CARBON 6 6 ( 2 0 1 4 ) 5 3 9 –5 4 6

Available at www.sciencedirect.com

ScienceDirect

journal homepage: www.elsevier.com/locate/carbon

Modulation of the band gap of graphene oxide: The

role of AA-stacking

A. Hunt a,*, D.A. Dikin b, E.Z. Kurmaev d, Y.H. Lee c, N.V. Luan c, G.S. Chang a,
A. Moewes a
a
Department of Physics and Engineering Physics, University of Saskatchewan, 116 Science Place, Saskatoon, Saskatchewan S7N 5E2, Canada
b
Department of Mechanical Engineering, Northwestern University, 2145 Sheridan Road, Evanston, IL 60208-3111, USA
c
Department of Energy Science (DOES), Sungkyunkwan University, Suwon 440-746, Republic of Korea
d
X-ray Emission Spectroscopy Lab, Institute of Metal Physics, RAS Ural Div., 18 Kovalevskoi Str., 620990 Yekaterinburg, Russia

A R T I C L E I N F O A B S T R A C T

Article history: The unique electronic properties of graphene make it an advantageous material for use in
Received 31 May 2013 many applications, except those that require a band gap. Much work has been done to
Accepted 13 September 2013 introduce an appropriately tuned band gap into graphene, including uniaxial strain and
Available online 23 September 2013 oxidation, with varying levels of success. We report here that the stacking configuration
of the sheets in multilayered graphene oxide can have a significant impact on the band
gap. Through comparison of X-ray absorption near-edge spectra of multilayered pristine
graphene sheets with spectra simulated using density functional theory, we have found
that AA-stacking pushes unoccupied states closer to the Fermi level than AB-stacking by
widening the p* resonance in both graphene oxide and graphene. If the near-Fermi states
have been removed such that the nearest unoccupied state to the Fermi level is the p* band,
then AA-stacked multilayered graphene oxide will have a smaller band gap than AB-
stacked graphene oxide. We have confirmed this by measuring the band gap of graphene
oxide and reduced graphene oxide indirectly using X-ray absorption near-edge spectros-
copy and X-ray emission spectroscopy. Controlling the stacking configuration of multilay-
ered graphene oxide may provide a novel method for tuning its band gap.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction conductor to replace the ubiquitous ITO [5], and as an inter-

mediate conducting layer in tandem solar cells [6]. Many tech-
Graphene is a monolayer of graphite wherein the carbon niques have been developed to make graphene sheets, such
atoms form a two-dimensional (2D) hexagonal structure as mechanical cleavage of graphite, chemical vapor deposi-
(the honeycomb lattice). The material is of great interest be- tion, and thermal fusion of polycyclic aromatic hydrocarbons.
cause of its exceptional electrical, mechanical and thermal Among them, the most promising method for mass produc-
properties. Graphene has been suggested for use in techno- tion of graphene is exfoliation of graphite oxide in a polar sol-
logical applications such as nanoelectronics, sensors, sup- vent, followed by reduction of the harvested graphene oxide
ercapacitors, and hydrogen storage [1–3]. In applications to (GO) sheets, although many methods have been proposed to
photovoltaics, graphene-based materials have been used as accomplish this [7,8]. However, more than just a precursor
a light collector in organic solar cells [4], as a transparent material for graphene, GO itself has very interesting

* Corresponding author: Fax: +1 306 966 6400.

E-mail address: [email protected] (A. Hunt).
0008-6223/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.carbon.2013.09.036
540 CARBON 6 6 ( 2 0 1 4 ) 5 3 9 –5 4 6

properties and can be used in polymer composites [9] and not possess, and the transition probabilities are not modified
dielectric layers in electronic devices [10]. Indeed, GO has by Franck–Condon factors like optical gap measurements,
shown to be a better transparent conducting film than graph- allowing for a detailed and unambiguous study of the carbon
ene in solar cell applications [11,12]. GO can be applied used states that populate the near-Fermi energy regime.
in various biological applications [13], for example, GO–
Fe3O4 nanoparticles can be used for drug delivery and release 2. Experimental
[14].
For electronic device applications [15], a band gap is re- 2.1. XES measurements
quired to control the type and concentration of carriers, how-
ever, graphene is a zero-gap semiconductor [16]. There are The XES measurements on the carbon 1s edges were per-
numerous methods by which one may introduce a band gap formed at Beamline 8.0.1 at the Advanced Light Source at
into graphene [17,18]. However, a natural point where one the Lawrence Berkeley National Laboratory using the soft
may study how to engineer the band gap of graphene is GO, X-ray fluorescence (SXF) endstation [25]. Emitted radiation
given its role in the preparation of graphene. GO has a finite was measured using a Rowland circle type spectrometer
electronic band gap due to oxidization of the carbon basal with a large spherical grating and a photon-counting area
plane by oxygen-containing functional groups: phenol, hy- detector. The total experimental resolution was 0.3 eV
droxyl, and epoxide groups on the plane and carboxylic FWHM. The fluorescence measurements were made using
groups at the lateral edge [19,20]. Interestingly, the effect that a depolarized configuration, which means that the vector E
the stacking arrangement of the graphene oxide sheets has of the incidence beam lies at the scattering plane, i.e.
on the band gap of multilayered GO has not been studied, de- p-polarization was used. The SXF endstation is configured
spite the fact that p–p overlap between graphene and func- such that the path of the emitted photons that can be
tionalizing adsorbed aromatics is known to alter electrical detected by the spectrometer and the incident beam are
characteristic of graphene [21]. There are many possible perpendicular to each other.
graphene stacking configurations, but two of the most impor-
tant are the AB-stacking (Bernel stacking) and the AA-stack- 2.2. XANES measurements
ing configurations. The two stacking configurations are
shown in Fig. 1. The images in Fig. 1 were made using VESTA High-resolution XANES spectra on the C 1s edge were mea-
[22]. AB-stacking is the lowest-energy configuration, and this sured at the Spherical Grating Monochromator (SGM) beam-
is the dominant stacking mode found in highly-oriented pyro- line at the Canadian Light Source [26]. The GO X-ray
lytic graphite (HOPG). It is known that the way in which absorption near-edge spectra (XANES) were measured in total
graphene sheets stack has a significant impact on the vibron- electron yield (TEY) and total fluorescence yield (TFY) modes.
ic structure of multilayered graphene sheets [23]. Given that The graphene XANES spectra, however, were also measured
the states that frame the band gap in GO will also be located in partial fluorescence yield (PFY) mode. The PFY spectra were
in the interplanar region, it stands to reason that the way in measured using a silicon drift detector (SDD) that acted as a
which the layers interact will affect the band gap. This work very low-resolution spectrometer (FWHM  150 eV). Integrat-
studies that very interaction. ing all counts detected by the SDD would yield a TFY spec-
We have been studying the many facets of the electronic trum, whereas isolating only the emission line from carbon
and physical structure of graphene oxide [24]. Here, we probe produces a C 1s PFY spectrum. This was necessary in this case
the band gap of multilayered GO (simply referred to as GO because the graphene and multilayered graphene samples
henceforth) and GO that has been reduced with hydrazine hy- were mounted on SiO2 substrates. Second order contamina-
drate (rGO). The band gap is studied indirectly through a com- tion from the monochromator allows for near-resonant exci-
bination of X-ray absorption near-edge fine structure tation of the O sites in the SiO2 substrate, as the O 1s
(XANES), to probe the unoccupied partial density of states absorption threshold lies at nearly twice the energy of the C
(pDOS), and non-resonant X-ray emission spectroscopy 1s edge. Although second order light from the monochroma-
(XES), to probe the occupied pDOS. Plotted on a common en- tor is significantly weaker than the first order light that reso-
ergy scale with respect to the core hole, the band gap is the nantly excites the graphene, the SiO2 substrate has orders of
energy separation between XES and XANES. These techniques magnitude more oxygen than the graphene layers has carbon.
boast site- and chemical-specificity that other techniques do Thus, O Ka emission from the substrate strongly competes
with the C Ka emission from the graphene.
For normalization, in all cases, the PFY, TFY, and TEY spec-
tra were normalized to the current generated in a photodiode
when this photodiode was exposed to the incident beam. This
photodiode current spectrum was not taken simultaneously
with the sample spectrum, but rather directly afterwards.
This technique allows one to directly measure the light inten-
sity hitting the sample as a function of energy. This method
allows one to correctly account for intensity fluctuations in
the incident beam due to carbon contamination on all opti-
Fig. 1 – The AB- and AA-stacking configurations for graphite cally active beamline components. Such contamination can
and multilayered graphene. introduce false features into carbon spectra.
 CARBON 6 6 ( 2 0 1 4 ) 5 3 9 –5 4 6 541

2.3. Graphene synthesis

Eight samples of pristine graphene mounted on SiO2 sub-

strates were prepared: two each of graphene, bilayered graph-
ene, multilayered graphene with four to six layers, and
multilayered graphene with ten or more layers. One sample
from each of the four pairs was annealed at 900 °C. The
graphene samples were grown using the atmospheric pres-
sure chemical vapor deposition (APCVD) method. Copper foil
(from Nilaco, 99.96%) with a thickness of 100 lm was used as
the substrate for monolayer graphene growth; bilayer and
multilayer graphene was grown on 70 lm copper foil. The
chamber was preheated to 1060 °C with 1000 sccm of Ar and
200 sccm of H2 and annealed for 40 min. After that, the rate
of gas injection was altered depending on the desired charac-
teristics of the graphene film. In the case of monolayer graph-
ene, the H2 gas injection rate was reduced to 10 sccm from
200 sccm, and 3 sccm of methane gas was injected. For bilayer
graphene, the H2 and methane gas injection rates were
70 sccm and 30 sccm, respectively. The growth time for both
mono and bilayer graphene was 5 min. For multilayered
graphene, we maintained 200 sccm of H2 gas flow and in-
jected only 3 sccm of methane. Growth time was increased
to 15 min. Methane and H2 gas inflows were then stopped
and the sample was cooled down to room temperature. After
growth, the finished samples were transferred to SiO2 sub-
strates using copper etchant.

3. Results

The GO sample was prepared as in the manner of GO paper

from water filtration. The rGO sample was prepared as the
Fig. 2 – Experimental determination of the band gap of GO
first, but during the process, the water was substituted with
and rGO. In each panel, an inset is provides a magnified look
DMF and hydrazine was added. The sample was heated and
at the states near the Fermi level. Within these insets, the
reduced in solution, then finally filtered. Further details on
second derivatives for each of the TEY, TFY, and XES spectra
the general synthesis technique are published elsewhere
are shown. The three horizontal arrows in the inset of panel
[27,28]. These samples were chosen because reduction of GO
(a) indicate the three possible band gaps. They are labeled a,
is a necessary step if one wishes to achieve graphene from
b, and c in order of increasing gap size. Note that the relative
graphene oxide. Incomplete reduction can provide a way to
scaling of the TEY, TFY, and XES spectra is different in the
modulate the oxidization level, and thus the band gap of GO.
insets as compared to the main panel; this was done purely
As mentioned earlier, we probed the electronic states of
for aesthetic purposes. (A colour version of this figure can be
the two samples using XES and XANES for the purposes of
viewed online.)
determining the band gap. The XES spectra were measured
at Beamline 8.0.1 at the Advanced Light Source [25], while
the XANES spectra were measured at the SGM beamline at sample via Auger decay, whereas the TFY technique requires
the Canadian Light Source [26]. Fig. 2 shows the results of one to count the photons emitted from the sample via radia-
our experiments. In this Figure, the non-resonant C Ka XES tive decay. Ideally, the intensity of emission of both electrons
spectrum and the C 1s XANES spectrum from each sample and photons scales linearly with the number of core holes
is plotted on the same energy axis, each in a separate panel. created. When this holds true, TEY and TFY are both direct
In terms of the energy axis, both the C 1s XANES and the C measures of the true absorption profile. In practice, however,
Ka XES spectra are both measured with respect to the C 1s there are two important caveats. The escape depth of Auger
core state. Thus, aligning the XANES and XES spectra on the electrons is very shallow, which means that TEY spectra are
same energy axis simply requires each to be calibrated to a very surface-sensitive. TFY is much more bulk-sensitive due
known reference. All XANES spectra shown in this work were to the longer escape depth of the emitted photons. However,
aligned to a sequentially measured XANES spectrum of HOPG. when exciting the sample on a strong resonance, such as
All XES spectra were similarly aligned to HOPG. the C 1s-2p* feature in graphite, the rate of core hole creation
The XANES spectra were measured in both total electron does not increase linearly with a concomitant increase in
yield (TEY) and total fluorescence yield (TFY) modes. The absorption cross-section, and the TFY signal saturates at
TEY technique involves counting electrons emitted from the strong resonance features [29]. This leads to a characteristic
542 CARBON 6 6 ( 2 0 1 4 ) 5 3 9 –5 4 6

‘squishing’ of high-intensity features. In order to correct for shown by their second derivatives? (2) Which local minimum
saturation effects, the displayed TFY spectra have been trea- does one use for the determination of the band gap of GO?
ted using the correction method of Eisebitt et al. [30,31]. Neither question can be answered until it is known what is
The second derivative method was chosen to analyze the causing the double-dip structure in the TFY of both samples.
band gap of GO. The second derivative is particularly useful As we shall show later, the answers to both of these questions
for analyzing the pre-edge of a spectrum because weak but involves the AA-stacking mode, and in particular, how this
sharp features show much more prominently in the second configuration behaves when subjected to pressure within
derivative than broad yet strong features. This property helps the bulk of a material.
to extract details from the near-Fermi portions of the XES and
XANES spectra. Other researchers have successfully deter- 3.1. Pristine graphene
mined the band gaps of various materials indirectly using a
combination of XANES and XES [32–34]. Calculating the sec- The first step to answer the two questions stated above is to
ond derivatives of XANES and XES spectra has proven to be determine whether the double-dip structure is due to func-
an accurate and highly reproducible method of finding the tionalization, or if one can see this structure in pristine mul-
band onsets, and thus the band gaps [35–38]. tilayered graphene. To test this possibility, eight samples of
Minima in the second derivative correspond to features pristine graphene mounted on SiO2 substrates were prepared:
(maxima) in the spectrum. The band gap is therefore the en- two each of graphene, bilayered graphene, multilayered
ergy separation between the highest-energy minimum in the graphene with four to six layers, and multilayered graphene
second derivative of an XES spectrum (denoted d2XES/dE2) with ten or more layers, hereafter referred to as Gr, 2Gr, 4Gr,
and the lowest-energy minimum in second derivative of an and 10Gr. The first set taken as is, and the second was an-
XANES spectrum. Depending on how the XANES spectrum nealed at 900°; further details of the synthesis can be found
is measured (TEY or TFY), the second derivative is denoted
d2TEY/dE2 or d2TFY/dE2.
With this understanding of the second derivative method
in mind, let us turn to the d2XES/dE2, d2TEY/dE2, and d2TFY/
dE2 spectra of GO and rGO, displayed within the insets of each
panel in Fig. 2. Before a discussion concerning the band gaps
of the samples can begin, however, the differences in the
d2TEY/dE2 and d2TFY/dE2 spectra of each sample is striking,
particularly in the 284.5–286.5 eV range. For both samples,
the d2TEY/dE2 shows only one minimum, at about 285.2 eV,
whereas the d2TFY/dE2 shows a double-dip structure, with
one minimum at 284.6 eV, the other at 285.6 eV. This double-
dip structure in the TFY is not due to saturation, as saturation
cannot introduce structure into a spectrum [29]. Neither is the
difference in shape due to a normalization artifact from a
contaminated mesh, as all XANES spectra were normalized
using an asynchronously measured photodiode spectrum of
the incident light to act as a background reference [39]. Thus
the difference in shape between the TEY and TFY spectra is
real and inherent. For rGO, there is a similar structure in the
284.5–286.5 eV range as indicated by dashed lines in Fig. 2,
but the problem does not affect the determination of the band
gap because both d2TFY/dE2 and d2TEY/dE2 have minima at
283.3 eV. This leads to a band gap of about 0.3 eV. However,
the determination of the band gap for GO is not so simple,
as it is not obvious, at this point, which minima of the three
available one should use. The three possible band gaps, based
upon the three candidate minima at 284.6 eV, 285.2 eV, and
285.6 eV, are labeled with arrows in panel (a) in Fig. 2. Fig. 3 – XANES spectra of Gr, 2Gr, 4Gr, and 10Gr, measured in
Note that Fig. 2 shows many spectral differences between TEY and PFY modes. XANES spectra of HOPG, measured in
the spectra of GO and rGO, particularly in the range between TEY and TFY mode, is included as a reference. The graphene
287 eV and 290 eV. This energy range is home to the spectral spectra were all measured with an angle of incidence of 80°
signatures of oxidizing functional groups [40–42,24]. The from the sample surface normal. The top two panels show
287–290 eV energy range holds key information about the the spectra measured from samples that were not annealed;
chemical character of the sample in question, but it is not rel- the bottom two panels show spectra that were measured
evant to a study of the band gap of graphene oxide. Therefore, from samples annealed at 900 °C. The second derivatives of
we do not discuss it. each group of spectra are shown in the bottom halves of
We thus present two fundamental questions: (1) Why do each of the panels. (A colour version of this figure can be
the TEY and TFY spectra have such different structure, as viewed online.)
 CARBON 6 6 ( 2 0 1 4 ) 5 3 9 –5 4 6 543

in the Experimental section. Fig. 3 displays XANES spectra 4. Discussion and analysis
measured from the samples with different numbers of pris-
tine graphene layers, as well as HOPG. The spectra of the mul- The double-dip structure has now been seen in GO, rGO, and
tilayered graphene samples were recorded in both TEY and multilayered pristine graphene if one stacks up 10 or more
PFY modes; the fluorescence spectrum of HOPG was recorded layers. Given that this structure occurs in pristine graphene
in TFY mode. The reason for the difference in fluorescence and HOPG, it is not due to oxidation. This conclusion is sup-
measurement technique is explained in Section 2. ported by the fact that all prevalent functional groups identi-
As one can clearly see, there are very few features in the fied on graphene oxide, namely carboxyl, hydroxyl, epoxide,
287–290 eV energy range in the graphene spectra, particularly and carbonyl, all have their characteristic resonances in the
for the annealed samples. This marks a sharp contrast with oxidation zone between 286 eV and 290 eV, and this energy re-
the GO samples, but the difference is expected. No spectral gion quite devoid of features. If the double-dip structure is not
features within this energy range indicate that the samples a chemical artifact, could it be due to structural differences,
have not been functionalized. However, there is another such as the interplanar stacking order?
marked difference between the graphene and GO samples We present a theoretical study of the effect of the stacking
worthy of mention: the PFY and TEY spectra of the graphene order on the electronic structure of multilayered graphene, in
samples are very similar in shape. This similarity is to be ex- which we explicitly consider the AA- and AB-stacking config-
pected because of the thinness of the graphene samples. The urations. Fig. 4 shows simulated XANES spectra of several
saturation effects that strongly distort the TFY spectra of GO structures, including graphene, AB- and AA-stacked bilayer
in Fig. 2 are only problematic for samples that are thicker graphene, and AB- and AA-stacked graphite; all simulations
than the penetration depth of the incident radiation, which were performed using the WIEN2k code using the local den-
is on the order of 100–200 nm for soft X-rays. sity approximation (LDA) exchange–correlation functional.
Saturation effects do not distort the PFY spectra of the For all multilayered configurations, the interplanar spacing
graphene samples and thus the PFY and TEY spectra are was optimized. The AB-stacked graphite had an optimized
much similar in shape. However, the PFY and TEY spectra interplanar spacing that was very close to the experimental
are not identical, particularly when comparing the PFY and value of 3.348 Å, which is the reason why the LDA functional
TEY spectra of the multilayered graphene samples. One was chosen. The Perdew–Burke–Ernzerhof generalized gradi-
can immediately see that features b at 284.8 eV and c at ent approximation (PBE-GGA) exchange–correlation func-
285.4 eV that constitute the double-dip structure in the tional did not report an optimal interplanar spacing. The
d2TFY/dE2 of GO shown earlier are seen again. The double AB-stacked bilayer graphene also had a calculated optimal
dip is seen most strongly in the d2PFY/dE2 spectra of both interplanar spacing very close to the known value for bulk
10Gr samples. It would seem that, as one adds layers, the graphite. This result agrees with previous studies [43]. How-
double-dip structure becomes more prominent in the ever, the AA-stacked configurations all preferred to be much
d2PFY/dE2 spectra measured from the sample. This trend is further apart. Separate calculations at 3.348 Å, the interplanar
strongly supported by the HOPG results. However, another spacing preferred by AB-stacked graphite, and at the opti-
pattern also emerges. In the second derivative, the mini- mized spacing of 3.6 Å were performed.
mum at 284.8 eV becomes prominent while the largely posi- The simulations in Fig. 4 show three very important re-
tive feature at 286.3 eV (feature d) is suppressed; the two sults. First, the p* band of graphene, as well as that of bilayer
spectral changes happen simultaneously, suggesting a com- and infinite layer stacked graphene (graphite) in the AB-stack-
mon cause. As layers are added, spectral weight is being ing configuration, has one major feature. In the second deriv-
added at both 284.8 eV and 286.3 eV. ative, this feature produces one minimum in each case.
It is therefore clear that, as one adds more graphene lay- Second, the p* band of bilayer and infinite layer graphene
ers, more states are added at both 284.8 eV and 286.3 eV. In- (graphite) shows a much broader lineshape; in the second
deed, HOPG also shows prominent dips in its d2TFY/dE2 at derivative, this yields two local minima. Third, the relative
these energies, a sample that is highly non-functionalized spacing of the AA-stacked minima (and hence the width of
and has been subjected to none of the chemicals involved the p* band) increases with an increasing number of layers
in either the creation of GO, or the transfer of graphene. An and a decreasing interplanar spacing. If one were to increase
as-yet unidentified layer-dependant phenomenon is affecting the spacing to infinity, the double peak structure in the sec-
these changes. Note that other researchers have noticed fea- ond derivatives of the bilayer graphene simulated XANES
tures in the 284.6–286.3 eV energy range. Indeed, the d2TEY/ spectra would eventually coalesce into one feature, as one ex-
dE2 and d2PFY/dE2 spectra of the non-annealed monolayer pects from graphene. In the AA-stacking bulk limit, when the
graphene sample also show a feature a dip 284.9 eV. In this interplanar distance equals that of AB-stacked graphite of
case, the feature at 284.8 eV is likely caused by another effect, 3.348 Å, the minima in the second derivative of the simulation
such as scattering from point defects, or doping from ad- line up quite well with the minima at 284.7 eV and 286.4 eV of
sorbed gases or solvents. Whatever the source, it is mitigated the d2TFY/dE2 of highly-oriented pyrolytic graphite (HOPG), as
by annealing, as this feature is sharply reduced in the spectra shown in Fig. 4. Thus, it would seem that they layer-depen-
from the annealed monolayer graphene. Spectral weight in dant effect that produces spectral weight at 284.7 eV and
the 284.6–284.9 eV energy range is thus likely a composite 286.4 eV is due to interactions between AA-stacked graphene
from many contributions, but it is the layer-dependent one sheets, whereas the minimum at 285.4 eV is due to AB-
that we seek here. stacking.
544 CARBON 6 6 ( 2 0 1 4 ) 5 3 9 –5 4 6

controlling the stacking configuration of a multilayered GO

samples gives the device fabricator another means by which
the band gap may be tuned, in addition to simply controlling
the oxidation level [44,45]. In short, the stacking order adjusts
the width of the p* band, which therefore controls the band
gap if the sample is oxidized and the near-Fermi states have
been removed.
AA-stacked GO therefore has a smaller band gap than AB-
stacked GO. However, there is more to the story. The theoret-
ical results displayed in Fig. 4 clearly show that the p* band
widens as one brings sheets of graphene together in an AA-
stacked order, which is markedly different behavior than that
exhibited by the AB-stacked variety. The implications of this
are immediate. If the p* band widens if AA-stacked sheets
are brought closer to one another, then if one has a primarily
AA-stacked sample of multilayered graphene oxide, then one
may widen the p* band and thus narrow the band gap simply
by applying external pressure along the c-axis. In short, AA-
stacking may provide a way to externally control the band
gap of GO.
Indeed, the sensitivity of the width of the p* feature in AA-
stacked multilayered graphene to pressure along the c-axis
explains why it is that the characteristic AA-stacked signal
is not seen in TEY spectra. As mentioned earlier, multilayered
graphene, if stacked in the AA configuration, is in its lowest
energy state if the sheets are farther apart (about 3.58 Å) than
if stacked in the AB configuration. The AA-stacked zones do
not transition to AB-stacking; there is no phase change.
Rather, the AA-stacked zones near the surface have the free-
dom to expand and have larger interplanar spacing. This lar-
ger spacing, in keeping with the results shown in Fig. 4, allows
for a smaller p* bandwidth, and thus the near-surface AA-
stacked regions show XANES spectra that look much like
the AB-stacked zones. TEY is highly surface sensitive, and
would see only the expanded AA-stacked zones. Deeper with-
in the sample, however, the freedom to expand would not ex-
ist for any AA-stacked region. Such AA-stacked zones would
be forced by the neighboring AB-stacked majority to assume
a nominal layer separation of 3.348 Å, which leads to the
Fig. 4 – Simulated XANES spectra for graphene, bilayer detectable widening of the p* resonance feature in the bulk-
graphene in AA- and AB-stacking configurations, and bulk sensitive TFY and PFY spectra of multilayered graphene,
graphite in AA- and AB-stacking. These simulations are HOPG, and GO.
compared to experimental spectra of graphene and HOPG. However, there is still an issue to address. The double-dip
Vertical lines highlight important features in the structure that is supposedly due to AA-stacking is an impor-
experimental spectra; for consistency, the features are tant component of the GO TFY spectra shown in Fig. 2 de-
labeled identically to those in Fig. 3. (A colour version of this spite the fact that multilayered GO is known to have much
figure can be viewed online.) greater interplanar spacing to accommodate the attached
functional groups. The mechanism that allows for AA-stack-
The evidence is clear that that double dip structure seen in ing characteristics to appear in multilayered GO with large
the d2PFY/dE2 of multilayered graphene, as well as that seen interplanar spacing is not understood at this time. However,
in the d2TFY/dE2 of multilayered GO, is due to spectral contri- it is clear that after reduction, minima at 284.6 eV and
butions from both AA- and AB-stacking. We are now ready to 286.5 eV become very prominent in the d2TFY/dE2 of rGO.
answer the question concerning which local minimum one Minima at these energies are due to AA-stacking, as we have
should use out of the three possibilities seen in the d2TEY/ already shown. It would appear that AA-stacking is perhaps
dE2 and d2TFY/dE2 spectra of GO shown in Fig. 2: one should the dominant stacking order in GO, much more so than in
use the lowest peak in the TFY due to AA-stacking (the band multilayered graphene. The reason for the preferred AA-
gap illustrated by arrow b). GO therefore shows a band gap of stacking order, as well as the extent of the interplanar inter-
1.8 eV. The AA-stacking order provides more states closer to action in highly oxidized GO samples, are subjects that re-
the Fermi level, the implication of which is immediate: quire further study.
 CARBON 6 6 ( 2 0 1 4 ) 5 3 9 –5 4 6 545

5. Conclusions Science, Office of Basic Energy Sciences, of the US Depart-

ment of Energy under Contract No. DE-AC02-05CH11231. We
We have found that the stacking configuration of the sheets in also acknowledge Compute Canada for the use of their com-
multilayered graphene and GO has a strong effect on the putational resources, because the electronic structures of
shape of the main p* feature. Fluorescence yield XANES mea- graphene, as well as AA- and AB-stacked bilayer graphene,
surements of graphene, multilayered graphene, and HOPG were simulated on the Grex cluster at the University of Man-
have shown that spectral weight at 286.3 eV and 284.7 eV itoba. The Grex cluster is part of Westgrid.
arises as a function of number layers. Through density func-
tional theory simulations, we have shown this spectral weight
is due to AA-stacking. Although there are other effects that R E F E R E N C E S
can introduce spectral weight at either 286.3 eV or 284.7 eV,
AA-stacking can explain both simultaneously, which is
important, given that they arise simultaneously. AA-stacking [1] Geim AK, Novoselov KS. The rise of graphene. Nat Mater
widens the normally sharp p* feature in multilayered GO as 2007;6(3):183–91.
well, bringing more states closer to the Fermi edge. If the [2] Schedin F, Geim AK, Morozov SV, Hill EW, Blake P, Katsnelson
MI, et al. Detection of individual gas molecules adsorbed on
near-Fermi states have been lost due to oxidation, then one
graphene. Nat Mater 2007;6(9):652–5.
can tune the band gap by controlling the stacking. The stack-
[3] Dai L, Chang DW, Baek JB, Lu W. Carbon nanomaterials for
ing configuration allows for an additional degree of freedom advanced energy conversion and storage. Small
when designing an appropriate band gap. 2012;8(8):1130–66.
The sensitivity of the band structure of AA-stacked multi- [4] Liu Q, Liu Z, Zhong X, Yang L, Zhang N, Pan G, et al. Polymer
layered graphene, and by extension the band structure of AA- photovoltaic cells based on solution-processable graphene
stacked multilayered GO, to interplanar spacing leads to an and P3HT. Adv Funct Mater 2009;19(6):894–904.
[5] Mattevi C, Eda G, Agnoli S, Miller S, Mkhoyan KA, Celik O,
exciting possibility: a band gap that is easily modulated by
et al. Evolution of electrical, chemical, and structural
the external pressure along the c-axis. There has been much properties of transparent and conducting chemically
research into opening a band gap in graphene or bilayer derived graphene thin films. Adv Funct Mater
graphene using stress, but most efforts have been to stress 2009;19(16):2577–83.
the sheet in a direction within the basal plane [46–49]. There [6] Tong SW, Wang Y, Zheng Y, Ng MF, Loh KP. Graphene
also has been some theoretical work into introducing a band intermediate layer in tandem organic photovoltaic cells. Adv
Funct Mater 2011;21(23):4430–5.
gap in bilayer graphene using stress parallel to the surface
[7] Tung VC, Allen MJ, Yang Y, Kaner RB. High-throughput
normal, but the strains necessary were in the range of 30–
solution processing of large-scale graphene. Nat Nanotechnol
50% [50]. However, we suggest going in the opposite direction: 2009;4(1):25–9.
beginning with a wide bang gap, apply stress to close it. [8] Liu J, Lin Z, Liu T, Yin Z, Zhou X, Chen S, et al. Multilayer
According to our results, a device constructed of few-layered stacked low-temperature-reduced graphene oxide films:
GO, if constructed in an AA-stacking regime, would have a preparation, characterization, and application in polymer
tuneable band gap. This gap may close as the layers are com- memory devices. Small 2010;6(14):1536–42.
[9] Stankovich S, Dikin DA, Dommett GHB, Kohlhaas KM,
pressed in the c-axis direction, because adding external com-
Zimney EJ, Stach EA, et al. Graphene-based composite
pression on the AA-stacked GO would push more states
materials. Nature 2006;442(7100):282–6.
towards the Fermi level, as the simulations in Fig. 4 suggest. [10] Myung S, Park J, Lee H, Kim KS, Hong S. Ambipolar memory
Comparing the two AA-stacked graphite calculations (one devices based on reduced graphene oxide and nanoparticles.
with an interplanar spacing of 3.348 Å, the other with Adv Mater 2010;22(18):2045–9.
3.578 Å), a strain of 7% could potentially close the band gap [11] He H, Li X, Wang J, Qiu T, Fang Y, Song Q, et al. Reduced
by 0.2 eV. graphene oxide nanoribbon networks: a novel approach
towards scalable fabrication of transparent conductive films.
Small 2013;9(6):820–4.
Acknowledgments [12] Yin Z, Wu S, Zhou X, Huang X, Zhang Q, Boey F, et al.
Electrochemical deposition of ZnO nanorods on transparent
All samples were made by Sasha Stankovich (at the Depart- reduced graphene oxide electrodes for hybrid solar cells.
ment of Chemistry, Northwestern University, USA). We Small 2010;6(2):307–12.
[13] Sun X, Liu Z, Welsher K, Robinson JT, Goodwin A, Zaric S,
acknowledge support by the Natural Sciences and Engineer-
et al. Nano-graphene oxide for cellular imaging and drug
ing Research Council of Canada (NSERC), the Canada Re- delivery. Nano Res 2008;1(3):203–12.
search Chair program, Canada Foundation for Innovation [14] Yang X, Zhang X, Ma Y, Huang Y, Wang Y, Chen Y.
(CFI), and the Russian Foundation for Basic Research (Projects Superparamagnetic graphene oxide–Fe3O4 nanoparticles
14-02-00006). We further acknowledge the financial support hybrid for controlled targeted drug carriers. J Mater Chem
from the WCU program (R31-2008-000-10029-0) through the 2009;19(18):2710–4.
NRF funded by MEST of Korea. The Canadian Light Source is [15] Avouris P, Chen Z, Perebeinos V. Carbon-based electronics.
Nat Nanotechnol 2007;2(10):605–15.
supported by NSERC, the National Research Council (NSC)
[16] Wallace P. The band theory of graphite. Phys Rev
Canada, the Canadian Institutes of Health Research (CIHR), 1947;71(9):622–34.
the Province of Saskatchewan, Western Economic Diversifica- [17] Siegel DA, Zhou SY, El Gabaly F, Fedorov AV, Schmid AK,
tion Canada, and the University of Saskatchewan. The Ad- Lanzara A. Self-doping effects in epitaxially grown graphene.
vanced Light Source is supported by the Director, Office of Appl Phys Lett 2008;93(24):243119.
546 CARBON 6 6 ( 2 0 1 4 ) 5 3 9 –5 4 6

[18] Li X, Cai W, An J, Kim S, Nah J, Yang D, et al. Large-area [34] Pao C, Babu P, Tsai H, Chiou J, Ray S, Yang S, et al. Electronic
synthesis of high-quality and uniform graphene films on structures of group-III-nitride nanorods studied by X-ray
copper foils. Science 2009;324(5932):1312–4. absorption, X-ray emission, and Raman spectroscopy. Appl
[19] Lerf A, He H, Forster M, Klinowski J. Structure of graphite Phys Lett 2006;88(22):223113.
oxide revisited. J Phys Chem B 1998;102(23):4477–82. [35] Kurmaev EZ, Wilks RG, Moewes A, Finkelstein LD, Shamin
[20] He H, Klinowski J, Forster M, Lerf A. A new structural model SN, Kunes J. Oxygen X-ray emission and absorption spectra
for graphite oxide. Chem Phys Lett 1998;287(1–2):53–6. as a probe of the electronic structure of strongly correlated
[21] Sreeprasad TS, Berry V. How do the electrical properties of oxides. Phys Rev B 2008;77(16):165127.
graphene change with its functionalization? Small [36] Braun C, Seibald M, Boerger SL, Oeckler O, Boyko TD, Moewes
2013;9(3):341–50. A, et al. Material properties and structural characterization
[22] Momma K, Izumi F. VESTA 3 for three-dimensional of M(3)Si(6)O(12)N(2):Eu(2+) (M = Ba, Sr) – a comprehensive
visualization of crystal, volumetric and morphology data. J study on a promising green phosphor for pc-LEDs. Chem Eur J
Appl Crystallogr 2011;44(Part 6):1272–6. 2010;16(31):9646–57.
[23] Ferrari AC, Meyer JC, Scardaci V, Casiraghi C, Lazzeri M, Mauri [37] Boyko TD, Bailey E, Moewes A, McMillan PF. Class of tunable
F, et al. Raman spectrum of graphene and graphene layers. wide band gap semiconductors gamma-(GexSi1-x)(3)N-4.
Phys Rev Lett 2006;97(18):187401. Phys Rev B 2010;81(15):155207.
[24] Hunt A, Dikin DA, Kurmaev EZ, Boyko TD, Bazylewski P, [38] Boyko TD, Zvoriste CE, Kinski I, Riedel R, Hering S, Huppertz
Chang GS, et al. Epoxide speciation and functional group H, et al. Anion ordering in spinel-type gallium oxonitride.
distribution in graphene oxide paper-like materials. Adv Phys Rev B 2011;84(8):085203.
Funct Mater 2012;22:3950–7. [39] Watts B, Thomsen L, Dastoor P. Methods in carbon K-edge
[25] Jia JJ, Callcott TA, Yurkas J, Ellis AW, Himpsel FJ, Samant MG, NEXAFS: experiment and analysis. J Electron Spectrosc Relat
et al. Argonne Natl Lab, Adv Photon Source; Stanford Phenom 2006;151(2):105–20.
Synchrotron Radiat Lab; Assoc Univ Inc. First experimental [40] Lee V, Whittaker L, Jaye C, Baroudi KM, Fischer DA, Banerjee
results from IBM/TENN/TULANE/LLNL/LBL undulator S. Large-area chemically modified graphene films:
beamline at the Advanced Light Source [Proceedings Paper]. electrophoretic deposition and characterization by soft X-ray
Rev Sci Instrum. 1995;66(2, Part 2):1394–1397. In: Fifth absorption spectroscopy. Chem Mater 2009;21(16):3905–16.
international conference on synchrotron radiation [41] Lee V, Park C, Jaye C, Fischer DA, Yu Q, Wu W, et al. Substrate
instrumentation, Stony Brook, NY, July 18–22, 1994. hybridization and rippling of graphene evidenced by near-
[26] Regier T, Krochak J, Sham TK, Hu YF, Thompson J, Blyth RIR. edge X-ray absorption fine structure spectroscopy. J Phys
Performance and capabilities of the Canadian Dragon: the Chem Lett 2010;1(8):1247–53.
SGM beamline at the Canadian Light Source. Nucl Instrum [42] Pacile D, Meyer JC, Rodriguez AF, Papagno M, Gomez-Navarro
Methods A 2007;582(1):93–5. C, Sundaram RS, et al. Electronic properties and atomic
[27] Dikin DA, Stankovich S, Zimney EJ, Piner RD, Dommett GHB, structure of graphene oxide membranes. Carbon
Evmenenko G, et al. Preparation and characterization of 2011;49(3):966–72.
graphene oxide paper. Nature 2007;448(7152):457–60. [43] Trickey SB, Muller-Plathe F, Diercksen GHF, Boettger JC.
[28] Stankovich S, Dikin DA, Piner RD, Kohlhaas KA, Interplanar binding and lattice relaxation in a graphite
Kleinhammes A, Jia Y, et al. Synthesis of graphene-based dilayer. Phys Rev B 1992;45(8):4460–8.
nanosheets via chemical reduction of exfoliated graphite [44] Jung I, Dikin DA, Piner RD, Ruoff RS. Tunable electrical
oxide. Carbon 2007;45(7):1558–65. conductivity of individual graphene oxide sheets reduced at
[29] de Groot F, Arrio M, Sainctavit P, Cartier C, Chen C. low temperatures. Nano Lett 2008;8(12):4283–7.
Fluorescence yield detection – why it does not measure the [45] Jeong HK, Jin MH, So KP, Lim SC, Lee YH. Tailoring the
X-ray absorption cross-section. Solid State Commun characteristics of graphite oxides by different oxidation
1994;92(12):991–5. times. J Phys D Appl Phys 2009;42(6):65418.
[30] Eisebitt S, Boske T, Rubensson J, Eberhardt W. Determination [46] Frank O, Bousa M, Riaz I, Jalil R, Novoselov KS, Tsoukleri G,
of absorption coefficients for concentrated samples by et al. Phonon and structural changes in deformed bernal
fluorescence detection. Phys Rev B 1993;47(21):14103–9. stacked bilayer graphene. Nano Lett 2012;12(2):687–93.
[31] Achkar AJ, Regier TZ, Wadati H, Kim YJ, Zhang H, Hawthorn [47] Ni ZH, Wang HM, Ma Y, Kasim J, Wu YH, Shen ZX. Tunable
DG. Bulk sensitive X-ray absorption spectroscopy free of self- stress and controlled thickness modification in graphene by
absorption effects. Phys Rev B 2011;83(8):081106. annealing. ACS Nano 2008;2(5):1033–9.
[32] Dong C, Persson C, Vayssieres L, Augustsson A, Schmitt T, [48] Ni ZH, Yu T, Lu YH, Wang YY, Feng YP, Shen ZX. Uniaxial
Mattesini M, et al. Electronic structure of nanostructured strain on graphene: Raman spectroscopy study and band-gap
ZnO from X-ray absorption and emission spectroscopy and opening. ACS Nano 2008;2(11):2301–5.
the local density approximation. Phys Rev B [49] Li Y, Jiang X, Liu Z, Liu Z. Strain effects in graphene and
2004;70(19):195325. graphene nanoribbons: the underlying mechanism. Nano
[33] McGuinness C, Stagarescu C, Ryan P, Downes J, Fu D, Smith K, Res 2010;3(8):545–56.
et al. Influence of shallow core-level hybridization on the [50] Verberck B, Partoens B, Peeters FM, Trauzettel B. Strain-
electronic structure of post-transition-metal oxides studied induced band gaps in bilayer graphene. Phys Rev B
using soft X-ray emission and absorption. Phys Rev B 2012;85(12):125403.
2003;68(16):165104.