See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/51744745

Point Defects on Graphene on Metals

Article in Physical Review Letters · September 2011

DOI: 10.1103/PhysRevLett.107.116803 · Source: PubMed

CITATIONS READS
126 100

7 authors, including:

Ivan Brihuega Antonio J. Martínez-Galera

Universidad Autónoma de Madrid Universidad Autónoma de Madrid
42 PUBLICATIONS 1,849 CITATIONS 34 PUBLICATIONS 477 CITATIONS

SEE PROFILE SEE PROFILE

Rubén Pérez
Universidad Autónoma de Madrid
134 PUBLICATIONS 4,121 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Electron transport through proteins View project

All content following this page was uploaded by Ivan Brihuega on 28 May 2014.

The user has requested enhancement of the downloaded file.

 Point defects on graphene on metals

M. M. Ugeda,1 D. Fernández-Torre,2 I. Brihuega,1, ∗ P. Pou,2 A.J.

Martı́nez-Galera,1 Rubén Pérez,2 and J. M. Gómez-Rodrı́guez1
1
D. de Fı́sica de la Materia Condensada,Universidad Autónoma de Madrid, E-28049 Madrid, Spain
2
D. de Fı́sica Teórica de la Materia Condensada,
Universidad Autónoma de Madrid, E-28049 Madrid, Spain
Understanding the coupling of graphene with its local environment is critical to be able to integrate
arXiv:1104.1594v1 [cond-mat.mes-hall] 8 Apr 2011

it in tomorrow’s electronic devices. Here we show how the presence of a metallic substrate affects the
properties of an atomically tailored graphene layer. We have deliberately introduced single carbon
vacancies on a graphene monolayer grown on a Pt(111) surface and investigated its impact in the
electronic, structural and magnetic properties of the graphene layer. Our low temperature scanning
tunneling microscopy studies, complemented by density functional theory, show the existence of a
broad electronic resonance above the Fermi energy associated with the vacancies. Vacancy sites
become reactive leading to an increase of the coupling between the graphene layer and the metal
substrate at these points; this gives rise to a rapid decay of the localized state and the quenching of
the magnetic moment associated with carbon vacancies in free-standing graphene layers.

PACS numbers: 73.22.Pr, 73.20.Hb, 68.37.Ef, 71.15.Mb

Exciting properties were supposed for graphene since graphene’s potential, very little is known about the influ-
long time ago [1]. However, it was not till 2004 [2] that ence of the metallic substrate in the properties of modi-
graphene ceased being a theoretical chimera to become fied graphene layers [19]. Here, we show that while prop-
the object of desire of the scientific community. In just erties of pristine graphene adsorbed on a weakly inter-
few years, most of these extraordinary properties have acting metal like Pt(111) are reasonably preserved, the
already been demonstrated [3–5] and many others are situation is dramatically different when point defects are
emerging as a result of the tremendous experimental and introduced in the graphene layer.
theoretical efforts devoted to this material [6, 7]. As a Our starting point is a perfectly clean graphene mono-
consequence, graphene has undoubtedly become one of layer adsorbed on Pt(111), one of the weakest interact-
the most promising candidates to play a key role in future ing graphene-metal systems [10, 17]. As a result of such
technology. Many of the experimental efforts have been a weak interaction, the graphene layer presents several
invested in growing larger and higher quality graphene orientations with respect to the Pt(111) surface, giv-
layers and also in understanding and controlling the cou- ing rise to various moiré patterns [17, 18, 20, 21]. In
pling of graphene with other materials, in particular with addition, recent photoemission experiments have shown
metals, a must to incorporate graphene to real devices. the existence of π bands presenting linear dispersion,
Epitaxial graphene on metals represents an ideal route to the same Fermi velocity as in free standing graphene
fulfill both requirements. Highly perfect graphene sheets (FSG) and a Dirac point slightly shifted to ∼ +300 meV
can be grown on various metals [8–10] and very recently [17]. We formed a complete graphene monolayer on the
macroscopic-sized graphene films have been grown and clean Pt(111) surface by chemical vapor deposition of
subsequently transferred to arbitrary substrates [11, 12]. ethylene in ultrahigh vacuum (UHV) environments at
The interaction of graphene layers with the different temperatures above 1275 K. Our experiments were per-
metallic substrates strongly depends on the metal itself formed at 6 K using a home-made low temperature scan-
[13, 14], but according to its strength two main groups ning tunneling microscope (LT-STM) operating in UHV
can be identified: strongly and weakly interacting sys- [22]. Figure 1a shows a large STM image of an atomi-
tems [10]. While properties of graphene monolayers be- cally perfect graphene/Pt(111) surface, where two moiré
longing to the first group can be quite different with re- patterns
spect to the free-standing case [15], in the weakly inter- √ √can be identified. In the upper-left corner an
( 21 × 21)R11◦ moiré is found, while the rest of the
acting systems the electronic structure of ideal graphene image shows a 3×3 moiré (both periodicities with respect
is basically preserved, as revealed by the experimental to the graphene lattice). Such a 3×3 superperiodicity is
observation of Dirac cones comparable to those of per- due to a 19.1◦ angle between graphene and the Pt(111)
fect graphene [16, 17]. surface, corresponding with a lattice misfit of 0.60% [20].
Till date, graphene on metals research has basically All the experimental results shown here are basically
focused on pristine graphene monolayers adsorbed on independent of the moiré periodicity, thus, for the sake
metal surfaces, which are at present quite well under- of simplicity and to facilitate a straightforward compar-
stood [10, 16, 17]. In contrast, despite the pressing chal- ison with theory, we will restrict ourselves to the 3×3
lenge to tailor graphene layers in order to fully exploit moiré (Fig. 1b), one of the most frequently found in the
 2

Our theoretical results capture the structural properties

of the graphene/Pt(111) system very well, in opposition
to other methodologies that include dispersion interac-
tions into standard DFT functionals [29]. In particular,
our calculated average distance between graphene and
the uppermost Pt layer is 3.35 Å, whereas LEED/LEEM
experiments yield 3.30 Å[17].
Experimental STM images of graphene adsorbed on
Pt(111) (Fig. 1b) can be understood in more depth
with the help of theory. In our approach, we use a non-
equilibrium Green’s function formalism to evaluate the
currents [30], using the OpenMX code [31] to map the
hamiltonian into a local orbital basis, and an idealized
FIG. 1. (color online) a) STM image of the pristine
graphene/Pt(111) surface showing two different moiré struc- Pt apex with a single dz2 orbital to represent the mi-
tures. b) Zoom in of the 3×3 region highlighted in a). Sample croscope tip. This model produces atomically resolved
bias: 50mV, tunneling current: 1.0nA for a) and b). STM images (see Fig. 1c) in good agreement with the exper-
data were measured and analyzed with WSXM [42]. c) Simu- imental results in Fig. 1b. Surprisingly, the brightest
lated STM image of 3×3 graphene/Pt at V=+100 mV. d) features in the calculated image correspond to C atoms
Calculated band structure of the 3×3 Graphene/Pt moiré that are lowest in the graphene sheet. Notice that in
(black lines) and pure graphene (red lines). Pure graphene
the 3×3 moiré the lattice mismatch is very small, and
bands have been shifted by +410 meV. e) STS measurement
of the LDOS of pristine graphene on Pt(111).f) Theoretical the topographic corrugation, measured as the height dif-
DOS of the 3×3 moiré projected on the p states of C atoms ference between the lowest and highest C atoms, is only
(s and d contributions are negligible in this energy range). 0.02 Å. In this case, the observed anticorrelation between
the simulated image and the atomic topography implies
that the STM corrugation is not a geometrical effect but
graphene/Pt(111) system [17]. Information about the lo- instead it can be explained as a purely electronic effect.
cal density of states (LDOS) of the sample was obtained To atomically tailor the graphene layer adsorbed on
with atomic precision by measuring at 6 K differential Pt(111), we have generated point defects on it by irradi-
conductance (dI/dV) spectra in open feedback loop mode ating the surface with 140 eV Ar+ ions, which are known
using the lock-in technique with frequency 2.3 kHz and ac to mainly produce single C vacancies on a graphite sur-
modulation of 1 mV. Our dI/dV spectra measured in the face [32]. Our LT-STM images show that the previously
pristine graphene surface show a clear dip at ∼ +300 mV pristine graphene sheet presents now a number of almost
accompanied by a V shaped rise at both sides, see Fig. 1e. identical bright features, associated with the amount of
Recent scanning tunneling spectroscopy (STS) experi- Ar+ ions reaching the surface, see Fig. 2. The exact
ments measured on a partially covered graphene/Pt(111) shape of these features slightly depends on its location
surface, have reported dI/dV spectra with a slight dip at with respect to the moiré pattern, as shown in Figs. 2b-
+150 mV [23], which was related to the unoccupied sur- e. We see an elongated protrusion occupying an exten-
face state existing in Pt(111) [24]. However, the clear V sion of two honeycomb lattice parameters, surrounded
shape of our spectra around the dip and its location at by a non trivial pattern of high LDOS intensity extend-
+300 mV, coinciding with the Dirac point energy the- ing less than 1nm. In ideal graphene, the presence of
oretically predicted [13] and estimated from photoemis- a point defect should generate √ short-wavelength
√ modu-
sion experiments [17], make us believe that the dip we lations of the LDOS with ( 3 × 3) R30◦ (R3 in the
observe at +300 mV is associated with the position of following) periodicity due to intervalley scattering [33].
the Dirac point in the graphene/Pt(111) surface. An- In well-decoupled graphene layers, as epitaxial graphene
other piece of evidence comes from the results of our on SiC or HOPG surfaces, such R3 patterns associated
density functional theory (DFT) calculations [25] based with atomic-size impurities have been observed by means
on VASP [26]. We have studied the 3×3 moiré using of low bias STM images, which are a measure of the
an empirically corrected version of the PBE functional LDOS at EF , [32, 34–37]. 2D Fourier transforms of our
[27] that includes the effects of van der Waals interac- low bias STM images measured in samples with a high
tions [28]. The calculated bands of the system are shown enough density of generated atomic vacancies, also al-
in Fig. 1d (in black). In the same figure we also in- lowed us to measure such R3 modulations and to esti-
clude the bands of pure graphene (in red) shifted by mate a Fermi wave-vector k F ∼ 0.5nm−1 , which is con-
+410 meV to make them overlap with their equivalent in sistent with ED =+300meV and a Fermi velocity 106 m/s
the moiré. Consistently, the calculated DOS of Fig. 1f, as recently measured by photoemission [17, 25].
which can be directly compared to the experiment in Fig. Contrary to the case of the graphite surface [32],
1e, presents a V-shaped minimum at that same energy. we cannot directly identify point defects generated by
 3

FIG. 2. (color online) a) STM topography, measured at 6 K,

showing the graphene/Pt(111) surface after the Ar+ irradi-
ation. Sample bias: -30mV, tunneling current: 0.8nA. b-e)
Zoom-ins showing the small variation in the bright features
shape for different moiré positions. All bright features on the FIG. 3. (color online) Relaxed single carbon vacancy in a 6×6
image can basically be identified with one of these 4 images. unit cell (4 cells of the 3×3 graphene/Pt(111) moiré ), with
the 2 C atoms that move towards the Pt highlighted: a) top
view, b) side view, c) calculated constant current STM image
at V = +100 mV with the ball and stick atomic atomic model
Ar sputtering in the graphene/Pt(111) surface with the superimposed. d) Experimental STM image of a vacancy from
three fold patterns predicted for single C vacancies on a panel 2a.
FSG layer [34]. Thus, we have to bring into play DFT
calculations in order to unravel the nature of the atomic
point defects observed in this system. In our simulations, and the vacancy itself are located, the dark region to the
we use a supercell with 2×2 units of the 3×3 moiré and out-of-plane C atoms, and the small, oval feature to the
a single C atom removed [25]. For this structure, several other atom in the pentagon ring which belongs to the
adsorption sites are possible, depending on the relative same sublattice as the weakly bonded pair. The modu-
location of the vacancy and the underlying Pt surface. lation of the charge density revealed by the STM image
We have studied some of these possibilities, finding that decays fast. However, we do not try to compare it quan-
the best match to STM experiments occurs for structures titatively with the experiment, since our unit cell is still
which reconstruct similarly to single vacancies on isolated small to model an isolated vacancy, and we believe the
graphene sheets [38]. In the reconstruction of free stand- intensity decay can be affected by interference between
ing graphene, two of the 3 undercoordinated C atoms neigbouring cells.
surrounding the vacancy move closer to each other and Introducing single C vacancies in an isolated graphene
become weakly bonded, forming a pentagon ring, and the layer has a profound impact in its properties. Accord-
third undercoordinated atom moves out of the graphene ing to many theoretical studies [39–41], they give rise to
plane by only ∼ 0.1Å. In graphene/Pt(111), the third quasi-localized states at the Fermi level, which can be as-
atom and one of its neigbours move out of the plane and sociated with the generation of local magnetic moments
towards the Pt by ∼ 1 Å, and form two new chemical around the C vacancies and produce a strong reduction
bonds with Pt atoms of the surface (Figs. 3a and b). of charge carriers’ mobility. Such theoretical expecta-
Besides, the average distance between the graphene sheet tions were recently confirmed by some of us by STS ex-
and the topmost Pt layer has decreased by ∼ 0.08 Å. It periments on C vacancies on the graphite surface, which
is thus clear that the previously inert graphene layer has revealed the presence of a very sharp resonance at the
become very reactive. The strong interaction between Fermi energy extending more than 3 nm away from each
the graphene with vacancy and the metal also results in single C vacancy [32]. It is thus crucial to understand
a quench of the possible magnetic moment of the system, how the coupling with the metallic substrate will affect
with the relaxed structure being non-magnetic. The sim- the properties of such C vacancies.
ulated STM images present a pattern like that in Fig. 3c. Our STS measurements on C vacancies in
The brightest features are a heart-shaped, elongated pro- graphene/Pt(111) show a strong increase of the
trusion and a small, oval protrusion next to it. Another LDOS starting at the Dirac point and reaching a
characteristic feature is a small dark area right next to maximum around +500 meV. As it can be seen in
the elongated protrusion. All these structures can be Fig. 4a, this is reflected in our dI/dV spectra as a
recognized in the experimental images (Fig. 3d). By broad electronic resonance centered at +500 meV with
comparing the simulated image with the position of the a FWHM ∼150 meV. dI/dV spectra acquired on C
atoms (Fig. 3c) we can associate the heart-shaped fea- vacancies located in different positions of the 3×3 moiré,
ture to the region where the two weakly bonded C atoms show that all C vacancies present this broad electronic
 4

Computer time provided by the Spanish Supercomputing

Network (RES), and by the Centro de Supercomputacion
y Visualizacion de Madrid (CeSViMa).

∗
Corresponding author.
Email address: ivan.brihuega@uam.es
FIG. 4. (color online) a) STS measurements of the LDOS on [1] P. R. Wallace, Physical Review 71, 622 (1947).
a C vacancy (blue) and on pristine graphene/Pt(111) (red). [2] K. S. Novoselov et al., Science 306, 666 (2004).
dI/dV were taken consecutively at 6 K with the same micro- [3] K. S. Novoselov et al., Nature 438, 197 (2005b).
scopic tip. b) dI/dV spectra measured on different C vacan- [4] Y. B. Zhang et al., Nature 438, 201 (2005).
cies, numbered 1 to 5. c) dI/dV map, measured at a sample [5] D. L. Miller et al., Science 324, 924 (2009).
bias of +500mV. 1 to 5 point to same vacancies as panel b). [6] A. H. Castro Neto et al., Rev. Mod. Phys. 81, 109 (2009).
[7] A. K. Geim, Science 324, 1530 (2009).
[8] A. T. N’Diaye et al., Phys. Rev. Lett. 97 215501 (2006).
resonance at +500 meV independently of its position [9] A. L. V. de Parga et al., Phys. Rev. Lett. 100 056807
(2008).
inside the moiré, which points to a small influence of
[10] J. Wintterlin, & M. L. Bocquet, Surf. Sci. 603, 1841
the moiré superstructure on its properties. The only (2009).
influence detected was the variation of the resonance [11] K. S. Kim et al., Nature 457, 706 (2009).
height for C vacancies on different moiré positions, see [12] X. S. Li et al., Science 324, 1312 (2009).
Figs. 4b-c. Analogous results were also obtained for [13] G. Giovannetti et al., Phys. Rev. Lett. 101 026803
different moiré patterns. Since the electronic resonance (2008).
associated with any vacancy lays at a fixed energy of [14] A. B. Preobrajenski et al., Phys. Rev. B 78 073401
(2008).
+500 meV, it is possible to map its spatial extension
[15] A. Gruneis, & D. V. Vyalikh, Phys. Rev. B 77 193401
by conductance images at +500 mV. Figure 4c shows (2008).
one of such conductance images measured on a 15×15 [16] I. Pletikosic et al., Phys. Rev. Lett. 102 056808 (2009).
nm2 region where 11 single C vacancies have been [17] P. Sutter, J. T. Sadowski, & E. Sutter, Phys. Rev. B 80
generated by Ar+ irradiation. As it can be observed, the 245411 (2009).
localized state only extends 0.6 nm away from the center [18] T. A. Land et al., Surf. Sci. 264 261 (1992).
of the vacancies. The conductance map also reflects [19] R. Balog et al., Nat. Mater. 9, 315 (2010).
[20] M. Enachescuet al., Phys. Rev. B 60, 16913 (1999).
the resonance height variation found for vacancies in
[21] G. Otero et al. Phys. Rev. Lett. 105, 216102 (2010).
different moiré locations. Our calculations show that [22] M. M. Ugeda, doctoral thesis, UAM (to be published).
this peak is mainly caused by the π-electrons of the two [23] N. Levy et al., Science 329, 544 (2010).
weakly bound atoms in the graphene reconstruction. [24] J. Wiebe et al., Phys. Rev. B 72 193406 (2005).
Thus, our results show that the electronic properties of [25] See supplementary material at
C vacancies in a graphene layer on top of a metal strongly http://link.aps.org/supplemental/XXX.
[26] G. Kresse & J. Furthmller, Phys. Rev. B 54, 11169
differ from the ones found for C vacancies in a well decou-
(1996); G. Kresse & D. Joubert, Phys. Rev. B 59, 1758
pled graphene layer as it is the graphite surface[32]. As (1999).
shown by our DFT calculations these differences can be [27] J.P. Perdew, K. Burke, & M. Ernzerhof, Phys. Rev. Lett.
explained as due to the increase of the interaction with 77, 3865 (1996).
the metallic surface after the C vacancy formation. [28] S. Grimme, J. Comp. Chem. 27, 1787 (2006).
In summary, our findings demonstrate that even in [29] M Vanin et al., Phys. Rev. B 81, 081408(R) (2010).
[30] J.M. Blanco, F. Flores, & R. Perez, Prog. Surf. Sci. 81,
weakly coupled graphene/metal systems the presence of 403 (2006).
the metal has to be seriously taken into account in order [31] http://www.openmx-square.org; T. Ozaki, Phys. Rev. B,
to controllably tune graphene properties by locally mod- 67 (2003), 155108).
ifying its structure. We have shown that the interaction [32] M. M. Ugeda et al., Phys. Rev. Lett. 104 096804 (2010).
with the metal strongly increases when single C vacan- [33] T. Ando, T. Nakanishi, and R. Saito, J. Phys. Soc. Jpn.
cies are introduced in the graphene layer, discarding the 67, 2857 (1998).
appealing possibility of inducing magnetic properties on [34] K. F. Kelly et al., Science 273, 1371 (1996).
[35] P. Mallet et al., Phys. Rev. B 76 (2007).
this graphene system by removing C atoms. [36] G. M. Rutter et al., Science 317, 219 (2007).
This work is supported by MAT2008-02929-NAN, [37] I. Brihuega et al., Phys. Rev. Lett. 101 206802 (2008).
MAT2010-14902, MAT2008-02939-E, CSD2010-00024 [38] O. V. Yazyev & L. Helm, Phys. Rev. b 75, 125408 (2007).
(MICINN, Spain), by PERG05-GA-2009-24209 (EU) and [39] V. M. Pereira et al., Phys. Rev. Lett. 96 036801 (2006).
S2009/MAT-1467 (Comunidad de Madrid, Spain). I.B [40] O. V. Yazyev, Phys. Rev. Lett. 101 037203 (2008).
[41] J. J. Palacios, J. Fernandez-Rossier, and L. Brey, Phys.
and P.P. were supported by the Ramón y Cajal program.
 5

Rev. B 77 195428 (2008).

[42] I. Horcas et al., Rev. Sci. Instrum. 78 013705 (2007).

View publication stats