Selective Oxidation of Carbon Nanotubes into Zigzag Graphene

Aug 18, 2010 - ... zigzag carbon nanotubes could not be broken into zigzag graphene nanoribbons by the same procedure. This result provides a possible...
0 downloads 0 Views 2MB Size
Download PDF
J. Phys. Chem. C 2010, 114, 14729–14733

14729

Selective Oxidation of Carbon Nanotubes into Zigzag Graphene Nanoribbons Yufeng Guo,* Zhuhua Zhang, and Wanlin Guo Institute of Nanoscience, Nanjing UniVersity of Aeronautics and Astronautics, Nanjing 210016, P. R. China ReceiVed: April 13, 2010; ReVised Manuscript ReceiVed: July 20, 2010

Fabrication of graphene nanoribbons of desired edge and width is a critical challenge for its practical applications. Our first-principles calculations and molecular dynamics simulations show that oxygen atoms are favorable to adsorb on the high curvature sidewalls of radially deformed armchair single-walled carbon nanotubes (SWNTs) and form unzipped C-O-C epoxy lines. With further oxidation, the unzipped epoxy lines are broken into carbonyl pairs and the armchair SWNTs transform into bilayer graphene nanoribbons of zigzag edges and approximately identical widths. Under the same radial deformation and oxidation, the formation of normal epoxy groups on the high curvature regions of zigzag SWNTs are more favorable because of different bond arrangement, and the zigzag carbon nanotubes could not be broken into zigzag graphene nanoribbons by the same procedure. This result provides a possible way to selectively produce bilayer graphene nanoribbons of controlled structure from carbon nanotubes by mechanical and chemical coupling actions. Introduction Graphene nanoribbon (GNR), one-dimensional structure or a strip of graphene of nanometer in width, exhibits remarkable physical and chemical properties because of its unique edge states and size effects.1-7 For actual applications in nanoscale system and devices, the structure and edge controllability of GNRs is a paramount issue but remains to be a challenge. Toward this goal, extensive efforts have been attempted to break carbon nanotubes into GNRs by various methods.8-19 In those experiments, oxidation is an important and effective way to unzip carbon nanotubes to obtain GNRs.10-13 The layer number and structure of the yielded GNRs from oxidative unzipping depend on the number of shells in the seed carbon nanotubes and the chirality of each shell. Carbon nanotube with controllable diameter becomes an important material to produce GNR with controlled geometry. When a single-walled carbon nanotube (SWNT) is radially compressed or under pressure, the cross-section deformation leads to two diametrically opposed regions in the nanotube with high curvature, rendering them especially favorable sites for chemical attack.20,21 Once diameter is larger than 2 nm, SWNTs will collapse radially under an external disturbing and be stabilized by van der Waals interaction between the two flattened regions in the tube wall.22,23 As a result, two high curvature edges of distinctly active chemical properties naturally appear on the nanotube opposite sidewalls. Carbon nanotube of radial deformation to its axis can be seen as a special graphenic geometry “biribbon”.20 Previous study reported that the bent section of a carbon nanotube induced by pressure is more reactive than the flat section, and the bent section is more favorable for O-adsorption as higher pressure is applied.24 The O atom bonds with two C atoms when it adsorbs on the carbon nanotube, and a C-O-C epoxy group is formed.24 Depending on the underlying C-C bond broken or not broken, the C-O-C epoxy group can be defined as the unzipped epoxy or normal epoxy group.25 Other study reported that the formation of the epoxy groups on radially compressed armchair SWNT will * To whom correspondence should be addressed: e-mail yfguo@ nuaa.edu.cn; Ph 86-25-84890513; Fax 86-25-84895827.

be enhanced along the high curvature regions.26 On the other hand, it has been shown that oxidizing the unzipped epoxy groups existed on flat graphene sheets leads to the formation of carbonyl groups, and on that site graphene sheets are broken up.27 Thus, radial deformation of SWNT combined with chemical oxidation raises the possibility of controllable fabrication of bilayer GNRs from carbon nanotubes. In this study, we show by density functional theory (DFT) calculations that the unzipped C-O-C epoxy lines are energetically favorable to form on the high curvature edges of armchair SWNTs, which are radially compressed or collapsed. The armchair nanotubes of radial deformation are broken up and transform into zigzag bilayer GNRs with further oxidizing pre-existed unzipped C-O-C epoxy lines into carbonyl pairs. As different C-C bond arrangement, oxygen atoms are prone to form normal epoxy groups on the high curvature regions of compressed zigzag SWNTs, and oxidation breaking the nanotube sidewall along the axis direction is not energetically favored. Thus, the open process does not occur on the radially compressed zigzag SWNTs. Upon compression, controllable fabrication of bilayer nanoribbons from carbon nanotubes is not only feasible but also selective. Model and Method We first study energetically favorable sites for initial Oadsorption on high curvature regions of radially compressed (8, 8) and (16, 0) SWNTs. The nanotubes are compressed into flattened tubes by limiting the y atomic coordinates of top and bottom C atoms, perpendicular to the tube axis, to a desired interlayer distance, as shown in Figure 1a,b. This compression leads to two diametrically opposed regions in the nanotube with high curvature. Theoretical calculations including first principles and empirical potential levels have identified that the structure with a flattened section is energetically minimum for the compressed carbon nanotubes.28-32 A unit cell with a length of 9.7 Å in the longitudinal z direction and 128 C atoms is chosen for the (8, 8) tube, and the length of the unit cell for the (16, 0) tube is 8.4 Å, which also includes 128 C atoms. The periodic boundary condition in the longitudinal direction is applied for the SWNTs. The vacuum regions in the x and y directions are

10.1021/jp103302w  2010 American Chemical Society Published on Web 08/18/2010

14730

J. Phys. Chem. C, Vol. 114, No. 35, 2010

Figure 1. (a, b) Relaxed geometry of radially compressed (8, 8) and (16, 0) SWNTs, which are compressed by limiting the y atomic coordinates. Two oxygen adsorption sites on the high curvature regions of the radially compressed (c) (8, 8) nanotube and (d) (16, 0) nanotube. The cyan and red dots are carbon and oxygen atoms, respectively.

20 Å in the supercell. The unit cells are large enough to avoid any inner cell interaction. The computations were performed using the VASP code with the ultrasoft pseudopotential and local density approximation (LDA) for the exchange-correlation potential.33-35 An energy cutoff of 500 eV and special 16 k points uniformly sampled along the 1D Brillouin zone are used to ensure an energy convergence of less than 1 meV/atom. The whole system is fully relaxed using a conjugate-gradient algorithm until the force on each atom is less than 0.1 eV/nm.

Guo et al.

Figure 2. Minimum-energy pathways of an O atom moving to the opposite end of an initial O atom adsorbing on the radially compressed (a) (8, 8) and (b) (16, 0) SWNTs. The insets show the initial and final configurations.

Figure 3. Radially compressed (8, 8) and (16, 16) SWNTs and collapsed (40, 40) SWNT with two unzipped C-O-C epoxy chains formed on the high curvature regions.

Results and Discussion The high curvature regions caused by radial deformation are easier to be oxidized than other parts of the nanotubes.24 We consider two O-adsorption sites on the high curvature regions of the nanotubes. For armchair nanotube after structural optimization (Figure 1c), the C-C distance beneath the O atom is 2.08 Å for the B site and 1.49 Å for the A site. The breaking of the C-C bond on the B site means the formation of the unzipped epoxy group, while the normal epoxy group is formed on the A site. The binding energy between the nanotube and O atom adsorbing on the B site is 0.86 eV lower than that on the A site. The C-C bonds on the B sites are just vertical to the nanotube axis. Radial deformation produces stronger bending effect on the sp2 hybridization of those vertical C-C bonds, making them more active for O-adsorption. The breaking of the C-C bond in the epoxy group is to maintain the sp2 hybridization, which is energetically favorable.36 For compressed zigzag nanotube (Figure 1d), our DFT calculations show that the C-C distances beneath the O atoms for the A and B site after structural optimization are 1.63 and 1.47 Å, respectively, and no bond breaking happens. Different from armchair tubes, oxygen atoms are prone to form the normal epoxy groups when they adsorb on the high curvature regions of the zigzag nanotubes. Moreover, the binding energy between the nanotube and O atom adsorbing on the A site is 0.02 eV lower than that on the B site. The C-C bonds on the B sites are slightly affected by the radial deformation because these bonds are parallel to the tube axis, while the C-C bonds on the A site which has an angle of 60° to the tube axis will be affected by the radial compression. However, the magnitude of such bending influence on the zigzag nanotubes is much lower than that of the bonds vertical to the tube axis on the armchair nanotubes. To determine the subsequent O-adsorption direction on the high curvature regions of the nanotubes, the minimum energy paths (MEPs) for the second O atom moving on the oxidized nanotubes are mapped out using the nudged elastic band method37 based on DFT. Figure 2 shows the MEPs of the second

O atom moving to the opposite end of the favorable unzipped epoxy site (B site) on the armchair (8, 8) nanotube (Figure 2a) and the favorable normal epoxy site (A site) on the zigzag (16, 0) nanotube (Figure 2b). For armchair nanotube, the final state is 1.6 eV lower than the initial state, and the energy barrier from the initial to final state is 0.83 eV. It should be mentioned that the O-adsorption configuration for the stable final state is still the unzipped epoxy group. For zigzag nanotube, the final state is 0.84 eV lower and the energy barrier is 1.32 eV. The stable oxidation configuration for the final state is the normal epoxy groups. Therefore, it can be concluded that on armchair nanotube oxidation induced unzipped epoxy groups are prone to form in a line along the high curvature regions, which is parallel to the tube longitudinal axis. In contrast, for zigzag nanotube the most stable configuration of the normal epoxy groups on the high curvature regions is not along the tube axis. According to the above results, for radially deformed armchair nanotubes of larger diameters it can be expected that upon oxidation two epoxy chains will form along the tube axis on the opposite edges, as shown in Figure 3. Our DFT calculations show that after structural optimization the C-C bonds of the epoxy chains on the high curvature regions are all broken, and the two epoxy chains are the unzipped epoxy groups. This is consistent with other theoretical predictions that unzipped epoxy groups on graphene sheets have a preference for aligning in a line.27,38 The unzipped C-O-C lines locating on the high curvature regions are still more chemically active than other parts of the nanotubes. Previous study revealed that oxidizing the unzipped epoxy chain already formed on graphene sheet will lead to the formation of carbonyl pairs and breaking of the sheet.27 Here we consider that two opposite unzipped epoxy chains formed on radially compressed armchair nanotubes shown in Figure 3 are attacked by other O atoms. To reduce computation, only a smallest unit cell of 2.42 Å is chosen for the armchair tubes. Figure 4 presents the energetic changes of the armchair

Zigzag Graphene Nanoribbons

J. Phys. Chem. C, Vol. 114, No. 35, 2010 14731

Figure 5. Initial and final structures of the radially compressed (8, 8) SWNT for other free O atoms not absorbing on the unzipped epoxy sites.

Figure 4. Average energetic change for a free O atom to bind and breaking the oxidized site of radially deformed armchair SWNT. Three insets in each panel are the initial (left), intermediate (middle), and final (right) configurations of the SWNT. For collapsed (40, 40) SWNT, only half structure is shown.

nanotubes on which free O atoms approach and bind with the unzipped epoxy lines and finally break the epoxy lines into carbonyl pairs. For the (8, 8) nanotube, it is shown from the insets that O atoms lengthen the original C-O bonds and bond with the carbon atoms therein. The unzipped epoxy lines are ultimately broken into carbonyl pairs and the nanotube transforms into a bilayer GNR of zigzag edges. According to our DFT calculations, the binding energies of an O2 molecular and a C-O bond are 7.42 and 5.5 eV, respectively. From the energetics viewpoint, the formation of the carbonyl pair (two C-O bonds formed) when a free O atom attacks the unzipped epoxy group is therefore energetically favorable. The open process includes the bonding of O atom and the releasing of strain energy when the C-O bonds are broken. The energy gain between the initial and final states is 5.3 eV per unit cell for the (8, 8) nanotube. As the radial compression induced structural deformation is usually symmetrical, the widths of the obtained top and bottom GNRs approximately equal to half circumference of the (8, 8) nanotube. For the (16, 16) and (40, 40) nanotubes, the oxidation induced breaking on the high curvature edges is similar to that of the (8, 8) nanotube, as shown by the insets in Figure 4, and the widths of the GNRs also approximately equal to half circumferences of the nanotubes. The energy gains from the radially deformed (16, 16) and (40, 40) nanotubes to zigzag bilayer GNRs are 4.65 and 5.03 eV, respectively. We have also studied the case where O attack happens far away from the unzipped epoxy line defects, as shown in Figure 5. The energy gain between the initial and final states of the radially compressed (8, 8) nanotube is only 3.47 eV, which is

Figure 6. Snapshots of MD simulations of (a) one free O atom and (b) another O atom interacting with the unzipped epoxy sites of the radially compressed (8, 8) SWNT at the temperature of 1000 K. (c) Snapshots of MD simulations of 0 and 0.2 ps for a free O atom interacting with a normal epoxy group forming on the radially compressed (16, 0) nanotube at temperature of 1000 K. To simulate compression, some top and bottom atoms of the (8, 8) and (16, 0) SWNTs are fixed to remain compression spacing unchanged during relaxation.

much lower than that (5.3 eV) of oxidative attacking on the unzipped epoxy line. This confirms that the unzipped epoxy line defect is more favorable for O adsorption and attacking than other parts of the nanotube. The high curvature regions of the radially deformed armchair nanotubes are then likely to be broken under oxidation circumstance. To confirm the feasibility of the transition from radially compressed armchair nanotube into bilayer GNR, we use DFTbased molecular dynamics (MD) simulation to study the oxidative reaction of the unzipped epoxy line forming on the high curvature region of the (8, 8) nanotube (Figure 6a,b). We first consider that the unzipped epoxy line is interacting with other free O atoms one by one. The temperature is scaled around 1000 K, and the time step is 1 fs. The whole system is fully relaxed in a canonical ensemble. Here the given temperature is high enough to activate the free O atoms to approach and interact with the unzipped epoxy line on the nanotube. After 0.6 ps relaxation the unzipped epoxy sites of the nanotube are broken into carbonyl pairs one by one, as shown in Figure 6a,b. The armchair nanotubes of radial deformation could be opened by oxidation along its high curvature sidewalls. Moreover, we have investigated whether the free O atom could break the normal epoxy group of the radially compressed zigzag (16, 0) SWNT with the same compression distance (Figure 1b). The same MD procedure is employed to study the oxidative reaction process of a normal epoxy group forming on a C-C bond 60° to the nanotube axis (the A site in Figure 1d) when interacting with another free O atom. As shown in Figure 6c, after 0.2 ps MD runs the O atom in the normal epoxy group is dragged out and forms oxygen molecular with the free O atom without any change on the nanotube. Compared with

14732

J. Phys. Chem. C, Vol. 114, No. 35, 2010

Figure 7. Snapshots of MD simulations of free O atoms interacting with the unzipped epoxy chains of the radially compressed (8, 8) and (16, 16) SWNTs and collapsed (40, 40) SWNT at the temperature of 1000 K.

armchair tube, the arrangement of the C-C bond 60° to the nanotube axis receives stronger constraint from other carbon atoms, and the sp2 hybridization of the C-C bond is less affected on the high curvature region of zigzag nanotube. Furthermore, the C-C bond of the normal epoxy group is slightly stretched, and the binding energy of a C-C bond is 10.9 eV in terms of our DFT calculation. If the normal epoxy group is broken into carbonyl pair by a free O atom, it will have to break the underlying C-C bond and a C-O bond. However, this process is not energetically favorable after carefully comparing the binding energies of C-C, C-O, and O-O bonds. As a consequence, the free O atom is favorable to bond with the O atom rather than the C atoms in the normal epoxy group. Those results indicate that the oxidation breaking on the nanotube sidewall could be selective and dependent on the tube chirality if the nanotube is under radial compression. Then we use the same MD simulations to study the case that two opposite unzipped epoxy chains forming on the high curvature regions of armchair nanotubes are simultaneously attacked by other O atoms, as shown in Figure 7. After 0.6 ps relaxation the (8, 8), (16, 16), and (40, 40) nanotubes are opened up and transform into bilayer GNRs of zigzag edges. The unzipped epoxy lines are broken into smooth carbonyl pairs that are formed at the edges. The obtained bilayer GNRs by this way are still stable even the system is annealed to room temperature. The widths of the two graphene nanoribbons are approximately identical because of the structural symmetry of the nanotubes. Actual carbon nanotubes may have various chiralities. To understand the chirality effect, Figure 8a shows the optimized structure of a radially compressed chiral (6, 3) nanotube on which high strain region an O atom adsorbs above a C-C bond nearly perpendicular to the tube axis. The C-C bond underlying the O atom is stretched to 2.07 Å, and an unzipped epoxy group is formed. This indicates that the compressed nanotube can be opened up at this site by other O atom attacking. However, those C-C bonds nearly perpendicular to the tube axis are not always distributing along the high strain edge, as shown in Figure 8a. The high curvature regions of chiral nanotubes under radial compression will be partially opened by oxidation. Radial compression leads to the transformation of the sp2 hybridization of the C-C bonds on the high curvature regions, which is perpendicular or nearly perpendicular to the tube axis, to the sp3 hybridization. On the other hand, the influence of bending on the sp2 hybridization of the C-C bonds can be enhanced by increasing the magnitude of compression as well.

Guo et al.

Figure 8. Geometry of radially compressed (a) (6, 3) and (b) (16, 0) SWNTs after structural optimization. Oxygen atoms adsorb at the high curvature sides of the nanotubes.

We have also studied such effect by considering that the (16, 0) zigzag tube is compressed to the structure of the collapsed (40, 40) tube shown in Figure 3c. The bonding states of the C-C bonds having an angle of 60° to the tube axis at the high curvature regions approximately become the sp3 hybridization. An O atom adsorbs on the same C-C bond (the A site) shown in Figure 1d. After structural relaxation, the underlying C-C bond is broken and the C-C distance becomes 2.16 Å, which means the formation of the unzipped epoxy group, as shown in Figure 8b. Thus, zigzag tubes can be oxidatively opened on the high strain regions under proper radial compression. Different from armchair tube, there are four C-C bonds of the same angle of 60° to the tube axis on the high strain regions and not along the same direction. The final opened edges by oxidation of the zigzag tubes may not be the same smooth as that of the armchair tubes. Nevertheless, those results present a way that carbon nanotubes can be selectively opened into bilayer GNRs by combining of chemical oxidation and mechanical compression. Conclusions In summary, DFT calculations and DFT-based molecular dynamics simulations demonstrate that radially deformed armchair carbon nanotubes could be opened by oxidization of the unzipped epoxy lines forming on the high curvature regions and transform into bilayer GNRs of zigzag edges and approximately identical widths. However, radially compressed zigzag carbon nanotubes could not be broken into zigzag bilayer GNRs by the same mechanical compression and oxidation procedure, as different C-C bond arrangement and smaller change of the sp2 hybridization on the high curvature regions. These results provide a new insight into making zigzag bilayer graphene nanoribbons with ultrasmooth edges and controllable width by using armchair carbon nanotubes. Acknowledgment. This work is supported by the 973 Program (2007CB936204), NSF (10732040, 10947156), Jiangsu NSF (BK2009365) of China, and Innovation Fund of NUAA. References and Notes (1) Novoselov, K. Nature Mater. 2007, 6, 720. (2) Han, M. Y.; Oezyilmaz, B.; Zhang, Y.; Kim, P. Phys. ReV. Lett. 2007, 98, 206805. (3) Wang, X.; Ouyang, Y.; Li, X.; Wang, H.; Guo, J.; Dai, H. J. Phys. ReV. Lett. 2008, 100, 206803. (4) Qian, H.; Negri, F.; Wang, C.; Wang, Z. J. Am. Chem. Soc. 2008, 130, 17970.

Zigzag Graphene Nanoribbons (5) Zhang, Z. H.; Chen, C. F.; Guo, W. L. Phys. ReV. Lett. 2009, 103, 187204. (6) Dutta, S.; Manna, A. K.; Pati, S. K. Phys. ReV. Lett. 2009, 102, 096601. (7) Dalosto, S. D.; Levine, Z. H. J. Phys. Chem. C 2008, 112, 8196. (8) Jiao, L. Y.; Zhang, L.; Wang, X. R.; Diankov, G.; Dai, H. J. Nature 2009, 458, 877. (9) Cano-Marquez, A. G.; et al. Nano Lett. 2009, 9, 1527. (10) Kosynkin, D. V.; Higginbotham, A. L.; Sinitskii, A.; Lomeda, J. R.; Dimiev, A.; Price, B. K.; Tour, J. M. Nature 2009, 458, 872. (11) Zhang, Z. X.; Sun, Z. Z.; Yao, J.; Kosynkin, D. V.; Tour, J. M. J. Am. Chem. Soc. 2009, 131, 13460. (12) Higginbotham, A. L.; Kosynkin, D. V.; Sinitskii, A.; Sun, Z. Z.; Tour, J. M. ACS Nano 2010, 4, 2059. (13) Cataldo, F.; Compagnini, G.; Patane´, G.; Ursini, O.; Angelini, G.; Ribic, P. R.; Margaritondo, G.; Cricenti, A.; Palleschi, G.; Valentini, F. Carbon 2010, 48, 2596. (14) Hirsch, A. Angew. Chem., Int. Ed. 2009, 48, 6594. (15) Terrones, M. ACS Nano 2010, 4, 1775. (16) Kim, K.; Sussman, A.; Zettl, A. ACS Nano 2010, 4, 1362. (17) Paiva, M. C.; Xu, W.; Proenca, M. F.; Novais, R. M.; Laegsgaard, E.; Besenbacher, F. Nano Lett. 2010, 10, 1764. (18) Elias, A. L.; et al. Nano Lett. 2010, 10, 366. (19) Jiao, L. Y.; Zhang, L.; Ding, L.; Dai, H. J. Nano. Res. 2010, 3, 387. (20) Stojkovic, D.; Lammert, P. E.; Crespi, V. H. Phys. ReV. Lett. 2007, 99, 026802. (21) Gu¨lseren, O.; Yildirim, T.; Ciraci, S. Phys. ReV. Lett. 2001, 87, 116802.

J. Phys. Chem. C, Vol. 114, No. 35, 2010 14733 (22) Chang, T. C. Phys. ReV. Lett. 2008, 101, 175501. (23) Gao, G. H.; Cagıny, T.; Goddard III, W. A. Nanotechnology 1998, 9, 184. (24) Zhang, Y. F.; Liu, Z. F. Carbon 2006, 44, 928. (25) Xiang, H. J.; Wei, S. H.; Gong, X. G. arXiv:1001.0171v1, 2009. (26) Alexandre, S. S.; Mazzoni, M. S. C.; Chacham, H. Phys. ReV. Lett. 2008, 100, 146801. (27) Li, Z. Y.; Zhang, W. H.; Luo, Y.; Yang, J. L.; Hou, J. G. J. Am. Chem. Soc. 2009, 131, 6320. (28) Chan, S. P.; Yim, W. L.; Gong, X. G.; Liu, Z. F. Phys. ReV. B 2003, 68, 075404. (29) Sluiter, M. H. F.; Kumar, V.; Kawazoe, Y. Phys. ReV. B 2002, 65, 161402. (30) Elliott, J. A.; Sandler, J. K. W.; Windle, A. H.; Young, R. J.; Shaffer, M. S. P. Phys. ReV. Lett. 2004, 92, 095501. (31) Sun, D. Y.; Shu, D. J.; Ji, M.; Liu, F.; Wang, M.; Gong, X. G. Phys. ReV. B 2004, 70, 165417. (32) Shen, L.; Li, J. Phys. ReV. B 2004, 69, 045414. (33) Kresse, G.; Hafner, J. Phys. ReV. B 1993, 47, 558. (34) Kresse, G.; Hafner, J. Phys. ReV. B 1994, 49, 14251. (35) Vanderbilt, D. Phys. ReV. B 1990, 41, 7892. (36) Froudakis, G. E.; Schnell, M.; Mu¨hlha¨user, M.; Peyerimhoff, S. D.; Andriotis, A. N.; Menon, M.; Sheetz, R. M. Phys. ReV. B 2003, 68, 115435. (37) Mills, G.; Jonsson, H.; Schenter, G. K. Surf. Sci. 1995, 324, 305. (38) Li, J. L.; Kudin, K. N.; McAllister, M. J.; Prud’homme, R. K.; Aksay, I. A.; Car, R. Phys. ReV. Lett. 2006, 96, 176101.

JP103302W