You are on page 1of 3

NANO LETTERS

Spontaneous Cross Linking of Small-Diameter Single-Walled Carbon Nanotubes


Hansong Cheng,* Guido P. Pez, and Alan C. Cooper
Computational Modeling Center and Corporate Science and Technology Center, Air Products and Chemicals, Inc., 7201 Hamilton BouleVard, Allentown, PennsylVania 18195-1501
Received December 23, 2002; Revised Manuscript Received March 20, 2003

2003 Vol. 3, No. 5 585-587

ABSTRACT
We report surprising findings of spontaneous cross-linking of small-diameter zigzag single-walled carbon nanotubes, with no applied external pressure, using ab initio density functional theory. Sunstantial differences in the cross-linking reactivity of zigzag and armchair single-walled carbon nanotubes are revealed by the calculations. The cross linking creates 2D sheetlike network structures of highly deformed nanotubes linked with covalent CC bonds. The cross-linking process can be directed by H2 chemisorption.

Single-walled carbon nanotubes (SWNT) exhibit unique chemical, physical, and electronic properties that are attributed to their curvature and the quasi-sp2 hybridized carbon atoms of the nanotube walls.1 SWNT with very small diameters, although graphitic, are far removed from graphite because the carbon atoms are forced to adopt C-C-C bond angles closer to those found for tetrahedron sp3 carbons while maintaining only 3-coordinates and thus should display pseudoradical bonding and higher chemical reactivity.2 Although few details of the properties of small (<8 ) carbon nanotubes have been reported, several recent publications have demonstrated that SWNT with diameters as small as 4 can be produced and studied.3,4 There have been theoretical predictions of SWNT interlinking at extremely high pressures5 and experimental evidence for SWNT coalescence by covalent C-C bond formation under electron beam radiation6 and at very high pressures (1.59.5 GPa) and temperatures (200-1500 C).7 A very recent computational study suggests that end-to-end coalescence of SWNT may occur at the hemifullerene caps.8 Here, we report surprising findings of spontaneous cross linking of the side walls of small-diameter zigzag SWNT, within hexagonal bundles, from quantum mechanical calculations. These small-diameter SWNT chemisorb hydrogen molecules (vs physical adsorption found for larger diameter SWNT9), and the cross-linking sites can be directed by this H2 chemisorption. The cross linking creates 2D sheetlike network structures of highly deformed nanotubes linked with covalent C-C bonds.
* Corresponding author. E-mail: chengh@apci.com. 10.1021/nl025970u CCC: $25.00 Published on Web 04/01/2003 2003 American Chemical Society

All calculations were done with the periodic density functional theory under the generalized gradient approximation as proposed by Perdew and Wang10 using a plane-wave basis set as implemented in the VASP package.11,12 Fully nonlocal ultrasoft pseudopotentials were used to describe the core electrons with a cutoff energy of 350 eV. The Brillouin zone integration was performed using the Monkhorst and Pack points of 2 2 4.13 All of the atoms in the unit cells and the cell parameters were fully optimized by employing the conjugated gradient and Newtons algorithms. The present study includes three small zigzag tubes, (5, 0), (6, 0), and (7, 0), with diameters of 4.0, 4.7, and 5.5 , respectively. At nanotube diameters above (7, 0) SWNT, the cross-linking process becomes less energetically favorable relative to the formation of standard nanotube bundles with circular SWNT cross sections (Table 1). For (5, 0) nanotubes, we selected a hexagonal supercell of 20 carbon atoms that contains a single nanotube, for structural optimization, and obtained a meta-stable structure that maintains a hexagonal nanotube bundle with little deformation. Upon further energy minimization, we obtained a minimum-energy structure that reveals a 2D cross-linked network with nanotubes connected by C-C bonds of 1.556 , a typical covalent C-C bond length (Figure 1a). The carbon atoms involved in cross linking adopt characteristic sp3 hybridization. In an oval-shaped, cross-linked nanotube structure, the bond-angle strain imposed on the remaining sp2 carbon atoms from the small diameter is significantly relaxed. The unit cell volume shows a substantial contraction of ca. 9.0% upon the cross-linking phase transition. The cross linking of (5, 0) nanotubes is a spontaneous, nearly barrierless

Table 1. Calculated Total Energies of the Nanotube Bundle and Cross-Linked Structuresa
nanotube type (5, 0) (6, 0) (7, 0)
a

H2 present? no yes no yes no yes

unit cell atoms 20 22 24 26 28 30

bundle (eV) -175.13 -181.99 -213.52 -220.60 -251.93 -259.02

cross-linked (eV) -177.36 -186.04 -214.43 -222.83 -251.26 -258.07

E (eV) -0.11 -0.20 -0.04 -0.09 0.02 0.03

Ephys (kcal/mol) -1.64 -6.71 -6.94

Echem (kcal/mol) -43.59 -37.13 -0.48

The energy change per carbon atom between the cross-linked structure and the nanotube bundle is defined as E.

Figure 1. (a) Cross-linked (5, 0) nanotube structure and (b) crosslinked (5, 0) nanotube structure with chemisorbed hydrogen after full structural optimization.

Figure 2. Potential energy changes along the optimization path of cross-linked (5, 0) nanotubes upon H2 chemisorption. Note that the one bad optimization step corresponds to a structure in which the cross-linking and hydrogenation processes are already completed. This bad step results from overshooting in the Newton optimization algorithm.

process with a 2.04-eV energy decrease per unit cell (0.1 eV/atom). Upon the insertion of one H2 molecule in the unit cell of the (5, 0) nanotube bundle structure followed by full structural optimization, we observed nearly spontaneous dissociation of the H-H bond, the formation of C-H bonds, and, subsequently, the cross linking of the adjacent nanotubes. The chemisorption creates carbon atoms with highly acute bond angles next to the adsorption sites, leading to the directed coalescence of the nanotubes. The optimization was initiated with a perfectly cylindrical nanotube bundle and one exohedral H2 molecule in the unit cell, giving the cross-linked structure shown in Figure 1b. The cross linking and C-H bond formation results in a 3.9% expansion of the unit cell volume. Once again, the process is almost barrierless, as demonstrated in Figure 2, which displays the potential energy changes along the optimization path. It is remarkable that the cleavage of the strong H-H bond in the cross-linking process requires virtually no activation energy. To determine whether the selected small unit cell has any effect on the surprising cross-linking phenomenon, we expanded the supercell by 2 2 1, which contains 80
586

carbon atoms and 8 hydrogen atoms, and performed full structural optimization, including the cell parameters, again. We observed essentially the same cross-linking structures. We therefore conclude that the calculated results are independent of the size of the unit cell. A very similar cross-linking process occurs with (6, 0) nanotubes in a unit cell that contains 24 carbon atoms. The C-C bond distance between nanotubes is 1.582 , slightly longer than the analogous bond in cross-linked (5, 0) nanotubes, which may reflect slightly less strain due to the larger diameter of the nanotubes. Accordingly, the driving force for (6, 0) cross linking is considerably reduced as indicated in Table 1. The cross linking results in a 7.3% contraction in the unit cell volume, somewhat less than what was found for (5, 0) nanotubes. Upon chemisorption of H2 and cross linking by (6, 0) nanotubes, the unit cell volume is expanded by 4.0%, very similar to what is found for (5, 0) nanotubes. Compared with (5, 0) nanotubes, the tendency to cross link is much weaker for (6, 0) nanotubes. The stability of nanotube bundles versus the cross-linked structure is reversed for (7, 0) nanotubes relative to the two smaller zigzag nanotubes. Here, the conventional nanotube
Nano Lett., Vol. 3, No. 5, 2003

bundle is slightly more stable. Contrary to what is observed for (5, 0) and (6, 0) nanotubes, hydrogenation of the (7, 0) nanotube does not provide an additional driving force for cross linking. This reversal of relative stability may be due to the larger diameter of (7, 0) nanotubes, which apparently makes the carbon atoms less reactive toward both cross linking and hydrogenation. These results clearly indicate that (7, 0) is at the boundary above which the zigzag nanotubes will adopt a nanotube bundle structure but below which the cross-linked structure will prevail. An interesting question is whether cross linking will occur for SWNT structural types other than zigzag. We have performed calculations using the same theoretical technique for two small-diameter (3, 3) and (4, 4) armchair nanotubes (4.1- and 5.4- diameter, respectively). Full structural and cell parameter optimization yields stable nanotube bundle structures, with slightly contracted unit cells, and no evidence of spontaneous cross linking. To promote cross linking, we applied external pressure to compress the nanotube bundles slightly, creating nanotube cross sections with an oval-like shape that could possibly facilitate the C-C bond formation because of smaller intertube distances.5 Subsequent energy minimization of the structures and cell parameters led not to cross-linked nanotubes but to a collapse of the nanotubes to a perfect graphite structure. These results indicate that the C-C bonds in these armchair tubes are extremely weak because of the large strain. For (3, 3) and (4, 4) nanotubes, the energy minimization results in 0.44 and 0.27 eV per carbon atom lower energy structures for graphite than in the nanotube bundle. The origin of the difference in cross-linking behavior between zigzag and armchair nanotubes is likely due to the following considerations. First, consider the relative orientation of carbon atoms in neighboring nanotubess the closest carbon atoms between two adjacent armchair tubes are aligned shoulder-by-shoulder whereas in the zigzag tubes they are virtually head-to-head and thus are predisposed to bond formation. Second, the zigzag architecture in general is less stable than that of the armchair, a well-known fact in planar polyaromatics14 because of poorer overlaps of the orbitals, and is thus more vulnerable to cross linking. Indeed, for the tube bundle structure, our calculation shows that on average the zigzag tube (5, 0) is about 0.43 kcal/mol per carbon atom smaller in binding energy than the armchair tube (3, 3), although the sizes of their tube diameters are similar. Finally, we point out that spontaneous dissociation of H2 occurs not only in the bundle of small-diameter zigzag nanotubes but also in isolated nanotubes of the same type. In fact, we performed energy minimization for a H2 molecule in an isolated single (5, 0) nanotube and observed the

spontaneous dissociation of H2. The optimized, partially hydrogenated nanotube exhibits an oval-shaped cross section, similar to what was observed in nanotube bundles. The new 2D cross-linked materials derived from smalldiameter carbon nanotubes should exhibit novel physical, chemical, and electronic properties. If the small-diameter SWNT previously synthesized in zeolite host materials3 can be isolated (free of host), then it is likely that these SWNT will exist in the more stable cross-linked form. As an analogue to useful materials formed from cross-linked polymer chains, aligned SWNT fibers may show enhanced mechanical properties and stability upon covalent cross linking. It remains to be seen whether other small-diameter nanotubes of any chirality, either single- or multiwalled, will undergo cross-linking processes. With the appropriate C-C orientation between adjacent carbon nanotubes and/or external pressure, cross linking should be anticipated. Acknowledgment. We acknowledge Air Products and Chemicals, Inc. for encouraging this work and for the publication of these results. Supporting Information Available: Cross-linked (6, 0) nanotube structure and cross-linked (6, 0) nanotube structure with chemisorbed hydrogen after full structural optimization and the trajectories along the optimization paths. This material is available free of charge via the Internet at http:// pubs.acs.org. References
(1) Saito, R.; Dresselhaus, G.; Dresselhaus, M. S. Physical Properties of Carbon Nanotubes; Imperial College Press: London, 1998. (2) Kostov, M.; Cheng, H.; Cooper, A. C.; Pez, G. P. Phys. ReV. Lett. 2002, 89, 6105. (3) Wang, N.; Tang, Z. K.; Li, G. D.; Chen, J. S. Nature (London) 2000, 408, 50. (4) Recent experiments suggest that these 4- nanotubes are predominantly of the (5, 0) type: Li, G. D.; Tang, Z. K.; Wang, N.; Wong, K. H.; Chen, J. S. Stud. Surf. Sci. Catal. 2001, 135, 3341. (5) Yildirim, T.; Gulseren, O.; Kilic, C.; Ciraci, S. Phys. ReV. B: Condens. Matter2000, 62, 12648. (6) Terrones, M.; Terrones, H.; Banhart, F.; Charlier, J.-C.; Ajayan, P. M. Science (Washington, D.C.) 2000, 288, 1226. (7) Khabashesku, V. N.; Gu, Z.; Brinson, B.; Zimmerman, J. L.; Margrave, J. L.; Davydov, V. A.; Kashevarova, L. S.; Rakhmanina, A. V. J. Phys. Chem. B 2002, 106, 11155. (8) Zhao, Y.; Yakobson, B. I.; Smalley, R. E. Phys. ReV. Lett. 2002, 88, 5501. (9) Cheng, H.; Pez, G. P.; Cooper, A. C. J. Am. Chem. Soc. 2001, 123, 5845. (10) Perdew, J. P.; Wang, W. Phys. ReV. B 1992, 45, 13244. (11) Kresse, G.; Hafner, J. Phys. ReV. B 1993, 47, 558. (12) Kresse, G.; Furthmuller, J. Phys. ReV. B 1996, 55, 11169. (13) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188. (14) Clar, E. Polycyclic Hydrocarbons; Academic Press: New York, 1964.

NL025970U

Nano Lett., Vol. 3, No. 5, 2003

587

You might also like