J. Phys. Chem. C 2007, 111, 9077-9080
9077
Vertically Aligned Large-Diameter Double-Walled Carbon Nanotube Arrays Having Ultralow Density Lijie Ci, Robert Vajtai, and P. M. Ajayan* Department of Material Science and Engineering, Rensselaer Polytechnic Institute, Troy, New York 12180 ReceiVed: March 16, 2007; In Final Form: April 25, 2007
We have selectively synthesized ultralow density (0.015 ( 0.002 g/cm3) double-walled carbon nanotube (DWNT) arrays. The low mass density is due to the large diameter and low-site density of DWNTs in the arrays. The diameter of the as-grown DWNTs varies from 4.5 to 14.8 nm with a mean value of 7.9 nm. The role of the thickness of Fe catalyst layer is evaluated for the selective growth of large-diameter DWNTs. Our DWNT arrays have promising applications in areas such as solar absorbers and super-hydrophobic surfaces.
Introduction The unique properties of carbon nanotubes (CNT) strongly depend on their structure, that is, diameter and chirality.1,2 Nanotubes with different structural properties may have different applications,3,4 and it is very important to tune nanotube properties by controlling their diameter for some specific applications. From the structural point of view, large-diameter nanotubes may be more stable because of the smaller curvature of graphene layer.5,6 The preparation of large-diameter SWNTs has been of great interest in the past years by the three main growth techniques: arc-discharge,7 laser-ablation,8 and chemical vapor deposition (CVD).9 In this paper, we report the selective synthesis of pure double-walled carbon nanotube (DWNT) arrays having large diameters via a modified CVD method. The density of the resulting DWNT arrays is around 15 mg/cm3, which makes our DWNT arrays the lightest nanotube material ever reported. We investigated the surface wettability of the DWNT samples and found that it is tunable by changing the growth parameters. Experimental Section Vertically aligned DWNT films were prepared by a waterassisted chemical vapor deposition process. A similar method has been reported to selectively fabricate aligned SWNT and DWNT arrays by Hata and co-workers10,11 and multiwalled CNT (MWNT) arrays by other group.12 We deposited 10 nm thick Al instead of Al2O310-12 as the buffer layer between Fe catalyst layer (1-5 nm thick) and Si wafer (with or without oxide layer) with an E-beam thermal evaporator. In most of the CVD growth, ethylene was used as carbon source, and Ar/H2 (15% H2 content) was used as a carrier gas. First, Ar/H2 (300 sccm) flowed through the alumina tube (43 mm inner diameter) while the furnace was heated up to the CNT growth temperature (750800 °C). At the growth temperature, the Ar/H2 flow was increased to 1300 sccm, and another route of Ar/H2 gas bubbled through a water bottle (which is kept at room temperature) with a flow rate of 80 sccm. Ethylene gas of 100 sccm was also flowing into the reactor at the same time. The CNT growth process lasted for 5 s to 30 min, depending on the CNT thickness * To whom correspondence should be addressed. Tel: 518-2766485. Fax: 518-2768554. E-mail:
[email protected].
requested. In the final stage, the furnace was cooled down to room temperature under Ar/H2 protection. Scanning electron microscopy (SEM, JEOL JSM-6330F) was performed to characterize nanotube samples and to measure the thickness of the resultant forest. High-resolution transmission electron microscopy (HRTEM, JEOL 2010) was performed to characterize the quality of nanotubes and to analyze their diameter distribution. Renishawl 1000 microscopic confocal Raman spectrometer with a laser excitation of 514 nm was used for micro-Raman scattering test. Results and Discussion With Fe catalyst layer thickness of 1.5 nm, millimeter-long DWNT arrays could be easily grown in 30 min. SEM images (Figure 1) show the well-aligned nanotube feature. Lowmagnification TEM observations indicate that our nanotube samples are very pure, and almost no Fe particle was found after scanning over the entire sample. HRTEM observations indicate that nearly all the nanotubes are double-walled. Diameter measurements from HRTEM images indicate a very large average diameter value, 7.9 nm, which is much larger than that of the DWNTs from Yamada et al. (3.7 nm).11 To know the stacking information of the large-diameter DWNT arrays, such as the mass density and the site density (or area density), we measured the mass density of the DWNT arrays for samples grown to different nanotube lengths (grown by controlling the duration of each CVD run), as seen in Figure 2a. We got a very low value of 0.015 ( 0.002 g/cm3. The density value is about 2 times lower than that of SWNT13 and DWNT11 arrays prepared by the similar CVD process and is very close to that of carbon nanofoam (0.002-0.010 g/cm3) that others have reported,14,15 which is the second lightest synthetic material. The lightest man-made solid listed in the Guinness Book of World Records is a kind of aerogel (0.0010.003 g/cm3).16 It is interesting to ask why the density of our DWNT arrays is so low. Two aspects should be considered: the density of individual nanotubes and the site density of nanotube arrays. To compare, two individual DWNTs with different diameters (7.9 and 3.7 nm) were modeled by Nanotube Modeler (version 1.2.9, JCrystalSoft), and the atomic and geometric information obtained was used to calculate the densities. On the basis of previous reports,13 we only considered the diameter factor for
10.1021/jp072123c CCC: $37.00 © 2007 American Chemical Society Published on Web 06/05/2007
9078 J. Phys. Chem. C, Vol. 111, No. 26, 2007
Ci et al.
Figure 1. Low-magnification SEM image of DWNT arrays (a) and high-resolution SEM image (b).
Figure 2. (a) Mass density (left Y-axis) and site density (right Y-axis) of DWNT arrays at different heights. (b) High-resolution SEM image shows the distribution of nanosized catalyst particles on substrate.
the comparison. The density of the DWNT with the larger diameter (7.9 nm, which is the mean value of our DWNT samples) is 0.784 g/cm3, and that of the smaller DWNT (3.7 nm, which corresponds to the average diameter of DWNT sample reported by Hata et al.11) is 1.647 g/cm3. The smaller tube is about 2 times denser than the larger one. The site density of our DWNT arrays was calculated according to the equation of Sd ) F/πr2Ft, where Sd stands for the site density, F is the mass density of the arrays, r is the mean nanotube diameter, and Ft is the mass density of an individual DWNT. We got a site density of 411 ( 56 tubes per µm2. On the basis on the site density, a tube-tube spacing value of around 50 nm is expected. The site density of our DWNT arrays is around an order of magnitude lower than that of the reported SWNT arrays (5200 ( 350 tubes per µm2).13 The site density value was also verified by our microscopic count of catalyst particles on substrate. Figure 2b shows a high-resolution SEM image of a substrate. The density of the catalyst particles is about 350 particles per µm2, and this number is close to the nanotube site density we obtained (the counting was done by a software, ImageJ, and the counts should be smaller than the real number because the method neglects some low-contrast particles). Assuming the mass density of individual DWNT with the diameter of 7.9 nm, we get the nanotube volume filling fraction of our DWNT arrays to be about 1.5∼2.0%. However, if the density of graphitic carbon (2.25 g/cm3) is considered instead, the carbon volume filling fraction would be just 0.5∼0.7%. Figure 3 a-d displays HRTEM images of DWNTs with different diameters from 4.5 to 14.8 nm. The nanotubes are very clean, and there is almost no amorphous carbon coating. Thinwalled nanotubes with relatively large diameter are supposed to be easily deflated or collapsed into ribbons. However, a cross section view of a DWNT with a diameter of 7.6 nm in Figure 3b indicates that the large-diameter DWNTs are still tubular, even though their cross sections are not perfectly circular. Some large-diameter DWNTs have varied radial dimension along the tube axis, like the nanotube shown in Figure 3d. Figure 3e-i displays typical DWNTs that are bent, buckled, curved, and collapsed suggesting more structural
defects compared to ideal SWNT or small-diameter DWNT.17,18 The low-growth temperature and rapid growth rate may be the main reasons for the frequent appearance of defects. Growth of vertically aligned DWNT arrays is very important for some engineering applications. DWNTs combine the advantage of structural features of SWNTs and MWNTs and have their novel physical and mechanical properties, which make DWNTs the best choice for certain applications, for example, field-emission displays (FEDs).19 To selectively grow only DWNTs, we varied the film thickness of Fe from 0.6 to 5 nm. The diameter and wall number distributions of each sample with different Fe thicknesses were analyzed under HRTEM observations. Figure 4 shows the histograms of diameter distribution and wall number of the nanotube samples from Fe films with different thickness. It is obvious that nanotube arrays consist of almost only DWNTs when the growth was carried out on the Fe catalyst with thickness of 1.5 nm. In fact, more growth results indicated that the Fe thickness window for DWNT array growth is from 1 to 2 nm. When Fe thickness is below 1 nm, growth is not well controlled, and a mixture of SWNT and DWNT is grown (as shown in Figure 4b, 49% SWNTs were grown when Fe is 0.6 nm thick). However, pure SWNT array growth seems not to be easy under our conditions. We also noticed that DWNTs are always the major tube type, even though the Fe thickness was changed from 1.5 to 5 nm. Most of the other nanotubes obtained are few walled (3∼6 walls). It is well-known that the size of catalyst particles is very critical for nanotube growth. The initial Fe thickness, together with other factors such as the interaction between Fe and substrate, will determine the size of catalyst particles. Our empirical results indicated that 1-2 nm thick deposited Fe film produced suitable catalyst size for DWNT growth. Why is the diameter of our as-grown DWNTs so large? The possible reason is that we used Al as a buffer layer between Fe and Si substrate (as described in our Experimental Section). Al will be oxidized after the substrates are removed from the vacuum and during CVD process. The interaction between Fe and Al is different
Large-Diameter Double-Walled Carbon Nanotubes
J. Phys. Chem. C, Vol. 111, No. 26, 2007 9079
Figure 5. First- and second-order Raman spectrum of the largediameter DWNTs. The first-order Raman spectrum is fitted by Lorentzian lines.
Figure 6. Surface wettability of DWNT arrays and its tunable feature by changing catalyst thickness: Fe 1 nm (a), Fe 1.5 nm (b), and Fe 4 nm (c).
Figure 3. (a-d) HRTEM images of individual DWNTs with different diameters; a cross section view of a DWNT is also shown in b. (e-i) HRTEM images of different deformed large-diameter DWNTs because of the existing congenital defects: bent (e), buckled (f), sharp diameter change (g), collapsed (h), and curved (i).
Figure 4. (a) Diameter histograms of nanotube samples grown from different Fe thicknesses. (b) Nanotube type (walls) histograms at different Fe thicknesses.
from that between Fe and Al2O3, and this may produce Fe catalyst particle size and distribution different from the reported work.11
The curvature effect should be neglectable for the largediameter DWNT, and their Raman spectrum is supposed to be similar to a two-layer graphene.20 Micro-Raman scattering was performed to characterize our DWNT samples. To avoid the strong light absorption during Raman test, aligned nanotubes were first dispersed and deposited on a glass substrate. Figure 5 shows a typical first- and second-order Raman spectrum of the large-diameter DWNTs. The first-order Raman spectrum shows similar profiles as that of aligned CVD grown MWNTs,21 which consist of a strong G line (1582 cm-1) and two disordered carbon-induced scatterings, D line (1349 cm-1) and D′ line (1617 cm-1). The intensity ratio between the G line and D line (IG/ID) is about 1.4, which is close to that of aligned MWNTs21 and is much lower than the reported value of DWNT arrays.11 The low IG/ID value indicates substantial defect concentration existing in the graphene layers. The 2D line (second order of D line) is located at 2696 cm-1, which is the same as that of twolayer graphene.20 The as-grown DWNT arrays appear very black, suggesting strong light absorption. Light reflection from the sample surface can be controlled and is increased remarkably with the thickness of the catalyst layer, and this change is also observable with the naked eyes from ultrablack to a little gray. It seems plausible to suppose the further enhancement in absorption degree by further controlling the diameter and spacing of the nanotubes in the arrays. Systematic experiments are underway to determine quantitatively optical absorption occurring in these arrays. Static contact angle (CA) measurements are generally used to judge the wettability of solid surfaces. CA measurements were performed on our nanotube arrays. The water CA on the DWNT arrays is around 137°; however, changing the Fe thickness resulted in changes in the wettability of the nanotube arrays. As Figure 6 shows, the nanotube samples grown from the thinner Fe catalyst layer show less hydrophobicity (CA is down to 100° when the Fe thickness is 1 nm), and the nanotube
9080 J. Phys. Chem. C, Vol. 111, No. 26, 2007 samples from thicker Fe layer show superhydrophobic behavior (CA is increased up to 170° when the Fe thickness is larger than 3 nm). Some applications of aligned nanotube arrays strongly depend on their surface wettability.22 Tuning surface wettability by direct growth leads to a solution different from the chemical surface modification process.23,24 Conclusions In summary, we have selectively synthesized DWNT arrays, and the resulting products have ultralow mass density because of the large diameter of DWNTs and the low-site density. HRTEM observations indicate that the large-diameter DWNTs are able to maintain their tubule shapes. The excellent light absorption and the tunable surface wettability of our DWNT arrays indicate potential applications in solar absorption and biotic techniques. Acknowledgment. The authors acknowledge the support of the Interconnect Focus Center, one of five research centers funded under the Focus Center Research Program, a Semiconductor Research Corporation program. References and Notes (1) Hamada, N.; Sawada, S. I; Oshiyama, A. Phys. ReV. Lett. 1992, 68, 1579-1581. (2) Saito, R.; Fujita, M.; Dresselhaus, G.; Dresselhaus, M. S. Appl. Phys. Lett. 1992, 60, 2204-2206. (3) Dresselhaus, M. S., Dresselhaus, G., Avouris, P., Eds.; Carbon nanotubes: Synthesis, Structure, Properties, and Applications; Topics in Applied Physics 80; Springer: New York, 2001. (4) Baughman, R. H.; Zakhidov, A. A.; De Heer, W. A. Science 2002, 297, 787-792. (5) Iijima, S.; Ichihashi, T. Nature 1993, 363, 603-605. (6) Robertson, D. H.; Brenner, D. W.; Mintminre, J. W. Phys. ReV. B 1992, 45, 12592-12595.
Ci et al. (7) Kiang, C. H. J. Phys. Chem. A 2000, 104, 2454-2456. (8) Lebedkin, S.; Schweiss, P.; Renker, B.; Malik, S.; Hennrich, F.; Neumaier, M.; Stoermer, C.; Kappes, M. Carbon 2002, 40, 417-423. (9) Yang, Q. H.; Bai, S.; Sauvajol, J. L.; Bai, J. B. AdV. Mater. 2003, 15, 792-795. (10) Hata, K.; Futaba, Don N.; Mizuno, K.; Namai, T.; Yumura, M.; Iijima, S. Science 2004, 306, 1362-1364. (11) Yamada, T.; Namai, T.; Hata, K.; Futaba, D. N.; Mizuno, K.; Fan, J.; Yudasaka, M.; Yumura, M.; Iijima, S. Nat. Nanotechnol. 2006, 1, 131136. (12) Zhu, L.; Xiu, Y.; Hess, D W.; Wong, C. P. Nano Lett. 2005, 5, 2641-2645. (13) Futaba, D. N.; Hata, K.; Namai, T.; Yamada, T.; Mizuno, K.; Hayamizu, Y.; Yumura, M.; Iijima, S. J. Phys. Chem. B 2006, 110, 80358038. (14) Rode, A. V.; Hyde, S. T.; Gamaly, E. G.; Elliman, R. G.; McKenzie, D. R.; Bulcock, S. Appl. Phys. A 1999, 69, S755-S758. (15) Rode, A. V.; Gamaly, E. G.; Luther-Davies, B. Appl. Phys. A 2000, 70, 135-144. (16) Novel Materials from Solgel Chemistry (accessed date: May 29, 2007); http://www-cms.llnl.gov/featured_science_archive/2005/07-05_ solgel.html. (17) Ci, L.; Rao, Z.; Zhou, Z.; Tang, D.; Yan, X.; Liang, Y.; Liu, D.; Yuan, H.; Zhou, W.; Wang, G.; Liu, W.; Xie, S. Chem. Phys. Lett. 2002, 359, 63-67. (18) Ren, W.; Li, F.; Chen, J.; Bai, S.; Cheng, H. M. Chem. Phys. Lett. 2002, 359, 196-202. (19) Seko, K.; Kinoshita, J.; Saito, Y. Jpn. J. Appl. Phys. 2005, 44, L743-L475. (20) Ferrari, A. C.; Meyer, J. C.; Scardaci, V.; Casiraghi, C.; Lazzeri, M.; Mauri, F.; Piscanec, S.; Jiang, D.; Novoselov, K. S.; Roth, S.; Geim, A. K. Phys. ReV. Lett. 2006, 97, 187401-4. (21) Li, W.; Zhang, H.; Wang, C.; Zhang, Y.; Xu, L.; Zhu, K.; Xie, S. Appl. Phys. Lett. 1997, 70, 2684-2686. (22) Zanello, L. P.; Zhao, B.; Hu, H.; Haddon, R. C. Nano Lett. 2006, 6, 562-567. (23) Li, H.; Wang, X.; Song, Y.; Liu, Y.; Li, Q.; Jiang, L.; Zhu, D. Angew. Chem., Int. Ed. 2001, 40, 1743-1746. (24) Lau, K. K. S.; Bico, J.; Teo, K. B. K.; Chhowalla, M.; Amaratunga, G. A. J.; Milne, W. I.; McKinley, G. H.; Gleason, K. K. Nano Lett. 2003, 3, 1701-1705.
