Electronic and transport properties of

kinked graphene
Jesper Toft Rasmussen, Tue Gunst, Peter Bøggild, Antti-Pekka Jauho
and Mads Brandbyge*

Full Research Paper Open Access

Address: Beilstein J. Nanotechnol. 2013, 4, 103–110.

Center for Nanostructured Graphene (CNG), Department of Micro- doi:10.3762/bjnano.4.12
and Nanotechnology (DTU Nanotech), Technical University of
Denmark, DK-2800 Kongens Lyngby, Denmark Received: 13 December 2012
Accepted: 31 January 2013
Email: Published: 15 February 2013
Mads Brandbyge* - [email protected]
This article is part of the Thematic Series "Physics, chemistry and biology
* Corresponding author of functional nanostructures".

Keywords: Guest Editors: P. Ziemann and T. Schimmel

adsorption and reactivity; curvature effects; DFT calculations;
electronic transport; graphene nanoribbons; graphene nanostructuring © 2013 Rasmussen et al; licensee Beilstein-Institut.
License and terms: see end of document.

Abstract
Local curvature, or bending, of a graphene sheet is known to increase the chemical reactivity presenting an opportunity for
templated chemical functionalisation. Using first-principles calculations based on density functional theory (DFT), we investigate
the reaction barrier reduction for the adsorption of atomic hydrogen at linear bends in graphene. We find a significant barrier
lowering (≈15%) for realistic radii of curvature (≈20 Å) and that adsorption along the linear bend leads to a stable linear kink. We
compute the electronic transport properties of individual and multiple kink lines, and demonstrate how these act as efficient barriers
for electron transport. In particular, two parallel kink lines form a graphene pseudo-nanoribbon structure with a semimetallic/semi-
conducting electronic structure closely related to the corresponding isolated ribbons; the ribbon band gap translates into a transport
gap for electronic transport across the kink lines. We finally consider pseudo-ribbon-based heterostructures and propose that such
structures present a novel approach for band gap engineering in nanostructured graphene.

Introduction
Nanostructures based on graphene have an enormous potential tronic applications: a band gap can be introduced by nanostruc-
for applications. Especially in future electronic devices compat- turing graphene.
ible with and extending silicon technology, due to the
outstanding electronic transport properties of graphene [1]. A common approach towards engineering the electronic struc-
However, it is crucial to modify the semimetallic electronic ture is to form quasi-1D graphene in the form of nanoribbons
structure of graphene to exploit its full potential for many elec- (GNR) [2]. The electronic structure of GNRs depends on width,

103
 Beilstein J. Nanotechnol. 2013, 4, 103–110.

direction and edge structure – all parameters that to some do not easily react with flat graphene when dosed from a single
degree can be controlled. GNRs can be formed by etching [2], side [22]. However, at positions where there is a substantial
by unzipping carbon nanotubes (CNTs) [3], or ultimately be local bending, rippling or strain of the graphene sheet the re-
grown with atomic-scale precision by using self-assembly of activity changes significantly [5,23]. So far there have been
precursor molecules on metal substrates [4]. However, for elec- only a few theoretical studies of the atomic geometry of hydro-
tronic applications this approach requires a structure-preserving genated ripple structures in unsupported, strain-induced,
means of releasing and transferring the structures to an insu- graphene ripples [24-26]. However, to the best of our knowl-
lating substrate. Bonding of H or other species to graphene with edge, no studies have addressed the reactivity of bends or the
large coverage opens an insulating band gap at the adsorption transport through hydrogenated ripples, or discussed the possi-
sites due to sp3 hybridisation [5]. Periodically ordered clusters bility of stabilising nonplanar structures by hydrogenation.
of adsorbed hydrogen can be formed on graphene in patterns
dictated by the Moiré lattice mismatch between graphene and In this paper we consider the reactivity of linear bends in a
the metal substrate, which opens a semiconducting band gap graphene sheet, and the electronic transport properties of kinks
[6]. Finally, regular perforations, known as a graphene antidot resulting from the hydrogenation of bends. Our starting point is
lattice (GAL) [7], or a nanoscale mesh of holes [8-10] can have the generic graphene structure shown in Figure 1a, which is
neck widths [10,11] down to 5 nm corresponding to band gaps inspired by the experimental observation of trench formation
of the order of 1 eV [2]. [15]. The bulging of this structure results from shortening the
distance between two separated, clamped regions in the sheet.
Graphene consists entirely of surface atoms and is thus exceed- The remaining sections of the paper are organised as follows:
ingly sensitive to the surroundings. In particular, the van der Section “System setup” describes our computational method
Waals (vdW) interaction with the substrate is of importance. and setup. In the subsequent section we present our results.
The substrate interactions, which make graphene cling to small First, we describe the adsorption barriers for the reaction with
features, may be exploited by manufacturing nanostructures in single atomic H on the graphene bend at positions with different
the substrate. Periodic steps in a Cu substrate has been used to local curvature (positions I–VIII in Figure 1a). Then we show
induce “wrinkles” or ripples in graphene with period and height how a linear bend transforms into a kink when decorated by H
on the order of 10 nm [12]. Recently, Hicks et al. [13] demon-
strated how arrays of 1D large band gap, semiconducting
graphene nanoribbons corresponding to a width of ≈1.4 nm can
be formed in graphene on a step-patterned SiC substrate. The
substrate interactions can clamp a graphene sheet while partly
suspended across small holes, so that a pressure difference
between the in- and outside leads to the formation of bubbles or
“blisters” in the sheet [14]. Also, linear folds, where the
graphene sheet is bulging up from the substrate, have been
induced for graphene suspended over trenches by using heat
treatment [15]. Thus, the sheet can obtain significant bends at
certain places induced by the substrate interaction, substrate
nanostructuring, and subsequent treatments [16]. Calculations
by Low et al. [17] showed how a sharp step of height 1 nm in a
SiC substrate, comparable to experimental values [13], can in-
duce a linear bend in the graphene sheet with a radius of curva-
ture down to around 1 nm.

The ability to accurately control sharp local curvatures of

graphene presents opportunities for strain-assisted modification
of the local electronic structure and chemical reactivity in the
Figure 1: (a) Smooth ripple-like structure where the first and last six
graphene sheet, and may open a route to band gap engineered rows of carbon-dimers are surface-clamped regions with a separation
devices [13,18-21]. Very recently, Wu et al. [21] showed how of L = 90 Å. Atomic hydrogen is adsorbed at positions I–VIII. (b) The
resulting kinked graphene structure after hydrogen is adsorbed in lines
graphene on a Si substrate decorated with SiO2 nanoparticles at the most reactive position (II) corresponding to the smallest local
induced local regions of strain and increased reactivity in a radius of curvature. The four kinks divide the structure into five
sections, S1–S5.
selective manner. Atomic hydrogen or other chemical species

104
 Beilstein J. Nanotechnol. 2013, 4, 103–110.

along the most reactive (most curved) line (Figure 1b), and we electronic conductance per unit-cell width transverse to the
present the electronic transmission through a single kink in the bend. To this end we attach semi-infinite flat graphene elec-
subsection “Single kink”. The kink acts as an effective barrier trodes to each side of the selected kinks, i.e., replace sections S1
with its transmission depending on the kink-angle, φ. In the and S2 in Figure 1b by semi-infinite electrodes in order to calcu-
subsection “Two kinks” we study how two parallel kinks lead to late the transmission through the single kink separating S1 and
the formation of a pseudo-ribbon-type electronic structure. S2. In the conductance calculations we employ a dense trans-
Finally, in subsection “Multiple kinks” we demonstrate the verse k-point grid of 400 points.
opening of a transport gap for multikink systems, such as the
one shown in Figure 1b. Results and Discussion
Adsorption barrier
System setup Adsorption of hydrogen on graphene involves a reaction barrier
The bend we consider in Figure 1a is created along the armchair that needs to be overcome before the single hydrogen atom
direction by fixing the first and last six rows of carbon atoms sticks to the graphene sheet. Several investigations based on
and shortening the separation L, while the rest of the atoms are DFT calculations show that atomic hydrogen adsorbs on-top on
allowed to relax. A separation of L = 90 Å is chosen in order to flat graphene with a barrier about 0.2 eV and binding energy in
obtain realistic curvatures [17]. We first assessed the reactivity the range of 0.7–1.0 eV [22,32-34]. Thus a minimum kinetic
of the structure at positions with different local curvature, see energy for the first hydrogen to react [32,33] is required, in an
positions I–VIII in Figure 1a. Subsequently we relaxed the out-of-equilibrium situation such as in an atomic beam [34].
structure where lines of hydrogen (Figure 1b) have been placed Casolo et al. [35] calculated the reaction barrier and adsorption
at the points of lowest radius of curvature, i.e., the points of energy for multiple hydrogen atoms on flat graphene. In agree-
highest local reactivity. This particular system is meant to illus- ment with other studies they found decreased barriers to
trate the potential of the hydrogen adsorption mechanism, and sticking for the second H atom, compared to the barrier for
to gain insight into the modification of the electronic properties adsorbing a single H atom on a clean surface [36].
due to the hydrogen lines. In a corresponding experimental
setup we can imagine placing graphene across a trench, which Here, we first focus on the trends in the change in adsorption
allows hydrogen adsorption on either side of the sheet. barrier as a function of the local curvature of the graphene
sheet. To this end we have considered atomic hydrogen absorp-
The atomic and electronic structure calculations are based on tion at the on-top carbon positions at points with different
density functional theory (DFT) using the SIESTA [27] code, curvature on the bent structure, see Figure 1a (positions I–VIII).
and the PBE-GGA exchange-correlation [28] functional. We The barrier is determined by calculating the total energy for
employ periodic boundary conditions (PBC) in the direction each position of hydrogen above graphene as the hydrogen is
along the bend with a cell-width of four carbon dimers, and 10 moved successively closer to the graphene. Following the
Monkhorst–Pack k-points. We use a mesh cut-off of 500 Ry adsorption investigations on flat graphene by Ivanovskaya et al.
throughout. When calculating reactivity in the form of reduced [37] we perform, in each step, a relaxation of the hydrogen-
reaction barriers the unit-cell is chosen so that the distance bonded carbon atom and its three nearest neighbours. Using the
between single hydrogen atoms is larger than 8.5 Å. This method described above we obtain a reaction barrier of 0.22 eV
ensures low hydrogen–hydrogen interaction, which is known to on locally flat graphene. This is comparable to results obtained
impact reaction barriers [29]. In the total energy calculations of by several groups using DZP or plane wave basis sets and the
relaxed atomic geometries and reaction barriers, we also use PW91 functional [22,29,32,33]. We find that a SZP basis set for
PBC transverse to the bend (5 k-points). We use a TZP basis-set the relaxed carbon atoms yields a reduced barrier height of
for hydrogen and a SZ basis for carbon, except in the reaction 0.18 eV (both basis sets with orbital range corresponding to an
barrier calculations where we compare calculations using a DZP energy shift of 0.01 eV). Hence, we use the SZP basis in the
and SZP basis for the four carbon atoms nearest to the following reaction-barrier calculations in order to lower the
hydrogen. In the barrier determinations we furthermore use computation time.
spin-polarised calculations because of unpaired electrons. For
the relaxed geometries a force tolerance of 0.01 eV/Å is used, For the positions (I-VIII) we obtain the reaction barrier for
and the final energies are corrected for basis-set-superposition adsorption of hydrogen as a function of the local radius of
errors (BSSE) [30]. curvature (RoC) shown in Figure 2. The second least curved
position (VIII), resulting in a large RoC, reduces the barrier by
Based on the computed atomic and electronic structures we roughly 3% compared to flat graphene (position I). The most
subsequently use the TranSIESTA [31] method to calculate the curved position (II) in our considered structure has a minimum

105
 Beilstein J. Nanotechnol. 2013, 4, 103–110.

RoC of ≈20 Å resulting in a barrier reduction of roughly 16%. tions for hydrogen adsorption in studies of flat graphene
For comparison, this RoC roughly corresponds to the radius of a [22,29,40]. With this in mind as well as the curvature-related
(25,25) nanotube. Experiments by Ruffieux et al. [38] compare reduction of reaction barriers, we conclude that the considered
hydrogen adsorption on C60 molecules, CNTs, and graphite to system allows the adsorption of hydrogen atoms in single lines
show that reactivity is increased with curvature. In our case we along armchair-edges. The kink in the atomic structure due to
find that the local electronic density of states changes little for the sp3-binding of a single H makes the graphene curve even
the atoms on the pristine bent graphene sheet (as in Figure 1a). more in its vicinity, which in turn, preferentially lowers the
Thus, we conclude that the lowering of the adsorption barrier barrier for absorption of a H along the linear bend. This
for the moderate RoC of about 20 Å is mainly due to the suggests a mechanism in which the hydrogen adsorption is
mechanical strain in the bend shifting the carbon atoms out of propagating and leads to the decoration of the entire linear bend
the graphene plane in the direction of the hydrogen. We note turning it into a kink line. It may be viewed as analogous to
that an additional increase in reactivity may result from the crack-formation mechanisms, where the breaking of a bond
change in electronic structure for highly bent graphene. Thus, increases the stress on neighbouring bonds; only in this case,
we expect an increase in reactivity for the graphene with a the graphene is hydrogenated rather than broken or destroyed.
linear bend, and a simple Arrhenius estimate using our data
yields a factor of 3–4 at room temperature (300 K). We have Single kink
also performed calculations using the less rigorous DFTB Next, we examine the energetic and transport properties of
method [39] and obtained results in agreement with the trend in kink-lines in the armchair direction. We first consider a single
reaction-barrier reduction obtained above. kink with angle φ, e.g., between sections S1 and S2 in Figure 1b.
The kink-angle φ is varied in the range 0°…90°, while the three
nearest unit cells on each side of the kink are allowed to relax.
The total energy per H is shown as a function of φ in the inset
of Figure 3, showing a minimum energy for φ ≈ 50°. This angle
roughly corresponds to the angle in an sp3 configuration where
2φ = 109.5°. The adsorption of H causes local changes in the
geometry, i.e., only the carbon atoms very close to the kink are
moved, while the remaining structure remains unperturbed. For
this reason, the adsorption of hydrogen atoms can be consid-
ered as a process that locally pins the bend.

Figure 2: Calculated reaction barriers for hydrogenation of bent

graphene as a function of the local radius of curvature (II–VIII in
Figure 1a). Flat graphene (position I) has an infinite radius of curva-
ture and is used to normalise the barriers. Calculations are spin-
polarised and allow for atomic relaxation.

We may understand the reaction barrier and its change with

curvature by considering the changes in carbon bond lengths.
The barrier is due to the fact that the reacting carbon atom has
to be pulled out of the graphene plane, stretching the strong
carbon–carbon bonds, when reacting with the incoming
hydrogen atom. When the graphene sheet is curved the carbon
atom is already slightly out of the plane, and thus the energy Figure 3: Electronic transmission through a single kink normalised by
the transmission of pristine graphene (T0) as a function of the kink
required to pull the atom further out of plane is decreased angle, φ, for electrons (E > 0) and holes (E < 0). The arrow indicates
compared to flat graphene. The ortho- and para-locations in the the normalised transmission at the equilibrium angle determined from
the total energy calculations shown in the inset.
graphene hexagon have been shown to be the preferred loca-

106
 Beilstein J. Nanotechnol. 2013, 4, 103–110.

The electron transmission per unit-cell width is linear in energy sion through two kinks separated by the corresponding aGNR
for pristine graphene in a energy range around the charge width (right panel). In the present case the initial width (or, kink
neutrality point (E = 0), e.g., T0 E. We find similarly that the separation) is N = 17 atomic lines of carbon, which shows a
calculated kink-transmission curves are also linear, and there- semimetallic behaviour in the transmission spectrum with a
fore we express the results for the transmissions in terms of the small transport gap. In accordance with isolated aGNRs the next
roughly energy-independent ratio T/T0 = const. The kink breaks two widths N = 18, 19 are semiconducting, while the last
the electron–hole symmetry, and we fit E > 0 and E < 0 sepa- investigated width N = 20 is semimetallic again (not shown).
rately, as shown in Figure 3. Larger kink angles result in an The close correspondence between the electronic band struc-
increase in the overall transmission, which may be attributed to ture for the GNR and the transmission gap for the double-kink
a better π-orbital overlap across the kink. For the equilibrium system allows us to consider the structure between two kinks as
angle, φ = 50°, the ratio T/T0 is close to 0.17 in both regions a pseudo-ribbon.
(indicated by the arrow in Figure 3), corresponding to a trans-
mission reduction of 83%. Thus, we see that the hydrogen- For the semiconducting pseudo-ribbons transport gaps
induced kinks in graphene can be used to form effective elec- surrounded by van Hove-type 1D behaviour are seen in the
tron barriers. We now turn to the effect of multiple barriers and transmission functions (Figure 4, right panel). The transport
periodic kink structures in order to examine resonant tunnelling gaps, Egap = 0.4 eV and Egap = 0.5 eV, are in reasonable agree-
phenomena and band gap formation. ment with the power-law scaling of Egap with width found for
aGNRs [41]. We note that the pseudo-ribbon breaks the elec-
Two kinks tron–hole symmetry: For the N = 18 case a larger van Hove
Band-structure calculations show that periodic nanoscale resonance is seen at the valence band edge, while for N = 19 a
rippling of the graphene is not sufficient to create a band gap larger resonance is seen at the conduction band edge. There are
[24] due to the low scattering by the elastic deformation [17]. In small transmission values within the electronic band gap due to
contrast, periodic arrangements of adsorbed hydrogen can leakage through the barriers, which we expect to introduce
indeed induce a semiconducting band gap [24,26]. The elec- shifts in the energies between the real and pseudo aGNR.
tronic band structures of hydrogen lines on flat graphene have
been examined by Chernozatonskii et al. [25,26], and recently Multiple kinks
also for nanoscale-rippled graphene [24]. Here we show how In order to illustrate the behaviour of systems with more kinks
two parallel kinks lead to a local electronic structure that resem- we consider a system consisting of four hydrogen-induced
bles that of an isolated GNR between the kinks. Such structures kinks, as illustrated in Figure 1b and Figure 5. The sections
could be produced experimentally by using the techniques S1/S5 are now replaced by left/right infinite-lead structures, and
described by Pan et al. [12]. Hydrogen-terminated armchair the “top” S3 pseudo-ribbon is connected to the leads via the
GNRs are semiconducting but have a small energy gap when “side” S 2 , S 4 pseudo-ribbons. We keep S 2 , S 3 identical for
the width in atomic lines is N = 3L − 1, where L is an integer simplicity and determine the transmission across the kinks (in
[41]. In Figure 4 we compare the electronic bandstructure for the z-direction in Figure 5), which is experimentally more
armchair GNRs (aGNRs) (left panel) to the electronic transmis- feasible. We now investigate how the different sections influ-

Figure 4: (Left) Band structures for H-passivated armchair ribbons with varying width, N. The ribbons are a zero-gap semiconductor, a semicon-
ductor with band gap EN+1 = 0.4 eV, and a semiconductor with band gap EN+2 = 0.5 eV for widths N, N + 1, N + 2, respectively. All widths are based
on the number of carbon atom lines N = 17. The band gaps are indicated by arrows and are highlighted. (Right) The electronic transmission functions
for the corresponding pseudo-ribbons, i.e., across two parallel kinks of varied separation as shown in the inset.

107
 Beilstein J. Nanotechnol. 2013, 4, 103–110.

Figure 5: Projected band structure and transmission through structures with multiple kinks. The top (section S3) and connecting pseudo-ribbons
(sections S2/S4) are varied in width, changing their electronic properties. The projected band structure along the pseudo-ribbons (i.e., kx) is shown by
using filled circles with radii proportional to the weight on section S3. Semimetallic (M) pseudo-ribbons corresponding to N = 17 are shown in blue,
while semiconducting (SC) ribbons of width N + 1 are shown in red. Band gaps of the top sections and connecting sections are high-
lighted. These are = 0.39 eV (b), = 0.42 eV (c), and = 0.35 eV, = 0.27 eV (d). The transmission T(E) (per simple transverse
unitcell) is determined across all kinks in the z-direction, and transmission gaps comparable to the band gaps are observed in subfigures b and d.

ence the total transport for the four possible combinations of seen with a transmission close to that of the metallic double-
semiconducting (SC) and semimetallic (M), corresponding to kink in the previous subsection. The SC/M/SC structure shows
the pseudo-ribbon widths N, N + 1 used in Figure 4. a transport gap similar to that of the single SC pseudo-ribbon
with van Hove resonances, while the S3-projected band struc-
In order to analyse the transmission we single out the band ture reveals isolated metallic states within the gap. We note that
structure projected on to the top section, S3 (excluding carbon the transmission at the resonances for the SC/M/SC structure is
and hydrogen atoms at the kink), in the band structure along the larger than the corresponding M/M/M transmission. For the
pseudo-ribbon direction. The weight on S3 is represented by a case of semiconducting top pseudo-ribbons in Figure 5c and
circle of radius Rnk, Figure 5d, we note that M/SC/M show a greatly reduced trans-
port gap compared to the single pseudo-ribbon case (also, note
the scaling of the transmission axis), while the SC/SC/SC struc-
(1) ture shows a complete extinction of the transmission in the elec-
tronic gap, as expected. Generally, we find that the main behav-
iour of the transmission is controlled by the connecting sections
Here Ψ is the wave function at kx (kz = 0) with n (i) being the S2, S4, i.e., there is a good correspondence between the side
band (orbital) index. The obtained projected band structures are section band gaps and the transmission gaps.
shown in the left parts of each subfigure in Figure 5. Generally,
some bands have no weight (no circles), while others have Conclusion
significant weight indicating that there is little mixing between The presented investigations show that linear kink-line struc-
the orbitals from section S3 and other sections. tures may form in graphene by reacting with atomic hydrogen
along a linear bend in the sheet. The adsorption barrier is
In Figure 5a and Figure 5b we consider pseudo-ribbons with lowered in the close vicinity of the bend, which can be
semimetallic top regions, namely S2/S3/S4 being M/M/M and exploited to form the kink. In particular, we have shown that a
SC/M/SC, respectively. For the all-metal pseudo-ribbons, radius of curvature of ≈20 Å reduces the hydrogen adsorption
M/M/M, an almost energy-independent transmission function is barrier by roughly 16% compared to H adsorption on pristine

108
 Beilstein J. Nanotechnol. 2013, 4, 103–110.

graphene. The calculations suggest that once a single hydrogen 4. Cai, J.; Ruffieux, P.; Jaafar, R.; Bieri, M.; Braun, T.; Blankenburg, S.;
Muoth, M.; Seitsonen, A. P.; Saleh, M.; Feng, X.; Müllen, K.; Fasel, R.
atom has been adsorbed, the induced local kink and resulting
Nature 2010, 466, 470–473. doi:10.1038/nature09211
increase in local curvature makes it easier for the following H to
5. Elias, D. C.; Nair, R. R.; Mohiuddin, T. M. G.; Morozov, S. V.; Blake, P.;
adsorb, thus creating a propagating kink formation. A full line Halsall, M. P.; Ferrari, A. C.; Boukhvalov, D. W.; Katsnelson, M. I.;
of hydrogen atoms pins the structure and divides the electronic Geim, A. K.; Novoselov, K. S. Science 2009, 323, 610–613.
systems into different regions. We have shown that the elec- doi:10.1126/science.1167130

tronic transmission through a single kink is reduced by 83% 6. Balog, R.; Jørgensen, B.; Nilsson, L.; Andersen, M.; Rienks, E.;
Bianchi, M.; Fanetti, M.; Lægsgaard, E.; Baraldi, A.; Lizzit, S.;
compared to pristine graphene, meaning that the kink-line acts
Sljivancanin, Z.; Besenbacher, F.; Hammer, B.; Pedersen, T. G.;
as an efficient barrier for electron motion. We have demon- Hofmann, P.; Hornekær, L. Nat. Mater. 2010, 9, 315–319.
strated how two close-by parallel kinks form a pseudo graphene doi:10.1038/NMAT2710
nanoribbon with similar behaviour of the electronic structure to 7. Pedersen, T. G.; Flindt, C.; Pedersen, J.; Mortensen, N. A.;
that for isolated nanoribbons. The transmission function Jauho, A.-P.; Pedersen, K. Phys. Rev. Lett. 2008, 100, 136804.
doi:10.1103/PhysRevLett.100.136804
displays transport gap features corresponding to the isolated
8. Kim, M.; Safron, N. S.; Han, E.; Arnold, M. S.; Gopalan, P. Nano Lett.
nanoribbon band gaps.
2010, 10, 1125–1131. doi:10.1021/nl9032318
9. Liang, X.; Jung, Y.-S.; Wu, S.; Ismach, A.; Olynick, D. L.; Cabrini, S.;
The present work thus suggests that it may be feasible to Bokor, J. Nano Lett. 2010, 10, 2454–2460. doi:10.1021/nl100750v
template functional electronic nanostructures by using the con- 10. Bai, J.; Zhong, X.; Jiang, S.; Huang, Y.; Duan, X. Nat. Nanotechnol.

formation of graphene, e.g., to the substrate, and that this in turn 2010, 5, 190–194. doi:10.1038/NNANO.2010.8
11. Shimizu, T.; Nakamura, J.; Tada, K.; Yagi, Y.; Haruyama, J.
induce changes in local reactivity. Our work clearly calls for
Appl. Phys. Lett. 2012, 100, 023104. doi:10.1063/1.3675547
extensions in a number of directions. First of all more calcula- 12. Pan, Z.; Liu, N.; Fu, L.; Liu, Z. J. Am. Chem. Soc. 2011, 133,
tions are needed in order to investigate the kink-line propaga- 17578–17581. doi:10.1021/ja207517u
tion reaction proposed by our results. To this end it is important 13. Hicks, J.; Tejeda, A.; Taleb-Ibrahimi, A.; Nevius, M. S.; Wang, F.;
to include a realistic description of the actual substrate. It is also Shepperd, K.; Palmer, J.; Bertran, F.; Le Fèvre, P.; Kunc, J.;
de Heer, W. A.; Berger, C.; Conrad, E. H. Nat. Phys. 2013, 9, 49–54.
interesting to consider other adsorbate species, possibly intro-
doi:10.1038/nphys2487
ducing doping of the pseudo-ribbons and electronic gating.
14. Bunch, J. S.; Dunn, M. L. Solid State Commun. 2012, 152, 1359–1364.
Finally, decoration and pinning of the edges of other geome- doi:10.1016/j.ssc.2012.04.029
tries such as “bubbles” or “blisters” is of interest, e.g., in order 15. Bao, W.; Miao, F.; Chen, Z.; Zhang, H.; Jang, W.; Dames, C.;
to produce GAL-like structures [7] without perforating the Lau, C. N. Nat. Nanotechnol. 2009, 4, 562–566.

graphene sheet. Calculations have shown how the adsorption of doi:10.1038/nnano.2009.191

16. Neek-Amal, M.; Peeters, F. M. Phys. Rev. B 2012, 85, 195445.
hydrogen is correlated over a length scale involving several of
doi:10.1103/PhysRevB.85.195445
the unit cells employed in this work [22,35,42,43]. Thus the 17. Low, T.; Perebeinos, V.; Tersoff, J.; Avouris, P. Phys. Rev. Lett. 2012,
adatom–adatom interaction will play a significant role in the 108, 096601. doi:10.1103/PhysRevLett.108.096601
kink-line propagation along the step and will be addressed in a 18. Levy, N.; Burke, S. A.; Meaker, K. L.; Panlasigui, M.; Zettl, A.;
future study. Guinea, F.; Castro Neto, A. H.; Crommie, M. F. Science 2010, 329,
544–547. doi:10.1126/science.1191700
19. Guinea, F.; Katsnelson, M. I.; Geim, A. K. Nat. Phys. 2010, 6, 30–33.
Acknowledgements doi:10.1038/NPHYS1420
We appreciate helpful discussion with Dr. H. Sevinçli. We 20. Ortolani, L.; Cadelano, E.; Veronese, G. P.; Degli Esposti Boschi, C.;
thank the Danish Center for Scientific Computing (DCSC) for Snoeck, E.; Colombo, L.; Morandi, V. Nano Lett. 2012, 12, 5207–5212.
providing computer resources. The Center for Nanostructured doi:10.1021/nl3023737

Graphene is sponsored by the Danish National Research Foun- 21. Wu, Q.; Wu, Y.; Hao, Y.; Geng, J.; Charlton, M.; Chen, S.; Ren, Y.;
Ji, H.; Li, H.; Boukhvalov, D. W.; Piner, R. D.; Bielawski, C. W.;
dation. JTR and MB thank the Lundbeck foundation for support
Ruoff, R. Chem. Commun. 2013, 49, 677–679.
(R95-A10510). doi:10.1039/C2CC36747E
22. Hornekær, L.; Rauls, E.; Xu, W.; Šljivančanin, Ž.; Otero, R.;
References Stensgaard, I.; Lægsgaard, E.; Hammer, B.; Besenbacher, F.
1. Novoselov, K. S.; Fal’ko, V. I.; Colombo, L.; Gellert, P. R.; Phys. Rev. Lett. 2006, 97, 186102.
Schwab, M. G.; Kim, K. Nature 2012, 490, 192–200. doi:10.1103/PhysRevLett.97.186102
doi:10.1038/nature11458 23. Srivastava, D.; Brenner, D. W.; Schall, J. D.; Ausman, K. D.; Yu, M.;
2. Han, M. Y.; Özyilmaz, B.; Zhang, Y.; Kim, P. Phys. Rev. Lett. 2007, 98, Ruoff, R. S. J. Phys. Chem. B 1999, 103, 4330–4337.
206805. doi:10.1103/PhysRevLett.98.206805 doi:10.1021/jp990882s
3. Xie, L.; Wang, H.; Jin, C.; Wang, X.; Jiao, L.; Suenaga, K.; Dai, H. 24. Wang, Z. F.; Zhang, Y.; Liu, F. Phys. Rev. B 2011, 83, 041403(R).
J. Am. Chem. Soc. 2011, 133, 10394–10397. doi:10.1021/ja203860a doi:10.1103/PhysRevB.83.041403

109
 Beilstein J. Nanotechnol. 2013, 4, 103–110.

25. Chernozatonskii, L. A.; Sorokin, P. B. J. Phys. Chem. C 2010, 114,

3225–3229. doi:10.1021/jp9100653 License and Terms
26. Chernozatonskii, L. A.; Sorokin, P. B.; Brüning, J. W. Appl. Phys. Lett.
2007, 91, 183103. doi:10.1063/1.2800889 This is an Open Access article under the terms of the
27. Soler, J. M.; Artacho, E.; Gale, J. D.; García, A.; Junquera, J.; Creative Commons Attribution License
Ordejón, P.; Sánchez-Portal, D. J. Phys.: Condens. Matter 2002, 14,
(http://creativecommons.org/licenses/by/2.0), which
2745–2779. doi:10.1088/0953-8984/14/11/302
permits unrestricted use, distribution, and reproduction in
28. Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77,
3865–3868. doi:10.1103/PhysRevLett.77.3865 any medium, provided the original work is properly cited.
29. Kerwin, J.; Jackson, B. J. Chem. Phys. 2008, 128, 084702.
doi:10.1063/1.2868771 The license is subject to the Beilstein Journal of
30. We note that our correction is not a proper BSSE correction, since we Nanotechnology terms and conditions:
only calculate the energy of a relaxed (bent) graphene sheet (as in
(http://www.beilstein-journals.org/bjnano)
Figure 1a) with nearby hydrogen ghost atoms, instead of the graphene
structure relaxed in the presence of hydrogen. Our analyses show that
this simpler approach underestimates the found reaction barriers by a The definitive version of this article is the electronic one
few tens of meV. which can be found at:
31. Brandbyge, M.; Mozos, J.-L.; Ordejón, P.; Taylor, J.; Stokbro, K. doi:10.3762/bjnano.4.12
Phys. Rev. B 2002, 65, 165401. doi:10.1103/PhysRevB.65.165401
32. Jeloaica, L.; Sidis, V. Chem. Phys. Lett. 1999, 300, 157–162.
doi:10.1016/S0009-2614(98)01337-2
33. Sha, X.; Jackson, B. Surf. Sci. 2002, 496, 318–330.
doi:10.1016/S0039-6028(01)01602-8
34. Xia, Y.; Li, Z.; Kreuzer, H. J. Surf. Sci. 2011, 605, L70–L73.
doi:10.1016/j.susc.2011.07.009
35. Casolo, S.; Løvvik, O. M.; Martinazzo, R.; Tantardini, G. F.
J. Chem. Phys. 2009, 130, 054704. doi:10.1063/1.3072333
36. Šljivančanin, Z.; Rauls, E.; Hornekær, L.; Xu, W.; Besenbacher, F.;
Hammer, B. J. Chem. Phys. 2009, 131, 084706.
doi:10.1063/1.3187941
37. Ivanovskaya, V. V.; Zobelli, A.; Teillet-Billy, D.; Rougeau, N.; Sidis, V.;
Briddon, P. R. Eur. Phys. J. B 2010, 76, 481–486.
doi:10.1140/epjb/e2010-00238-7
38. Ruffieux, P.; Gröning, O.; Bielmann, M.; Mauron, P.; Schlapbach, L.;
Gröning, P. Phys. Rev. B 2002, 66, 245416.
doi:10.1103/PhysRevB.66.245416
39. Aradi, B.; Hourahine, B.; Frauenheim, T. J. Phys. Chem. A 2007, 111,
5678–5684. doi:10.1021/jp070186p
40. Šljivančanin, Z.; Andersen, M.; Hornekær, L.; Hammer, B.
Phys. Rev. B 2011, 83, 205426. doi:10.1103/PhysRevB.83.205426
41. Barone, V.; Hod, O.; Scuseria, G. E. Nano Lett. 2006, 6, 2748–2754.
doi:10.1021/nl0617033
42. Buchs, G.; Krasheninnikov, A.; Ruffieux, P.; Gröning, P.; Foster, A. S.;
Nieminen, R. M.; Gröning, O. New J. Phys. 2007, 9, 275.
doi:10.1088/1367-2630/9/8/275
43. Boukhvalov, D.; Katsnelson, M.; Lichtenstein, A. I. Phys. Rev. B 2008,
77, 035427. doi:10.1103/PhysRevB.77.035427

110