- Email: [email protected]
Growth and electrical characterization of high-aspect-ratio carbon nanotube arrays
Carbon 44 (2006) 253–258 www.elsevier.com/locate/carbon
Growth and electrical characterization of high-aspect-ratio carbon nanotube arrays Lingbo Zhu a, Jianwen Xu b, Yonghao Xiu a, Yangyang Sun b, Dennis W. Hess a, C.P. Wong b,* a
School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, 311 Ferst Drive, Atlanta 30332, GA, USA b School of Materials Science and Engineering, Georgia Institute of Technology, 711 Ferst Drive, Atlanta 30332, GA, USA Received 30 April 2005; accepted 23 July 2005 Available online 13 September 2005
Abstract The remarkable properties of carbon nanotubes (CNTs) make them attractive for microelectronic applications, especially for interconnects and nanoscale devices. In this paper, we describe a microelectronics compatible process for growing high-aspect-ratio CNT arrays with application to vertical electrical interconnects. A lift-off process was used to pattern catalyst (Al2O3/Fe) islands to diameters of 13 or 20 lm. After patterning, chemical vapor deposition (CVD) was involved to deposit highly aligned CNT arrays using ethylene as the carbon source, and argon and hydrogen as carrier gases. The as-grow CNTs were characterized by transmission electron microscopy (TEM), scanning electron microscopy (SEM) and X-ray photoelectron spectroscopy (XPS). The results demonstrated that the CNTs have high purity, and form densely-aligned arrays with controllable array size and height. Two-probe electrical measurements of the CNT arrays indicate a resistivity of 0.01 X cm, suggesting possible use of these CNTs as interconnect materials. 2005 Elsevier Ltd. All rights reserved. Keywords: Carbon nanotubes; Chemical vapor deposition; Electron microscopy
1. Introduction Carbon nanotubes (CNTs) display a well-defined quasi one dimensional cylindrical structure, formed by rolling up graphene sheets of sp2 bonded carbon atoms [1]. The CNTs can be either metallic or semiconducting, depending on how the graphite layer is wrapped into a cylinder [2,3]. Metallic CNTs show ballistic conductivity at room temperature [4]. The ballistic conductivity, high thermal conductivity [5] and mechanical strength of CNTs make them ideal candidates for electrical interconnects in IC packaging and nanoscale devices.
*
Corresponding author. Tel.: +1 404 894 8391; fax: +1 404 894 9140. E-mail address: [email protected] (C.P. Wong). 0008-6223/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2005.07.037
The electronic properties of perfect multi-walled carbon nanotubes (MWCNTs) are similar to those of single-walled carbon nanotubes (SWCNTs) [6], due to weak coupling between the graphite cylinders (layer distance is P0.34 nm). Electrons transport ballistically (without scattering) in metallic SWCNTs and MWCNTs over reasonable lengths (1 lm), thereby enabling CNTs to carry very high currents (>109 A/cm2) [7] without electromigration failure. Phonons also propagate easily along the nanotubes [6]. The measured thermal conductivity of an individual MWCNT at room temperature is >3000 W/m K [5], which exceeds the conductivity of diamond (2000 W/m K). Based on these advantageous properties of CNTs, researchers have reported the integration of CNTs into electrical interconnect applications [8–11]. However, the resistance of a single ballistic SWCNT less than 1 lm long is about
254
L. Zhu et al. / Carbon 44 (2006) 253–258
6.5 kX with perfect contacts [12], while ballistic transport in MWCNTs with a resistance of 12.9 kX was also reported [13]. The high resistance of an individual CNT indicates that an array of thousands of parallel CNTs will be necessary for interconnect applications. Although there are many reports about the growth of the aligned CNT arrays with controlled areas and nanotube length [14–16], deliberate production of high-aspect-ratio nanotube structures was seldom reported. In this study, we report a simple, but efficient CVD method for the growth of aligned carbon nanotube arrays from a small area (<20 lm in diameter) but extending to a very long length (several hundreds of lm) without any bending or curling. These high-aspect-ratio nanotube arrays are particularly useful as electrical interconnects or mechanical strengthening backbones.
2. Experimental
(DI) water rinses. The patterned substrate was placed in the CVD chamber for highly-oriented CNT pillar growth. 2.2. CNT characterization Samples for transmission electron microscopy were prepared by sonicating a trace amount of synthesized carbon nanotubes in methanol for 30 min. A few drops of the suspension were placed onto a TEM grid. TEM (JEOL 400EX) analysis was performed at 400 kV. Scanning electron microscopy (SEM) was carried out on a JEOL 1530 equipped with a thermally assisted field emission gun operating at 10 keV. Spatially resolved energy-dispersive X-ray spectroscopy (EDS) was used to assess the purity of carbon nanotubes. X-ray photoelectron spectroscopy (XPS) was performed using a physical electronics (PHI) model 1600 XPS system equipped with a monochromator. The system used an Al Ka source (hm = 1486.8 eV).
2.1. CNT array growth Chemical vapor deposition (CVD) was performed in a horizontal alumina tube (3.8 cm diameter; 92 cm long) housed in a Lindberg Blue furnace. The substrates used in this study were (0 0 1) silicon wafers coated with SiO2 (400 nm) by thermal oxidation or plasma-enhanced CVD. The catalyst layers of Al2O3 (15 nm)/Fe (2 nm) were formed on the silicon wafer by a lift-off process. CVD growth of CNTs was carried out at 700–800 C with ethylene as the carbon source, and hydrogen and argon as carrier gases. A schematic of the process is shown in Fig. 1 and described briefly below. A thin layer of photoresist (Shipley 1813) was spincoated on a silicon substrate and subsequently exposed to UV light and the patterns developed. The catalyst layers of Al2O3/Fe were then deposited onto the substrate by a sequential electron-beam evaporation process. The photoresist was removed by acetone and the wafer cleaned using acetone, isopropyl alcohol and de-ionized
2.3. Resistance and capacitance measurements of CNT arrays Two-probe current–voltage measurements of the CNT arrays were performed using an Agilent 4285 A Precision LCR Meter. Capacitance–voltage (CV) responses were measured by an HP semiconductor analyzer at high frequency of 100 kHz. Before sample measurement, the analyzer was calibrated by touching probes on the two electrodes without CNT arrays on them to exclude the contribution of capacitance from the substrate. The nanotube array solution was placed on a substrate that had pre-patterned Au electrodes. Randomly, the two ends of some CNT arrays contacted electrodes. These samples were mounted on an electrical probe station, and two-terminal resistance measurements were performed by contacting the probes (12 lm tungsten needles) to these electrodes.
3. Results and discussion 3.1. CNT array growth
Fig. 1. Schematic of the CNT pillar growth.
Aligned CNTs grow readily from the catalyst patterns into well-defined vertical structures, as shown in Fig. 2, which shows SEM micrographs of CNT arrays with different pattern size, pitch and height. The synthesis conditions for Fig. 2a and b are for 3 min at 750 C and 700 C, respectively. The images demonstrate that nearly vertically-aligned CNT pillars are formed on the substrate with minimal entanglement with neighboring CNTs, but the arrays bend at the tips due to the gravity if the CNTs are too long. In Fig. 2a, the height of the CNT arrays reaches 420 lm with an aspect ratio
L. Zhu et al. / Carbon 44 (2006) 253–258
255
Fig. 2. SEM micrographs of highly-aligned CNT arrays. (a) SEM image of nanotube arrays synthesized on 12 lm by 12 lm catalyst patterns; (b) SEM image of nanotube arrays synthesized on 20 lm by 20 lm catalyst patterns; (c) SEM image of nanotube arrays synthesized on cylindrical patterns with diameter of 20 lm; (d) side view of the nanotube arrays in (c); (e) enlarged view of nanotube tip of the arrays in (c); (f) enlarged view of a nanotube root of the arrays in (d).
of 32, while in Fig. 2b the height of the CNT arrays reaches 300 lm with an aspect ratio of 15. Furthermore, the vertical alignment of pillars can be influenced by the array aspect ratio. With a pillar height <150 lm, no pillar bending was observed. In fact, further studies have demonstrated the CNT pillars maintain vertical growth up to an aspect ratio of 15. Fig. 2e shows a high magnification SEM image of a nanotube array tip with a diameter of 20 lm. Fig. 2f shows an enlarged image of the base of a free-standing nanotube array which indicates the well-defined edges of these CNT arrays. These
results show that the height and aspect ratios of CNT pillars can satisfy most interconnect applications. There are two levels of hierarchical alignment structure that exist in this CVD process. The primary alignment is a bundle of densely-packed CNTs to form the CNT pillar (Fig. 3a). The secondary alignment is the array of CNT pillars on the substrate, where the growth of CNTs is confined within the catalyst patterns. From the microscopic to the macroscopic level, the alignment of CNTs make possible novel interconnect structures that can be produced by methods compatible
256
L. Zhu et al. / Carbon 44 (2006) 253–258
Fig. 3. (a) SEM image of CNT arrays; (b) HRTEM image of CNT bundles.
with existing IC fabrication sequences. In Fig. 3a no carbon or other particles are observed on CNT walls, and the high density of CNT bundles is evident. We estimate that the CNT density could reach 1500 lm 2. EDS examination along the CNT bundles detected no elements except carbon, and HRTEM examination revealed no particles attached on the CNTs walls. These results indicate the high purity of the as-synthesized nanotubes. From HRTEM images, we also know that most of the CNTs synthesized by this CVD method are double-walled or triple-walled with an average diameter of 10 nm. Fig. 3a shows a double-walled CNT. 3.2. CNT growth mode As shown in Fig. 4, the thin Fe film of 2 nm segregated into small islands, after a heat treatment at
700 C or 750 C without ethylene gas introduction or carbon nanotube growth. Island diameters are between 5 and 15 nm, with an average diameter of 10 nm and a pitch (center-to-center spacing between adjacent particles) of 15 nm. These catalyst islands effectively stimulate the growth of nanotubes in this process. It could be assumed that the nanotubes were grown off every Fe particle based on the density, size and pitch of the assynthesized nanotubes. Previous studies suggested that the size of the catalysts could define the diameter of as-grown nanotubes [17,18], even though the CNTs are not aligned ones. In our case, it was also observed that the diameter of the as-grown aligned nanotubes is the same as that of the Fe islands, which suggests that the Fe particle size and distribution also determine the aligned CNT diameter and distributions. If the Fe particle size is further decreased, the denser CNT arrays can be formed. XPS survey scans were taken on aligned CNT films prepared by the same CVD procedure except that catalyst material covered the entire substrate. Fig. 5 clearly indicates that only carbon is detected and the XPS spectrum is identical with that of a standard graphite sample. These results confirm that the as-synthesized CNTs are highly graphitized and of high purity. EDS studies agree with the XPS spectra in that no elements other than carbon are detected. These data clearly demonstrate the base-growth mechanism for nanotube synthesis by the CVD process. The growth mode also suggests the continuous growth of CNTs throughout the whole pillar height, as there are no new catalysts added during the growth process. 3.3. Electrical measurements
Fig. 4. Catalyst nanoparticles formed on the substrate after a CVD run without ethylene gas or carbon nanotube growth.
Fig. 6 shows the optical micrograph of a nanotube bundle bridging two electrodes. The distance between two electrodes near the ends of the nanotube bundle is
L. Zhu et al. / Carbon 44 (2006) 253–258
215 lm. The measured electrical resistance is 250 X (Fig. 7a), which corresponds to a resistivity of 0.009 X cm. This resistivity value is similar to those reported previously. For instance, the resistivity of nanotube bundles at 300 K has been reported to be 0.0065 X cm [19], while resistivities between 8 · 10 4 X cm and 12 · 10 3 X cm have been measured for individual nanotubes [20]. For our measurements, the current–voltage (IV) characteristics indicate that contact resistance exists. The nearly linear relationship observed between current and voltage is good indication of metallic conductivity. Similarly, except for nanotube defects, the high measured resistance of metallic nanotubes is also due to high contact resistance [21] caused by the unique electronic structure of nanotubes, which gives rise to weak electronic coupling at the Fermi surface. Moreover, in our studies, the probes contacted only the sidewalls of the nanotubes during IV measurements. As a result, higher electrical resistance is observed, since the electrical resistance in directions vertical to the tube axis is much larger than that parallel to the tube axis [22]. The capacitance of the nanotube bundle was 2.55 pF as the voltage was scanned from 1 V to 1 V, as shown in Fig. 7b. The preliminary measurement results show that the CNT arrays by this process are promising for electrical interconnect applications, even though now the resistance and capacitance are still too large for direct CNT interconnect applications.
C1s
60000 50000
Counts/s
40000 30000 20000
C KL L
10000 0 1400
1200
1000
800
600
400
200
257
0
Binding Energy(eV) Fig. 5. XPS survey scan of CNT array surface.
4. Conclusions
Fig. 6. Optical micrographs of nanotube bundle of length 215 lm, bridging two electrodes.
We have demonstrated an effective and reproducible process for growth of high-quality CNT bundles with high aspect ratio and very high CNT density (>1500 lm 2). At 750 C, the nanotubes grow at a high
2.70 0.010
2.68 2.66 Capacitance (pF)
0.008
I (A)
0.006 0.004
2.64 2.62 2.60 2.58 2.56 2.54
0.002
2.52 0.000 0.0
a
2.50 0.5
1.0
V(V)
1.5
-1.0
2.0
-0.5
b
Fig. 7. (a) IV curve and (b) CV curve of CNT arrays of Fig. 6.
0.0
V(V)
0.5
1.0
258
L. Zhu et al. / Carbon 44 (2006) 253–258
average rate of 150 lm/min with high aspect ratio. HRTEM, XPS and SEM characterization demonstrate that the as-synthesized ACNTs have high purity without carbon particles attached. The feasibility of using such nanotube arrays for interconnect applications has been investigated by measuring the IV and CV curves using two-probe methods. The measured electrical resistance and capacitance are still too large for direct CNT interconnect applications. Currently, the challenges involve the development of methods to decrease the contact resistance between CNTs and other metal bonding pads/electrodes. Acknowledgements We would like to thank NSF for funding support (DMI-0422553). We also thank Dr. Yong Ding for HRTEM examinations and Dr. Jie Diao for XPS analysis. References [1] Dresselhaus MS, Dresselhaus G, Eklund PC. In: Science of fullerenes and carbon nanotubes. New York, NY, San Diego, CA: Academic Press; 1996. [2] Hamada N, Sawada S, Oshiyama A. New one-dimensional conductors—graphitic microtubules. Phys Rev Lett 1992;68(10): 1579–81. [3] Saito R, Fujita M, Dresselhaus G, Dresselhaus MS. Electronicstructure of chiral graphene tubules. Appl Phys Lett 1992;60(18): 2204–6. [4] Frank S, Poncharal P, Wang ZL, de Heer WA. Carbon nanotube quantum resistors. Science 1998;280(5370):1744–6. [5] Kim P, Shi L, Majumdar A, McEuen PL. Thermal transport measurements of individual multiwalled nanotubes. Phys Rev Lett 2001;87(21):215502-1–4. [6] Baughman RH, Zakhidov AA, de Heer WA. Carbon nanotubes— the route toward applications. Science 2002;297(5582):787–92.
[7] Wei BQ, Vajtai R, Ajayan PM. Reliability and current carrying capacity of carbon nanotubes. Appl Phys Lett 2001;79(8):1172–4. [8] Srivastava N, Banerjee K. Interconnect challenges for nanoscale electronic circuits. Jom 2004;56(10):30–1. [9] Li J, Ye Q, Cassell A, Ng HT, Stevens R, Han J, et al. Bottom-up approach for carbon nanotube interconnects. Appl Phys Lett 2003;82(15):2491–3. [10] Kreupl F, Graham AP, Duesberg GS, Steinhogl W, Liebau M, Unger E, et al. Carbon nanotubes in interconnect applications. Microelectron Eng 2002;64(1–4):399–408. [11] Hoenlein W, Kreupl F, Duesberg GS, Graham AP, Liebau M, Seidel R, et al. Carbon nanotubes for microelectronics: status and future prospects. Mater Sci Eng C—Biomim Supramol Syst 2003;23(6–8):663–9. [12] McEuen PL, Fuhrer M, Park H. Single-walled carbon nanotube electronics. IEEE Trans Nanotechnol 2002;1(1):78–85. [13] Dresselhaus MS, Dresselhaus G, Avouris P. In: Carbon nanotubes. Berlin: Springer; 2001. [14] Fan SS, Chapline MG, Franklin NR, Tombler TW, Cassell AM, Dai HJ. Self-oriented regular arrays of carbon nanotubes and their field emission properties. Science 1999;283(5401):512–4. [15] Ren ZF, Huang ZP, Xu JW, Wang JH, Bush P, Siegal MP, et al. Synthesis of large arrays of well-aligned carbon nanotubes on glass. Science 1998;282(5391):1105–7. [16] Wei BQ, Vajtai R, Jung Y, Ward J, Zhang R, Ramanath G, et al. Assembly of highly organized carbon nanotube architectures by chemical vapor deposition. Chem Mater 2003;15(8):1598–606. [17] Cheung CL, Kurtz A, Park H, Lieber CM. Diameter-controlled synthesis of carbon nanotubes. J Phys Chem B 2002;106(10): 2429–33. [18] Flahaut E, Laurent C, Peigney A. Catalytic CVD synthesis of double and triple-walled carbon nanotubes by the control of the catalyst preparation. Carbon 2005;43(2):375–83. [19] Song SN, Wang XK, Chang RPH, Ketterson JB. Electronic properties of graphitic nanotubes from galvanomagnetic effects. Phys Rev Lett 1994;72(5):697–700. [20] Dai HJ, Wong EW, Lieber CM. Probing electrical transport in nanomaterials: conductivity of individual carbon nanotubes. Science 1996;272(5261):523–6. [21] Tersoff J. Contact resistance of carbon nanotubes. Appl Phys Lett 1999;74(15):2122–4. [22] Wang XB, Liu YQ, Yu G, Xu CY, Zhang JB, Zhu DB. Anisotropic electrical transport properties of aligned carbon nanotube films. J Phys Chem B 2001;105(39):9422–5.
Growth and electrical characterization of high-aspect-ratio carbon nanotube arrays Lingbo Zhu a, Jianwen Xu b, Yonghao Xiu a, Yangyang Sun b, Dennis W. Hess a, C.P. Wong b,* a
School of Chemical and Biomolecular Engineering, Georgia Institute of Technology, 311 Ferst Drive, Atlanta 30332, GA, USA b School of Materials Science and Engineering, Georgia Institute of Technology, 711 Ferst Drive, Atlanta 30332, GA, USA Received 30 April 2005; accepted 23 July 2005 Available online 13 September 2005
Abstract The remarkable properties of carbon nanotubes (CNTs) make them attractive for microelectronic applications, especially for interconnects and nanoscale devices. In this paper, we describe a microelectronics compatible process for growing high-aspect-ratio CNT arrays with application to vertical electrical interconnects. A lift-off process was used to pattern catalyst (Al2O3/Fe) islands to diameters of 13 or 20 lm. After patterning, chemical vapor deposition (CVD) was involved to deposit highly aligned CNT arrays using ethylene as the carbon source, and argon and hydrogen as carrier gases. The as-grow CNTs were characterized by transmission electron microscopy (TEM), scanning electron microscopy (SEM) and X-ray photoelectron spectroscopy (XPS). The results demonstrated that the CNTs have high purity, and form densely-aligned arrays with controllable array size and height. Two-probe electrical measurements of the CNT arrays indicate a resistivity of 0.01 X cm, suggesting possible use of these CNTs as interconnect materials. 2005 Elsevier Ltd. All rights reserved. Keywords: Carbon nanotubes; Chemical vapor deposition; Electron microscopy
1. Introduction Carbon nanotubes (CNTs) display a well-defined quasi one dimensional cylindrical structure, formed by rolling up graphene sheets of sp2 bonded carbon atoms [1]. The CNTs can be either metallic or semiconducting, depending on how the graphite layer is wrapped into a cylinder [2,3]. Metallic CNTs show ballistic conductivity at room temperature [4]. The ballistic conductivity, high thermal conductivity [5] and mechanical strength of CNTs make them ideal candidates for electrical interconnects in IC packaging and nanoscale devices.
*
Corresponding author. Tel.: +1 404 894 8391; fax: +1 404 894 9140. E-mail address: [email protected] (C.P. Wong). 0008-6223/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2005.07.037
The electronic properties of perfect multi-walled carbon nanotubes (MWCNTs) are similar to those of single-walled carbon nanotubes (SWCNTs) [6], due to weak coupling between the graphite cylinders (layer distance is P0.34 nm). Electrons transport ballistically (without scattering) in metallic SWCNTs and MWCNTs over reasonable lengths (1 lm), thereby enabling CNTs to carry very high currents (>109 A/cm2) [7] without electromigration failure. Phonons also propagate easily along the nanotubes [6]. The measured thermal conductivity of an individual MWCNT at room temperature is >3000 W/m K [5], which exceeds the conductivity of diamond (2000 W/m K). Based on these advantageous properties of CNTs, researchers have reported the integration of CNTs into electrical interconnect applications [8–11]. However, the resistance of a single ballistic SWCNT less than 1 lm long is about
254
L. Zhu et al. / Carbon 44 (2006) 253–258
6.5 kX with perfect contacts [12], while ballistic transport in MWCNTs with a resistance of 12.9 kX was also reported [13]. The high resistance of an individual CNT indicates that an array of thousands of parallel CNTs will be necessary for interconnect applications. Although there are many reports about the growth of the aligned CNT arrays with controlled areas and nanotube length [14–16], deliberate production of high-aspect-ratio nanotube structures was seldom reported. In this study, we report a simple, but efficient CVD method for the growth of aligned carbon nanotube arrays from a small area (<20 lm in diameter) but extending to a very long length (several hundreds of lm) without any bending or curling. These high-aspect-ratio nanotube arrays are particularly useful as electrical interconnects or mechanical strengthening backbones.
2. Experimental
(DI) water rinses. The patterned substrate was placed in the CVD chamber for highly-oriented CNT pillar growth. 2.2. CNT characterization Samples for transmission electron microscopy were prepared by sonicating a trace amount of synthesized carbon nanotubes in methanol for 30 min. A few drops of the suspension were placed onto a TEM grid. TEM (JEOL 400EX) analysis was performed at 400 kV. Scanning electron microscopy (SEM) was carried out on a JEOL 1530 equipped with a thermally assisted field emission gun operating at 10 keV. Spatially resolved energy-dispersive X-ray spectroscopy (EDS) was used to assess the purity of carbon nanotubes. X-ray photoelectron spectroscopy (XPS) was performed using a physical electronics (PHI) model 1600 XPS system equipped with a monochromator. The system used an Al Ka source (hm = 1486.8 eV).
2.1. CNT array growth Chemical vapor deposition (CVD) was performed in a horizontal alumina tube (3.8 cm diameter; 92 cm long) housed in a Lindberg Blue furnace. The substrates used in this study were (0 0 1) silicon wafers coated with SiO2 (400 nm) by thermal oxidation or plasma-enhanced CVD. The catalyst layers of Al2O3 (15 nm)/Fe (2 nm) were formed on the silicon wafer by a lift-off process. CVD growth of CNTs was carried out at 700–800 C with ethylene as the carbon source, and hydrogen and argon as carrier gases. A schematic of the process is shown in Fig. 1 and described briefly below. A thin layer of photoresist (Shipley 1813) was spincoated on a silicon substrate and subsequently exposed to UV light and the patterns developed. The catalyst layers of Al2O3/Fe were then deposited onto the substrate by a sequential electron-beam evaporation process. The photoresist was removed by acetone and the wafer cleaned using acetone, isopropyl alcohol and de-ionized
2.3. Resistance and capacitance measurements of CNT arrays Two-probe current–voltage measurements of the CNT arrays were performed using an Agilent 4285 A Precision LCR Meter. Capacitance–voltage (CV) responses were measured by an HP semiconductor analyzer at high frequency of 100 kHz. Before sample measurement, the analyzer was calibrated by touching probes on the two electrodes without CNT arrays on them to exclude the contribution of capacitance from the substrate. The nanotube array solution was placed on a substrate that had pre-patterned Au electrodes. Randomly, the two ends of some CNT arrays contacted electrodes. These samples were mounted on an electrical probe station, and two-terminal resistance measurements were performed by contacting the probes (12 lm tungsten needles) to these electrodes.
3. Results and discussion 3.1. CNT array growth
Fig. 1. Schematic of the CNT pillar growth.
Aligned CNTs grow readily from the catalyst patterns into well-defined vertical structures, as shown in Fig. 2, which shows SEM micrographs of CNT arrays with different pattern size, pitch and height. The synthesis conditions for Fig. 2a and b are for 3 min at 750 C and 700 C, respectively. The images demonstrate that nearly vertically-aligned CNT pillars are formed on the substrate with minimal entanglement with neighboring CNTs, but the arrays bend at the tips due to the gravity if the CNTs are too long. In Fig. 2a, the height of the CNT arrays reaches 420 lm with an aspect ratio
L. Zhu et al. / Carbon 44 (2006) 253–258
255
Fig. 2. SEM micrographs of highly-aligned CNT arrays. (a) SEM image of nanotube arrays synthesized on 12 lm by 12 lm catalyst patterns; (b) SEM image of nanotube arrays synthesized on 20 lm by 20 lm catalyst patterns; (c) SEM image of nanotube arrays synthesized on cylindrical patterns with diameter of 20 lm; (d) side view of the nanotube arrays in (c); (e) enlarged view of nanotube tip of the arrays in (c); (f) enlarged view of a nanotube root of the arrays in (d).
of 32, while in Fig. 2b the height of the CNT arrays reaches 300 lm with an aspect ratio of 15. Furthermore, the vertical alignment of pillars can be influenced by the array aspect ratio. With a pillar height <150 lm, no pillar bending was observed. In fact, further studies have demonstrated the CNT pillars maintain vertical growth up to an aspect ratio of 15. Fig. 2e shows a high magnification SEM image of a nanotube array tip with a diameter of 20 lm. Fig. 2f shows an enlarged image of the base of a free-standing nanotube array which indicates the well-defined edges of these CNT arrays. These
results show that the height and aspect ratios of CNT pillars can satisfy most interconnect applications. There are two levels of hierarchical alignment structure that exist in this CVD process. The primary alignment is a bundle of densely-packed CNTs to form the CNT pillar (Fig. 3a). The secondary alignment is the array of CNT pillars on the substrate, where the growth of CNTs is confined within the catalyst patterns. From the microscopic to the macroscopic level, the alignment of CNTs make possible novel interconnect structures that can be produced by methods compatible
256
L. Zhu et al. / Carbon 44 (2006) 253–258
Fig. 3. (a) SEM image of CNT arrays; (b) HRTEM image of CNT bundles.
with existing IC fabrication sequences. In Fig. 3a no carbon or other particles are observed on CNT walls, and the high density of CNT bundles is evident. We estimate that the CNT density could reach 1500 lm 2. EDS examination along the CNT bundles detected no elements except carbon, and HRTEM examination revealed no particles attached on the CNTs walls. These results indicate the high purity of the as-synthesized nanotubes. From HRTEM images, we also know that most of the CNTs synthesized by this CVD method are double-walled or triple-walled with an average diameter of 10 nm. Fig. 3a shows a double-walled CNT. 3.2. CNT growth mode As shown in Fig. 4, the thin Fe film of 2 nm segregated into small islands, after a heat treatment at
700 C or 750 C without ethylene gas introduction or carbon nanotube growth. Island diameters are between 5 and 15 nm, with an average diameter of 10 nm and a pitch (center-to-center spacing between adjacent particles) of 15 nm. These catalyst islands effectively stimulate the growth of nanotubes in this process. It could be assumed that the nanotubes were grown off every Fe particle based on the density, size and pitch of the assynthesized nanotubes. Previous studies suggested that the size of the catalysts could define the diameter of as-grown nanotubes [17,18], even though the CNTs are not aligned ones. In our case, it was also observed that the diameter of the as-grown aligned nanotubes is the same as that of the Fe islands, which suggests that the Fe particle size and distribution also determine the aligned CNT diameter and distributions. If the Fe particle size is further decreased, the denser CNT arrays can be formed. XPS survey scans were taken on aligned CNT films prepared by the same CVD procedure except that catalyst material covered the entire substrate. Fig. 5 clearly indicates that only carbon is detected and the XPS spectrum is identical with that of a standard graphite sample. These results confirm that the as-synthesized CNTs are highly graphitized and of high purity. EDS studies agree with the XPS spectra in that no elements other than carbon are detected. These data clearly demonstrate the base-growth mechanism for nanotube synthesis by the CVD process. The growth mode also suggests the continuous growth of CNTs throughout the whole pillar height, as there are no new catalysts added during the growth process. 3.3. Electrical measurements
Fig. 4. Catalyst nanoparticles formed on the substrate after a CVD run without ethylene gas or carbon nanotube growth.
Fig. 6 shows the optical micrograph of a nanotube bundle bridging two electrodes. The distance between two electrodes near the ends of the nanotube bundle is
L. Zhu et al. / Carbon 44 (2006) 253–258
215 lm. The measured electrical resistance is 250 X (Fig. 7a), which corresponds to a resistivity of 0.009 X cm. This resistivity value is similar to those reported previously. For instance, the resistivity of nanotube bundles at 300 K has been reported to be 0.0065 X cm [19], while resistivities between 8 · 10 4 X cm and 12 · 10 3 X cm have been measured for individual nanotubes [20]. For our measurements, the current–voltage (IV) characteristics indicate that contact resistance exists. The nearly linear relationship observed between current and voltage is good indication of metallic conductivity. Similarly, except for nanotube defects, the high measured resistance of metallic nanotubes is also due to high contact resistance [21] caused by the unique electronic structure of nanotubes, which gives rise to weak electronic coupling at the Fermi surface. Moreover, in our studies, the probes contacted only the sidewalls of the nanotubes during IV measurements. As a result, higher electrical resistance is observed, since the electrical resistance in directions vertical to the tube axis is much larger than that parallel to the tube axis [22]. The capacitance of the nanotube bundle was 2.55 pF as the voltage was scanned from 1 V to 1 V, as shown in Fig. 7b. The preliminary measurement results show that the CNT arrays by this process are promising for electrical interconnect applications, even though now the resistance and capacitance are still too large for direct CNT interconnect applications.
C1s
60000 50000
Counts/s
40000 30000 20000
C KL L
10000 0 1400
1200
1000
800
600
400
200
257
0
Binding Energy(eV) Fig. 5. XPS survey scan of CNT array surface.
4. Conclusions
Fig. 6. Optical micrographs of nanotube bundle of length 215 lm, bridging two electrodes.
We have demonstrated an effective and reproducible process for growth of high-quality CNT bundles with high aspect ratio and very high CNT density (>1500 lm 2). At 750 C, the nanotubes grow at a high
2.70 0.010
2.68 2.66 Capacitance (pF)
0.008
I (A)
0.006 0.004
2.64 2.62 2.60 2.58 2.56 2.54
0.002
2.52 0.000 0.0
a
2.50 0.5
1.0
V(V)
1.5
-1.0
2.0
-0.5
b
Fig. 7. (a) IV curve and (b) CV curve of CNT arrays of Fig. 6.
0.0
V(V)
0.5
1.0
258
L. Zhu et al. / Carbon 44 (2006) 253–258
average rate of 150 lm/min with high aspect ratio. HRTEM, XPS and SEM characterization demonstrate that the as-synthesized ACNTs have high purity without carbon particles attached. The feasibility of using such nanotube arrays for interconnect applications has been investigated by measuring the IV and CV curves using two-probe methods. The measured electrical resistance and capacitance are still too large for direct CNT interconnect applications. Currently, the challenges involve the development of methods to decrease the contact resistance between CNTs and other metal bonding pads/electrodes. Acknowledgements We would like to thank NSF for funding support (DMI-0422553). We also thank Dr. Yong Ding for HRTEM examinations and Dr. Jie Diao for XPS analysis. References [1] Dresselhaus MS, Dresselhaus G, Eklund PC. In: Science of fullerenes and carbon nanotubes. New York, NY, San Diego, CA: Academic Press; 1996. [2] Hamada N, Sawada S, Oshiyama A. New one-dimensional conductors—graphitic microtubules. Phys Rev Lett 1992;68(10): 1579–81. [3] Saito R, Fujita M, Dresselhaus G, Dresselhaus MS. Electronicstructure of chiral graphene tubules. Appl Phys Lett 1992;60(18): 2204–6. [4] Frank S, Poncharal P, Wang ZL, de Heer WA. Carbon nanotube quantum resistors. Science 1998;280(5370):1744–6. [5] Kim P, Shi L, Majumdar A, McEuen PL. Thermal transport measurements of individual multiwalled nanotubes. Phys Rev Lett 2001;87(21):215502-1–4. [6] Baughman RH, Zakhidov AA, de Heer WA. Carbon nanotubes— the route toward applications. Science 2002;297(5582):787–92.
[7] Wei BQ, Vajtai R, Ajayan PM. Reliability and current carrying capacity of carbon nanotubes. Appl Phys Lett 2001;79(8):1172–4. [8] Srivastava N, Banerjee K. Interconnect challenges for nanoscale electronic circuits. Jom 2004;56(10):30–1. [9] Li J, Ye Q, Cassell A, Ng HT, Stevens R, Han J, et al. Bottom-up approach for carbon nanotube interconnects. Appl Phys Lett 2003;82(15):2491–3. [10] Kreupl F, Graham AP, Duesberg GS, Steinhogl W, Liebau M, Unger E, et al. Carbon nanotubes in interconnect applications. Microelectron Eng 2002;64(1–4):399–408. [11] Hoenlein W, Kreupl F, Duesberg GS, Graham AP, Liebau M, Seidel R, et al. Carbon nanotubes for microelectronics: status and future prospects. Mater Sci Eng C—Biomim Supramol Syst 2003;23(6–8):663–9. [12] McEuen PL, Fuhrer M, Park H. Single-walled carbon nanotube electronics. IEEE Trans Nanotechnol 2002;1(1):78–85. [13] Dresselhaus MS, Dresselhaus G, Avouris P. In: Carbon nanotubes. Berlin: Springer; 2001. [14] Fan SS, Chapline MG, Franklin NR, Tombler TW, Cassell AM, Dai HJ. Self-oriented regular arrays of carbon nanotubes and their field emission properties. Science 1999;283(5401):512–4. [15] Ren ZF, Huang ZP, Xu JW, Wang JH, Bush P, Siegal MP, et al. Synthesis of large arrays of well-aligned carbon nanotubes on glass. Science 1998;282(5391):1105–7. [16] Wei BQ, Vajtai R, Jung Y, Ward J, Zhang R, Ramanath G, et al. Assembly of highly organized carbon nanotube architectures by chemical vapor deposition. Chem Mater 2003;15(8):1598–606. [17] Cheung CL, Kurtz A, Park H, Lieber CM. Diameter-controlled synthesis of carbon nanotubes. J Phys Chem B 2002;106(10): 2429–33. [18] Flahaut E, Laurent C, Peigney A. Catalytic CVD synthesis of double and triple-walled carbon nanotubes by the control of the catalyst preparation. Carbon 2005;43(2):375–83. [19] Song SN, Wang XK, Chang RPH, Ketterson JB. Electronic properties of graphitic nanotubes from galvanomagnetic effects. Phys Rev Lett 1994;72(5):697–700. [20] Dai HJ, Wong EW, Lieber CM. Probing electrical transport in nanomaterials: conductivity of individual carbon nanotubes. Science 1996;272(5261):523–6. [21] Tersoff J. Contact resistance of carbon nanotubes. Appl Phys Lett 1999;74(15):2122–4. [22] Wang XB, Liu YQ, Yu G, Xu CY, Zhang JB, Zhu DB. Anisotropic electrical transport properties of aligned carbon nanotube films. J Phys Chem B 2001;105(39):9422–5.