Chih-Ming Hsu, Hong-Jen Lai, and Cheng-Tzu Kuo
Citation: Journal of Vacuum Science & Technology A 22, 1461 (2004); doi: 10.1116/1.1735908
View online: http://dx.doi.org/10.1116/1.1735908
View Table of Contents: http://scitation.aip.org/content/avs/journal/jvsta/22/4?ver=pdfcov
Published by the AVS: Science & Technology of Materials, Interfaces, and Processing
Articles you may be interested in
Field emission of carbon nanotubes on anodic aluminum oxide template with controlled tube density
Appl. Phys. Lett. 86, 123111 (2005); 10.1063/1.1886260
Growth of multiwalled-carbon nanotubes using vertically aligned carbon nanofibers as templates/scaffolds and improved field-emission properties
Appl. Phys. Lett. 86, 053110 (2005); 10.1063/1.1852730
Ink-jet printing of nanoparticle catalyst for site-selective carbon nanotube growth
Appl. Phys. Lett. 82, 811 (2003); 10.1063/1.1540726
Effect of catalyst on growth behavior of carbon nanotubes synthesized by microwave heating thermal chemical vapor deposition process
J. Vac. Sci. Technol. B 21, 400 (2003); 10.1116/1.1531150
Nucleation and growth of carbon nanotubes by microwave plasma chemical vapor deposition
Industrial Technology Research Institute, Materials Research Laboratories, 195 Sec. 4, Chung Hsing Road, Chutung, Hsinchu 310, Taiwan
Cheng-Tzu Kuo
National Chiao Tung University, Department of Materials Science and Engineering, 1001 Ta-Hsueh Road, Hsinchu 300, Taiwan
共Received 21 October 2003; accepted 8 March 2004; published 20 July 2004兲
The effect of the interfacial reaction of Co catalyst with a Si substrate on growth of carbon nanotubes 共CNTs兲 was investigated. Well-aligned multiwall CNTs 共MWCNTs兲 were synthesized and applied barrier layers by microwave plasma-enhanced chemical vapor deposition共MPECVD兲. Growth proceeded in a flowing mixture of H2, CH4, and N2 as precursors at a temperature of
600 °C and a ⫺200 V substrate bias; Co was sputtered as the catalytic material. Transmission electron microscopy共TEM兲 and x-ray diffraction were employed to examine the growth behavior of CNTs on Si 共100兲 substrates on which Co had been deposited by MPECVD. The TEM results indicate that discrete conical CoSi2 layers with兵111其 and 共100兲 faceted interfaces were formed on
a Si共100兲 substrate during CNTs growth. Direct evidence that the growth is by tip growth and base growth is presented. The results show that well-aligned CNTs exhibit a significant emission current. The field emission characteristics of CNTs are contributed to the relationship between the application of different barrier layers and the growth mode of CNTs. © 2004 American Vacuum
Society. 关DOI: 10.1116/1.1735908兴
I. INTRODUCTION
The identification of the structure of carbon nanotubes 共CNTs兲 by Iijima in 19911 was a scientific milestone. CNTs
have attracted intense interest in the field of nanotechnology because of their unique properties and wide range of appli-cations, such as nanoelectronics,2,3 scanning probes, and4 field emission displays 共FEDs兲.5,6 The selective growth of CNTs on Si wafers, catalyzed by Fe, Co, Ni, or other ele-ments, is important in the CNTs deposition. However, the metal silicide produced at high temperature in the CVD growth7–9and transition-metal silicides are crucial materials for the tremendous success of Si integrated circuits; titanium and cobalt silicide are the two more widely used self-aligned silicides.10 For this reason, CNTs have potential for use in interconnects in electronic devices, contact windows of sili-cide and field emitters. However, strong adhesion between CNTs, electron-emitting layers and substrates were required for fabricating practical cold cathodes in applying electron emitting devices. The interaction between metal catalysts and substrate, in growth processes of CNTs presented by several groups,7–9can ensure excellent adhesion and field emission properties. This study considers the interfacial reactions among CNTs, cobalt, and the Si substrate. The relationship between interfacial reactions and field emission characteris-tics on the growth of CNTs is also investigated.
II. EXPERIMENT
CNTs were synthesized on Co 共75 nm兲-coated Si 共100兲 wafers using microwave plasma-enhanced chemical vapor deposition 共MPECVD兲, in which the Co film was generated by physical vapor deposition. The synthetic process as fol-lows:共1兲 A 500 W H2plasma pretreatment for Co film for 10
min at 9 Torr, formed well-distributed nanosized catalysts; 共2兲 CH4/N2 process gases were introduced in a ratio of 10/
100 sccm, at a deposition pressure of 16 Torr and a micro-wave power of 800 W, for 10 min. Interfacial reaction be-tween CNTs and Si substrate were obtained by x-ray diffraction 共XRD兲 to identify crystal phases. Scanning elec-tron microscopy共SEM兲 was used to characterize the growth morphologies of the CNTs, and the cross sections and micro-structures of the CNTs were observed by transmission elec-tron microscopy 共TEM兲; the depth profiles of the catalytic film were obtained by Auger electron spectra 共AES兲. Field emission from the arrays of CNTs and the relationships be-tween the buffer layers were measured in a vacuum chamber equipped with electrical stepper and applied a high voltage between the sample and anode. The typical spacing from the tip of the CNT arrays to the anode is 100 m. All of the current measurements were taken in a vacuum at a pressure of 10⫺6 Torr with an electrometer 共Keithley 237兲 and re-corded by a personal computer.
III. RESULTS AND DISCUSSION
Aligned CNTs were successfully synthesized on a Co-deposited Si substrate. Figure 1共a兲 shows the typical SEM morphology of CNTs. The figure indicates that aligned but a兲Author to whom correspondence should be addressed; electronic mail:
tangled CNTs cover the substrate. The length of the CNTs arrays increases with growth time, reaching 10m after 10 min. A highly magnified SEM image, displayed in Fig. 1共b兲, depicts CNTs, growing separately and perpendicularly out of the substrate in bundles of CNTs. Observation of the inter-face between the CNTs and the Si substrate reveals a cross section of the specimen prepared using a diamond saw and polishing. Peeled out CNTs with a collapsed bottom-end are observed in this specimen, shown in Fig. 2共a兲. CNTs grown by this method should have defects, resulting in crooked and entangled shapes. The SEM image关Fig. 2共b兲兴 of the bottom end of the stripped-out CNT arrays includes more CNTs with distorted and collapsed bottom-ends, corresponding to the continued presence of the cone-shaped Co particles on the Si substrate, as presented in Fig. 2共c兲. The cone-shaped Co par-ticles have been examined by XRD and AES共Figs. 3 and 4兲. Well-distributed conic particles 共70–200 nm兲 were embed-ded in the Si substrate. Figure 2共d兲 illustrates the ends of CNTs without catalytic particles, after they had been re-moved from Si substrate. Consequently, Figs. 2共a兲–2共d兲
shown a typical base growth mode. The growth mechanism of CNTs is very similar to that suggested by other research groups,11in whose work graphite layers generated by CVD on catalytic particles were transformed into nanotubes that enclosed those the catalytic particles.12Various methods, in-cluding spin coating of a metal precursor and the sputtering of metal film have been used to derive a uniform distribution of catalytic particles to grow CNTs on a Si substrate. Gener-ally, the deposited metal film is pretreated by H2 or NH3
plasma to remove oxide from the surface, forming clear and small metallic particles examined by plasma treatment. Figure 3 plots the XRD patterns of the Co-deposited Si sub-strate after being treated with H2 plasma. The figure shows
that CoSi2 forms instead of CoSi formation. The latter is an
intermetallic compound formed during annealing at high temperature (⬎600 °C) in the Co/Si system. Cobalt silicide has recently emerged as a preferred choice for use as a con-tact and interconnect material in microelectronics, because it has low resistivity 共10–15 ⍀兲 and excellent chemical sta-bility. Moreover, in the upper spectrum of Fig. 3, Co metal is
FIG. 1. SEM morphologies of CNTs grown applied a SiO2barrier layer:共a兲
well aligned CNTs cover the substrate;
共b兲 a highly magnified SEM image of
the CNTs arrays increases the length with growth time, reaching 10m af-ter 10 min.
FIG. 2. SEM morphologies of CNTs grown without a SiO2barrier layer:共a兲
well aligned CNTs perpendicular to the substrate over a large area;共b兲 bot-tom of CNTs separated from Si and catalytic particles; 共c兲 ends of CNTs without catalytic particles, removed from the Si substrate, and共d兲 cone-shaped catalytic particle embedded in the Si substrate.
found after CNTs are grown, and in the growth of CNTs, the discrete CoSi2 layers seem to grow into the Si substrate
be-low the continuous CoSi2 layer, reducing the concentration
of Co, according to a depth of the AES profile plotted in Fig. 4. Figure 5共a兲 presents bamboo-like multiwalled CNTs 共MWCNTs兲 microstructures, which are similar to those ob-served by other research groups.12 The diameter of the MWCNTs is around 90 nm, and the root and tips of CNTs are opened and closed, respectively. Catalytic particles are fixed on the substrate; implying that Co particles easily form the silicide.13–15Aligned CNTs, as presented in Fig. 6共a兲 are the glue used in preparing the cross-section TEM specimen. The triangular cobalt silicide (CoSix) is formed by the
reac-tion between cobalt particles and Si during the growth pro-cess. Moreover, Fig. 6共a兲 clearly displays the interfaces be-tween Co particles and the Si substrate. The Co diffuses into the Si substrate to form cobalt silicide with a facet interface. This reaction is believed to explain the strengthening of the adhesion between the Co catalyst and the substrate by sili-cide formation which would explain why the metal catalysts
adhere strongly to the substrate during the growth of CNTs. The inset in Fig. 6共a兲 is highly magnified; it clearly presents Co particles and CoSix. The interfacial reaction between
CNTs/Co and the Si substrate were examined, showing base-growth mode. Figures 6共b兲 and 6共c兲 show selected area dif-fraction 共SAD兲 patterns and the numbers refer to different areas of diffraction, as shown in the inset in Fig. 6共a兲. Figure 6共b兲 presents the SAD (zone⫽关01-1兴) pattern of Co cata-lytic particles together with CNTs, which indicates that the diffuse 共002兲 diffraction spots of CNT are all in directions
FIG. 3. XRD patterns of the base-growth sample under various conditions;
共a兲 flowing H2plasma pretreatment, and共b兲 after growth of CNTs.
FIG. 4. AES depth profile of base-growth sample after growth of CNTs.
FIG. 5. TEM morphologies of no evident catalytic particles in CNTs ends.
FIG. 6. Cross-sectional TEM images of共a兲 interfacial region of CNTs/Co/Si.
Bamboo-shaped CNTs were clearly grown from a Co nanoparticle;共b兲 SAD pattern around the CNTs/Co interface, and共c兲 SAD pattern around the tri-angular CoSi2/Si interface.
between Co 共200兲 and Co 共1-1-1兲. Figure 6共c兲 shows the SAD pattern of triangular CoSi2with the Si substrate and the
zone axis 关01-1兴. The diffracted spots of 共200兲 from CoSi2
and Si coincide with each other, implying that a fully coher-ent relationships 共A-type interface兲 exist between the orien-tation of CoSi2 and that of Si.16Clearly, a triangular CoSi2 layer with兵111其 and 共100兲 facets is locally formed in the Si substrate. The nucleated CoSi2 may be grown on a Si sub-strate with both 兵111其 and 共100兲 planes at the CoSi2/Si in-terface because the兵111其 interfacial energy of CoSi2is lower
than the 共100兲 interfacial energy. However, the precise be-havior associated with epitaxial growth from amorphous car-bonic cobalt is not yet fully understood, and warrants further study with reference to the potential applications in ultralarge scale integration.
A high-resolution TEM 共HRTEM兲 micrograph along the 具100典 zone axis is obtained to assess the coherency of the CNTs, CoSi2, and Si layer. Figure 7共a兲 presents a lattice
structure image. An epitaxial共100兲 CoSi2 layer with a small
兵111其-faceted interface is formed on the 共100兲 substrate with a sharp silicide–silicon interface. Figure 7共b兲 shows a sharp silicide–silicon interface, clearly demonstrating the presence of an epitaxial layer. Several studies have found that the close lattice match between the CoSi2 and Si crystal
matri-ces, which exhibits only a ⬃1.2% lattice mismatch, enables the epitaxial growth of CoSi2 on Si 共100兲.
17
The epitaxial CoSi2 on Si兵111其 exhibits coherence. Figure 7共c兲 presents a
lattice image of Co particles and CNTs, and the inset in Fig. 7共c兲 shows the fast Fourier transform image of a 共002兲 graph-ite layer of CNTs. Clearly, the graphgraph-ite lattice grows from the
Co particles. All the evidencesupports the complete matching of the lattice across the interface, implying that the CoSi2
layer is fully coherent with the Si substrate. Therefore, a CoSi2 layer can be said to be epitaxially grown on a共100兲 Si
substrate during the growth of CNTs, without an intermedi-ate layer.
The catalysts usually remain at the tips of CNTs grown using plasma-enhanced CVD. However, in this work, the base-growth mechanism was observed and CoSi2 was found
to pin the CNTs into the Si substrate. This mechanism is supposed to be able to solve the conductive and adhesive problems in relation of the CNT–FED, in which the CNTs are easily pulled from the cathode toward the anode when applied high current. Figure 8 plots the emission current den-sity curves of these base-grown CNT arrays, and compares them to the curves obtained from another tip-grown sample; the CNTs with diameters of 5 nm are well-aligned on the Si substrate and the Co particles are present at the tips of CNTs. Further details of the tip-grown CNTs array will be discussed in another article.18The turn-on voltage is 2.2 V/m关linear limit of the Fowler–Nordheim 共F–N兲 curve兴 and the maxi-mum emission current density is 40 mA/cm2 at 4.9 V/um, results that are sufficient for FED. The inset in Fig. 8 shows a good linear fit, implying that the emission current of both samples exhibits F–N behavior but with different slopes. Ac-cording to the F–N equation, J⬀(E2/)exp(⫺B3/2/E)
with B⫽6.83⫻109 (VeVm⫺1), the field enhancement factor 共兲 can be calculated from the slope of F–N curve, provided that the work function of the carbon nanotube CNTs is 5 eV, the same as for graphite. In this study, the field enhancement factor 共兲 of this bottom-growth carbon nanotube is 1530, which is much higher than that obtained from the tip-growth CNTs array (⫽800) obtained from Fig. 8. Some factors, such as the diameters of the CNTs and the work function of the prepared emitter used for field emission effect, implying that the Co catalyst at the tips of CNTs, although having a small diameter. According to numerous reports, the metal tip
FIG. 7. HRTEM images of 共a兲 interfacial region of Co/CoSi2/Si and共b兲
interfacial region of CoSi2/Si; 共c兲 HRTEM lattice image of the interface
region, showing MWCNTs grown from a Co nanoparticle.
FIG. 8. Typical field emission J – E curves of the vertically aligned tip-grown CNTs and of base-tip-grown CNTs.
IV. CONCLUSIONS
Aligned CNTs are grown by MPECVD in a Si substrate on which Co had been deposited. Interdiffusion between Co and the Si substrate strongly affect the growth characteristics and the F–E properties of CNT arrays. Cobalt silicides are shown to be formed with a full coherent interface with the Si substrate. The presence of the conical Co particles is also shown to be critical in growing aligned CNTs; the base-growth mode is demonstrated for the formation of CNTs in this work. Furthermore, the presence of the catalytic particles at the tips of the CNTs is detrimental to field emission appli-cations. The CNTs onto whose CoSi2 is pinned on Si
sub-strate exhibit favorable emission characteristics; the F–N plots show highly enhanced field factors, implying that such an array of CNTs are important in FED.
ACKNOWLEDGMENTS
The authors would like to thank the Industrial Technology Research Institute 共Contract Nos. A321XS9410 and A311XS31E1兲, the National Science Council 共Contract Nos. NSC90-2216-E-009-034, NSC90-2216-E-009-035, and NSC90-2216-E-009-040兲 and the Ministry of Education of Taiwan 共Contract No. MOE89-E-FA06-1-4兲 for financially
Univerity is appreciated for preparing the HRTEM specimens.
1
S. Iijima, Nature共London兲 354, 56 共1991兲.
2L. Kouwenhoven, Science 275, 1896共1997兲.
3M. Bockrath, D. H. Cobden, P. L. McEuen, N. G. Chopra, A. Zettl, A.
Thess, and R. E. Smalley, Science 275, 1922共1997兲.
4
J. H. Hafner, C. L. Cheung, and C. M. Lieber, Nature共London兲 398, 761
共1999兲.
5A. G. Rinzler, J. H. Hafner, P. Nikolaev, L. Lou, S. G. Kim, D. Tomanek,
D. Colbert, and R. E. Smalley, Science 269, 1550共1995兲.
6W. B. Choi et al., Appl. Phys. Lett. 75, 3129共1999兲. 7
T. Arcos, F. Vonau, M. G. Garnier, V. Thommen, H.-G. Boyen, and P. Oelhafen, Appl. Phys. Lett. 80, 2383共2002兲.
8L. P. Biro et al., Appl. Phys. Lett. 76, 706共2000兲.
9M. Chhowalla, K. B. K. Teo, C. Ducati, N. L. Rupesinghe, and G. A. J.
Amaratunga, J. Appl. Phys. 90, 5308共2001兲.
10
H. S. Rhee, T. W. Jang, and B. T. Ahn, Appl. Phys. Lett. 74, 1003共1999兲.
11Z. F. Ren, Z. P. Huang, J. W. Xu, J. H. Wang, P. Bush, M. P. Siegal, and
P. N. Provencio, Science 282, 1105共1998兲.
12C. J. Lee and J. Park, Appl. Phys. Lett. 77, 3397共2000兲. 13
H. S. Rhee, B. T. Anh, and D. K. Sohn, J. Appl. Phys. 86, 3452共1999兲.
14Y. Roichman, A. Berner, R. Brener, C. Cytermann, D. Shilo, E.
Zolotoy-abko, and M. Eizenberg, J. Appl. Phys. 87, 3306共2000兲.
15G. B. Kim, H. K. Baik, and S. M. Lee, Appl. Phys. Lett. 69, 3498共1996兲. 16
A. Alberti and E. Rimini, Appl. Phys. Lett. 75, 2924共1999兲.
17
N. G. Einspruch and G. B. Larrabee, VLSI Electronic: Microstructure
Science共Academic, New York, 1983兲, Vol. 6, p. 330.
18C. M. Hsu, C. H. Lin, H. J. Lai, and C. T. Kuo, Thin Solid Films
