration with silicon based technology. The authors thank Dr. Warren MoberlyChan for the help with TEM and Dr. JuanCarlos Cheang for the help with Ru- therford backscattering spectroscopy. This work is supported by the DARPA/AFOSR and the NSF. 1 S. Iijima, Nature ͑London͒ 354, 56 ͑1991͒. 2 Carbon Nanotubes: Synthesis, Structure, Properties, and Applications, edited by M. S. Dresselhaus, G. Dresselhaus, and P. Avouris ͑Springer, Berlin, 2001͒. 3 W. Z. Li, S. S. Xie, L. X. Qian, B. H. Chang, B. S. Zou, W. Y. Zhuo, R. A. Zhao, and G. Wang, Science 274, 1701 ͑1996͒. 4 Z. F. Ren, Z. P. Huang, J. W. Xu, J. H. Wang, P. Bush, M. P. Siegal, and P. N. Provencio, Science 282, 1105 ͑1998͒. 5 S. Fan, M. G. Chapline, N. R. Franklin, T. W. Tombler, A. M. Cassell, and H. Dai, Science 283, 512 ͑1999͒. 6 J. Kong, H. T. Soh, A. M. Cassell, C. F. Quate, and H. J. Dai, Nature ͑London͒ 395, 878 ͑1998͒. 7 J. Li, C. Papadopoulos, J. M. Xu, and M. Moskovits, Appl. Phys. Lett. 75, 367 ͑1999͒. 8 A. M. Cassell, N. R. Franklin, T. W. Tombler, E. M. Chan, J. Han, and H. Dai, J. Am. Chem. Soc. 121, 7975 ͑1999͒. After finishing this manuscript, we learned of similar results about growth of nanotubes on electrodes using catalysts from solutions. N. R. Franklin, Q. Wang, T. W. Tombler, A. Javey, M. Shim, and H. Dai, Appl. Phys. Lett. 81, 913 ͑2002͒. 9 Z. F. Ren, Z. P. Huang, D. Z. Wang, J. G. Wen, J. W. Xu, J. H. Wang, L. E. Calvet, J. Chen, J. F. Klemic, and M. A. Reed, Appl. Phys. Lett. 75, 1086 ͑1999͒. 10 Y. Y. Wei, G. Eres, V. I. Merkulov, and D. H. Lowndes, Appl. Phys. Lett. 78, 1394 ͑2001͒. 11 J. I. Sohn, C. Choi, S. Lee, and T. Seong, Appl. Phys. Lett. 78, 3130 ͑2001͒. 12 Note: An early version of this letter was finished in March 2002. Recently, there was a work on using nickel–iron thin film catalysts to grow small diameter nanotubes with ethylene gas. P. M. Campbell, E. S. Snow, and J. P. Novak, Appl. Phys. Lett. 81, 4586 ͑2002͒. 13 Y. Zhang, A. Chang, J. Cao, Q. Wang, W. Kim, Y. Li, N. Morris, E. Yenilmez, J. Kong, and H. Dai, Appl. Phys. Lett. 79, 3155 ͑2001͒. 14 J. Chen, M. A. Reed, A. M. Rawlett, and J. M. Tour, Science 286, 1550 ͑1999͒. 15 N. M. Rodriguez, J. Mater. Res. 80, 3233 ͑1993͒. 16 H. Kanzow and A. Ding, Phys. Rev. B 60, 11180 ͑1999͒. 17 R. T. K. Baker, Carbon 27, 315 ͑1989͒. 18 Y. Li, W. Kim, Y. Zhuang, M. Rolandi, D. Wang, and H. Dai, J. Phys. Chem. B 105, 11424 ͑2001͒. 19 Y. Avigal and R. Kalish, Appl. Phys. Lett. 78, 2291 ͑2001͒. Downloaded 09 Dec 2005 to 128.103.60.225. Redistribution subject to AIP license or copyright, see http://apl.aip.org/apl/copyright.jsp