Configuration and electronic properties of graphene nanoribbons on Si(2 1 1) surface

Applied Surface Science 257 (2011) 2474–2480

Contents lists available at ScienceDirect

Applied Surface Science journal homepage: www.elsevier.com/locate/apsusc

Configuration and electronic properties of graphene nanoribbons on Si(2 1 1) surface W. Wang, L.Z. Sun ∗ , C. Tang, X.L. Wei, J.X. Zhong Laboratory for Quantum Engineering and Micro-Nano Energy Technology, Xiangtan University, Yuhu Road, Xiangtan 411105, Hunan, China

a r t i c l e

i n f o

Article history: Received 12 February 2010 Received in revised form 27 September 2010 Accepted 1 October 2010 Available online 23 October 2010 Keywords: Si(2 1 1) surface GNRs Adsorption Electronic structure

a b s t r a c t We perform first-principles calculations based on density functional theory to study the configuration ¯ and [1¯ 1 1] and electronic properties of graphene nanoribbons (GNRs) on Si(2 1 1) surface. Both [0 1 1] adsorption orientations of Si(2 1 1) surface are considered. We find that the adsorption energy is determined not only by the edge states of GNRs, but also by the ribbon width and the orientation of the substrate. Bridge and M-shape adsorption configurations appear gradually as the ribbon width increases. The substrate effectively affects the edge states of GNRs and tends to depress the metallic nature of zigzag GNRs (Z-GNRs) and metallize the armchair GNRs (A-GNRs). © 2010 Elsevier B.V. All rights reserved.

1. Introduction Graphene nanoribbons (GNRs) [1–3], quasi-one-dimensional materials can be patterned from graphene, have recently attracted intense interest as a fascinating building block for nano-electronic and spintronic devices due to their unique transport properties [4] and electronic structure [5–7]. Moreover, their transport properties and electronic structure can be modulated by their chirality, geometry or chemical modification. For example, the H-passivated zigzag GNRs (Z-GNRs) are metallic, whereas the H-passivated armchair GNRs (A-GNRs) show metallic or semiconducting nature depending on their ribbon width [8]. Lots of prototypic devices, such as field effect transistors, have been proposed [9] and fabricated [10] by the GNRs. The interaction between graphene/GNRs and substrate has also drawn a great deal of attention. Mattausch et al. [11] and Varchon et al. [12] reported the adsorption of graphene on SiC. However, they found that the graphene adsorbed on SiC surface does not exhibit the nature of single-layer graphite film until add up to two or multiple layers. Sorkin et al. [13] reported that on the Si-terminated SiC(0 0 0 1) surface the initial planar shape of GNRs is substantially distorted by the underlying substrate, however, as for the C-terminated SiC(0 0 0 1) substrates the planar shape of GNRs reserves. Zhang et al. [14] found that the electronic properties of Z-GNRs adsorbed on Si(0 0 1) substrate strongly depend on their ribbon width and adsorption orientation.

∗ Corresponding author. Tel.: +86 731 5829199. E-mail addresses: [email protected] (L.Z. Sun), [email protected] (J.X. Zhong). 0169-4332/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.apsusc.2010.10.004

In view of the important role of the silicon material in the future nano-technology, the combination of GNRs and silicon is a significant issue for the practical applications of GNRs. There are a lot of experimental and theoretical studies for the adsorption of materials on the silicon substrate [15–19]. However, to our knowledge, the theoretical studies on the interaction between GNRs and the silicon substrate are few. Moreover, the stability and electronic properties of GNRs adsorbed on Si(2 1 1) surface has not been reported as yet. The main purpose of this paper is to explore the adsorption configuration and electronic structure of GNRs adsorbed on Si(2 1 1) substrate. The main reason we choose the Si(2 1 1) surface is that it is a High-Miller-index surface with step reconstruction [20]. The nature of the surface can efficiently modulate the structure and the electronic properties of GNRs adsorbed on it. In our present paper, we find that the stability of GNRs on the Si(2 1 1) surface is mainly determined by the sp3 hybridization between the edge carbon atoms of GNRs and the silicon atoms of the substrate. Consequently, the adsorption energy of GNRs on the surface relies on the ribbon width and the orientation of the substrate. We also find that the substrate tends to depress the metallic nature of Z-GNRs and metallize the A-GNRs. 2. Computational details The structures of Z-GNRs and A-GNRs are shown in Fig. 1a and b. Nz and Na (here we adopt the nomenclature of reference [21]) denote the number of zigzag carbon chains and carbon dimer lines of Z-GNRs and A-GNRs, respectively. In our present work, in order to keep the ribbon width increasing by a complete honey-

W. Wang et al. / Applied Surface Science 257 (2011) 2474–2480

2475

Fig. 1. Schematic diagram of the structures for Z-GNRs (a), A-GNRs (b) and top view of Si(2 1 1) unit cell (c), respectively. (d and e) Two different side views of Si(2 1 1) unit cell. Gray, yellow and blue balls represent carbon, silicon and hydrogen atom, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

combs gradually, the number of Nz increases from 4 to 13 by 1 and Na increases from 7 to 23 by 2. The width is in the range of 7.10–26.28 Å and 7.38–27.06 Å for Z-GNRs and A-GNRs, respectively. Both H-terminated and H-free GNRs are considered here because they have different electronic properties. Fig. 1c–e shows the top view and two different orientated side views of the unit cell of Si(2 1 1) surface. The marked symbols of the bond lengths and the bond angles of the surface adopted from the report of Grein et al. [22]. The optimized bond lengths and the bond angles of the surface (the computational details as follows) are summarized in Table 1. The optimized reconstruction configuration of Si(2 1 1) is in agreement with the report of Grein et al. [22] to some extent. All optimized bond lengths and the bond angle of C1 –A1 –A1 differ from Grein’s within 3.50% and 0.60%, respectively. However, in comparison with the structure reported by Grein et al. [22], the bond angles of C1 –A1 –C3 and A1 –C3 –A3 are enlarged up to 19.15% and 7.66%, and the bond angle of A2 –C1 –A1 is lessened 7.53%, respectively. The deflection mainly derives from the different methods between Grein’s and ours. Based on the two dimensional (2 × 1) reconstructed Si(2 1 1) surface unit cell obtained above, we extend it 4 times along the [1¯ 1 1]

Table 1 Coefficients of Si(2 1 1) surface. The units of bond length and bond angle are angstrom (Å) and degree(◦ ), respectively. d is the change ratio between ours and the results from the previous work [22]. Atoms

Previous work (Ref. [19])

Present work

d

C1 –A1 A1 –A1 A1 –C3 C3 –A3 A3 –C2 C2 –A2 A2 –C1 A2 –C1 –A1 C1 –A1 –C3 C1 –A1 –A1 A1 –C3 –A3

2.33 2.39 2.39 2.37 2.32 2.33 2.35 94.3 108.1 107.2 112.0

2.26 2.34 2.31 2.36 2.38 2.37 2.37 87.2 128.8 107.8 120.6

−3.13% −2.09% −3.35% −0.34% 2.46% 1.68% 0.69% −7.53% 19.15% 0.59% 7.66%

¯ direction to build up the direction and 5 times along the [0 1 1] 4 × 1 and 1 × 5 supercells, respectively. The 4 × 1 and 1 × 5 super¯ and cells are corresponding to the adsorption orientation of [0 1 1] [1¯ 1 1], respectively. 9 and 7 monolayers are chosen for the 4 × 1 and 1 × 5 supercells, respectively. The dangling bonds of silicon atoms at the bottom layer are passivated by hydrogen, and two adjacent image surfaces are separated by a vacuum region around 10–13 Å. The existence of lattice mismatch between GNRs and silicon substrate is inevitable. To reduce the lattice mismatch, we stretch or contract the lattice of the GNRs and the surface trivially. Such adjustments of the lattice may affect the absolute results of each system, but it does not affect the evolution of the adsorption stability and the electronic structure emphasized in our present work. The computational model in our present work is a slab with two surfaces embedded with vacuum regions in a supercell imposed periodic boundary conditions. The dipole correction for such supercell should be considered. However, as for our present model, the slab as shown in Fig. 1 is not a polarized system. The dipole creates by the electrostatic potential between the periodic slabs is typically small. We have used the dipole correction method introduced by Neugebauer et al. [23] test several cases of our models, the dipole correction only slightly affects the adsorption configurations. Moreover, though such correction influences the absolute total energy of each systems, the relative one, namely the adsorption energy derived from the difference between two systems, is unchanged. In view of such validation, the dipole correction is not considered in our present work. Our calculations are performed with the density functional package VASP [24–26]. Ionic potentials are represented by projected augmented wave (PAW) [27], and the exchange correlation potential is described with generalized gradient approximationsPBE functional [28]. In our present work, the largest supercell contains 269 atoms. The plane-wave cutoff energy is chosen as 400 eV. Brillouin zone integrations are performed with Gaussian broadening [29] of 0.1 eV. The Brillouin zone is sampled by using 1 × 1 × 1 Gamma centered Monkhorst-Pack grids for the calculations of relaxation. Gamma centered grids with 1 × 3 × 1 and

2476

W. Wang et al. / Applied Surface Science 257 (2011) 2474–2480

¯ (a) and [1¯ 1 1] (b), H-free N19 -A-GNRs along [0 1 1] ¯ (c) and [1¯ 1 1] (d), H-free N23 -A-GNRs along Fig. 2. Equilibrium adsorption structure for H-free N11 -Z-GNRs along [0 1 1] ¯ (h), H-terminated N10 -Z-GNRs along [0 1 1] ¯ (i) and [1¯ 1 1] (j) direction. ¯ (e), H-free N13 -A-GNRs along [1¯ 1 1] (f), H-free N13 -Z-GNRs (g) and H-free N21 -A-GNRs along [0 1 1] [0 1 1] Gray, yellow and blue balls represent carbon, silicon and hydrogen atom, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

2 × 1 × 1 k-point mesh are used to investigate the electronic struc¯ and the [1¯ 1 1] ture and the adsorption energy along the[0 1 1] direction, respectively. For the optimized calculations, all atoms are allowed to fully relax without any symmetry assumptions. The criteria of energy and force convergence are less than 1 × 10−4 eV per unit cell and 0.05 eV/Å, respectively. 3. Results and discussions 3.1. Structural analysis In order to obtain the stable adsorption configuration, we gradually moved the GNRs normal to the adsorption direction, and then fully relaxed it. By comparing the total energy of each possible ¯ direction the most stable adsorption sites, we find that along [0 1 1] configuration for H-free GNRs is the right edge of GNRs on the Tr or T sites and the left edge of GNRs on the E or T sites, as shown in Fig. 2a, c and g. In these figures we can see that the silicon atoms of Tr or T sites are favorable for sp3 hybridization with the right edge carbon atoms of GNRs and the silicon atoms of E or T sites are favorable for sp3 hybridization with the left edge carbon atoms of GNRs. The sp3 hybridization releases much more energy to stabilize the adsorption configuration. Because the covalent bonding orientation

of E and Tr site atoms restrict the effective sp3 hybridization, the E and Tr sites are unstable for the adsorption of right and left edge of GNRs, respectively. We find that the adsorption configurations of GNRs show a bridge-like structure as the ribbon width N less than 12. The edges of GNRs are obviously bent down because of the restriction of the covalent bonding orientation of the E, Tr and T site atoms. The corrugation of the GNRs in turn increases the system’s energy. The final stable adsorption configuration is determined by the competing effect between the covalent bonding and the corrugation of the GNRs. As the atoms of one edge covalently bond with the substrate, the right edge for instance, the bonding characteristic of another edge is determined by the mismatch extent between the edge and the favorable adsorption sites. The more mismatch the more corrugation of the GNRs is needed. If the elastic energy derived from the corrugation is larger than that of covalent bonding effect, the covalent bond between the left edge atoms and the sub¯ direction, one edge strate will not occur. For example, along [0 1 1] of H-free N13 -, N15 -, N17 - and N23 -A-GNRs can not covalently bond with the substrate, and the adsorption configuration of N23 -A-GNRs ¯ direction is shown in Fig. 2e. Namely, the adsorption along [0 1 1] ¯ direction depends on the ribbon of A-GNRs on Si(2 1 1) along [0 1 1] width. However, as the ribbon width N increases, the adsorption configuration obviously changes to m-shaped structure, the results

W. Wang et al. / Applied Surface Science 257 (2011) 2474–2480

2477

Fig. 3. Electron density difference between the silicon atoms and the dropped carbon atoms of the m-shape N13 -Z-GNRs (a), the middle carbon atoms of the n-shape N7 -Z-GNRs (b) and the edge atoms of the N12 -Z-GNRs(c).

¯ direction are of N13 -Z-GNRs and N21 -A-GNRs adsorbed along [0 1 1] shown in Fig. 2g and h. The results indicate that GNRs exhibit more flexibility and their adsorption show less selectivity as the ribbon width increases. Moreover, our previous studies [30] found that GNRs adsorbed on 6H–SiC(0 0 0 1) surface show a periodic ripple structure. Based on the above results, we can predict that GNRs adsorbed on Si(2 1 1) surface should also show likewise periodic ripple structure. The most stable adsorption configuration along the [1¯ 1 1] direction for H-free GNRs is that their two edges above 4 sites and sp3 hybridize with 4 and 1 site silicon atoms, respectively. The adsorption configuration of GNRs also changes from the bridge-like structure to the m-shaped structure as the ribbon width increases. However, due to the less of atom-parallel steps along [1¯ 1 1] direction of the surface, the edges of GNRs are deformed as they adsorb along this direction, as shown in Fig. 2b, d and f. We also find that Z-GNRs more easily adsorb along [1¯ 1 1] direction than A-GNRs due to their more flexible edges. As for A-GNRs, except for N9 -, N11 -, N15 - and N21 -A-GNRs, one or even two edges can not bond with the substrate, the configuration of N13 -A-GNRs is shown in Fig. 2f. The adsorption configurations of the H-terminated N10 -Z¯ and [1¯ 1 1] are shown in Fig. 2i and GNRs on Si(2 1 1) along [0 1 1] ¯ are favorable for j, respectively. The edges of GNRs along [0 1 1] sp3 hybridization with the silicon atoms of E and Tr site. However, the interaction between the H-terminated GNRs and the substrate along [1¯ 1 1] direction is so weak that the H-terminated GNRs can only be adsorbed on Si(2 1 1) surface physically along [1¯ 1 1] direction. To investigate the bonding nature between GNRs and the Si(2 1 1) surface, we calculated the electron density difference (EDD) of the system. The EDD is an efficient tool to describe the redistribution of the electrons along with the bonding process. The EDD is defined as the difference between the total charge density of the relaxed structure and the sum of neutral atomic charge densities placed at atomic sites, i.e. (r) = (r) −



˛ (r)

(1)

˛

where (r) is the total charge density of the equilibrium structure and ˛ (r) is the charge density of the neutral atom ˛. Therefore, the EDD represents the net charge redistribution as atoms are brought together to form the system. Through the analysis of the EDD, we can clearly express the bonding nature between GNRs and the substrate. We focus the analysis on the interaction between the silicon atoms of Si(2 1 1) surface and three different site carbon atoms of GNRs: the middle carbon atoms of m-shaped, the middle carbon

atoms of n-shaped and the edge carbon atoms of GNRs. Three typical results are shown in Fig. 3. The negative values in the figures denote the electrons moving out relatively to the superposition of the atomic electrons along with the bonding process. Fig. 3a and b shows that there are not obviously charge exchange between the silicon atoms and the dropped carbon atoms of the m-shape N13 -ZGNRs and the middle atoms of n-shape N7 -Z-GNRs. However, the edge carbon atoms of GNRs and the E site silicon atoms form a typical polar covalent bond, as shown in Fig. 3c. The analysis of the EDD of the three cases indicates that the main adsorption sites of GNRs on the substrate locate at their edges. Moreover, the vertical bonding nature between the edge atoms of GNRs and the silicon atoms of the substrate produces the upswept structure of GNRs. 3.2. Adsorption energy To describe the stability of GNRs on Si(2 1 1) surface, we calculated the total adsorption energy (TAE). The TAE is given by: Ead = E(a+s) − Ea − Es

(2)

Here E(a+s) , Ea and Es are the total energy of the system of GNRs on Si(2 1 1), free standing GNRs and isolated Si(2 1 1) surface, respectively. Then the average adsorption energy (AAE) can be obtained through the TAE (Ead ) of GNRs dividing by the total atom number of GNRs. The dependence of the TAE and the AAE on the ribbon width of GNRs is shown in Fig. 4. The results indicate that the AAE reduce gradually as the ribbon width of GNRs increases along both adsorption directions. The AAE changes from −0.61 to −0.18 eV and from −0.44 to −0.15 eV as the ribbon width of Z-GNRs increases from ¯ and [1¯ 1 1] adsorption N4 -Z-GNRs to N13 -Z-GNRs along the [0 1 1] direction, respectively. The AAE of A-GNRs also decreases with the increasing of the ribbon width. The change of the AAE is in the range ¯ and [1¯ 1 1] of −0.37 to −0.06 eV and −0.17 to −0.01 eV along [0 1 1] adsorption direction, respectively. As mentioned above, the adsorption is determined by the interaction between the edge atoms of GNRs and the substrate. So, the total adsorption energy mainly comes from the released binding energy between the edge carbon atoms of GNRs and the silicon atoms of the substrate. Meanwhile, the edge states slightly vary with the increasing of the ribbon width, so the AAE reduces with the increasing of the ribbon width. The AAE of A-GNRs presents a few fluctuations along with the increasing of the ribbon width which is different from the case of Z-GNRs. The TAE and AAE fluctuations of A-GNRs are mainly derived from their bonding condition with the substrate. For example, because the poor bonding characteristics between the edge atoms and the substrate, as shown in Fig. 4, the TAE and AAE of N13 - and N19 -A-GNRs

2478

W. Wang et al. / Applied Surface Science 257 (2011) 2474–2480

0.1

(a)

(b)

0.0 -0.1 -0.2 -0.3

AAE (eV)

-0.5

-11

-11

-12

-12

TAE (eV)

TAE (eV)

-0.4

-13 -14 -15 -16

-0.6

-13 -14 -15 -16 -17

-17 3 4 5 6 7 8 9 10 11 12 13 14

4 5 6 7 8 9 10 11 12 13

N

N

-0.7 0.1

(c)

(d)

0.0 -0.1 -0.2

TAE (eV)

-0.4 -0.5 -0.6

2 0 -2 -4 -6 -8 -10 -12 -14 -16

TAE (eV)

-0.3

8 10 12 14 16 18 20 22 24

2 0 -2 -4 -6 -8 -10 -12 -14 -16

8 10 12 14 16 18 20 22 24

N

N

-0.7 6

8

10

12

14

16

18

20

22

24 6

8

10

12

14

16

18

20

22

24

N ¯ (a) and [1¯ 1 1](b) and H-free A-GNRs along [0 1 1] ¯ (c) and [1¯ 1 1] (d) direction. The insets are the Fig. 4. Average adsorption energy (AAE) for H-free Z-GNRs along [0 1 1] corresponding total adsorption energy (TAE).

along [1¯ 1 1] direction are so small that the adsorption behave as a physical adsorption. The adsorption energy of H-terminated GNRs is also constrained by the chirality of GNRs, the orientation of Si(2 1 1) surface and the ¯ adsorption direction, the average width of GNRs. Along the [0 1 1] adsorption energy rang from −0.01 to −0.12 eV and from −0.02 to −0.01 eV for H-terminated Z-GNRs and H-terminated A-GNRs, respectively. Along the [1¯ 1 1] adsorption direction, the average adsorption energy rang from −0.05 to −0.01 eV and from +0.00 to +0.06 eV for H-terminated Z-GNRs and H-terminated A-GNRs, respectively. The average adsorption energy also decreases with the increasing of the ribbon width. The results indicate that Hterminated A-GNRs can only physically adsorbed on the Si(2 1 1) surface along [1¯ 1 1] direction. 3.3. Density of states To investigate the electronic properties of the most favorable adsorption configurations, we calculate the density of states (DOS) of GNRs on the Si(2 1 1) surface. Fig. 5a shows the DOS of H-free N10 ¯ direction. The DOS peaks around Z-GNRs adsorbed along the [0 1 1] Fermi level induced by the edge states of GNRs reduce due to the coupling effect between GNRs and the substrate. Fig. 5c shows the DOS of H-free N10 -Z-GNRs adsorbed along the [1¯ 1 1] direction. Now the interaction between GNRs and Si(2 1 1) surface is weak compar¯ direction, and the DOS peak around Fermi ing with that of the [0 1 1] level is nearly un-changed. Fig. 5b and d shows that the DOS peaks around Fermi level induced by the edge states of H-terminated N10 -Z-GNRs adsorbed along both directions reduce. Although the free-standing H-free and H-terminated N21 -A-GNRs both show a small energy gap about 0.2 eV, DOS peaks appear around their Fermi level induced by the substrate, as shown in Fig. 6. To briefly sum-

marize, the substrate tends to reduce the metallic characteristics of ZGNRs and metallize the AGNRs. 3.4. A brief discussion: H-contaminated Si(2 1 1) The modulation effects of ideal Si(2 1 1) surface on the structural and electronic properties of GNRs have been discussed above. However, such ideal Si(2 1 1) surface can only survive under extreme, such as high vacuum situation. The surface inevitably interacts with the surrounding environment due to its active dangling bonds. In view of hydrogen is one of the most ubiquitous and simplest pollutants on the Si surface, we take the H-contaminated Si(2 1 1) surface, H-free and H-terminated N11 -Z-GNRs and N21 -A-GNRs as examples to briefly discuss the interaction of the GNRs with the Hcontaminated Si(2 1 1) surface. Fig. 7c and d shows the adsorption ¯ direction of the H configurations of N11 -Z-GNRs along the [0 1 1] contaminated Si(2 1 1) surface. The bridge like adsorption configuration is reserved. The figures indicate that the edge carbon atoms of both H-free and H-terminated N11 -Z-GNRs adsorb on the substrate and form C–Si bonds. The average bond length is 1.98 Å and 2.03 Å for H-free and H-terminated N11 -Z-GNRs, respectively. The AAE is −0.138 eV and +0.128 eV for H-free and H-terminated N11 -ZGNRs, respectively. The adsorption process is exothermal for H-free N11 -Z-GNRs and endothermal for H-terminated N11 -Z-GNRs. The Tr sites of H contaminated Si(2 1 1) surface are still favorable for sp3 hybridization with the right edge carbon atoms of GNRs, and the T sites are favorable for sp3 hybridization with left edge carbon atoms of GNRs. The binding energy between C–Si is 1.73 eV and 1.17 eV larger than that of Si–H and C–H, respectively, the sp3 hybridization between the edge carbon atoms of GNRs and the silicon atoms of substrate releases more energy and stabilizes the adsorption configuration. Some contaminated hydrogen atoms around the edge of GNRs are even pulled away. The results of the DOS of N11 -Z-GNRs

W. Wang et al. / Applied Surface Science 257 (2011) 2474–2480

(a)

(b)

(c)

(d)

2479

30

20

DOS (arb.units)

10

30

20

10

0 -4

-2

0

2

-4

-2

0

2

Energy ¯ direction and density of states for H-free (c) and H-terminated (d) N10 -ZFig. 5. Density of states for H-free (a) and H-terminated (b) N10 -Z-GNRs adsorbed along [0 1 1] GNRs adsorbed along [1¯ 1 1] direction. Real and dashed lines represent the density of states of GNRs adsorbed on the substrate and the corresponding free-standing GNRs, respectively. Fermi level is set to zero.

on the H contaminated Si(2 1 1) surface as shown in Fig. 7a indicate that the H contaminated Si(2 1 1) surface further depressed their metallic nature in comparison with the clean one. Fig. 7a also shows that the peak of the DOS of H-free N11 -Z-GNRs around the Fermi level is lower than that of the H-terminated one. The equilibrium

60

adsorption configurations of H-free and H-terminated N21 -A-GNRs on the H contaminated Si(2 1 1) surface are shown in Fig. 7e and f. The edge carbon atoms of H-free N21 -A-GNRs form C–Si bonds with the substrate and the average bond length is 1.93 Å. The AAE is −0.001 eV. As for H-terminated N21 -Z-GNRs, the edge carbon atoms

(a)

(b)

(c)

(d)

45

30

DOS (arb.units)

15

60

45

30

15

0 -4

-2

0

2

-4

-2

0

2

Energy (eV) ¯ direction and density of states for H-free (c) and H-terminated (d) N21 -AFig. 6. Density of states for H-free (a) and H-terminated (b) N21 -A-GNRs adsorbed along [0 1 1] GNRs adsorbed along [1¯ 1 1] direction. Real and dashed lines represent the density of states of GNRs adsorbed on the substrate and the corresponding free-standing GNRs, respectively. Fermi level is set to zero.

2480

W. Wang et al. / Applied Surface Science 257 (2011) 2474–2480

Fig. 7. Real and dashed lines in (a) and (b) represent the DOS of H-free and H-terminated N11 -Z-GNRs and N21 -A-GNRs adsorbed along the [0 1 − 1] direction of H contaminative Si(2 1 1) surface, respectively. The configurations corresponding to H-free and H-terminated N11 -Z-GNRs and N21 -A-GNRs are shown in (c)–(f), respectively. Gray, yellow and blue balls represent carbon, silicon and hydrogen atom, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of the article.)

only form incomplete C–Si bonds with the substrate whose adsorption energy of each carbon atom is +0.130 eV. Fig. 7b shows that the peak of the DOS of N21 -A-GNRs adsorbed on the H contaminated Si(2 1 1) surface is slightly changed in comparison with those of adsorbed on the clean one. It is worth to noting that due to the passivation of the H on the surface, the adsorption energy of GNRs on the H contaminated Si(2 1 1) surface is higher than those adsorbed on the clean ones.

Fund of Hunan Provincial Education Department (Grant No. 10K065).

4. Conclusion

[6] [7] [8] [9] [10]

Using first-principles method, we find that the adsorption of GNRs on the Si(2 1 1) surface depend not only on the orientation of the substrate but also on the ribbon width. The adsorption energy is mainly determined by the sp3 hybridization between the edge carbon atoms of GNRs and the silicon atoms on the surface. The adsorption configurations show bridge or M-shape structures depending on the ribbon width. The substrate effectively affects the edge states of GNRs and tends to reduce the metallic characteristics of ZGNRs and metallize AGNRs. On the H contaminated Si(2 1 1) surface the metallic nature of ZGNRs is further depressed. Moreover, the adsorption energy of GNRs on the H contaminated Si(2 1 1) surface is higher than those adsorbed on the clean surface. Acknowledgments This work is supported in part by the Specialized Research Fund for the Doctoral Program of Higher Education (Grant No. 20070530008), the Cultivation Fund of the Key Scientific and Technical Innovation Project, Ministry of Education of China (Grant No. 708068), the National Natural Science Foundation of China (Grant Nos. 10874143 and 10774127) and the Scientific Research

References [1] [2] [3] [4] [5]

[11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

K. Nakada, M. Fujita, Phys. Rev. B 54 (1996) 17954. D. Jiang, B.G. Sumpter, S. Dai, J. Chem. Phys. 126 (2007) 134701. V. Barone, O. Hod, G.E. Scuseria, Nano Lett. 6 (2006) 2748. A.N. Andriotis, M. Menon, Appl. Phys. Lett. 92 (2008) 042115. A. Yamashiro, Y. Shimoi, K. Harigaya, K. Wakabayashi, Phys. Rev. B 68 (2003) 193410. K. Wakabayashi, M. Fujita, H. Ajiki, M. Sigrist, Phys. Rev. B 59 (1999) 8271. Y.W. Son, M.L. Cohen, S.G. Louie, Nature 444 (2006) 347. T. Kawai, Y. Miyamoto, O. Sugino, Y. Koga, Phys. Rev. B 62 (2000) R16349. D.A. Areshkin, C.T. White, Nano Lett. 7 (2007) 3253. Xinran Wang, Xiaolin Li, Li Zhang, P.K. Youngki Yoon, Weber, Hailiang Wang, Jing Guo, Hongjie Dai, Science 324 (2009) 768. A. Mattausch, O. Pankratov, Phys. Rev. Lett. 99 (2007) 076802. F. Varchon, R. Feng, J. Hass, X. Li, B.N. Nguyen, C. Naud, P. Mallet, J.Y. Veuillen, C. Berger, E.H. Conrad, L. Magaud, Phys. Rev. Lett. 99 (2007) 126805. V. Sorkin, Y.W. Zhang, Phys. Rev. B 81 (2010) 085435. Z.H. Zhang, W.L. Guo, Appl. Phys. Lett. 95 (2009) 023107. A.S. Foster, M.A. Goslvez, T. Hynninen, R.M. Nieminen, K. Sato, Phys. Rev. B 76 (2007) 075315. B.C. Gupta, I.P. Batra, S. Sivananthan, Phys. Rev. B 71 (2005) 075328. K.A. Ritter, J.W. Lyding, Nanotechnology 19 (2008) 015704. Y. Huang, X.S. Chen, H. Duan, W. Lu, J. Electron. Mater. 36 (2007) 925. E. Cappelli, S. Orlandob, M. Servidoric, C. Scilletta, Appl. Surf. Sci. 254 (2007) 1273. D.J. Chadi, Phys. Rev. B 29 (1984) 785. Y.W. Son, M.L. Cohen, S.G. Louie, Phys. Rev. Lett. 97 (2006) 216803. C.H. Grein, J. Cryst. Growth 180 (1997) 54. Neugebauer, Scheffler, Phys. Rev. B 46 (1992) 16067. G. Kresse, J. Furthmuller, Phys. Rev. B 54 (1996) 11169. G. Kresse, J. Hafner, Phys. Rev. B 47 (1993) 558. G. Kresse, J. Hafner, Phys. Rev. B 49 (1994) 14251. P.E. Blöchl, Phys. Rev. B 50 (1994) 17953. J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865. C. Elsässer, M. Fähnle, C.T. Chan, K.M. Ho, Phys. Rev. B 49 (1994) 13975. C. Tang, L. Meng, L.Z. Sun, K.W. Zhang, J.X. Zhong, J. Appl. Phys. 104 (2008) 113536.