pubs.acs.org/Langmuir © 2009 American Chemical Society
Large-Area Patterning of Carbon Nanotube Ring Arrays Saloome Motavas, Badr Omrane, and Chris Papadopoulos* Department of Electrical and Computer Engineering, University of Victoria, P.O. Box 3055, Victoria, British Columbia V8W 3P6, Canada Received July 29, 2008. Revised Manuscript Received January 28, 2009 Single-walled carbon nanotubes were assembled into large-area arrays of nanoscale rings on silicon via direct patterning with a self-assembled colloidal polystyrene sphere mask. Nanotubes from liquid suspension gathered at the base of each sphere in the mask to form rings, and the resulting arrays consisted of wellordered nanotube rings with diameters of 203 ( 21 nm and 97 ( 14 nm for rings formed with 780 and 450 nm colloidal spheres, respectively. Ring heights were found to be 4.7 ( 1.8 nm and 5.9 ( 1.4 nm for 780 and 450 nm sphere masks, respectively. A first-order geometric model was proposed to account for the observed ring diameters. The approach presented demonstrates an efficient and straightforward path for patterning carbon nanotubes into well-defined surface distributions on various substrates with highly controlled and tunable dimensions.
I. Introduction Controlling the distribution of carbon nanotubes (CNTs) on surfaces is of paramount importance in taking advantage of their outstanding physical properties in a wide range of fields.1 CNTs that possess ringlike geometries are a particularly interesting class of structures that have attracted increasing attention as a result of their unique diameter-dependent properties and potential applications.2-6 Previous approaches to CNT ring fabrication have relied on liquid-tube surface interactions,7 chemical vapor deposition,8 and chemical modification.9 Although offering simplicity and parallel fabrication, these techniques lack direct control over ring diameter. Using more complex processing, scanning probe methods have recently been shown to allow the formation of CNT rings with controllable diameters on chemically modified gold substrates.10 In this work, we employ a different approach to CNT ring fabrication, which allows the direct fabrication of well-ordered CNT ring arrays with highly controlled and tunable diameters on a variety of substrates as required for applications and fundamental studies while retaining simplicity and high throughput.
* Corresponding author. E-mail:
[email protected]. (1) Saito, R.; Dresselhaus, G.; Dresselhaus, M. S. Physical Properties of Carbon Nanotubes; Imperial College Press: Singapore, 1998. (2) Shea, H. R.; Martel, R.; Avouris, P. Phys. Rev. Lett. 2000, 84, 4441– 4444. (3) Tamura, R.; Ikuta, M.; Hirahara, T.; Tsukada, M. Phys. Rev. B. 2005, 71, 0454181–0454187. (4) Haddon, R. C. Nature 1997, 388, 31–32. (5) Watanabe, H.; Manabe, C.; Shigematsu, T.; Shimizu, M. Appl. Phys. Lett. 2001, 78, 2928–2930. (6) Latil, S.; Roche, S.; Rubio, A. Phys. Rev. B. 2003, 67, 1654201– 1654209. (7) Martel, R.; Shea, H. R.; Avouris, P. J. Phys. Chem. B 1999, 103, 7551– 7556. (8) Song, L.; Ci, L. J.; Sun, L. F.; Jin, C.; Liu, L.; Ma, W.; Liu, D.; Zhao, X.; Luo, S.; Zhang, Z.; Xiang, Y.; Zhou, J.; Zhou, W.; Ding, Y.; Wang, Z. L.; Xie, S. Adv. Mater. 2006, 18, 1817–1821. (9) Sano, M.; Kamino, A.; Okamura, A.; Shinkai, S. Science 2001, 293, 1299–1301. (10) Zou, S.; Maspoch, D.; Wang, Y.; Mirkin, C. A.; Schatz, G. C. Nano Lett. 2007, 7, 276–280.
Langmuir 2009, 25(8), 4655–4658
II.
Experimental Section
Our process is based on colloidal lithography,11 a general and efficient low-cost patterning technique based on using self-assembled colloidal sphere monolayers as nanoscale masks. This method has been shown to be effective for patterning large areas with periodic arrays of nanostructures whose sizes can be controlled by employing different-diameter colloidal spheres.12,13 Figure 1 shows a schematic of the procedure used to form the CNT ring arrays: For mask formation (i), polystyrene spheres in an aqueous suspension (Interfacial Dynamics Corp., surfactantfree carboxyl white, 780 or 450 nm diameter, used as received) were deposited onto Æ100æ Si wafers (Silicon Quest International Inc., n type, 10-20 Ω 3 cm) to form hexagonal close-packed monolayer regions. For CNT deposition (ii), three sources of single-walled carbon nanotubes (SWNTs) were tried: (A) SigmaAldrich, 1.2-1.5 nm diameter, 2-5 μm length; (B) Alfa-Aesar, 1.4 nm average diameter, 1 μm average length; and (C) SouthWest NanoTechnologies, 1.0 ( 0.3 nm diameter, 1 μm average length. The SWNTs were dispersed in methanol via mild bath sonication at concentrations of approximately 0.5-1 mg/mL and allowed to settle before being deposited onto the colloidal masks in 1.5 μL drops taken from the supernatant solution. The samples were then allowed to dry in an ambient environment, followed by heating in an oven at 90 C for 5 min. During evaporation, the liquid retracts toward the bottom of the spheres as a result of capillary forces,14 which results in the formation of circular ring structures consisting of CNTs around the base of each sphere (iii). After the removal of the spheres using adhesive tape, ordered arrays of CNT rings remain on the substrate (iv).
III.
Results
We characterized our samples using scanning electron microscopy (SEM, Hitachi S-4700), atomic force microscopy (AFM, Nanonics MV-1000), and Raman spectroscopy (Renishaw inVia). Figure 2a shows an SEM image of a typical (11) Li, Y.; Cai, W.; Duan, G. Chem. Mater. 2008, 20, 615–624. (12) Hulteen, J. C.; Van Duyne, R. P. J. Vac. Sci. Technol., A 1995, 13, 1553–1558. (13) Weekes, S. M.; Ogrin, F. Y.; Murray, W. A. Langmuir 2004, 20, 11208–11212. :: (14) Boneberg, J.; Burmeister, F.; Schafle, C.; Leiderer, P.; Reim, D.; Fery, A.; Herminghaus, S. Langmuir 1997, 13, 7080–7084.
Published on Web 2/24/2009
DOI: 10.1021/la803633w 4655
Article
Motavas et al.
SWNT deposit following the removal of a 780 nm sphere mask. Large-area patterning of nanoscale rings with good uniformity arranged in hexagonal close-packed arrays is displayed. In Figure 2b, we show results of a fabrication trial using 450-nm-diameter spheres. Compared to the larger sphere masks, these CNT rings are approximately one-half the size, which demonstrates the control over CNT ring diameter afforded by colloidal patterning. In some cases, we observed clear defects consisting of different diameter rings within the arrays. An example of this is seen in the SEM image of Figure 2c, which depicts two very small diameter rings. We also found instances in which the defects had diameters significantly larger than the rest of the rings in the array (discussed further in section IV).
Figure 1. Schematic of CNT ring array formation: (i) colloidal mask formation, (ii) deposition of CNTs onto the colloidal mask, (iii) drying, and (iv) sphere removal exposing CNT rings.
Figure 3. (a) AFM image and cross-section profile of a CNT ring formed on silicon using 780-nm-diameter spheres and CNT type B. (b) AFM image showing SWNTs emanating from ring edges. (c) Close-up view of a ring with overlapping ends composed of nanotube bundle approximately 700 nm in length. (d) Portion of a CNT ring array (450 nm spheres, CNT type C) displaying SWNTs spanning between rings. (e) Cross-section profile of the image in d, indicating individual SWNT branching off of a small nanotube bundle.
Figure 2. (a) SEM image of a large-area array of SWNT rings on silicon made using a 780-nm-diameter colloidal sphere monolayer mask and CNT type A. The inset shows a close-up view of an individual ring from a similar sample made using CNT type B. (b) SEM image of an SWNT ring array formed using a 450-nm-diameter colloidal sphere monolayer mask and CNT type C. (c) Portion of an array containing very small diameter CNT ring defects. 4656
DOI: 10.1021/la803633w
Langmuir 2009, 25(8), 4655–4658
Motavas et al.
Article
Figure 4. Raman spectra obtained from a CNT ring array sample (780 nm spheres, CNT type B) displaying the G band and the radial breathing mode. The inset expands the spectrum to show the D band near 1300 cm-1.
For a detailed analysis of the CNT ring dimensions, we used AFM (intermittent contact mode) in order to avoid artifacts that may be present in the SEM images as a result of electrostatic effects. Figure 3a contains a typical AFM image and cross-section of a ring formed using 780-nm-diameter spheres that has a diameter of approximately 200 nm and a height of 4 nm. From an analysis of several such images, we found that the mean diameter ( standard deviation values were 203 ( 21 and 97 ( 14 nm for CNT rings formed with 780 and 450 nm colloidal spheres, respectively. Similarly, ring heights were found to be 4.7 ( 1.8 and 5.9 ( 1.4 nm for 780 and 450 nm spheres, respectively, which is comparable to previous studies.8,10 The AFM scans also allowed us to gain further insight into ring structure and formation. For example, as shown in Figure 3b-d, we can observe SWNTs emanating from and bridging between rings. This is consistent with ring formation being caused by liquid retracting and collecting CNTs at the base of the spheres, which can result in longer CNTs being extended on the substrate between spheres. The intertwined, overlapping structure of tubes making up the rings is evident in these images. In addition, the ring shown in Figure 3c displays two distinct ends that overlap to close the circle. Such structures result from relatively short SWNT bundles present in the liquid suspension whose curved ring geometries are stabilized by the interaction between their two ends. To characterize the ring arrays further, Raman spectroscopy was performed with 785 nm excitation at room temperature. Figure 4 shows a typical result obtained on the patterned region of a sample made using CNT type B. Two main bands related to CNTs are displayed: The G band related to tangential mode vibrations of C atoms along the tube wall peaks near 1592 cm-1 with a shoulder peak at approximately 1568 cm-1. A small peak near 1300 cm-1 (D band, Figure 4 inset) was also observed. The characteristic CNT radial breathing mode (RBM) was found at 160 cm-1, indicating a tube diameter of 1.4-1.5 nm15 consistent with the tubes deposited from suspension.
(15) Dresselhaus, M. S.; Dresselhaus, G.; Jorio, A. J. Phys. Chem. C 2007, 111, 17887–17893.
Langmuir 2009, 25(8), 4655–4658
Figure 5. Ideal geometry used to relate CNT ring diameter DR, ring height t, and sphere diameter DS.
IV. Discussion The factors determining the shape and size of the CNT rings, particularly their diameter, are important to consider. From our observations with spheres of two distinct diameters, we find that the ring diameter DR is related to the sphere diameter DS by DR ≈ 0.26DS and DR ≈ 0.215DS for rings formed with 780- and 450-nm-diameter colloidal spheres, respectively. Perhaps the simplest model that could account for the observed ring diameter is an ideal geometric one, as depicted in Figure 5. From this diagram, one can obtain the following relation 2
DR
31=2 ( ) π DR ð1Þ ¼ 42DS 2 -2DS 2 cos -cos -1 -4t2 5 2 DS
where t is the nominal ring height. Substituting the empirical diameter ratio data into eq 1 gives a ring height of approximately 12.7 nm for the 780 nm sphere masks and 5.2 nm for the 450 nm sphere masks. The latter agrees quite well with the experimentally observed ring height whereas the former overestimates the ring height by almost a factor of 3. Equation 1 should be applied with care, however, because it does not take into account any deviation from ideal spherical geometry when the colloidal particles interact with the surface and DOI: 10.1021/la803633w 4657
Article
Motavas et al.
Figure 6. (a) SEM image of residual particles sometimes observed to form rings (780 nm sphere sample). (b) AFM image of a control sample showing a residual thin film without ring formation (450 nm sphere sample). adhere to it,16 which can lower the ring height for a given diameter. This effect may be more pronounced for larger spheres where the ring diameter has a greater range and is more sensitive to changes in height. Thus, for a given ring height eq 1 can be considered to give a lower limit for CNT ring diameter as a function of the sphere size used. The ring height itself will be influenced by the size and distribution of CNT bundles in suspension and by intertube/ interbundle interactions. Typically, a combination of individual tubes and bundles 2-5 nm in diameter exists in SWNT suspensions,17 and this is consistent with our AFM observations. In addition, we observed that a larger variation in tube length (e.g., CNT type A) usually resulted in less uniform rings, likely caused by the subsequent variation in ring height and cross-section. Considering defects more generally, sphere packing within the colloidal mask also plays a role. For example, small-diameter defects such as those shown in Figure 2c may be caused by spheres that are slightly elevated from the surface relative to their neighbors. Conversely, larger defects could arise from loose sphere packing that changes the local symmetry and liquid surface tension during evaporation. Sphere packing may also need to be taken into account when considering the sphere deformation discussed above. The CNT ring-formation process will ultimately depend on the strength of the capillary force driving fluid flow at the base of the spheres and the bending energy of the tubes,18 in addition to tube interactions with the sphere/substrate surface and each other. In Figure 6, we examine possible artifacts in the CNT ring array patterning process presented. Figure 6a shows an SEM image of an array of rings consisting of fine particles that are likely ashes or other carbonaceous/catalytic particles present in the CNT dispersion. Such particle rings were quite rare and easily distinguished from the CNT rings. Another potential artifact might occur during the sphere lift-off process in which polystyrene residue could be left behind on the substrate. Figure 6b shows an AFM image of the result obtained from a control sample in which pure methanol solution was used in the colloidal patterning process instead of CNT suspension. In this case, we do not observe any rings, (16) Tabor, D. J. Colloid Interface Sci. 1977, 58, 2–13. (17) Chen, J.; Rao, A. M.; Lyuksyutov, S.; Itkis, M. E.; Hamon, M. A.; Hu, H.; Cohn, R. W.; Eklund, P. C.; Colbert, D. T.; Smalley, R. E.; Haddon, R. C. J. Phys. Chem. B 2001, 105, 2525–2528. (18) Avouris, Ph.; Hertel, T.; Martel, R.; Schmidt, T.; Shea, H. R.; Walkup, R. E. Appl. Surf. Sci. 1999, 141, 201–209.
4658
DOI: 10.1021/la803633w
but instead a fairly uniform residual thin film remains around the original sphere positions, likely from the organic solvent and/or the initial sphere suspension. Thus, polystyrene sphere residues are unlikely to contribute to the CNT ring-formation process.
V. Conclusions The patterning of CNTs from liquid suspension using a colloidal mask is a convenient and effective means of processing nanotubes into well-defined thin film ring arrays over large areas in a parallel manner. This method is not substratespecific, and changing the size of the spheres making up the colloidal mask allows CNT rings with tunable diameters to be generated. Different types of CNTs in various solvents and concentrations should be examined in order to optimize the resulting nanotube distribution and also ascertain the ultimate limits of our approach as a function of colloidal sphere size. Because the physical properties of the CNT rings depend critically on their size, the fabrication of rings with small, tunable diameters is of interest in applications and for uncovering new phenomena in these nanoscale structures. To our knowledge, the colloidal approach is able to produce the smallest SWNT rings observed thus far using any approach. Potential applications of the CNT ring array patterns include nanoelectronics,5 spin-based devices,19 nanophotonics,20 and biosensors.21 In summary, our results demonstrate the controlled assembly of large-area SWNT ring arrays on silicon via colloidal lithography. The facile and flexible nature of this approach coupled with its high throughput and inherent tunability allows the novel properties of nanoscale CNT rings to be studied and developed for applications. Acknowledgment. This work was supported by the Canada Foundation for Innovation, the British Columbia Knowledge Development Fund, and the Natural Science and Engineering Research Council of Canada. We thank the UBC BioImaging Facility for SEM use and Western Economic Diversification Canada for Raman spectroscopy. :: (19) Foldi, P.; K alm an, O.; Benedict, M. G.; Peeters, F. M. Nano. Lett. 2008, 8, 2556–2558. :: (20) Aizpurua, J.; Hanarp, P.; Sutherland, D. S.; Kall, M.; Bryant, G. W.; Garcıa de Abajo, F. J. Phys. Rev. Lett. 2003, 90, 057401-1-4. (21) Kim, S.; Jung, J.-M.; Choi, D.-G.; Jung, H.-T.; Yang, S.-M. Langmuir 2006, 22, 7109–7112.
Langmuir 2009, 25(8), 4655–4658
