synthesis of aligned carbon nanotubes using thermal chemical...

29 October 1999

Ž .Chemical Physics Letters 312 1999 461–468www.elsevier.nlrlocatercplett

Synthesis of aligned carbon nanotubes using thermal chemicalvapor deposition

Cheol Jin Lee a, Dae Woon Kim a, Tae Jae Lee a, Young Chul Choi b,Young Soo Park b, Young Hee Lee b,c,), Won Bong Choi d, Nae Sung Lee d,

Gyeong-Su Park e, Jong Min Kim d

a Department of Electrical Engineering, Kunsan National UniÕersity, Kunsan 573-701, South Koreab Department of Semiconductor Science and Technology, and Semiconductor Physics Research Center, Chonbuk National UniÕersity,

Chonju 561-756, South Koreac Department of Physics, Chonbuk National UniÕersity, Chonju 561-756, South Korea

d Display Laboratory, Samsung AdÕanced Institute of Technology, P.O. Box 111, Suwon 440-600, South Koreae Analytical Engineering Laboratory, Samsung AdÕanced Institute of Technology, P.O. Box 111, Suwon 440-600, South Korea

Received 14 June 1999; in final form 2 September 1999

Abstract

Aligned carbon nanotubes have been synthesized on transition metal-coated silicon substrates with C H using thermal2 2

chemical vapor deposition. It was found that nanotubes can be mostly vertically aligned on a large area of plain Si substrateswhen the density of metal domains reaches a certain value. Pretreatment of Co–Ni alloy by HF dipping and etching withNH gas prior to the synthesis is crucial for vertical alignment. Steric hindrance between nanotubes at an initial stage of3

growth forces nanotubes to align vertically. Nanotubes are grown by a catalyst-cap growth mechanism. Applications to fieldemission displays are demonstrated with emission patterns. q 1999 Elsevier Science B.V. All rights reserved.

Ž .Since the discovery of carbon nanotubes CNTw x1 , synthesis of carbon nanotubes for mass produc-tion has been achieved by several methods such as

w x w xlaser vaporization 2 , arc discharge 3 , and pyroly-w xsis 4 . However, controllibility of diameters, lengths,

and chiralities has never been easily accessible withsuch approaches. In spite of considerable efforts for

w xCNTs to apply for electron field emitters 5–8 ,practical approaches are still limited by the intricatesample-preparation processes and large area synthe-sis. In particular, synthesis of vertically aligned na-

) Corresponding author. Fax: q82-652-270-3585; e-mail:[email protected]

notubes is of great importance for applications tofield emitters. Recently, vertically aligned CNTs havebeen synthesized on glass by plasma-enhanced

Ž . w xchemical vapor deposition PECVD 9 . Alignedw xCNTs can be also grown on mesoporous silica 10

w xand Fe-patterned porous silicon 11 using the chemi-Ž .cal vapor deposition CVD method. Despite such

breakthroughs in the synthesis, the growth mecha-nism of the alignment are still far from being clearlyunderstood. Furthermore, flat-panel emission dis-plays on a large area has not been clearly demon-strated yet at a practical level. Here, we have synthe-sized vertically aligned CNTs on a large area ofCo–Ni codeposited Si substrates by the thermal CVDusing C H gas. We demonstrate that CNTs can be2 2

0009-2614r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.Ž .PII: S0009-2614 99 01074-X

( )C.J. Lee et al.rChemical Physics Letters 312 1999 461–468462

mostly vertically aligned on catalyzed Si substratewhen the domain density reaches a certain value.Controlling the domain density of Co–Ni alloy byHF dipping and dry etching with NH gas prior to3

the deposition is of crucial importance. Steric hin-drance between nanotubes at an initial stage of thegrowth forces nanotubes to align vertically. Na-notubes are then further grown by the catalyst-capgrowth mechanism. We further demonstrate the fieldemission patterns and I–V characteristics obtainedfrom vertically aligned carbon nanotubes.

We used p-type silicon substrates with a resistiv-ity of 15 V-cm, since it is of importance to usesilicon substrates for future Si-based optoelectronic

Ž .devices. Co–Ni CorNi, 1:1.5 metal alloys with100 nm in thickness were thermally evaporated atroom temperature in a vacuum of 10y6 torr on

Ž .oxidized Si 100 substrates. These samples were fur-ther annealed at 4008C in Ar ambient for 20 min,which enhanced the adhesion of metal layers to theoxidized Si substrates. The annealed samples were

Ždipped for 100–200 s in HF solution HFrDI water,.1:12 and were then loaded on the quartz boat inside

the CVD quartz reactor of 60 mm in diameter. Inorder to prevent the oxidation of Co–Ni alloys, Argas was flowed into the quartz reactor during in-creasing the temperature to a desirable value. Sam-ples were pretreated using NH gas with a flow rate3

of 80 sccm for 5–20 min at 800–9008C. Then, CNTswere synthesized using C H gas with a flow rate of2 2

15–40 sccm for 10–20 min at the same temperatureas the NH pretreatment. The reactor was cooled3

down slowly to room temperature in Ar ambientafter the synthesis.

Fig. 1a–d shows scanning electron microscopeŽ . Ž .SEM JEOL JSM-6400, 20 kV images of CNTswith various magnifications. CNTs are well aligned

Ž 2 .over the large area 20=30 mm of the substrate,as shown in Fig. 1a. The surface morphologies ofCNTs are clean and uniform with the length of about5 mm and the diameter of about 200 nm. The whitebundles of CNTs shown in Fig. 1a are peeled off dueto scratches. Note that top of the aligned nanotubesis terminated by a transition-metal cap, the whitespots, as shown in Fig. 1b, which is similar to the

w xprevious reports 9,10 . Fig. 1c reveals high densityof nanotubes, where nanotubes are curly at the bot-tom but rather straight near the top of nanotubes. The

density of nanotubes is about 3=108 cmy2 , about100 times larger than the typical density of microtipsin conventional Spindt-type field emission arrays.Some nanotubes are kinky at the initial stage ofgrowth, as shown in Fig. 1d. We evaluated the wallstructures by the transmission electron microscopyŽ . Ž .TEM Hitachi H-9000NA, 300 kV , as shown inFig. 1e,f. Peeled nanotubes were dispersed on acopper carbon-microgrid. TEM images show longand straight multi-wall nanotube with an hollow

Ž .inside Fig. 1e . The lattice images are shown in Fig.1f. However, long-range lattice images are notformed due to relatively low growth temperature.This was also confirmed in our Raman scatteringmeasurement that a relatively large defective peak at1295 cmy1 was observed with the main graphiticG-peak at 1595 cmy1. Fig. 2 shows the Ramanspectrum of alinged carbon nanotubes using FT-Ra-

Ž .man spectrometer BRUKER RFS 100rS with theŽ .excitation laser of Nd:YAG wavelength: 1064 nm .

Besides two main peaks, several peaks at lowwavenumbers were also observed in our multiwallnanotubes, although these were usually observed in

w xsingle-wall nanotubes 12–14 .Nanotubes were reproducibly grown with the

above conditions. We have also tried various growthw xconditions 14 . CNTs were grown well independent

of growth temperatures within the temperature rangeof 800–9008C, since C H gases are highly decom-2 2

posed in this range by the catalytic assistance ofw xtransition metals 10 . Tube lengths increased with

increasing growth time. However, some carbona-ceous particles appeared with longer growth time.The growth rate was 30 mmrh, similar to that from

w xthe CNT growth on porous Si 10 . We note that theHF dipping and NH pretreatment are crucial steps3

to obtain high density of nucleation sites. WithoutNH pretreatment, CNTs were grown uniformly but3

not vertically aligned. Therefore, the NH pretreat-3

ment is a very critical step in vertically aligningcarbon nanotubes on Si substrates.

Fig. 3 shows the SEM images of surface morphol-ogy for Co–Ni alloy. The HF dipping etches metalsurface and increases the surface roughness withincreasing the HF dipping time, as shown in Fig.3a,b. The subsequent NH pretreatment further etches3

the surface and forms small domains inside the metalcluster, as shown in Fig. 3c, acting as nucleation

( )C.J. Lee et al.rChemical Physics Letters 312 1999 461–468 463

Fig. 1. SEM images of carbon nanotubes grown with gas flow rate of 40 sccm at 8508C for 10 min. Edges are vividly visualized by peel-offŽ . 2 Ž .with a razor. a A uniformly distributed morphology of vertically aligned carbon nanotubes in large area of 20=30 mm . b Magnified

Ž . Ž . Ž . Ž .top view. c Magnified tilted view of carbon nanotubes at the peeled edge. d Magnified view of c . e TEM image of a single carbonŽ .nanotube in higher resolution. f The lattice image of a nanotube.


Fig. 2. Raman spectrum of vertically aligned carbon nanotubes.

seeds for vertical alignment of nanotubes. The sizeof small domains within the metal cluster in Fig. 3cis about 200 nm, the same as the diameter of CNTs.This strongly suggests that carbon diffusion withinthe small domain is limited by the domain wall. Withthe absence of NH pretreatment, nanotubes were3

laid down with a low density. Both HF dipping andNH pretreatment were necessary to control the3

surface morphology to align nanotubes vertically incase of using the Co–Ni alloy.

Nanotubes grown on Ni-coated substrates had adiameter of about 100 nm, whereas diameters ofnanotubes grown on Co-coated substrates were about200 nm. However, nanotubes were not vertically

w xaligned in both cases 14 . In case of the Co–Nialloy, Ni particles have faster etching rate than Coparticles during dry NH etching. We found that Ni3

particles were etched away during dry etching ofNH , which resulted in the increase of the density of3

nucleation sites within Co particles. This was con-Ž .firmed by the energy-dispersive X-ray EDX spec-

Ž .trum Fig. 3d . The amount of Co left on the surfaceafter NH treatment was about six times larger than3

that of Ni. Note that the amount of Ni was 1.5 timeslarger than that of Co in as-deposited samples. When

the density of nucleation sites reaches a certainvalue, nanotubes grown in directions other than ver-tical direction are prohibited from growing due to thesteric hindrance from the adjacent nanotubes andthen the growth direction is changed to further growvertically. This can be confirmed from some kinkynanotubes at the early growth stage, as can be clearlyseen in Fig. 1d.

Fig. 4 illustrates our proposed growth model.C H molecules are adsorbed on the metal domains.2 2

As carbon atoms are further supplied, a carbon–metaleutectic alloy can be formed, decreasing the meltingtemperature of the alloy. Formation of carbon–metaleutectic alloy promotes the diffusion of carbon in themetal alloy, initiating carbon aggregations followedby the nanotube formation, as shown in Fig. 4a. Thecarbon diffusion is limited by the domain size withinthe metal particles and thus the diameter of na-notubes should not be larger than the domain size.As the nanotube grows further, part of the metaldomain is pushed upward, forming a metal cap, asshown in Fig. 4b. The metal cap saturates danglingbonds of the nanotube at the edge, stabilizing nan-otube edges. Presuming sufficient supply of carbonatoms and high density of domains within the metal


Ž . Ž . Ž .Fig. 3. SEM images of surface morphology for Co–Ni alloy. a No HF dipping, b HF dipping for 140 s, and c HF dipping for 140 sfollowed by NH pretreatment for 10 min.3

cluster, nanotubes are forced to align vertically, asshown in Fig. 4c. Some nanotubes at the edge of thecluster may not be aligned vertically at the initialstage but will be eventually aligned vertically by thesteric hindrance from other nanotubes for furthergrowth, as shown in Fig. 4c. As CNTs becomelonger, C H gases may not easily reach the bottom2 2

of the substrate due to the compact CNTs highlypopulated on the surface. Instead, they will have abetter chance to reach the top metal cap and then,adsorbed carbon atoms can diffuse into the edge ofnanotubes. Formation of complete hexagons are cata-lytically promoted by the assistance of the metal capw x15 , giving rise to the continuous cap growth. This

model is able to provide uniform height of nanotubesin large area and control the height with growth time.Uniform growth in height may not be achieved oncatalyst-patterned substrates with the model of base

w xgrowth suggested in another report 11 . Our modelcan also explain the existence of the metal caps atthe top of nanotubes after the growth, as observedfrom our SEM and TEM images and other reportsw x9,10 .

The field emission patterns were observed fromvertically aligned carbon nanotubes. Fig. 5 illustratesthe emission patterns from 1.4=1.6 cm2 in size

Ž . Ž . Ž .with bias conditionings for a 30, b 60 and c 90min. Fig. 5d shows typical I–V curve from Fig. 5c


Ž . Ž . Ž .Fig. 4. Schematic diagram of our growth model. a Formation of nucleation process, b cap growth, and c mechanism for verticalalignment at the initial stage.


Ž . Ž . Ž . Ž .Fig. 5. Emission patterns from 1.4=1.6 cm in size with bias conditionings for a 30 min, b 60 min, and c 90 min, and d typical I–VŽ . Ž .curve from c . Open squares in d show emission currents from patterned graphite powder-epoxy composite. The inset shows

Fowler–Nordheim plot from vertically aligned carbon nanotubes.

along with the corresponding Fowler–Nordheim plotŽ . Ž .inset . The anode, an ITO indium tin oxide -phos-phor-coated glass, was separated from the bottomlayer of nanotubes by a spacer in distance of 570

mm. The chamber was maintained at 1=10y6 Torrduring the I–V measurements. Although sampleswere clean, as seen in Fig. 1a, they were annealedinitially by repeatedly applying 100 and 1000 V in


order to burn unintentionally protruded nanotubes.This improves the emission patterns as shown in Fig.5a–c. The turn-on voltage was about 1.2 Vrmmwith a current of 10 nA, which is much smaller thanthe other reports obtained from patterned nanotube–

w xepoxy composite 8 and from single-wall nanotubesw xproduced by arc-discharge 16 . Turn-on voltage also

varies with the resistance of the transition metals.For comparison, we have also fabricated patternedgraphite powder–epoxy composite, which resulted inhigher turn-on voltage, as shown in Fig. 5d. In spiteof smaller effective emission area from CVD-grownsamples, the emission current was much larger thanthat of graphite. The current increases sharply afterturn-on voltage and saturates near 3 Vrmm. Theinset clearly shows that this field emission followsthe Fowler–Nordheim equation. The presence ofmetal particles at the top of nanotubes prohibitedefficient emission and some protruded nanotubesgave rise to non-uniform emission patterns. A keencontrol with bias annealings is required to have largeemission currents by removing metal caps and uni-form emission patterns over a large area. We empha-size that the present approaches are easily integrablewithin the current semiconductor process.

In summary, we have synthesized aligned carbonnanotubes on plain Si substrates using thermal CVDmethod. Vertically aligned nanotubes are uniformlysynthesized on a large area of Si substrates. A newmetal-cap growth mechanism is proposed based onthe experimental observations. This approach shouldbe easily applicable to a large area synthesis withoutloss of generality. We have also demonstrated thefield emission patterns from the aligned nanotubes.

Acknowledgements

We acknowledge the financial support by theKorea Science and Engineering Foundation through

the Semiconductor Physics Research Center at Chon-buk National University.

References

w x Ž .1 S. Iijima, Nature 354 1991 56.w x2 A. Thess, R. Lee, P. Nikolaev, H. Dai, P. Petit, J. Robert, C.

Xu, Y.H. Lee, S.G. Kim, D.T. Colbert, G. Scuseria, D.Ž .Tomanek, J.E. Fisher, R.E. Smalley, Science 273 1996 483.´

w x3 C. Journet, W.K. Maser, P. Bernier, A. Loiseau, M. Lamy dela Chapelle, S. Lefrant, P. Deniard, R. Lee, J.E. Fischer,

Ž .Nature 388 1997 756.w x4 M. Terrones, N. Grobert, J. Olivares, J.P. Zhang, H. Ter-

rones, K. Kordatos, W.K. Hsu, J.P. Hare, P.D. Townsend, K.Prassides, A.K. Cheetham, H.W. Kroto, D.R.M. Walton,

Ž .Nature 388 1997 52.w x Ž .5 W.A. de Heer, A. Chatelain, D. Ugarte, Science 270 1995

1179.w x Ž .6 P.G. Collins, A. Zettl, Phys. Rev. B 55 1997 9391.w x7 Y. Saito, K. Hamaguchi, K. Hata, K. Uchida, Y. Tasaka, F.

Ikazaki, M. Yumura, A. Kasuya, Y. Nishina, Nature 389Ž .1997 554.

w x8 Q.H. Wang, A.A. Setlur, J.M. Lauerhaasi, J.Y. Dai, E.W.Ž .Seelig, R.P. Chang, Appl. Phys. Lett. 72 1998 2912.

w x9 Z.F. Ren, Z.P. Huang, J.W. Xu, J.H. Wang, P. Bush, M.P.Ž .Siegal, P.N. Provencio, Science 282 1998 1105.

w x10 W.Z. Li, S.S. Xie, L.X. Qian, B.H. Chang, B.S. Zou, W.Y.Ž .Zhou, R.A. Zhao, G. Wang, Science 274 1996 1701.

w x11 S. Fan, M.G. Chapline, N.R. Franklin, T.W. Tombler, A.M.Ž .Casell, H. Dai, Science 283 1999 512.

w x12 J.M. Holden, P. Zhou, X. Bi, P.C. Eklund, S. Bandow, R.A.Jishi, K.D. Chowdhury, G. Dressehaus, M.S. Dresselhaus,

Ž .Chem. Phys. Lett. 220 1994 186.w x13 A.M. Rao, E. Richer, S. Bandow, B. Chase, P.C. Eklund,

K.A. Williams, S. Fang, K.R. Subbaswamy, M. Menon, A.Thess, R.E. Smally, G. Dresselhaus, M.S. Dresselhaus, Sci-

Ž .ence 275 1997 187.w x14 C.J. Lee, D.W. Kim, T.J. Lee, Y.C. Choi, Y.S. Park, W.S.

Kim, W.B. Choi, N.S. Lee, J.M. Kim, Y.G. Choi, S.C. Yu,Ž .Y.H. Lee, Appl. Phys. Lett. 75 1999 1721.

w x15 Y.H. Lee, S.G. Kim, D. Tomanek, Phys. Rev. Lett. 78´Ž .1997 2393.

w x16 J.M. Bonard, J.P. Salvetat, T. Stockli, W.A. de Heer, L.Ž .Forro, A. Chatelain, Appl. Phys. Lett. 73 1998 2113.

synthesis of aligned carbon nanotubes using thermal chemical...

Documents