IOP PUBLISHING Semicond. Sci. Technol.

23 (2008) 075026 (5pp)

SEMICONDUCTOR SCIENCE AND TECHNOLOGY

doi:10.1088/0268-1242/23/7/075026

Tight-binding description of patterned graphene

Behnaz Gharekhanlou, Marjan Alavi and Sina Khorasani
School of Electrical Engineering, Sharif University of Technology, PO Box 11365-9363, Tehran, Iran E-mail: khorasani@sharif.edu

Received 2 November 2007, in nal form 13 April 2008 Published 23 May 2008 Online at stacks.iop.org/SST/23/075026 Abstract The existence of an energy gap of graphene is vital as far as nano-electronic applications such as nano-transistors are concerned. In this paper, we present a method for introducing arbitrary energy gaps through breaking the symmetry point group of graphene. We investigate the tight-binding approximation for the dispersion of and electronic bands in patterned graphene including up to ve nearest neighbors. As we show by applying special defects in graphene structure, an energy gap appears at Dirac points and the effective mass of fermions also becomes a function of the number of defects per unit cell. (Some gures in this article are in colour only in the electronic version)

1. Introduction
Graphene as one of the important crystalline forms of carbon has attracted much interest over the recent years. This is mainly because of its unique transport properties, as well as relative ease of fabrication, which makes graphene an ideal medium for many applications ranging from nano-electronics, opto-electronics, superconductivity, Josephson junctions, and even study of ballistic transport and modeling massless Dirac fermions [1]. In 2004, a team led by de Heer [2, 3] did a breakthrough research on the fabrication of extremely high-quality ultrathin graphene layers, and successfully demonstrated the possibility of patterning graphene sheets using conventional lithographic techniques. They suggested possible application of patterned graphene in a new class of high-mobility graphene-based nano-electronic devices. Later on, the same team reported a coherence length of beyond 1 m at 4 K in patterned graphene structures grown on SiC substrates using the vacuum graphitization technique [4]. Son et al [5] calculated the energy band structure of graphene nano-ribbons and made estimations for the corresponding energy gap as a function of the width of nano-ribbons. They showed that the non-ribbons with both zig-zag and arm-chair boundaries would exhibit energy gap. Ostrovsky et al [6] considered the transport of electrons in disordered graphene layers and observed that the characteristics of the transport would strongly depend on the
0268-1242/08/075026+05$30.00

nature of disorders. Also, Khveshchenko [7] reported the Coulomb interaction of Dirac fermions in disordered graphene mono-layers. Connement of Dirac fermions in graphene has been addressed by Peres et al [8]. Optical transitions in graphene in the infrared spectrum have been studied by Sadowski et al [9], and it has been noted that a single-particle model could correctly describe the transition rates, in remarkable agreement with experiments. Recently, Gunlycke et al [10] have showed possible ballistic transport of electrons in nano-ribbons of graphene having a width of the order of 10 nm with zig-zag edges. Their calculations reveal that electronphonon interactions are signicantly suppressed due to the boundary conditions at the edges of ribbons, and the expected mean free path for electrons can be as large as 70 m at the room temperature. Using a tight-binding calculation, Yao et al [11] have theoretically shown that the spinorbit coupling of electrons in graphene leads to the removal of degeneracy at Dirac points, where a very small energy gap of the order of 1 eV is expected. Ohta et al [12] have conducted a signicant experimental study on the electronic band structure of a bi-layer graphene and observed that an electronic band gap at the Dirac point can be induced and controlled using an externally applied Coulomb potential. Pisani et al [13] in a recent paper discuss the strong dependence of electronic structure of nano-ribbons on a magnetic eld, and conclude the possible application of nano-ribbons of graphene in the controlled transport of spin and spintronics. Similarly, Silvestrov and Efetov [14]
1
2008 IOP Publishing Ltd Printed in the UK

Semicond. Sci. Technol. 23 (2008) 075026

B Gharekhanlou et al

Figure 1. Effect of patterning on the graphene band structure at the Dirac point.

have discussed the quantum connement and the existence of bound states of quasi-particles in graphene by the application of externally applied voltage to a localized region of nanoribons. Finally, in the breakthrough research by Bunch et al [15], the application of suspended monolayer graphene ribbons as tunable nano-electromechanical resonators is reported. The authors have measured charge sensitivities in excess of 1.2 1022 C Hz1/2 . Furthermore, Bostwick et al [16] have recently made an extensive series of experiments on graphene using angle-resolved photoelectron spectroscopy and veried the accuracy of quasi-particle picture and Dirac points in the electronic band structure. Despite its excellent transport properties, unfortunately, graphene nds limited use as a basis for electronic devices mainly due to lack of an energy gap [17]. This prohibits the use of graphene in transistors and other planar electronic devices. A very recent paper by Zhou et al [18] discusses a novel method for introducing an energy gap to graphene through substrateinduced strain when epitaxially grown on SiC substrates. It has been experimentally shown that an energy gap as large as 0.26 eV can be expected from strained mono-layer graphene lattices. However, the gap totally disappears when the number of layers increases only to four. In this paper, we report the results of computation of electronic band structure in an orderly patterned graphene mono-layer, in the absence of external electric and magnetic elds. We suppose that the graphene layer is patterned in such a way that carbon atoms are periodically set out of the lattice and replaced by vacancies, while leaving the symmetry point group of the two-dimensional honeycomb structure either unchanged as C6v , or reduced to the sub-group C3v . When the symmetry group is reduced, one normally expects the removal of degeneracies according to Tinkham [19]. Calculations are done using the standard tight-binding method [20], neglecting the spinorbit interactions. We also show opening of an energy gap and the removal of degeneracy at Dirac points, and a dependence of Fermion effective mass as well as energy gap to the distance between vacancies. The dependence of effective mass on other parameters including energy gap compares to two earlier reports by Pennington and Goldsman [21], where a similar effect in single-walled carbon nanotubes has been discussed. As will be discussed, an energy gap is predicted to open due to the removal of degeneracy and the reduction of the
2

Figure 2. Hexagonally-arranged carbon atoms in graphene, unit basis vectors and the three nearest neighbors in the tight-binding model.

symmetry group. Hence, at the same time one may expect this energy gap to persist within the same order regardless of the number of deposited mono-layers as long as the symmetry group is reduced.

2. Calculations
Upon opening a gap in graphene as illustrated in gure 1, the expression for the anisotropic two-dimensional effective mass can be written as m x,y 1 = h2 c,v 2 Ec,v , 2 x,y (1)

where E is the energy of electron, c and v represent the conduction ( ) and valence () bands and is the absolute value of Bloch wavevector measured from the nearest Dirac point along the prescribed directions. Also, the second derivative may be evaluated at the same Dirac point in the reciprocal lattice where = 0. Although the band structure of graphene is not isotropic, however, the effects of anisotropy may be disregarded based on the C6 v symmetry point group of the two-dimensional crystal (2D); however, when discussing the patterned structure, the band structure can become well anisotropic. Within a parabolic approximation to the band

Semicond. Sci. Technol. 23 (2008) 075026

B Gharekhanlou et al

Figure 3. Showing a Dirac point in the band structure of graphene by taking rst neighbors.

Figure 4. Opening of an energy gap at Dirac points due to vacancies.

structure of patterned graphene, E() may be expanded around any of the Dirac points located at K as Ec () = + Ec + h2 1 mc,x |x Kx |2 2 + m 1 |y Ky |2 , c,y h m 1 |x Kx |2 v,x 2 1 2 + mv,y |y Ky | ,
2

(2a)

Ev () = + Ev +

(2b)

respectively, for conduction and valence bands. Here, Ec,v corresponds to the shifts in the peaks of conduction and valence bands due to patterning. Hence, the energy gap Eg will be simply given by Eg = Ec + Ev . (3)
3

Please note the difference of (2a), (2b) to the band structure of unpatterned graphene where the energy dispersion around the Dirac points is given by the linear expression E() = V | K|, with V being the 2D Fermi velocity h [22]. The energy band structure (2a), (2b) of graphene would eventually determine not only the effective mass of electrons and holes, but also the effective velocity of Fermionic Dirac particles which normally propagate in unpatterned graphene at a speed of c/300, with c being the velocity of light in vacuum [1]. Also, as we later observe, there will be a noticeable anisotropy in the effective mass in the patterned graphene. In the tight-binding model applied here, it is assumed that the full Hamiltonian of the system may be approximated by the Hamiltonian of an isolated atom centered at each lattice point, and only the effects of the rst-nearest neighbors are involved.

Semicond. Sci. Technol. 23 (2008) 075026

B Gharekhanlou et al

Figure 5. Typical patterning model of graphene with reduced symmetry (rst- and second-nearest neighbors removed from the lattice).

Hence, the atomic orbitals n (r), which are eigenfunctions of the single-atom Hamiltonian Hat are assumed to be very small at distances exceeding the lattice constant. This is what is normally meant by the tight-binding method. We further assume that any corrections to the atomic potential U (r), which are required to obtain the full Hamiltonian H of the system, are appreciable only when the atomic orbitals are small. The solution to the time-independent single-electron Schr dinger equation (r) is then assumed to be a linear o combination of atomic orbitals (r) =
n

bn

n (r).

(4)

This leads to a matrix equation for the coefcients bm and Bloch energies () of the form m + R=0 m (R) exp(i R) () = Em . (5) bm + R=0 m (R) exp(i R) Here, Em is the energy of the mth atomic level, and m = m (R) = m (R) =
m (r)

U (r)(r) d3 r, R) d3 r,

(6a) (6b) (6c)

m (r)(r m (r)

U (r)(r R) d3 r,

are the corresponding transfer and overlap integrals. Before including the vacancy defects in graphene structure, we rst calculate the band structure of graphene up to ve nearest neighbors using the described tight-binding approach. Then we omit the terms related to the absent atoms from the summation of atomic potential terms in the Hamiltonian. Here, as is normally done in the standard tightbinding approximation, m and m are neglected in (5), and just have applied the effect of the last overlap integral. Also, we note that in graphene, each of the two nearest atoms have atomic orbital n (r) with opposite signs for the overall atomic orbitals as pz . The overlapping orbital obtained by the interference of pz orbitals corresponds to the energy band
4

structure (), and in particular and electronic bands as shown in gure 1. In gure 2, the arrangement of carbon atoms in graphene sheet is illustrated together its unit basis vectors, and the three nearest neighbors surrounding a central atom. As can be seen here, there are three nearest, six second-nearest, three thirdnearest neighbors, etc. In gure 3 Dirac point is clearly seen (marked by the black circle) in the side view from the band structure of graphene. The calculations are done here by taking only the rst-nearest neighbors. As is evident from this gure, and electronic bands are almost completely conical across the Dirac point. We also have calculated the band structure of graphene again this time with taking ve nearest neighbors into account. Since the computational cell size is articially increased, the band structure squeezes in the reciprocal space. Nevertheless, the symmetry and degeneracy at Dirac points are still preserved. In gure 4 the band structure of a patterned graphene is shown, with some atoms periodically taken out of the lattice. As can be easily seen, there is no further degeneracy at the previously Dirac points remains. In other words, an energy gap is opened in justication of the diagram in gure 1. As the number of defects in a cell is increased, by taking the next set of nearest neighbors, the corresponding energy gap varies signicantly as in table 1. Here, the energy gap reaches a maximum, when two of the second-nearest neighbors are deleted. For instance, for the structure illustrated in gure 5, the symmetry point group is reduced from the original C6 v point group to its C3 v sub-group. The new super cell is marked with the dashed hexagon, which now includes 1 + 3 1 + 6 1 + 6 1 = 9 atoms in each cell. 2 3 In contrast, Zhou et al [18] have recently obtained an energy gap of 0.26 eV, a value close to that of our structure, by removing the translational symmetry. They have achieved this symmetry breaking through lattice-mismatched expitaxial growth of graphene on SiC, which results in a strained graphene lattice. Finally, the dependence of the Fermion effective mass on the number of defects is also tabulated in table 1 along the principal orthogonal directions. There is a

Semicond. Sci. Technol. 23 (2008) 075026

B Gharekhanlou et al

Table 1. Energy gap and fermion effective mass along the principal directions versus the number of defects in the graphene structure. Number of neighboring vacancies 0 1 2a 2b 3c 4d 5e 6f
a b

References
[1] Novoselov K S, Geim A K, Morozov S V, Jiang D, Katsnelson M I, Grigorieva I V, Dubonos S V and Firsov A A 2005 Nature 438 197 [2] Berger C et al 2004 J. Phys. Chem. B 108 19912 [3] Berger C et al 2006 Science 312 1191 [4] Hass J, Feng R, Li T, Li X, Zong Z, de Heer W A, First P N, Conrad E H, Jeffrey C A and Berger C 2006 Appl. Phys. Lett. 89 143106 [5] Son Y-W, Cohen M L and Louie S G 2006 Phys. Rev. Lett. 97 216803 [6] Ostrovsky P M, Gornyi I V and Mirlin A D 2006 Phys. Rev. B 74 235443 [7] Khveshchenko D V 2006 Phys. Rev. B 74 161402 [8] Peres N M R, Castro Neto A H and Guinea F 2006 Phys. Rev. B 73 241403 [9] Sadowski M L, Martinez G, Potemski M, Berger C and de Heer W A 2006 Phys. Rev. Lett. 97 266405 [10] Gunlycke D, Lawler H M and White C T 2007 Phys. Rev. B 75 085418 [11] Yao Y, Ye F, Qi X-L, Zhang S-C and Fang Z 2007 Phys. Rev. B 75 041401 [12] Ohta T, Bostwick A, Seyller T, Horn K and Rotenberg E 2006 Science 313 951 [13] Pisani L, Chan J A, Montanari B and Harrison N M 2007 Phys. Rev. B 75 064418 [14] Silvestrov P G and Efetov K B 2007 Phys. Rev. Lett. 98 016802 [15] Bunch J S, van der Zande A M, Verbridge S S, Frank I W, Tanenbaum D M, Parpia J M, Craighead H G and McEuen P L 2007 Science 315 490 [16] Bostwick A, Ohta T, Seyller T, Horn K and Rotenberg E 2007 Nature Phys. 3 36 (Preprint cond-mat/0609660) [17] Novoselov K 2007 Nature Mater. 6 720 [18] Zhou S Y, Gweon G-H, Fedorov A V, First P N, de Heer W A, Lee D-H, Guinea F, Castro Neto A H and Lanzara A 2007 Nature Mater. 6 770 [19] Tinkham M 1964 Group Theory and Quantum Mechanics (New York: McGraw-Hill) [20] Reich S, Maultzsch J, Thomsen C and Ordej n P 2002 Phys. o Rev. B 66 035412 [21] Pennington G and Goldsman N 2003 Phys. Rev. B 68 045426 Pennington G and Goldsman N 2005 Phys. Rev. B 71 205318 [22] Hwang E H and Das Sarma S 2007 Phys. Rev. B 75 205418 (Preprint cond-mat/0610561)

Gap location K K M K K K K K

Energy gap (eV) 0 0.0081 4.1094 0.0103 0 0.01 0.01 0.0104

Fermion effective mass mc/m0 0.02 0.9312 0.0206 0.0534 0.0206 0.0205 0.0212 mv /m0 0.0201 1.5145 0.0207 0.0531 0.0207 0.0206 0.0214

Two of the nearest neighbors removed. One of the rst and one of the second-nearest neighbors removed. c All three nearest neighbors removed. d One of the rst and three of the second-nearest neighbors removed. e One of the rst and four of the second-nearest neighbors removed. f One of the rst and ve of the second-nearest neighbors removed.

remarkable anisotropy in the effective mass, mainly because of the newly reduced point group of the structure.

3. Conclusions
We have presented the electronic band structure of an orderly patterned graphene mono-layer, in the absence of external electric and magnetic elds, under the constraint that the graphene layer is patterned in such a way that carbon atoms are periodically replaced by vacancies. This has been shown to either leave the symmetry point group of the two-dimensional honeycomb structure either unchanged or reduced to the one of its sub-groups. We have demonstrated an adjustable energy gap due to the removal of degeneracy at Dirac points, and a dependence of Fermion effective mass as well as energy gap to the distance between vacancies.