Characterization of single-walled carbon nanotubes ... - Duke University
Full Paper Received: 10 October 2009
Revised: 15 November 2009
Accepted: 22 November 2009
Published online in Wiley Interscience:
(www.interscience.com) DOI 10.1002/aoc.1610
Characterization of single-walled carbon nanotubes synthesized using iron and cobalt nanoparticles derived from self-assembled diblock copolymer micelles Qiang Fua , Luke Reeda, Jie Liub and Jennifer Lua∗ We present a comparative study of single-walled carbon nanotubes grown using iron and cobalt nanoparticles as catalysts via the chemical vapor deposition approach. Monodispersed iron and cobalt oxide nanoparticles with an average size of 2 nm were prepared using a polystyrene-b-poly (4-vinylpyridine) diblock copolymermicelle template. The 2 nm iron oxide nanoparticles generated single-walled carbon nanotubes with an average diameter of 1.5 nm while 2 nm cobalt oxide nanoparticles produced single-walled carbon nanotubes with an average diameter of 1.0 nm. To achieve high growth yield using iron nanoparticles as catalyst, higher carbon feed rate is required. These findings demonstrate the importance of the synergic interaction between catalyst and carbon precursor in single-walled carbon nanotube formation. It also elucidates the important role of catalyst c 2010 John Wiley & Sons, Ltd. chemical composition on carbon nanotube properties. Copyright Keywords: self-assembly; nanocatalysts; diblock copolymers; carbon nanotubes
Introduction Single-walled carbon nanotubes (SWCNTs) have attracted a great deal of mention due to their remarkable properties.[1] The chemical vapor deposition technique (CVD) has become the primary method for substrate-based applications, because of its scalability, versatility and unparalleled control over growth processes, such as nanotube locations and orientations.[2 – 4] In a CVD process, nanoparticles are generally used as nanocatalysts for carbon nanotube (CNT) growth and thus their properties will be largely influenced by these nanocatalysts. This is manifested by many experimental results suggesting that there is a close correlation between the CNT diameter and the size of catalyst nanoparticle.[5 – 7] Many single metallic and bimetallic nanoparticles, such as iron,[5 – 9,10b] cobalt,[9,10b] gold,[10a] copper[11] and nickel,[9a,10b] have been used as Nanocatalysts to decompose carbon vapor precursors. Among them, iron and cobalt nanoparticles have been regarded as two of the most effective catalyst systems due to their ability to produce relatively high yield CNTs with wide process window. It is known that a catalyst nanoparticle not only serves as a heterogeneous nucleation site but also interacts with source precursors to facilitate decomposition. Carbon has different solubilities and compound formation tendencies with iron and cobalt.[12] Such difference in affinity with carbon can result in different catalytic activities. To examine the effect of composition on CNT growth, substrates with identical size and density of iron and cobalt nanoparticles were prepared to eliminate the influence of catalyst size and density effect on CNT growth. It was found that iron and cobalt nanoparticles require different carbon feed rates to achieve the highest yield of CNTs. The finding was verified using either patterned or non-patterned nanoparticle arrays. Iron nanoparticles need higher carbon concentation to initiate CNT
Appl. Organometal. Chem. (2010)
growth than cobalt nanoparticles. Furthermore, 2 nm iron oxide nanoparticles generated SWCNTs with an average diameter of 1.5 nm while 2 nm cobalt oxide nanoparticles produced SWCNTs with an average diameter of 1.0 nm. These results demonstrate that the synergistic interaction between catalyst and carbon precursor is important for achieving high-yield growth. They also elucidate the critical role of catalyst chemistry in CNT properties.
Experimental Nanoparticle Preparation Polystyrene (53400)-b-poly(4-vinylpyridine) (8800) (PS508 -bP4VP84 ) (Polymer Source, Quebec, Canada) was dissolved in toluene to form a 0.25 wt% solution. CoCl2 ·6H2 O or FeCl2 ·4H2 O (Sigma-Aldrich, St Louis, MO, USA) was then added to PS508 -bP4VP84 solutions to give a molar ratio of 0.25 between the metal ions and pyridine groups. The solutions were stirred overnight at room temperature to fully sequester metal species. Nanoparticles were formed after spin-coating the metal-modified polymer solutions at 4000 rpm/s on silicon substrates with a 100 nm thermally grown oxide layer followed by oxygen plasma to remove the polymer template. To pattern the nanoparticles, a bilayer lift-off process reported previously was used.[13] Briefly, a photoresist film with 60 nm
∗
Correspondence to: Jennifer Lu, School of Engineering, University of California, Merced, 5200 North Lake Drive, Merced, CA 95348, USA. E-mail: [email protected]
a School of Engineering, University of California, Merced, Merced, CA 95348, USA b Department of Chemistry, Duke University, Durham, NC 27708, USA
c 2010 John Wiley & Sons, Ltd. Copyright
Q. Fu et al. LOL1000 and 800 nm OCG 825 was spun onto a thermal-oxidecoated silicon wafer. After the photoresist patterning, the metal modified PS508 -b-P4VP84 micelle solutions were deposited by spin-coating. A lift-off process was carried out followed by oxygen plasma to give rise to patterned nanoparticle arrays. To keep the number of catalyst nanoparticles similar between patterned and non-patterned substrates under the same growth condition, the sizes of substrates were adjusted purposely.
A
B
Chemical Vapor Deposition of SWCNTs The catalyst-loaded substrates were put into a 1-inch quartz tube and transferred into a furnace. The substrates were first annealed at 700 ◦ C in air for 10 min. After cooling to room temperature, the substrates were reheated to 900 ◦ C under a 100 sccm H2 flow. Once the furnace temperature reached 900 ◦ C, a flow of carbon precursors, which was a mixture of methane and ethylene, was introduced to grow CNTs. The methane flow rate was set to be 750, 800 and 850 sccm with the ethylene flow rate varied from 0 to 30 sccm. After 10 mins, the growth was stopped by switching off the carbon precursor flow. Lastly, the furnace was cooled down to room temperature under the protection of H2 . Nanoparticle and SWCNT Characterization The nanoparticles and SWCNTs were characterized with a Nanoscope IIIa Atomic Force Microscope (AFM, Vecco, Plainview, USA). Scanning electron microscopy (SEM) characterization was carried out on a XL30 scanning electron microscope (FEI Company, Oregon, USA) with an acceleration voltage of 1 kV. Raman spectra were measured on a 1 µm2 illumination area with a 632.8 nm and 1 mW laser with a Renishaw InVia Raman microscope.
Results and Discussion Nanoparticles with Similar Sizes and Densities Although nanoparticles can be produced with controlled size and composition using the colloid approach,[14,15] there is no means of controlling spacing and order between nanoparticles on a surface. Self-assembled block copolymers have been widely employed to create periodically ordered nanoparticle arrays on various substrates. Iron, cobalt, nickel and gold nanoparticles prepared with the block copolymers have been successfully used for the synthesis of SWCNTs.[7 – 10,13,16,17] It has been demonstrated that metal nanoparticle arrays with adjustable spacing, size and composition can be prepared using a solution micelle template formed by polystyrene-b-poly(vinylpyridine). These nanoparticles are excellent catalysts enabling the growth of high-quality and low-defect SWCNTs.[8 – 10,16,17] In this set of experiment, PS508 -b-P4VP84 , which consists of two immiscible blocks, a long hydrophobic PS block and a short hydrophilic P4VP block, was used. In a selective solvent such as toluene, PS508 -b-P4VP84 molecules self-assemble into micelles. In toluene, P4VP blocks are self-assembled to form the core while the corona consists of PS blocks. Because of the strong complexation tendency of 4-vinylpyridyl groups with transition metals, iron and cobalt species can be selectively sequestered into the cores. By spin-coating, the metal loaded solution micelles can be transferred and form a monolayer of hexagonally ordered surface micelles on substrates. After the removal of the polymer template, nanoparticles formed on surfaces. Figure 1 is a set of
www.interscience.wiley.com/journal/aoc
Figure 1. AFM images of nanoparticles produced using PS508 -b-PVP84 diblock copolymer. (A) Iron oxide nanoparticles. (B) Cobalt oxide nanoparticles (image area 1 × 1 µm2 ; height scale 10 nm).
AFM images showing that highly ordered nanoparticle arrays have been generated. According to the AFM height analysis of roughly 200 nanoparticles, the average size was 2 nm with a minimum size of 1.5 nm and a maximum size of 2.2 nm for as-synthesized iron nanoparticles, while the average size was 1.9 nm and ranged from 1.4 to 2.2 nm for cobalt nanoparticles. By exploiting the block copolymer template approach, nanoparticles with similar sizes and densities can be prepared. Therefore the effects of nanoparticle size and density on growth can be eliminated. Direct comparison of different catalyst compositions on CNT growth characteristics can be carried out. CNT Growth Results Both iron and cobalt nanoparticles enable the growth of SWCNTs. By systematically varying the ratio of methane and ethylene gases, the growth condition for achieving the highest CNT yield in a methane and ethylene mixture was 800 sccm of methane and 20 sccm of ethylene for iron nanoparticles and 800 sccm of methane only for cobalt nanoparticles. Figure 2(A, B) shows SEM images of SWCNT networks grown using cobalt and iron nanoparticles under the growth conditions which afford the highest yield. Because the catalyst nanoparticles are well dispersed on the substrates, the SWCNT networks have uniform density across entire surfaces. The iron nanoparticles produced short and straight tubes while cobalt nanoparticles gave long and curvy tubes. Figure 2(C, D) shows the representative Raman spectrum of CNTs synthesized using iron and cobalt nanoparticles respectively. The miniscule D band and narrow and high intensity of G band indicate that the CNTs synthesized from cobalt and iron nanoparticles have high quality with very limited number of defects. Figure 3 is a set of histogram of CNT diameters according to AFM height analysis showing that the diameters of CNTs synthesized from cobalt nanoparticles are from 0.6 to 1.5 nm with an average of 1 nm, and those from iron nanoparticles are from 0.9 to 2.5 nm with an average of 1.5 nm. This result was corroborated by Raman spectroscopy analysis that showed that, comparing CNT diameters, iron nanoparticles result in CNTs with larger diameters. The representative Raman spectra are displayed as inserts in Fig. 3, the majority of CNTs grown from iron nanoparticles oscillating around 170 cm−1 with CNTs grown from iron nanoparticles resonating at higher frequency around 210 cm−1 . Single-walled CNTs produced from iron and cobalt nanoparticles are different in several respects. First, the growth condition for the highest yield is different for iron and cobalt nanoparticles. Using iron nanoparticles as catalysts, the highest yield growth
c 2010 John Wiley & Sons, Ltd. Copyright
Appl. Organometal. Chem. (2010)
Characterization of single-walled carbon nanotubes
A
B
C
D
Figure 2. SEM image of SWCNTs grown using (A) cobalt nanoparticles, methane = 800 sccm and (B) iron nanoparticles, methane = 800 sccm, ethylene = 20 sccm (B) (scale bar: 5 µm). A representative Raman analysis of CNTs synthesized using cobalt nanoparticles and (C) iron nanoparticles (D).
B
A 25
25 Frequency
20 Frequency
30
15 10 5
20 15 10 5
0 0
1
2
3
4
0
0
Diameter (nm)
1
2
3
4
Diameter (nm)
Figure 3. AFM height analysis results; inserts are the representative Raman spectrum indicating the CNT radially resonating near 210 and 170 cm−1 using (A) cobalt nanoparticles and (B) iron nanoparticles.
condition is 800 sccm methane and 20 sccm ethylene. Similar conditions were also reported by other research groups.[18] Using cobalt nanoparticles, the growth condition which offers the highest yield uses only methane at 800 sccm flow rate. Since ethylene is more easily to be decomposed, this result indicates that higher carbon concentration is required for iron nanocatalysts. Furthermore, the diameter and length of SWCNTs are different. The iron nanoparticles produced short and straight SWCNTs with diameter around 1.5 nm, while the cobalt nanoparticles resulted in long and curvy SWCNTs with average diameter around 1 nm.
Appl. Organometal. Chem. (2010)
These observations can be rationalized by considering carbon solubility and carbide formation capability of catalyst nanoparticles. It is generally believed that CNT growth starts with hydrocarbon decomposition by catalyst nanoparticles at elevated temperature. The disassociated carbon species dissolve in the catalyst nanoparticles. As more hydrocarbons are decomposed, the carbon concentration increases. Oversatuation of carbon leads to CNT growth. According to phase diagrams, carbon has a relatively high solubility in iron. As a result, a high carbon concentration is required to initiate SWCNT growth. With the increase in carbon concentration, Fe3 C is formed prior to carbon precipitation,
c 2010 John Wiley & Sons, Ltd. Copyright
www.interscience.wiley.com/journal/aoc
Q. Fu et al.
A
Acknowledgments
B
This work is supported by a start-up fund at the University of California, Merced and a NSF-CBET fund. We also would like to acknowledge the help of the Agilent technologies for Raman analysis.
References
Figure 4. SEM images of SWCNTs grown from catalyst nanoparticle patterns: (A) cobalt nanoparticles; (B) iron nanoparticles (scale bar: 20 µm).
which may slow down carbon diffusion, leading to lower growth rate.[19,20] This is consistent with our observation of shorter SWCNTs and higher carbon feed rate for iron nanoparticles. In comparison, carbon has lower solubility in cobalt, and does not form stable compounds with cobalt.[21] Therefore, oversaturation will lead to carbon directly precipitating out. Consequently, SWCNT nucleation and growth can take place at a higher speed with lower carbon feed rate. The experimental observation that cobalt nanoparticles require lower carbon concentration, which can be achieved using highly stable methane as a carbon precursor, is supported by this argument. Maintaining similar metal nanoparticle population between substrates with or without patterning and using the same growth condition for the highest growth yield, patterned CNTs were generated. Figure 4 shows the SEM image of SWCNTs grown from patterned cobalt and iron nanoparticle arrays on a SiO2 substrate. Cobalt nanoparticles resulted in long and small diameter SWCNTs, which are particularly important for many applications. Since nanoparticles synthesized by this method do not agglomerate, CNT growth yield can be calculated.[9a] The growth yield is defined by the number of tubes vs. the number of nanoparticles. Approximately 4% of cobalt nanoparticles are able to initiate carbon nanotube growth while 1% of iron nanoparticles are active to initiate growth.
Conclusions SWCNTs synthesized using iron and cobalt nanocatalysts exhibit different properties. Using 2 nm iron nanoparticles, short and straight SWCNTs with an average diameter of 1.5 nm were synthesized. On the other hand, 2 nm cobalt nanoparticles generated long and wavy SWCNTs with a diameter around 1.0 nm. The optimized growth condition for achieving maximum CNT yield for iron nanoparticles is a mixture of methane and ethylene, with higher carbon concentration, while for cobalt nanoparticles it is methane alone, with lower carbon concentration. This is due to the different tendency in terms of carbon solubility and carbide formation ability. These results highlight the importance of synergetic interaction between nanocatalysts and carbon feedstock for SWCNT growth. They reveal the important role of catalyst chemical composition in CNT properties.
www.interscience.wiley.com/journal/aoc
[1] a) H. J. Dai, Surface Sci. 2002, 500(1–3), 218; b) J. Liu, S. S. Fan, H. J. Dai, Recent advances in methods of forming carbon nanotubes, MRS Bull. 2004, 29(4), 244; c) J. Robertson, Mater. Today 2004, October, 46; d) H. Dai, J. Acc. Chem. Res. 2002, 35, 1035; e) M. S. Dresselhaus, G. Dresselhaus, P. Avouris, Carbon Nanotubes: Synthesis, Structure, Properties, and Applications, Springer: Berlin, 2001. [2] S. M. Huang, B. Maynor, X. Y. Cai, J. Liu, Adv. Mater. 2003, 15(19), 1651. [3] J. Kong, H. T. Soh, A. M. Cassell, C. F. Quate, H. J. Dai, Nature 1998, 395(6705), 878. [4] a) S. J. Kang, C. Kocabas, T. Ozel, M. Shim, N. Pimparkar, M. A. Alam, S. V. Rotkin, J. A. Rogers, Nat. Nanotechnol. 2007, 2(4), 230; b) Z. J. Zhang, B. Q. Wei, G. Ramanath, P. M. Ajayan, Appl. Phys. Lett. 2000, 77(23), 3764; c) A. Ismach, D. Kantorovich, E. Joselevich, J. Am. Chem. Soc. 2005, 127, 11554. [5] C. L. Cheung, A. Kurtz, H. Park, C. M. Lieber, J. Phys. Chem. B 2002, 106(10), 2429. [6] Y. M. Li, W. Kim, Y. G. Zhang, M. Rolandi, D. W. Wang, H. J. Dai, J.Phys. Chem. B 2001, 105(46), 11424. [7] J. Q. Lu, D. A. Rider, E. Onyegam, H. Wang, M. A. Winnik, I. Manners, Langmuir 2006, 22(11), 5174. [8] Q. Fu, S. M. Huang, J. Liu, J. Phys. Chem. B 2004, 108(20), 6124. [9] a) J. Lu, J. Phys. Chem. C 2008, 112, 10344; b) J. Lu, S. S. Yi, T. Kopley, C. Qian, J. Liu, E. Gulari, J. Phys. Chem. B 2006, 110, 6655. [10] a) S. Bhaviripudi, E. Mile, S. A. Steiner, A. T. Zare, M. S. Dresselhaus, A. M. Belcher, K. Jing, J. Am. Chem. Soc. 2007, 129(6), 1516; b) S. Bhaviripudi, A. Reina, J. F. Qi, J. Kong, A. M. Belcher, Nanotechnology 2006, 17, 5080. [11] W. W. Zhou, Z. Y. Han, J. Y. Wang, Y. Zhang, Z. Jin, X. Sun, Y. W. Zhang, C. H. Yan, Y. Li, Nano Lett. 2006, 6, 2987; b) L. Ding, D. N. Yuan, J. Liu, J. the Am. Chem. Soc. 2008, 130, 5428. [12] a) C. P. Deck, K. Vecchio, Carbon 2006, 44, 267; b) H. Okamoto, Binary Diagrams for Binary Alloys, ASM International: Materials Park, OH, 2000; c) T. B. Massalski, Binary Alloy Phase Diagrams, ASM International: Materials Park, OH, 1992. [13] J. Q. Lu, T. E. Kopley, N. Moll, D. Roitman, D. Chamberlin, Q. Fu, J. Liu, Chem. Mater. 2005, 17(9), 2227. [14] a) C. B. Murray, S. H. Sun, W. Gaschler, H. Doyle, T. A. Betley, C. R. Kagan, IBM J. Res. Devl. 2001, 45(1), 47; b) S. Sun, C. B. Murray, D. Weller, L. Folks, A. Moser, Science 2000, 287, 1989. [15] Y. Xia, B. Gates, Y. Yin, Y. Lu, Adv. Mater. 2000, 12(10), 693. [16] a) R. D. Bennett, A. J. Hart, R. E. Cohen, Adv. Mater. 2006, 18(17), 2274; b) R. D. Bennett, G. Y. Xiong, Z. F. Ren, R. E. Cohen, Chem. Mater. 2004, 16(26), 5589. [17] X. Liu, T. P. Bigioni, Y. Xu, A. M. Cassell, B. A. Cruden, J. Phys. Chem. B 2006, 110(41), 20102. [18] W. Kim, H. C. Choi, M. Shim, Y. M. Li, D. W. Wang, H. J. Dai, Nano Lett. 2002, 2(7), 703. [19] C. J. Lee, J. Park, J. A. Yu, Chem. Phys. Lett. 2002, 360(3–4), 250. [20] E. V. Levchenko, A. V. Evteev, I. V. Belova, G. E. Murch, Acta Mater. 2009, 57(3), 846. [21] A. F. Guillermet, Z. Metallk. 1987, 78(10), 700.
c 2010 John Wiley & Sons, Ltd. Copyright
Appl. Organometal. Chem. (2010)
Revised: 15 November 2009
Accepted: 22 November 2009
Published online in Wiley Interscience:
(www.interscience.com) DOI 10.1002/aoc.1610
Characterization of single-walled carbon nanotubes synthesized using iron and cobalt nanoparticles derived from self-assembled diblock copolymer micelles Qiang Fua , Luke Reeda, Jie Liub and Jennifer Lua∗ We present a comparative study of single-walled carbon nanotubes grown using iron and cobalt nanoparticles as catalysts via the chemical vapor deposition approach. Monodispersed iron and cobalt oxide nanoparticles with an average size of 2 nm were prepared using a polystyrene-b-poly (4-vinylpyridine) diblock copolymermicelle template. The 2 nm iron oxide nanoparticles generated single-walled carbon nanotubes with an average diameter of 1.5 nm while 2 nm cobalt oxide nanoparticles produced single-walled carbon nanotubes with an average diameter of 1.0 nm. To achieve high growth yield using iron nanoparticles as catalyst, higher carbon feed rate is required. These findings demonstrate the importance of the synergic interaction between catalyst and carbon precursor in single-walled carbon nanotube formation. It also elucidates the important role of catalyst c 2010 John Wiley & Sons, Ltd. chemical composition on carbon nanotube properties. Copyright Keywords: self-assembly; nanocatalysts; diblock copolymers; carbon nanotubes
Introduction Single-walled carbon nanotubes (SWCNTs) have attracted a great deal of mention due to their remarkable properties.[1] The chemical vapor deposition technique (CVD) has become the primary method for substrate-based applications, because of its scalability, versatility and unparalleled control over growth processes, such as nanotube locations and orientations.[2 – 4] In a CVD process, nanoparticles are generally used as nanocatalysts for carbon nanotube (CNT) growth and thus their properties will be largely influenced by these nanocatalysts. This is manifested by many experimental results suggesting that there is a close correlation between the CNT diameter and the size of catalyst nanoparticle.[5 – 7] Many single metallic and bimetallic nanoparticles, such as iron,[5 – 9,10b] cobalt,[9,10b] gold,[10a] copper[11] and nickel,[9a,10b] have been used as Nanocatalysts to decompose carbon vapor precursors. Among them, iron and cobalt nanoparticles have been regarded as two of the most effective catalyst systems due to their ability to produce relatively high yield CNTs with wide process window. It is known that a catalyst nanoparticle not only serves as a heterogeneous nucleation site but also interacts with source precursors to facilitate decomposition. Carbon has different solubilities and compound formation tendencies with iron and cobalt.[12] Such difference in affinity with carbon can result in different catalytic activities. To examine the effect of composition on CNT growth, substrates with identical size and density of iron and cobalt nanoparticles were prepared to eliminate the influence of catalyst size and density effect on CNT growth. It was found that iron and cobalt nanoparticles require different carbon feed rates to achieve the highest yield of CNTs. The finding was verified using either patterned or non-patterned nanoparticle arrays. Iron nanoparticles need higher carbon concentation to initiate CNT
Appl. Organometal. Chem. (2010)
growth than cobalt nanoparticles. Furthermore, 2 nm iron oxide nanoparticles generated SWCNTs with an average diameter of 1.5 nm while 2 nm cobalt oxide nanoparticles produced SWCNTs with an average diameter of 1.0 nm. These results demonstrate that the synergistic interaction between catalyst and carbon precursor is important for achieving high-yield growth. They also elucidate the critical role of catalyst chemistry in CNT properties.
Experimental Nanoparticle Preparation Polystyrene (53400)-b-poly(4-vinylpyridine) (8800) (PS508 -bP4VP84 ) (Polymer Source, Quebec, Canada) was dissolved in toluene to form a 0.25 wt% solution. CoCl2 ·6H2 O or FeCl2 ·4H2 O (Sigma-Aldrich, St Louis, MO, USA) was then added to PS508 -bP4VP84 solutions to give a molar ratio of 0.25 between the metal ions and pyridine groups. The solutions were stirred overnight at room temperature to fully sequester metal species. Nanoparticles were formed after spin-coating the metal-modified polymer solutions at 4000 rpm/s on silicon substrates with a 100 nm thermally grown oxide layer followed by oxygen plasma to remove the polymer template. To pattern the nanoparticles, a bilayer lift-off process reported previously was used.[13] Briefly, a photoresist film with 60 nm
∗
Correspondence to: Jennifer Lu, School of Engineering, University of California, Merced, 5200 North Lake Drive, Merced, CA 95348, USA. E-mail: [email protected]
a School of Engineering, University of California, Merced, Merced, CA 95348, USA b Department of Chemistry, Duke University, Durham, NC 27708, USA
c 2010 John Wiley & Sons, Ltd. Copyright
Q. Fu et al. LOL1000 and 800 nm OCG 825 was spun onto a thermal-oxidecoated silicon wafer. After the photoresist patterning, the metal modified PS508 -b-P4VP84 micelle solutions were deposited by spin-coating. A lift-off process was carried out followed by oxygen plasma to give rise to patterned nanoparticle arrays. To keep the number of catalyst nanoparticles similar between patterned and non-patterned substrates under the same growth condition, the sizes of substrates were adjusted purposely.
A
B
Chemical Vapor Deposition of SWCNTs The catalyst-loaded substrates were put into a 1-inch quartz tube and transferred into a furnace. The substrates were first annealed at 700 ◦ C in air for 10 min. After cooling to room temperature, the substrates were reheated to 900 ◦ C under a 100 sccm H2 flow. Once the furnace temperature reached 900 ◦ C, a flow of carbon precursors, which was a mixture of methane and ethylene, was introduced to grow CNTs. The methane flow rate was set to be 750, 800 and 850 sccm with the ethylene flow rate varied from 0 to 30 sccm. After 10 mins, the growth was stopped by switching off the carbon precursor flow. Lastly, the furnace was cooled down to room temperature under the protection of H2 . Nanoparticle and SWCNT Characterization The nanoparticles and SWCNTs were characterized with a Nanoscope IIIa Atomic Force Microscope (AFM, Vecco, Plainview, USA). Scanning electron microscopy (SEM) characterization was carried out on a XL30 scanning electron microscope (FEI Company, Oregon, USA) with an acceleration voltage of 1 kV. Raman spectra were measured on a 1 µm2 illumination area with a 632.8 nm and 1 mW laser with a Renishaw InVia Raman microscope.
Results and Discussion Nanoparticles with Similar Sizes and Densities Although nanoparticles can be produced with controlled size and composition using the colloid approach,[14,15] there is no means of controlling spacing and order between nanoparticles on a surface. Self-assembled block copolymers have been widely employed to create periodically ordered nanoparticle arrays on various substrates. Iron, cobalt, nickel and gold nanoparticles prepared with the block copolymers have been successfully used for the synthesis of SWCNTs.[7 – 10,13,16,17] It has been demonstrated that metal nanoparticle arrays with adjustable spacing, size and composition can be prepared using a solution micelle template formed by polystyrene-b-poly(vinylpyridine). These nanoparticles are excellent catalysts enabling the growth of high-quality and low-defect SWCNTs.[8 – 10,16,17] In this set of experiment, PS508 -b-P4VP84 , which consists of two immiscible blocks, a long hydrophobic PS block and a short hydrophilic P4VP block, was used. In a selective solvent such as toluene, PS508 -b-P4VP84 molecules self-assemble into micelles. In toluene, P4VP blocks are self-assembled to form the core while the corona consists of PS blocks. Because of the strong complexation tendency of 4-vinylpyridyl groups with transition metals, iron and cobalt species can be selectively sequestered into the cores. By spin-coating, the metal loaded solution micelles can be transferred and form a monolayer of hexagonally ordered surface micelles on substrates. After the removal of the polymer template, nanoparticles formed on surfaces. Figure 1 is a set of
www.interscience.wiley.com/journal/aoc
Figure 1. AFM images of nanoparticles produced using PS508 -b-PVP84 diblock copolymer. (A) Iron oxide nanoparticles. (B) Cobalt oxide nanoparticles (image area 1 × 1 µm2 ; height scale 10 nm).
AFM images showing that highly ordered nanoparticle arrays have been generated. According to the AFM height analysis of roughly 200 nanoparticles, the average size was 2 nm with a minimum size of 1.5 nm and a maximum size of 2.2 nm for as-synthesized iron nanoparticles, while the average size was 1.9 nm and ranged from 1.4 to 2.2 nm for cobalt nanoparticles. By exploiting the block copolymer template approach, nanoparticles with similar sizes and densities can be prepared. Therefore the effects of nanoparticle size and density on growth can be eliminated. Direct comparison of different catalyst compositions on CNT growth characteristics can be carried out. CNT Growth Results Both iron and cobalt nanoparticles enable the growth of SWCNTs. By systematically varying the ratio of methane and ethylene gases, the growth condition for achieving the highest CNT yield in a methane and ethylene mixture was 800 sccm of methane and 20 sccm of ethylene for iron nanoparticles and 800 sccm of methane only for cobalt nanoparticles. Figure 2(A, B) shows SEM images of SWCNT networks grown using cobalt and iron nanoparticles under the growth conditions which afford the highest yield. Because the catalyst nanoparticles are well dispersed on the substrates, the SWCNT networks have uniform density across entire surfaces. The iron nanoparticles produced short and straight tubes while cobalt nanoparticles gave long and curvy tubes. Figure 2(C, D) shows the representative Raman spectrum of CNTs synthesized using iron and cobalt nanoparticles respectively. The miniscule D band and narrow and high intensity of G band indicate that the CNTs synthesized from cobalt and iron nanoparticles have high quality with very limited number of defects. Figure 3 is a set of histogram of CNT diameters according to AFM height analysis showing that the diameters of CNTs synthesized from cobalt nanoparticles are from 0.6 to 1.5 nm with an average of 1 nm, and those from iron nanoparticles are from 0.9 to 2.5 nm with an average of 1.5 nm. This result was corroborated by Raman spectroscopy analysis that showed that, comparing CNT diameters, iron nanoparticles result in CNTs with larger diameters. The representative Raman spectra are displayed as inserts in Fig. 3, the majority of CNTs grown from iron nanoparticles oscillating around 170 cm−1 with CNTs grown from iron nanoparticles resonating at higher frequency around 210 cm−1 . Single-walled CNTs produced from iron and cobalt nanoparticles are different in several respects. First, the growth condition for the highest yield is different for iron and cobalt nanoparticles. Using iron nanoparticles as catalysts, the highest yield growth
c 2010 John Wiley & Sons, Ltd. Copyright
Appl. Organometal. Chem. (2010)
Characterization of single-walled carbon nanotubes
A
B
C
D
Figure 2. SEM image of SWCNTs grown using (A) cobalt nanoparticles, methane = 800 sccm and (B) iron nanoparticles, methane = 800 sccm, ethylene = 20 sccm (B) (scale bar: 5 µm). A representative Raman analysis of CNTs synthesized using cobalt nanoparticles and (C) iron nanoparticles (D).
B
A 25
25 Frequency
20 Frequency
30
15 10 5
20 15 10 5
0 0
1
2
3
4
0
0
Diameter (nm)
1
2
3
4
Diameter (nm)
Figure 3. AFM height analysis results; inserts are the representative Raman spectrum indicating the CNT radially resonating near 210 and 170 cm−1 using (A) cobalt nanoparticles and (B) iron nanoparticles.
condition is 800 sccm methane and 20 sccm ethylene. Similar conditions were also reported by other research groups.[18] Using cobalt nanoparticles, the growth condition which offers the highest yield uses only methane at 800 sccm flow rate. Since ethylene is more easily to be decomposed, this result indicates that higher carbon concentration is required for iron nanocatalysts. Furthermore, the diameter and length of SWCNTs are different. The iron nanoparticles produced short and straight SWCNTs with diameter around 1.5 nm, while the cobalt nanoparticles resulted in long and curvy SWCNTs with average diameter around 1 nm.
Appl. Organometal. Chem. (2010)
These observations can be rationalized by considering carbon solubility and carbide formation capability of catalyst nanoparticles. It is generally believed that CNT growth starts with hydrocarbon decomposition by catalyst nanoparticles at elevated temperature. The disassociated carbon species dissolve in the catalyst nanoparticles. As more hydrocarbons are decomposed, the carbon concentration increases. Oversatuation of carbon leads to CNT growth. According to phase diagrams, carbon has a relatively high solubility in iron. As a result, a high carbon concentration is required to initiate SWCNT growth. With the increase in carbon concentration, Fe3 C is formed prior to carbon precipitation,
c 2010 John Wiley & Sons, Ltd. Copyright
www.interscience.wiley.com/journal/aoc
Q. Fu et al.
A
Acknowledgments
B
This work is supported by a start-up fund at the University of California, Merced and a NSF-CBET fund. We also would like to acknowledge the help of the Agilent technologies for Raman analysis.
References
Figure 4. SEM images of SWCNTs grown from catalyst nanoparticle patterns: (A) cobalt nanoparticles; (B) iron nanoparticles (scale bar: 20 µm).
which may slow down carbon diffusion, leading to lower growth rate.[19,20] This is consistent with our observation of shorter SWCNTs and higher carbon feed rate for iron nanoparticles. In comparison, carbon has lower solubility in cobalt, and does not form stable compounds with cobalt.[21] Therefore, oversaturation will lead to carbon directly precipitating out. Consequently, SWCNT nucleation and growth can take place at a higher speed with lower carbon feed rate. The experimental observation that cobalt nanoparticles require lower carbon concentration, which can be achieved using highly stable methane as a carbon precursor, is supported by this argument. Maintaining similar metal nanoparticle population between substrates with or without patterning and using the same growth condition for the highest growth yield, patterned CNTs were generated. Figure 4 shows the SEM image of SWCNTs grown from patterned cobalt and iron nanoparticle arrays on a SiO2 substrate. Cobalt nanoparticles resulted in long and small diameter SWCNTs, which are particularly important for many applications. Since nanoparticles synthesized by this method do not agglomerate, CNT growth yield can be calculated.[9a] The growth yield is defined by the number of tubes vs. the number of nanoparticles. Approximately 4% of cobalt nanoparticles are able to initiate carbon nanotube growth while 1% of iron nanoparticles are active to initiate growth.
Conclusions SWCNTs synthesized using iron and cobalt nanocatalysts exhibit different properties. Using 2 nm iron nanoparticles, short and straight SWCNTs with an average diameter of 1.5 nm were synthesized. On the other hand, 2 nm cobalt nanoparticles generated long and wavy SWCNTs with a diameter around 1.0 nm. The optimized growth condition for achieving maximum CNT yield for iron nanoparticles is a mixture of methane and ethylene, with higher carbon concentration, while for cobalt nanoparticles it is methane alone, with lower carbon concentration. This is due to the different tendency in terms of carbon solubility and carbide formation ability. These results highlight the importance of synergetic interaction between nanocatalysts and carbon feedstock for SWCNT growth. They reveal the important role of catalyst chemical composition in CNT properties.
www.interscience.wiley.com/journal/aoc
[1] a) H. J. Dai, Surface Sci. 2002, 500(1–3), 218; b) J. Liu, S. S. Fan, H. J. Dai, Recent advances in methods of forming carbon nanotubes, MRS Bull. 2004, 29(4), 244; c) J. Robertson, Mater. Today 2004, October, 46; d) H. Dai, J. Acc. Chem. Res. 2002, 35, 1035; e) M. S. Dresselhaus, G. Dresselhaus, P. Avouris, Carbon Nanotubes: Synthesis, Structure, Properties, and Applications, Springer: Berlin, 2001. [2] S. M. Huang, B. Maynor, X. Y. Cai, J. Liu, Adv. Mater. 2003, 15(19), 1651. [3] J. Kong, H. T. Soh, A. M. Cassell, C. F. Quate, H. J. Dai, Nature 1998, 395(6705), 878. [4] a) S. J. Kang, C. Kocabas, T. Ozel, M. Shim, N. Pimparkar, M. A. Alam, S. V. Rotkin, J. A. Rogers, Nat. Nanotechnol. 2007, 2(4), 230; b) Z. J. Zhang, B. Q. Wei, G. Ramanath, P. M. Ajayan, Appl. Phys. Lett. 2000, 77(23), 3764; c) A. Ismach, D. Kantorovich, E. Joselevich, J. Am. Chem. Soc. 2005, 127, 11554. [5] C. L. Cheung, A. Kurtz, H. Park, C. M. Lieber, J. Phys. Chem. B 2002, 106(10), 2429. [6] Y. M. Li, W. Kim, Y. G. Zhang, M. Rolandi, D. W. Wang, H. J. Dai, J.Phys. Chem. B 2001, 105(46), 11424. [7] J. Q. Lu, D. A. Rider, E. Onyegam, H. Wang, M. A. Winnik, I. Manners, Langmuir 2006, 22(11), 5174. [8] Q. Fu, S. M. Huang, J. Liu, J. Phys. Chem. B 2004, 108(20), 6124. [9] a) J. Lu, J. Phys. Chem. C 2008, 112, 10344; b) J. Lu, S. S. Yi, T. Kopley, C. Qian, J. Liu, E. Gulari, J. Phys. Chem. B 2006, 110, 6655. [10] a) S. Bhaviripudi, E. Mile, S. A. Steiner, A. T. Zare, M. S. Dresselhaus, A. M. Belcher, K. Jing, J. Am. Chem. Soc. 2007, 129(6), 1516; b) S. Bhaviripudi, A. Reina, J. F. Qi, J. Kong, A. M. Belcher, Nanotechnology 2006, 17, 5080. [11] W. W. Zhou, Z. Y. Han, J. Y. Wang, Y. Zhang, Z. Jin, X. Sun, Y. W. Zhang, C. H. Yan, Y. Li, Nano Lett. 2006, 6, 2987; b) L. Ding, D. N. Yuan, J. Liu, J. the Am. Chem. Soc. 2008, 130, 5428. [12] a) C. P. Deck, K. Vecchio, Carbon 2006, 44, 267; b) H. Okamoto, Binary Diagrams for Binary Alloys, ASM International: Materials Park, OH, 2000; c) T. B. Massalski, Binary Alloy Phase Diagrams, ASM International: Materials Park, OH, 1992. [13] J. Q. Lu, T. E. Kopley, N. Moll, D. Roitman, D. Chamberlin, Q. Fu, J. Liu, Chem. Mater. 2005, 17(9), 2227. [14] a) C. B. Murray, S. H. Sun, W. Gaschler, H. Doyle, T. A. Betley, C. R. Kagan, IBM J. Res. Devl. 2001, 45(1), 47; b) S. Sun, C. B. Murray, D. Weller, L. Folks, A. Moser, Science 2000, 287, 1989. [15] Y. Xia, B. Gates, Y. Yin, Y. Lu, Adv. Mater. 2000, 12(10), 693. [16] a) R. D. Bennett, A. J. Hart, R. E. Cohen, Adv. Mater. 2006, 18(17), 2274; b) R. D. Bennett, G. Y. Xiong, Z. F. Ren, R. E. Cohen, Chem. Mater. 2004, 16(26), 5589. [17] X. Liu, T. P. Bigioni, Y. Xu, A. M. Cassell, B. A. Cruden, J. Phys. Chem. B 2006, 110(41), 20102. [18] W. Kim, H. C. Choi, M. Shim, Y. M. Li, D. W. Wang, H. J. Dai, Nano Lett. 2002, 2(7), 703. [19] C. J. Lee, J. Park, J. A. Yu, Chem. Phys. Lett. 2002, 360(3–4), 250. [20] E. V. Levchenko, A. V. Evteev, I. V. Belova, G. E. Murch, Acta Mater. 2009, 57(3), 846. [21] A. F. Guillermet, Z. Metallk. 1987, 78(10), 700.
c 2010 John Wiley & Sons, Ltd. Copyright
Appl. Organometal. Chem. (2010)
