NANO LETTERS
Well-Aligned Open-Ended Carbon Nanotube Architectures: An Approach for Device Assembly
2006 Vol. 6, No. 2 243-247
Lingbo Zhu,† Yangyang Sun,‡ Dennis W. Hess,*,† and Ching-Ping Wong*,‡ School of Chemical & Biomolecular Engineering, Georgia Institute of Technology, 311 Ferst DriVe, Atlanta, Georgia 30332, and School of Materials Science & Engineering, Georgia Institute of Technology, 771 Ferst DriVe, Atlanta, Georgia 30332 Received November 5, 2005; Revised Manuscript Received December 6, 2005
ABSTRACT To circumvent the high carbon nanotube (CNT) growth temperature and poor adhesion with the substrates that currently plague CNT implementation, we proposed using CNT transfer technology enabled by open-ended CNTs. The process is featured with separation of CNT growth and CNT device assembly. Field emission testing of the as-assembled CNT devices is in good agreement with the Fowler−Nordheim (FN) equation, with a field enhancement factor of 4540. This novel technique shows promising applications for positioning CNTs on temperaturesensitive substrates and for the fabrication of field emitters, electrical interconnects, and thermal management structures in microelectronics packaging.
The simulated properties of carbon nanotubes (CNTs) are extraordinary, such as extremely high thermal conductivities,1 very high electrical conductivities, and outstanding mechanical properties.2 Combined with their one-dimensional molecular structures, CNTs offer many promising applications for future electronic materials and device structures.3-6 However, the resistance of an individual ballistic singlewalled CNT (SWCNT) less than 1 µm long is ∼6.5 kΩ assuming perfect contacts,7 whereas ballistic transport in multiwalled CNTs (MWCNTs) displays a resistance of 12.9 kΩ.8 The high resistance of an individual CNT indicates that an array of thousands of parallel CNTs will be necessary for interconnect applications. Growth of aligned CNT arrays with controlled areas and nanotube lengths is well-known.9-12 However, none of these processes can produce well-aligned CNTs with open ends. If open-ended aligned CNT films/ arrays can be formed, then improved electrical conductance of MWCNTs should result because recent studies have demonstrated that the internal walls of MWCNTs can participate in electrical transport, thereby enabling large current-carrying capacity.13 These considerations imply that multichannel ballistic transport could be achieved if the caps * Corresponding authors. Dennis W. Hess. Tel: +86-1-404-894-5922. Fax: +86-1-404-894-2866. E-mail:
[email protected]; Dr. Ching-Ping Wong. Tel: +86-1-404-894-8391. Fax: +86-1-404-894-9140. E-mail:
[email protected]. † School of Chemical & Biomolecular Engineering, Georgia Institute of Technology. ‡ School of Materials Science & Engineering, Georgia Institute of Technology. 10.1021/nl052183z CCC: $33.50 Published on Web 01/04/2006
© 2006 American Chemical Society
of the CNTs are removed; CNT electrical conductance should therefore be improved dramatically. Such achievements may then allow CNTs to serve as conductive nanowires and thus replace copper and aluminum films used in state-of-the-art circuits; such nanowires are less susceptible to electromigration under high current density than are Cu and Al. It is also expected that the hollow cavity of CNTs will allow the wicking of solders, such as Sn/Pb and Sn/Ag/Cu, due to capillary forces. As a result, interconnects of CNTs with metal electrodes by solders should be possible and the limited wetting of solders on CNT films would be eliminated.14 The novel process described below includes aligned open-ended CNT architecture growth and CNT transfer technology. The success of this methodology is reflected in the performance of the assembled CNT field emitters. This process may offer a new paradigm for transferring and integrating CNTs onto integrated circuits (ICs) as well as other moduli in microelectronic packaging systems because the approaches used circumvent the high CNT growth temperature and poor adhesion that currently plague CNT implementation. CNTs fabricated by arc-discharge techniques,15 laser ablation, and chemical vapor deposition (CVD) are usually capped at the ends in the form of hemispherical or polyhedral domes.16 The nanotubes can be opened by etching away the caps using oxygen17 and carbon dioxide18; however, the nanotube walls are inevitably damaged. As a result, CNT electrical and mechanical properties are degraded. Realization of high-quality open-ended CNT synthesis requires either a
Figure 1. (a) Cross-sectional SEM image of eight-layer CNT films. The films have been partially scratched to show the existence of the CNT layers. (b) High-magnification SEM image of the CNT films in a. (c) HRTEM image to show the open-ended structures of as-grown CNTs. (d) Single-layered CNT film to show the uniformity of the nanotube films up to 375 µm.
Figure 2. Schematic diagram of “CNT transfer technology”. UBM: under bump metallization. See the text for detailed explanations.
novel posttreatment process to open CNTs or an alteration of the CVD process to allow in situ high-quality open-ended CNT growth without the need for subsequent processes. In situ growth of open-ended CNTs is desirable because it is cost-effective. Our intent is to develop a novel process to open the nanotubes in situ in order to study the corresponding CNT properties while maintaining CNT film alignment. Toward this end, we recently reported a CVD-growth strategy to form CNT stacks by water-assisted selective etching;19 Figure 1a shows an example of stacked eight-layer CNT films. The films were partially scratched by tweezers to demonstrate the existence of layered structures. Figure 1b is a high-magnification SEM image of the individual CNT layer, which clearly demonstrates that no particles are 244
observed on CNT walls and that the high density of CNT films has been formed. EDS investigations along the CNT bundles detected no elements except carbon. More interestingly, from HRTEM examinations, we find that the CNTs synthesized by this CVD method are open-ended (Figure 1c). In this process, parts per million levels of water etch the ends of the nanotubes because of the higher concentration of structural defects that exist in the CNT ends. The pentagon structures in the caps result in curvature and correspondingly larger strains and thus higher chemical reactivity compared with unstrained bonds (nanotube walls). Therefore, the nanotube ends are most susceptible to chemical attack by water molecules. This simple process offers a novel way to synthesize aligned open-ended CNT films. To explore the Nano Lett., Vol. 6, No. 2, 2006
Figure 3. (a) Photograph of an open-ended CNT film transferred onto the copper substrate coated with a eutectic tin-lead solder. The clear silicon chip shows that the CNTs are effectively transferred to the substrate and connected by the solder. (b) Photograph of a closedended CNT film that has only partially transferred onto the copper substrate coated with the eutectic tin-lead solder. Most CNTs still remain on the silicon chip, indicating poor wetting properties of solders on the CNT films. (c) A cross-sectional SEM image of well-aligned open-ended CNT films transferred onto the copper substrate. The eutectic tin-lead solder is used to interconnect CNTs and the copper substrate. (d) Side view of the interface between the CNTs and the solders at angle of 30°.
unique properties shown by open-ended CNTs, only singlelayer CNT films are grown as shown in Figure 1d, which indicates the uniformity of the film height to a thickness of 375 µm. The substrates used in this study were (001) silicon wafers coated with SiO2 (500 nm) by thermal oxidation. The catalyst layers of Al2O3 (15 nm)/Fe (3 nm) were formed on the silicon wafer by sequential e-beam evaporation. CVD growth of CNTs was carried out at 775 °C with ethylene (150 sccm) as the carbon source and hydrogen (180 sccm) and argon (350 sccm) as carrier gases. The water vapor concentration in the CVD chamber was controlled by bubbling a small amount of argon gas through water held at 22 °C. Ethylene flowed into the CVD system for a preset time, after which the flow was terminated; it followed by 5 min of only water, argon, and hydrogen flow, which was used to selectively etch the nanotube tips and carbon atoms at the interfaces between the nanotubes and catalyst particles. Additional studies have demonstrated that the CNT film height can remain uniform to more than 500 µm if CVD Nano Lett., Vol. 6, No. 2, 2006
process conditions are adequately controlled. The length uniformity of nanotube films is important for the CNT assembly process described below because a conformal geometry is required to guarantee that a maximum number of CNT interconnects contact solder that will be placed on the substrates. For electronic device applications, chemical vapor deposition (CVD) methods are particularly attractive because of characteristic CNT growth features such as selective spatial growth, large area deposition capability, and aligned CNT growth. However, the CVD technique suffers from several drawbacks. One of the main challenges for applying CNTs to circuitry is the high growth temperature (>600 °C). Such temperatures are incompatible with microelectronic processes, which are typically performed below 400-500 °C in back-end-of-line sequences. Another issue is the poor adhesion between CNTs and the substrates, which will result in long term reliability issues and high contact resistance. To fabricate microelectronic devices that incorporate CNT 245
Figure 4. (a) Emission pattern of the as-assembled CNTs by applying an electrical field of 3.0 V/µm. (b) Field emission measurements of the CNT films in a at room temperature. The inset shows a Fowler-Nordheim plot, which indicates that the transferred CNTs demonstrate good field emission characteristics.
blocks, the CNTs should be selectively positioned and interconnected to other materials such as metal electrodes or bonding pads. The common practices for CNT growth on such substrates involve the deposition catalysts such as Fe or Ni on metal layers such as Ti or Ti/Au. Unfortunately, the electrical contact is not necessarily improved because the catalyst nanoparticles are capped by the nanotubes. To overcome these disadvantages, we propose a methodology that we term “CNT transfer technology”, which is enabled by open-ended CNT structures. This technique is similar to flip-chip technology as illustrated schematically in Figure 2. The substrates can be FR-4 boards coated with copper foil or other materials and moduli, such as heat sinks. To improve the adhesion and wetting of solder on the substrates, the under bump metallization (UBM) layers are sputtered onto the substrate metallization. The eutectic tin-lead paste (100 µm) is then stencil-printed on the UBM. After reflow, the tin-lead solder is polished to 30 µm. The silicon substrates with CNTs are flipped and aligned to the corresponding copper substrates and reflowed in a seven-zone BTU reflow oven at higher temperatures (peak temperature at 270 °C) than those typically used (220 °C) to simultaneously form electrical and mechanical connections. This process is straightforward to implement and offers a strategy for both assembling CNT devices and scaling up a variety of devices fabricated using nanotubes (e.g., flat panel displays). This process could overcome the serious obstacles of integration of CNTs into integrated circuits and microelectronic device packages by offering low process temperatures and improved adhesion of CNTs to the substrates. Figure 3a indicates that the entire CNT film (1.5 × 1.5 cm2) is transferred to the substrate (2.54 × 2.54 cm2) because no trace amount of CNTs is evident on the silicon chip. This result is in stark contrast to the same process wherein the CNTs are closedended in that only a fraction of the CNT film is transferred to the substrate (shown in Figure 3b), indicating that the 246
adhesion of the attached nanotube films is poor. Figure 3c is a cross-sectional SEM image of well-aligned open-ended CNT films transferred onto the copper substrate. Any detachment of the CNT film from the substrate is not found. Furthermore, the close observations of the interface between CNTs and the solders show that CNTs are well connected by the solders, as shown in Figure 3d. In addition, upon drop testing, the closed-ended CNT films detached from the substrate, whereas open-ended CNTs did not. Such observations indicate that the wetting properties of the solders on the aligned open-ended CNT film have been improved. We believe that the open channels of nanotubes assist the adhesion between the nanotubes and the solders because of strong capillary forces that draw the solder inside the CNTs. In the molten state, the eutectic tin-lead solder has a similar surface tension as does elemental lead. Indeed, experiments have shown that lead could fill into carbon nanotube cavities by capillary forces and that the filling at least partially wets the inside surface of the nanotubes.20 At present, we are not certain if compound formation has occurred by reaction of the solder with carbon at the nanotube tips. To explore the electrical properties of CNTs connected by solders on the copper substrates, field emission characterization of the as-prepared assembly has been performed. The height of the nanotube films is ∼323 µm with diameters in the range of 10 to 20 nm. By changing the catalyst thickness, reaction temperature, and time, the nanotube film thickness can be controllably varied. We measured the (cathodic) electron emission from 1.5 × 1.5 cm2 well-aligned open-ended CNT films shown in Figure 3a at room temperature and in a vacuum chamber below 10-5 Torr. The spacing between the CNT tip and the anode (phosphor-coated ITO glass) was ∼180 µm and was maintained by a poly (tetafluoroethylene) (PTFE) spacer. The phosphor screen became quite homogeneous over the whole area when the applied field was 3.0 V/µm, as shown in Figure 4a. The measured current density (mA/cm2) as a function of electric Nano Lett., Vol. 6, No. 2, 2006
field (V/µm) is shown in Figure 4b. A typical turn-on field, which produces a current density of 10 µA/cm2, is ∼1.8V/ µm, whereas the emission current density of 1 mA/cm2 requires an applied field of ∼2.74 V/µm. The small turn-on field is consistent with literature data of 1.5-2 V/µm observed in CVD-grown dense CNT films.21 At an electric field of 3.4 V/µm, the assembled CNT field emitters emit a current density of 5 mA/cm2, which is a remarkably large value considering the distance between the cathode and the anode. A plot of ln(I/V2) versus 1/V yields a straight line in a good agreement with the Fowler-Nordheim (FN) expression; this agreement demonstrates that the current originates from field emission (i.e., field emission process).22,23 Furthermore, the quality of fit to the Fowler-Nordheim expression implies good nanotube/substrate electrical contact.24 Finally, these observations clearly show an additional advantage to CNT transfer technology: a very small voltage drop can be achieved along CNT/substrate interfaces. The slope of the FN plot can be used to calculate the field enhancement factor, β. The Fowler-Nordheim equation can be written as24 I)A
() ( ) (
)
10.4 1.42 × 10-6 2 V 2 Bφ1.5d β exp exp φ d βV xφ
where I is the emission current (A), A is the emission area (m2), V is the applied voltage, d is the distance between the CNT tips and the anode (m), φ is the work function (eV), and B is a constant (6.44 × 109, VeV-1.5m-1). When ln(I/ V2) is plotted versus 1/V, the slope of this linear formulation is given by -Bφ1.5d/β. Assuming that the work function is 5.0 eV25 the derived field enhancement factor is calculated to be 4540, which is sufficient for application in field emission displays. In summary, we have demonstrated an efficient method for manufacturing well-aligned open-ended CNTs. The openended structures are the key to the successful assembly of CNTs on substrates by a solder reflow process. This process is compatible with current microelectronics fabrication sequences and technology. The distinctive CNT-transfer technology features are separation of CNT growth and CNT device assembly. Overall, the advantages of CNT transfer technology are embodied in the low process temperature, adhesion improvement, and the feasibility of transferring CNTs to different substrates. Field emission testing of the as-assembled CNT devices indicates good field emission characteristics, with a field enhancement factor of 4540. The testing results are in agreement with the Fowler-Nordheim
Nano Lett., Vol. 6, No. 2, 2006
expression, which implies good electrical contact between the CNTs and the solder, and a very small voltage drop across CNT/solder interfaces. CNT transfer technology, enabled by open-ended CNTs, shows promising applications for positioning of CNTs on temperature-sensitive substrates, and for the fabrication of field emitters, electrical interconnects, and thermal management structures in microelectronics packaging. Acknowledgment. We thank the National Science Foundation (NSF) for funding support under contract DMI0422553, and we also thank Dr. Yong Ding for HRTEM examinations. References (1) Berber, S.; Kwon, Y. K.; Toma´nek, D. Phys. ReV. Lett. 2000, 84, 4613. (2) Ajayan, P. M. Chem. ReV. 1999, 99, 1787. (3) Graham, A. P.; Duesberg, G. S.; Hoenlein, W.; Kreupl, F.; Liebau, M.; Martin, R.; Bajasekharan, B.; Pamler, W.; Seidel, R.; Steinhoegl, W.; Unnger, E. Appl. Phys. A 2005, 80, 1141. (4) Kreupl, F.; Graham, A. P.; Duesberg, G. S.; Steinho¨gl, W.; Liebau, M.; Unger, E.; Ho¨nlein, W. Microelectron. Eng. 2002, 64, 399. (5) Srivastava, N.; Banerjee, K. JOM 2004, 56, 30. (6) Xu, H. Nat. Mater. 2005, 4, 649. (7) McEuen, P. L.; Fuhrer, M.; Park, H. IEEE T. Nanotechnol. 2002, 1, 78. (8) Dresselhaus, M. S.; Dresselhaus, G.; Avouris, P. Carbon Nanotubes; Springer: Berlin, 2001. (9) Fan, S. S.; Chapline, M. G.; Franklin, N. R.; Tombler, T. W.; Cassell, A. M.; Dai, H. J. Science 1999, 283, 512. (10) Ren, Z. F.; Huang, Z. P.; Xu, J. W.; Wang, J. H.; Bush, P.; Siegal, M. P.; Provencio, P. N. Science 1998, 282, 1105. (11) Wei, B. Q.; Vajtai, R.; Jung, Y.; Ward, J.; Zhang, R.; Ramanath, G.; Ajayan, P. M. Mater. Chem. 2003, 15, 1598. (12) Zhu, L.; Xu, J.; Xiu, Y.; Sun, Y.; Hess, D. W.; Wong, C. P. Carbon 2006, 44, 253. (13) Li, H. J.; Lu, W. G.; Li, J. J.; Bai, X. D.; Gu, C. Z. Phys. ReV. Lett. 2005, 95, 086601. (14) Dujardin, E.; Ebbesen, T. W.; Hiura, H.; Tanigaki, K. Science 1994, 265, 1850. (15) Ebbesen, T. W.; Ajayan, P. M. Nature 1992, 358, 220. (16) Ugarte, D.; Sto¨ckli, T.; Bonard, J. M.; Chaˆtelain, A.; de Heer, W. A. Appl. Phys. A 1998, 67, 101. (17) Ajayan, P. M.; Ebbesen, T. W.; Ichihashi, T.; Iijima, S.; Tanigaki, K.; Hiura, H. Nature 1993, 362, 522. (18) Tsang, S. C.; Harris, P. J. F.; Green, M. L. H. Nature 1993, 362, 520. (19) Zhu, L.; Xiu, Y.; Hess, D. W.; Wong, C. P. Nano Lett. 2005, 5, 2641. (20) Ajayan, P. M.; Iijima, S. Nature 1993, 361, 333. (21) Yu, W. J.; Cho, Y. S.; Choi, G. S.; Kim, D. Nanotechnology 2005, 16, S291. (22) Sohn, J. I.; Lee, S.; Song, Y. H.; Choi, S. Y.; Cho, K. I.; Nam, K. S. Appl. Phys. Lett. 2001, 78, 901. (23) de Jonge, N.; Allioux, M.; Doytcheva, M.; Kaiser, M.; Te, K. B. K.; Lacerda, R. G.; Milne, W. I. Appl. Phys. Lett. 2004, 85, 1607. (24) Bonard, J. M.; Klinke, C.; Dean, K. A.; Coll, B. F. Phys. ReV. B 2003, 67, 115406. (25) Lee, O. J.; Lee, K. H Appl. Phys. Lett. 2003, 82, 3770.
NL052183Z
247
