Dual path mechanism in the thermal reduction of graphene oxide

Article abstract of DOI:10.1021/ja205168x

Graphene is easily produced by thermally reducing graphene oxide. However, defect formation in the C network during deoxygenation compromises the charge carrier mobility in the reduced material. Understanding the mechanisms of the thermal reactions is essential for defining alternative routes able to limit the density of defects generated by carbon evolution. Here, we identify a dual path mechanism in the thermal reduction of graphene oxide driven by the oxygen coverage: at low surface density, the O atoms adsorbed as epoxy groups evolve as O₂ leaving the C network unmodified. At higher coverage, the formation of other O-containing species opens competing reaction channels, which consume the C backbone. We combined spectroscopic tools and ab initio calculations to probe the species residing on the surface and those released in the gas phase during heating and to identify reaction pathways and rate-limiting steps. Our results illuminate the current puzzling scenario of the low temperature gasification of graphene oxide.

Full text of DOI:10.1021/ja205168x

ARTICLE
pubs.acs.org/JACS
Dual Path Mechanism in the Thermal Reduction of Graphene Oxide
,
†
‡,§
‡,§
||
^,#
Rosanna Larciprete,* Stefano Fabris, Tao Sun, Paolo Lacovig, Alessandro Baraldi, and
,
||
Silvano Lizzit*
†
CNR-Institute for Complex Systems, via Fosso del Cavaliere 100, I-00133 Roma, Italy
CNR-IOM DEMOCRITOS, Theory@Elettra group, S.S. 14 km 163.5, I-34149 Trieste, Italy
‡
§
SISSA ꢀ Scuola Internazionale Superiore di Studi Avanzati, via Bonomea 265, I-34136 Trieste, Italy
Sincrotrone Trieste S.C.p.A., Strada Statale 14 Km 163.5, I-34149 Trieste, Italy
^
Physics Department and CENMAT, University of Trieste, Via Valerio 2, I-34127 Trieste, Italy
Laboratorio TASC CNR-IOM, S.S. 14 Km 163.5, I-34149 Trieste, Italy
#
S Supporting Information
b
ABSTRACT:
Graphene is easily produced by thermally reducing graphene oxide. However, defect formation in the C network during
deoxygenation compromises the charge carrier mobility in the reduced material. Understanding the mechanisms of the thermal
reactions is essential for defining alternative routes able to limit the density of defects generated by carbon evolution. Here, we
identify a dual path mechanism in the thermal reduction of graphene oxide driven by the oxygen coverage: at low surface density, the
O atoms adsorbed as epoxy groups evolve as O leaving the C network unmodified. At higher coverage, the formation of other
2
O-containing species opens competing reaction channels, which consume the C backbone. We combined spectroscopic tools and ab
initio calculations to probe the species residing on the surface and those released in the gas phase during heating and to identify
reaction pathways and rate-limiting steps. Our results illuminate the current puzzling scenario of the low temperature gasification of
graphene oxide.
1
3,14
’
INTRODUCTION
topological defects and C vacancies
and to the permanent
1
1,12,15,16
,17,18
inclusion of foreign atoms in the graphene lattice5.
The possibility of efficiently reducing graphene oxide (GO)
Currently, GO is deoxygenated via chemical
as well as
while maintaining optimal transport properties is a challenging
issue that needs to be faced in view of widening the versatility of
graphene in electronics and optoelectronics.
GO of being solution processable and easy to handle is provided
by O atoms attached to the graphene scaffold in several oxidizing
groups, with a relative abundance controlled by the preparation
8,10,19ꢀ21
thermal routes.
In this context, the appeal of using
thermal processing is reinforced by the possibility to remove
oxygen with submicrometric lateral resolution by handling
1
ꢀ3
The property of
22,23
nanosized heat sources.
reduction of GO occurs around 450 K,
H O, CO, and CO in the gas phase. The latter process originates
It is well assessed that thermal
8,10,19
with the release of
2
,4ꢀ6
2
2
methods.
Graphene oxide is an insulating material, and
vacancies in the graphene basal plane and consumes the edges.
Thermal desorption of molecular oxygen is instead either not
progressive elimination of oxygen results in a gradual increase
of the electronic mobility.
of reduced GO are far from being comparable to those of
ideal dopant-free graphene because the sp conjugation is only
7
ꢀ10
However, the transport properties
19,21
10
detected
or present only in traces. Moreover, the GO mass
2
1
1,12
partially retrieved.
Actually, materials resulting from the oxo-
Received: June 4, 2011
reduction cycle present severe limitations due to the generation of
Published: August 16, 2011
r 2011 American Chemical Society
17315
dx.doi.org/10.1021/ja205168x
|J. Am. Chem. Soc. 2011, 133, 17315–17321

Journal of the American Chemical Society
ARTICLE
Figure 1. Surface composition and thermal annealing of graphene with low O coverage. (a,b) Top: O1s and C1s core level spectra measured on
graphene/Ir(111) with an oxygen coverage θ = 0.03 ML. The best fit curves and the spectral components are indicated by the blue lines and the colored
peaks. Bottom: Coded density plots of the O1s and C1s spectral intensities versus temperature. The C1s image is saturated to highlight the evolution of
the minor spectral components. (c,d) O1s (c) and C1s (d) component intensities versus temperature measured at a rate of 0.2 K/s. The two bars in (c)
mark the inflection points of the O1 and O2 curves. (e) TPD curves measured during annealing of the sample at a rate of 2 K/s. The curves are
background subtracted and vertically shifted for clarity.
loss attributed indistinctly to the decomposition of functional
groups anchored directly to GO or attached to oxidized debris
reactions leading to molecular oxygen and CO/CO evolution
are always assisted by epoxide diffusion and have comparable
energy barriers.
2
5,10,12,19,20
non covalently bonded to GO
still needs to be clarified.
Carbon gasification during GO reduction is determined by the
nature and concentration of the oxidizing groups and the manifold
of surface reactions thermally activated that compete with each
other. This scenario is made rather complex by the intriguing
chemical composition of GO that contains epoxy and hydroxyl
groups on the basal plane as well as carbonyls, carboxyls, ethers,
quinones, lactols, and phenols attached at vacancy and edge
We combined high-resolution fast X-ray photoelectron spec-
troscopy (XPS) with synchrotron radiation and temperature
programmed desorption (TPD) with numerical modeling based
on Density Functional Theory (DFT) to reveal the species
residing on oxidized graphene and released in the gas phase at
different temperatures, to address the energetics and diffusion of
the surface groups, and to identify the reaction pathways and the
rate-limiting steps for the reduction. The oxidized graphene was
prepared by dosing oxygen atoms onto a graphene layer grown
on the Ir(111) surface. The reason of this choice is 2-fold: (i) the
graphene layer supported on Ir(111) is highly ordered with a very
2
4ꢀ27
sites.
For OH groups as well as for the oxidizing species
attached to vacancies and edges, the thermal behavior is somehow
28,29
assessed and a stability scale commonly accepted.
On the
contrary, for epoxy groups the analogous information concerning
surface diffusion, stability, reaction routes, and desorption tem-
perature is quite vague. This is likely because the reduction of
graphitic C is traditionally investigated on materials oxidized at
39
low concentration of defects, which may drastically affect the
oxidation/deoxidation processes and, (ii) due to the very weak
interaction with the underlying substrate, graphene can be seen
3
0,32
40,41
temperatures above 500ꢀ600 K
that appear too high for the
as almost freestanding.
Indeed, the lattice mismatch between
retention of epoxy oxygens, which have been reported to desorb
graphene and Ir(111) gives rise to the formation of a moirꢀe
periodic corrugation of the carbon layer, where graphene regions
almost completely detached from the substrate coexist with
regions that are closer to the Ir surface, thus feeling a slightly
stronger interaction. We find that the thermal behavior of the
majority of the O-carrying surface species is not affected by this
corrugation that, as shown below, is just reflected in the forma-
tion of a small amount of CꢀO bonds most probably residing in
the depressions of the graphene layer.
2
4,33
below 400 K.
However, basal plane O atoms have been often
3
4ꢀ38
theoretically predicted
to be reactive agents capable of
promoting the formation of more complex oxidized structures
and of catalyzing surface reactions, and might represent the key
species in the low temperature gasification of GO.
Here, we identify a dual path mechanism for the reduction of
oxidized graphene governed by the coverage of oxygen atoms.
The thermal reduction of graphene oxidized by a low density of
epoxy groups proceeds by releasing solely molecular oxygen via a
cycloaddition reaction from epoxyꢀepoxy pairs. It is the forma-
tion of etherꢀepoxy pairs at high O coverage that promotes the
’
RESULTS AND DISCUSSION
elimination of CO/CO mixtures, thus determining surface
Low O Coverage. The graphene layer was exposed to con-
2
defects and C loss. We demonstrate that the competing surface
trolled doses of oxygen atoms at room temperature, leading to a
1
7316
dx.doi.org/10.1021/ja205168x |J. Am. Chem. Soc. 2011, 133, 17315–17321

Journal of the American Chemical Society
ARTICLE
Figure 2. Calculated minimum energy paths for epoxy diffusion, clustering, and recombination to molecular oxygen. (a) Relative energy of the most
0
stable structures involving epoxy pairs as well as calculated activation energies for epoxy diffusion and formation of epoxyꢀepoxy (C, C ), epoxyꢀether
(
D), and etherꢀether (E) groups. (b) Calculated minimum energy path for the recombination of an epoxyꢀepoxy pair to molecular oxygen via a
cycloaddition reaction. All energy values are in eV.
coverage of θ = 0.03 monolayers (ML). With respect to the clean
that are adsorbed in the moir ꢀe regions closer to the metal
substrate in analogy with the H adsorbate.
2
45
graphene on Ir(111) surface, which shows a single sp peak
4
1
(
284.14 eV) in the C1s spectrum and the components due to
No oxygen adatom is detected above 520 K. At these tempera-
tures, the C1s and Ir4f7/2 spectra resemble those of pristine
graphene, therefore showing that after complete reduction the C
bulk (Ir , 60.84 eV) and first layer Ir atoms (Ir , 60.32 eV) in the
B
S
4
1
Ir4f7/2 spectrum (Figure S1), new features in the XPS spectra
are observed after oxidation (Figures 1a,b). The oxygen atoms
induce a dominant O1 component (531.1 eV) in the O1s
spectrum and a C1 component (285.8 eV) in the C1s spectrum,
2
atoms recover the full sp hybridization and the bonding with the
Ir support is lost. In addition, the integral area of the C1s peak is
comparable to the one measured before oxidation, indicating the
lack of C loss during the oxidationꢀreduction cycle.
42ꢀ44
both due to the formation of epoxy structures.
The appear-
ance of the components C2 (284.6 eV) in the C1s and of a new
component (60.66 eV) in the Ir4f7/2 spectrum (Figure S1)
supports the emergence of a CꢀIr interaction induced by the
The decrease in the O-related XPS peaks between 350 and
520 K correlates with the TPD curves measured between room
temperature and 750 K (Figure 1e). The data relative to masses
28 (CO), 32 (O ), and 44 (CO ) demonstrate that only
44
adsorption of O atoms on graphene. Moreover, the Ir-sup-
ported graphene shows novel O adspecies, so far unreported in
oxidized graphite or carbon nanotubes, originating the compo-
nent O2 (528.8 eV) in the O1s spectrum.
2
2
molecular oxygen is released by the oxidized graphene, whereas
the quantity of CO and CO remains below the detection limit, in
2
agreement with the integrity of the reduced graphene indicated
by XPS. These results unambiguously prove that O atoms can be
reversibly adsorbed and desorbed from the basal plane of perfect
graphene layers without leading to vacancy formation in the C
network.
The thermal evolution of the C1s and O1s spectra
(
Figures 1aꢀd) acquired while heating the oxidized graphene
described above sets the onset for reduction at T ≈ 350 K. This
temperature activates the O desorption from the epoxy groups
only, as shown by the decrease of the O1 and C1 intensities at
The two desorption rate maxima appearing in the O TPD
2
∼
350 K, whereas the O2 and C2 peaks start to decrease at ∼400
curve at Tmax = 480 and ∼520 K with a relative shift of ∼40 K can
be attributed to the desorption of the two species represented by
the O1 and O2 components, which exhibit a similar separation
between their Tmax values (Figure 1c). Note that the upshift of
about 50 K of the maximal desorption rates shown by the TPD
curves with respect to those derived from the XPS data is due to
K. In particular, the curves relative to O1 and O2 exhibit
inflection points, maximal desorption rates, at Tmax = 430 and
4
75 K, 45 K apart from each other (Figure 1c). The similarity
between the desorption temperatures measured for O1 and O2
disproves the recent attribution of the O2 component to O atoms
intercalated between the metal substrate and the graphene
47
the different heating rates used during the XPS (0.2 K/s) and
TPD (2 K/s) experiments, which were suitably selected to
optimize the signal-to-noise ratio for each technique. From the
Tmax values of 480 and 520 K, we estimate desorption energies of
1.21 ( 0.02 and 1.33 ( 0.02 eV when assuming a second-order
4
4
layer, as O atoms bonded to Ir(111), even if under the
graphene cover, should desorb at a temperature a few hundred
46
kelvin higher than that measured for O2. Moreover, the high
perfectness of graphene grown on Ir(111), with a low density of
defects or vacancies, which are the only places where the oxygen
atoms could intercalate to bind to the metal substrate, further
excludes this attribution. Instead, the slightly higher thermal
stability of the O adatoms originating O2 can reasonably derive
from the availability of adsorption sites where the oxygen atoms
can bind more strongly than the epoxy species generating O1.
We guess that the O2 component belongs to the oxygen atoms
13
ꢀ1 45
desorption and a pre-exponential factor of 10
s .
Our DFT calculations show that the desorption of molecular
oxygen is governed by a cycloaddition reaction between two
epoxy groups and is mediated by epoxy group diffusion. The
calculated barrier for the diffusion of an epoxy O atom on
38
graphene is 0.73 eV (Figure 2a). According to the transition
13
ꢀ1
state theory (assuming a prefactor of 10
s
T = 350 K), this
1
7317
dx.doi.org/10.1021/ja205168x |J. Am. Chem. Soc. 2011, 133, 17315–17321

Journal of the American Chemical Society
ARTICLE
Figure 3. Surface composition and thermal annealing of graphene with high O coverage. (a,b) Top: O1s and C1s core level spectra measured on
graphene/Ir(111) with an oxygen coverage θ = 0.25 ML. The best fit curves and the spectral components are indicated by the blue lines and the colored
peaks. The component C3, C4, and C5 are attributed to ethers, semiquinones, and lactones, respectively, whereas the weak peak C6 is due to C atoms
neighboring lattice vacancies. In the O1s spectrum, the weak component O3 indicates CdO double bonds, whereas the component O4 at 532.4 eV due
to ethers cannot be disentangled from the intense O1 peak and becomes visible only above 350 K after epoxy desorption (Figure S2). Bottom: Coded
density plots of the O1s and C1s spectral intensities versus temperature. The C1s image is saturated to highlight the evolution of the minor spectral
components. (c,d) O1s (c) and C1s (d) component intensities versus temperature measured at a rate of 0.2 K/s. The bar in (c) marks the inflection point
of the curve relative to the O1s area. (e) TPD curves measured during the thermal annealing of the sample illustrated above at a rate of 2 K/s. The curves
are background subtracted and vertically shifted for clarity.
2
ꢀ1
activation energy corresponds to rates of about 10 s , and
hence the O adsorbates are mobile in the relevant T range.
In these conditions, the calculations predict the clustering of the
(283.6 eV) partially responsible for the broadening of the main
2 42
sp peak. Double CdO bonds produce the O3 component
(530.0 eV) in the O1s spectrum, whereas the presence of ethers
0
44
epoxy groups in second-nearest neighbor configurations(C and C
that generate the component O4 around 532.4 eV (Figure S2)
in Figure 2a), with an energy gain of almost 0.5 eV. The recom-
bination of these epoxyꢀepoxy pairs into adsorbed molecular
oxygen proceeds via a cycloaddition reaction and requires an
activation energy of 1.13 eV (Figure 2b), in close agreement with
the experimental value of 1.21 eV resulting from the TPD data.
The adsorbed oxygen molecules then desorb to the gas phase with
a barrier of just 0.53 eV. Note that the next- and third-nearest
neighbor configurations of O adatoms (epoxyꢀether and etherꢀ
ether, D and E in Figure 2a, respectively) cannot be formed
by epoxide diffusion, because this requires activation energies
cannot be disentangled from the intense O1 component.
The thermal evolution of the C1s and O1s spectra shows that,
similarly to the previous low coverage case, epoxide reduction is
activated at ∼300ꢀ350 K and is completed below 550 K
(Figure 3aꢀd). A residual quantity of O is still detected at
500 K in the form of ether and semiquinone and is reduced after
annealing to T > 640 K (Figure S2). The comparison between
the C1s spectral intensities measured on pristine and reduced
graphene shows that 10ꢀ15% of C atoms are lost during the
oxidationꢀdesorption cycle. This is at variance with graphene
oxidized only with epoxy groups, which are reversibly reduced to
the pristine defect-free graphene.
48
(
∼1.3 eV) larger than that needed for the epoxide recombination.
We have also verified that O desorption from the cycloaddition
2
intermediate is energetically preferred over its surface rearrange-
ment into a stable pair of semiquinones (Figure S3).
The TPD curves reported in Figure 3e show that the desorp-
tion of molecular oxygen between 350 and 550 K is paralleled by
High O Coverage. The O1s and C1s spectra measured on the
graphene surface with an oxygen coverage θ=0.25MLareshownin
Figure 3a and b. Epoxy groups, which are still largely dominating, are
accompanied by a low concentration (∼8%) of C atoms forming
single, double, and triple bonds with O atoms. We assign the
the removal of comparable quantities of CO and CO . Hence,
2
the kinetics of the three reactions leading to O , CO, and CO
2
2
should be controlled by rate-limiting steps having similar activa-
tion energies. Desorption energies of 1.19ꢀ1.21 eV are esti-
mated for the three molecules from the T values all comprised
max
49
component C3 (285.3 eV) to ethers, whereas the components C4
286.8 eV) and C5 (288.0 eV) can be attributed to C atoms forming
double (quinones) and triple (lactones) bonds with O atoms,
in the 445ꢀ455 K range. As in the previous case, the T values
max
(
are up-shifted by ∼40 K with respect to that derived from the
thermal behavior of the total O1s signal, which shows the
inflection point at 410 K (Figure 3c).
43
respectively. Carbon vacancies originate the small C6 component
1
7318
dx.doi.org/10.1021/ja205168x |J. Am. Chem. Soc. 2011, 133, 17315–17321

Journal of the American Chemical Society
ARTICLE
mechanisms having nevertheless equivalent activation energy
barriers. The reduction of graphene oxidized only by isolated
epoxy groups or by epoxyꢀepoxy pairs will produce exclusively
molecular oxygen. Higher O coverages will necessarily generate a
sizable fraction of nearest-neighbor O adsorbates (etherꢀ
epoxy), which easily converts into the lactoneꢀether pairs,
precursors for CO/CO desorption. It is worth noticing that
2
once the consumption of the graphene lattice has started,
diffusing epoxy groups can bind at vacancies forming additional
ethers. Moreover, the thermal decomposition of the lactoneꢀ
ether pairs leaves C vacancies saturated by ether groups
(
Figure 4). Both processes, which explain the slightly increasing
ether fraction observed during annealing (C3 in Figure 3), favor
the formation of additional etherꢀepoxy pairs.
Despite the lack of molecular oxygen in the TPD spectra of
10,19,21
GO,
our study proves that O desorption is actually
2
Figure 4. Formation of surface precursors for GO decomposition.
Calculated minimum energy path for the formation of the lactoneꢀ
ether surface precursors from the epoxyꢀether pairs assisted by a
diffusing neighboring epoxide. Energy differences and barriers are given
in eV.
thermodynamically favored provided that the epoxy groups are
mobile on the GO basal plane. Actually, the ratios of CO /O
2
2
and CO/O released in the gas phase are directly related to the
2
initial surface density of epoxy species: if the concentration of
ethers is enhanced by the presence of defects in the graphene
29
Desorption of CO and CO at temperatures as low as
lattice, the rate of the O cycloaddition reaction will be damped
2
2
3
GO,
50ꢀ400 K, has been observed during the reduction of
in favor of competing gasification reactions.
1
0,19
50,51
or of C materials exposed to atomic oxygen.
On
the contrary, these species are typically released above 600ꢀ670
’
CONCLUSIONS
3
1,30
32
K
or even at higher temperatures when C-based materials
Our results show that the key factor controlling the onset of
are oxidized by molecular oxygen and the O atoms are pre-
dominantly bonded to graphitic edges or vacancy sites. Here, the
low temperature erosion of the graphene network indicates that
C atoms are lost via a reaction that benefits from the reservoir of
O adatoms diffusing on the basal plane.
lattice damage in GO is the surface density of epoxy species: at
low coverage, these undergo reversible desorption, while with
increasing coverage they create the precursors for the release of
CO/CO mixtures. Although chemically synthesized GO con-
2
tains other functional groups, hydroxyls attached to the basal
plane and a variety of species decorating the graphene edges,
which will necessarily open other reaction channels, the role
identified here for basal plane O atoms in catalyzing the evolution
of C at low temperature remains essential.
This opens a new scenario for the thermal reduction of GO.
To move forward, novel chemical strategies need to be found to
scavenge the epoxy oxygens before they start to diffuse on the
surface and trigger the extensive disruption of the graphene
network. This will limit the lattice damage to that induced by
the removal of O bonded in other functional groups such as
Recent calculations have shown that the relevant precursor for
38
GO gasification are etherꢀlactone pairs. Actually, the high
barrier (>1.8 eV) required to dissociate an isolated lactone to
molecular CO or CO can be reduced to as low as 0.6 eV by the
2
assistance of O adatoms, which diffuse close to the lactone and
38
form the etherꢀlactone pair. On the basis of the calculated
energetics, we propose that the key structure to nucleate the
etherꢀlactone precursor is a pair of two nearest-neighboring O
adatoms, whose equilibrium geometry forms an etherꢀepoxy pair
(
D in Figure 2). The formation of these structures from epoxide
diffusion requires an activation energy (E ) larger than 1.3 eV and
a
ethers and quinones, which desorb as CO and CO at higher
is therefore disfavored over the epoxy recombination and desorp-
2
temperature.
tion (E = 1.13 eV, see Figure 2). However, long graphene
a
exposures to atomic O increase the probability for the adsorption
of O adatoms as first neighbors of other epoxy groups already
residing on graphene, thus forming stable etherꢀepoxy pairs.
Once formed, these etherꢀepoxy pairs are shown to attract
the diffusing epoxides (A in Figure 4) and can convert via a two-
step reaction mechanism into the lactoneꢀether precursors,
which are more than 3 eV lower in energy than the initial surface
structure (epoxyꢀether and a diffusing epoxy). The rate-limiting
step of the whole process has an activation energy of 1.10 eV
’
EXPERIMENTAL METHODS
The experiments were performed in the ultra high vacuum chamber
ꢀ
11
(
base pressure 8 ꢁ 10
mbar) of the SuperESCA beamline at the
Elettra synchrotron radiation facility (Trieste, Italy). The Ir substrate
+
was prepared by cycles of Ar sputtering at 1.5 keV, oxygen treatment
between 600 and 1000 K, and flash annealing at 1473 K. Temperatures
were measured via a K-type thermocouple spot-welded to the edge of the
crystal. The quality of the Ir (111) surface was checked by measuring the
Ir4f7/2 core level spectrum and by monitoring the LEED pattern.
Graphene was grown by ethylene (C H ) pyrolysis in several cycles
consisting of saturating the surface at 620 K followed by flash annealing
to 1423 K.
Atomic oxygen was produced by a radio frequency plasma source
(TECTRA, Gen2) equipped with an ion suppressing grid and an ion
trap. The atomic source current and the O pressure were kept constant
(
Figure 4), in very good agreement with the value of ∼1.20 eV
deduced by TPD. This barrier, ultimately governing the CO/
2
4
CO desorption, is comparable to that of epoxy recombination
2
4
1
via cycloaddition (1.13 eV) leading to molecular oxygen. The
relative position of the ether with respect to the lactone in
the lactoneꢀether precursor (C and D in Figure 4) determines
38
the CO or CO products.
2
2
ꢀ
6
The thermal reduction of oxidized graphene will produce
at 20 mA and 8 ꢁ 10 mbar, respectively. The sample was kept at
molecular oxygen as well as CO and CO via competing reaction
room temperature and exposed to atomic oxygen for increasing time.
2
1
7319
dx.doi.org/10.1021/ja205168x |J. Am. Chem. Soc. 2011, 133, 17315–17321

Journal of the American Chemical Society
ARTICLE
Oxidation was performed with the O source facing the back of the
sample to minimize any possible damage by energetic ions. Measure-
ments were carried out at oxygen coverage of 0.03 ML (“low O
coverage”) and 0.25 ML (“high O coverage”), that is, on the graphene
surface with a lower oxygen content with respect to GO prepared by wet
intermediates. This material is available free of charge via the
Internet at http://pubs.acs.org.
’
AUTHOR INFORMATION
5
2
processing, whose composition varies between C H O to C H O .
8
2
3
8 4 5
Corresponding Author
rosanna.larciprete@isc.cnr.it; lizzit@elettra.trieste.it
For the thermal desorption experiments, the sample was heated at 0.2 K/s
while acquiring O1s and C1s spectra by fast XPS. Alternatively, the sample
was heated in steps, and each time high-resolution photoemission spectra
were acquired at room temperature. At the end of each experiment, the Ir
surface was cleaned, and a fresh graphene layer was prepared.
’ ACKNOWLEDGMENT
Ir4f7/2, C1s and O1s core level spectra were measured at photon
energy of 130, 400, and 650 eV, respectively, with an overall energy
resolution ranging from 40 to 150 meV. For each spectrum, the binding
energy was calibrated by measuring the Fermi level position of the Ir
substrate. The measurements were performed with the photon beam
impinging at grazing incidence (70ꢀ) while photoelectrons were col-
lected at normal emission angle. The core level spectra were fitted using
S.F. and T.S. acknowledge Regione FVG for funding the
project NANOCAT and the ISCRA program of CINECA for
computational resources.
’ REFERENCES
(1) Park, S.; Ruoff, R. S. Nat. Nanotechnol. 2009, 4, 217–224.
ꢁ
DoniachꢀS ꢂu nji ꢀc functions convoluted with Gaussians. The oxygen
(2) Eda, G.; Chhowalla, M. Adv. Mater. 2010, 22, 2392–2415.
coverage was estimated by evaluating in the C1s spectrum the percen-
tage of C atoms forming single, double, and triple bonds with O atoms
and deducing the corresponding amount of adsorbed O.
(3) Loh, K. P.; Bao, Q.; Eda, G.; Chhowalla, M. Nat. Chem. 2010,
2
, 1015–1024.
4) Dikin, D. A.; Stankovich, S.; Zimney, E. J.; Piner, R. D.;
Dommett, G. H. B.; Evmenenko, G.; Nguyen, S. T.; Ruoff, R. S. Nature
007, 448, 457–460.
5) Stankovich, S.; Dikin, D. A.; Piner, R. D.; Kohlhaas, K. A.;
Kleinhammes, A.; Jia, Y.; Wu, Y.; Nguyen, S. T.; Ruoff, R. S. Carbon
007, 45, 1558–1565.
6) Kim, J.; Cote, L. J.; Kim, F.; Yuan, W.; Shull, K. R.; Huang, J.
(
The TPD data were recorded with a quadrupole mass spectrometer
53
equipped with a “Feulner cup” with a sample-size opening. Before each
measurement, the sample was adjusted in front of the cup and was
heated with a rate of 2 K/s. To compensate for the difference in
ionization efficiency of each molecular species, the measured partial
pressures were divided by the corresponding relative sensitivity factors.
2
(
2
(
The TPD curve of CO was corrected for the contribution due to CO
cracking.
2
J. Am. Chem. Soc. 2010, 132, 8180–8186.
(7) G ꢀo mez-Navarro, C.; Weitz, R. T.; Bittner, A. M.; Scolari, M.;
Mews, A.; Burghard, M.; Kern, K. Nano Lett. 2007, 7, 3499–3503.
(8) Jung, I.; Dikin, D. A.; Piner, R. D.; Ruoff, R. S. Nano Lett. 2008,
8, 423–487.
All simulations were based on the density functional theory (PBE
functional) and were performed in the framework of ultrasoft pseudo-
potentials and plane-waves as implemented in the PWscf code of the
54
(
9) Eda, G.; Mattevi, C.; Yamaguchi, H.; Kim, H.; Chhowalla, M.
J. Phys. Chem. C 2009, 113, 15768–15771.
10) Jung, I.; Field, D. A.; Clark, N. J.; Zhu, Y.; Yang, D.; Piner, R. D.;
Quantum-ESPRESSO package. The energy cutoffs for the electron
wave function and augmented charge density were set to 30 and 300 Ry,
respectively. Oxygen adsorption and mobility were studied on the free-
standing graphene surface, which was modeled with a 5 ꢁ 5 supercell
together with a 4 ꢁ 4 ꢁ 1 k-point mesh. Convergence in energy
difference and transition states with respect to supercell size, basis set,
(
Stankovich, S.; Dikin, D. A.; Geisler, H.; Ventrice, C. A., Jr.; Ruoff, R. S.
J. Phys. Chem. C 2009, 113, 18480–18486.
(11) Mattevi, C.; Eda, G.; Agnoli, S.; Miller, S.; Mkhoyan, K. A.;
38
Celik, O.; Mastrogiovanni, D.; Granozzi, G.; Garfunkel, E.; Chhowalla,
M. Adv. Funct. Mater. 2009, 19, 2577–2583.
and k-point sampling was checked to be below 0.02 eV.
Selected calculations were also performed for the O adsorption and
mobility on graphene supported by the Ir(111) surface. The metal
surface was modeled with a three-layer 9 ꢁ 9 slab at the equilibrium
lattice constant of bulk Ir (3.895 Å) matching a 10 ꢁ 10 graphene sheet.
In this case, the k-point sampling was restricted to the Γ point only. All
slabs were separated in the perpendicular direction by more than
(
12) Becerril, H. A.; Mao, J.; Liu, Z.; Stoltenberg, R. M.; Bao, Z.;
Chen, Y. ACS Nano 2008, 2, 463–470.
13) Schniepp, H. C.; Li, J.-L.; McAllister, M. J.; Sai, H.; Herrera-
(
Alonso, M.; Adamson, D. H.; Prud’homme, R. K.; Car, R.; Saville, D. A.;
Aksay, A. J. Phys. Chem. B 2006, 110, 8535–8539.
(14) G ꢀo mez-Navarro, C.; Meyer, J. C.; Sundaram, R. S.; Chuvilin, A.;
Kurasch, S.; Burghard, M.; Kern, K.; Kaiser, U. Nano Lett. 2010, 10,
1144–1148.
(15) Yang, D.; Velamakanni, A.; Bozoklu, G.; Park, S.; Stoller, M.;
Piner, R. D.; Stankovich, S.; Jung, I.; Field, D. A.; Ventrice, C. A., Jr.;
Ruoff, R. S. Carbon 2009, 47, 145–152.
14 Å of vacuum. As reported in the Supporting Information, the effect
of the underlying Ir surface on the mobility and recombination of the
epoxy groups is negligible.
Reaction mechanisms and transition barriers were studied with the
55
climbing-image nudged-elastic-band (NEB) method employing be-
tween 8 and 10 system replicas. These time-consuming calculations were
performed with a 3 ꢁ 3 ꢁ 1 k-point mesh. Stable structures and
transition states were verified by normal-mode analysis. Further details
are reported in the Supporting Information.
(
16) Bagri, A.; Mattevi, C.; Acik, M.; Chabal, Y. J.; Chhowalla, M.;
Shenoy, V. B. Nat. Chem. 2010, 2, 581–587.
17) Li, X.; Wang, H.; Robinson, J. T.; Sanchez, H.; Diankov, G.; Dai,
H. J. Am. Chem. Soc. 2009, 131, 15939–15944.
18) Moon, I. K.; Lee, J.; Ruoff, R. S.; Lee, H. Nat. Commun. 2010,
, 1–6.
19) McAllister, M. J.; Li, J.-L.; Adamson, D. H.; Schniepp, H. C.;
(
(
1
’
ASSOCIATED CONTENT
(
Abdala, A. A.; Liu, J.; Herrera-Alonso, M.; Milius, D. L.; Car, R.;
Prud’homme, R. K.; Aksay, I. A. Chem. Mater. 2007, 19, 4396–4404.
S
Supporting Information. Full author lists of refs 45 and
b
5
3, spectroscopy of the Ir4f core level, further details on the
(20) Rourke, J. P.; Pandey, P. A.; Moore, J. J.; Bates, M.; Kinloch,
thermal annealing of graphene with high O coverage, additional
information concerning O adsorption, mobility, and cycloaddi-
tion on Ir(111)-supported graphene, dependence of calcu-
lated activation barriers on O coverage, and stability of reaction
I. A.; Young, R. J.; Wilson, N. R. Angew. Chem., Int. Ed. 2011, 50,
3173–3177.
ꢀ
(21) Barroso-Bujans, F; Alegrìa, A; Colmenero, J. J. Phys. Chem. C
2010, 114, 21645–21651.
1
7320
dx.doi.org/10.1021/ja205168x |J. Am. Chem. Soc. 2011, 133, 17315–17321

Journal of the American Chemical Society
22) Wei, Z.; Wang, D.; Kim, S.; Kim, S.-Y.; Hu, Y.; Yakes, M. K.;
ARTICLE
(
Laracuente, A. R.; Dai, Z.; Marder, S. R.; Berger, C.; King, W. P.; de Heer,
W. A.; Sheehan, P. E.; Riedo, E. Science 2010, 328, 1373–1376.
(
23) Zhou, Y.; Bao, Q.; Varghese, B.; Tang, L. A. L.; Tan, C. K.; Sow,
C.-H.; Loh, K. P. Adv. Mater. 2010, 22, 67–71.
24) Lerf, A.; He, H.; Forster, M.; Klinowski, J. J. Phys. Chem. B 1998,
02, 4477–4482.
25) Casabianca, L. B.; Shaibat, M. A.; Cai, W. W.; Park, S.; Piner, R.;
Ruoff, S. R.; Ishii, Y. J. Am. Chem. Soc. 2010, 132, 5672–5676.
26) Gao, W.; Alemany, L. B.; Ci, L.; Ajayan, P. M. Nat. Chem. 2009,
, 403–408.
27) Cai, W.; Piner, R. D.; Stadermann, F. J.; Park, S.; Shaibat, M. A.;
(
1
(
(
1
(
Ishii, Y.; Yang, D.; Velamakanni, A.; An, S. J.; Stoller, M.; An, J.; Chen,
D.; Ruoff, R. S. Science 2008, 321, 1815–1817.
(
28) Marchon, B.; Tysoe, W. T.; Carrazza, J.; Heinemann, H.;
Somorjai, G. A. J. Phys. Chem. 1998, 92, 5744–5749.
29) Carlsson, J. M.; Hanke, F.; Linic, S.; Scheffler, M. Phys. Rev. Lett.
009, 102, 166104.
(
2
(
(
30) Kelemen, S. R.; Freund, H. Energy Fuels 1988, 2, 111–118.
31) Ma, M. C.; Brown, T. C.; Haynes, B. S. Surf. Sci. 1993, 297,
3
6
1
12–326.
(32) Zhuang, Q.; Kyotani, T.; Tomita, A. Energy Fuels 1995, 9,
30–634.
(33) He, H.; Klinowski, J.; Forster, M.; Lerf, A. Chem. Phys. Lett.
998, 287, 53–56.
(
34) Chen, N.; Yang, R. T. J. Phys. Chem. A 1998, 102, 6348–6356.
(35) S ꢀa nchez, A.; Mondrag ꢀo n, F. J. Phys. Chem. C 2007, 111,
6
12–617.
36) Orrego, J. F.; Zapata, F.; Truong, T. T.; Mondrag ꢀo n, F. J. Phys.
Chem. A 2009, 113, 8415–8420.
(
(37) Radovic, L. R. J. Am. Chem. Soc. 2009, 131, 17166–75.
(38) Sun, T.; Fabris, S.; Baroni, S. J. Phys. Chem. C 2011, 115,
4
730–4737.
39) Coraux, J.; N’Diaye, A.; Busse, C.; Michely, T. Nano Lett. 2008,
, 565–570.
40) N’Diaye, A.; Bleikamp, S.; Feibelman, P. J.; Michely, T. Phys.
Rev. Lett. 2006, 97, 215501.
41) Lacovig, P.; Pozzo, M.; Alf ꢂe , D.; Vilmercati, P.; Baraldi, A.;
Lizzit, S. Phys. Rev. Lett. 2009, 103, 166101.
42) Barinov, A.; Malcioglu, O. B.; Stefano Fabris, S.; Sun, T.;
Gregoratti, L.; Dalmiglio, M.; Kiskinova, M. J. Phys. Chem. C 2009,
(
8
(
(
(
1
13, 9009–9013.
43) Larciprete, R.; Gardonio, S.; Petaccia, L.; Lizzit, S. Carbon 2009,
7, 2579–2589.
44) Vinogradov, N. A.; Schulte, K.; Ng, M. L.; Mikkelsen, A.;
Lundgren, E.; Martensson, N.; Preobrajenski, A. B. J. Phys. Chem. C
011, 115, 9568–9577.
(
4
(
2
(
45) Balog, R.; et al. Nat. Mater. 2010, 9, 315–319.
(46) Ivanov, V. P.; Boreskov, G. K.; Savchenko, V. I.; Egelhoff, W. F.;
Weinberg, W. H. Surf. Sci. 1976, 61, 207–220.
(
(
(
47) Redhead, P. Vacuum 1962, 12, 203–211.
48) Cabrera-Sanfelix, P. J. Phys. Chem. A 2009, 113, 493–498.
49) Zhang, W.; Carravetta, V.; Li, Z.; Luo, Y.; Yang, J. J. Chem. Phys.
2
5
3
1
009, 131, 244505.
50) Marsh, H.; O’Hair, T. E.; Wynne-Jones, L. Carbon 1969, 7,
55–566.
(
(51) Wicke, B.; Wong, C.; Grady, K. Combust. Flame 1986, 66,
7–46.
(52) Boukhvalov, D. W.; Katsnelson, M. I. J. Am. Chem. Soc. 2008,
30, 10697–10701.
(53) Feulner, P.; Menzel, D. J. Vac. Sci. Technol. 1980, 17, 662–663.
(54) Giannozzi, P.; et al. J. Phys.: Condens. Matter 2009, 21, 395502.
(55) J ꢂo nsson, H.; Mills, G.; Jacobsen, K. W. In Classical and
Quantum Dynamics in Condensed Phase Simulations; Berne, B. J., Ciccotti,
G., Coker, D. F., Eds.; World Scientific: New Jersey, 1998; Chapter 16.
1
7321
dx.doi.org/10.1021/ja205168x |J. Am. Chem. Soc. 2011, 133, 17315–17321