View Online / Journal Homepage / Table of Contents for this issue

PCCP Dynamic Article Links

Cite this: Phys. Chem. Chem. Phys., 2011, 13, 20836–20843

www.rsc.org/pccp PAPER
Graphene CVD growth on copper and nickel: role of hydrogen in kinetics
and structure
Maria Losurdo,* Maria Michela Giangregorio, Pio Capezzuto and Giovanni Bruno
Received 19th July 2011, Accepted 27th September 2011
Published on 17 October 2011 on http://pubs.rsc.org | doi:10.1039/C1CP22347J

DOI: 10.1039/c1cp22347j

Understanding the chemical vapor deposition (CVD) kinetics of graphene growth is important for
advancing graphene processing and achieving better control of graphene thickness and properties.
In the perspective of improving large area graphene quality, we have investigated in real-time the
CVD kinetics using CH4–H2 precursors on both polycrystalline copper and nickel. We highlighted
Downloaded by Brown University on 13 July 2012

the role of hydrogen in differentiating the growth kinetics and thickness of graphene on copper
and nickel. Specifically, the growth kinetics and mechanism is framed in the competitive
dissociative chemisorption of H2 and dehydrogenating chemisorption of CH4, and in the
competition of the in-diffusion of carbon and hydrogen, being hydrogen in-diffusion faster in
copper than nickel, while carbon diffusion is faster in nickel than copper. It is shown that
hydrogen acts as an inhibitor for the CH4 dehydrogenation on copper, contributing to suppress
deposition onto the copper substrate, and degrades quality of graphene. Additionally, the
evidence of the role of hydrogen in forming C–H out of plane defects in CVD graphene on Cu is
also provided. Conversely, resurfacing recombination of hydrogen aids CH4 decomposition in
the case of Ni. Understanding better and providing other elements to the kinetics of graphene
growth is helpful to define the optimal CH4/H2 ratio, which ultimately can contribute to
improve graphene layer thickness uniformity even on polycrystalline substrates.

Introduction rapid cooling of the substrate occurs.12 Conversely, growth on

The unique properties of graphene,1–3 which make it suitable
for a large variety of applications, are motivating the research
for a scalable synthetic route of high-quality graphene. Chemical
vapor deposition (CVD) has emerged as a reliable technological
process for fabricating large area graphene on transition metals
like copper (Cu)4–8 and nickel (Ni).8,9
Although CVD has been used to produce graphene that can
reach already sizes as large as 30 inches and that can be easily
transferred to other substrates,10 polycrystalline graphene
grains separated by grain boundaries that are deleterious for
mobility11 are typically achieved. In order to better understand
the dependence of the shape of graphene grains on various
growth conditions and achieve better control over the graphene
nucleation, a fundamental insight into the growth mechanisms
is needed.
The CVD of graphene on the Ni catalyst has been described
as a two-step mechanism including a first stage of carbon
atoms incorporation into the Ni substrate, followed by out-
diffusion onto the Ni surface to form graphene layers when
Institute of Inorganic Methodologies and of Plasmas, IMIP-CNR,
via Orabona 4, 70126 Bari, Italy.
E-mail: maria.losurdo@ba.imip.cnr.it; Fax: +39 080 5443562;
Tel: +39 080 5443562
polycrystalline Cu substrates has been regarded more viable to
obtain monolayer graphene mainly by a surface controlled
process.13 Even first principles modeling of graphene growth
on different metals shows that the Cu catalyzed process differs
from the growth on other metals.14 Primarily the difference in
the growth kinetics and mechanism between Ni and Cu has
been ascribed to a very low carbon solubility (o0.001 atomic%)
in Cu compared to the higher solubility of carbon in Ni (>0.1
atomic%).7,12 Indeed, is the different carbon solubility the
only factor to be considered in explaining the different graphene
growth on the Ni and Cu catalysts? This is the main question we
are addressing here.
Hydrogen, coming either by the carbon precursor, typically,
CH4, and by the H2 used as diluent gas (mixtures of CH4–H2
are reported in the literature with various ratios) can also have
a role in the graphene CVD growth.15 Furthermore, hydrogen
is also used in the pretreatment (cleaning and crystallization)
of the Ni and Cu substrates, and its interaction with those
substrates can affect the subsequent CH4 chemisorption kinetics.
Although hydrogen has been considered an inactive species16
recombining and desorbing from the catalyst surface, here we
complement the existing vision with further evidence of the
role of hydrogen in the graphene kinetics, since it can actually
perform a number of important processes, i.e., (i) H2 and/or
atomic H can in-diffuse into the catalyst and compete with

20836 Phys. Chem. Chem. Phys., 2011, 13, 20836–20843 This journal is c the Owner Societies 2011

CH4 for chemisorption; (ii) atomic H creates sites for hydro-
carbon and carbon radicals on the surface by subsequent
H-abstraction reactions, removing hydrogen from the surface;
it is generally believed that the main growth species in CVD
growth is the CH3 radical, which on the surface undergoes
successive hydrogen abstraction by H atoms (see discussion
below); (iii) hydrogen can passivate defects and grain bound-
aries; (iv) hydrogen can be active in the competition of CHx
deposition/C-etching, and (v) it can play an important role in
the C sp3 - sp2 transition.
Indication of an important role of hydrogen in determining
the graphene growth kinetics and in limiting the graphene
thickness to a monolayer comes from previous observations
Published on 17 October 2011 on http://pubs.rsc.org | doi:10.1039/C1CP22347J
that when the fraction of CH4 with respect to H2 is increased,
the graphene growth on Cu is no longer self-limiting.16 Further-
more, a deleterious effect of H2 on the quality of graphene on
Cu increasing the density of wrinkles with the increase of H2 has
been reported.17
Additionally, there is an indication of a different effective-
Downloaded by Brown University on 13 July 2012
ness of the CH4 dehydrogenation reaction on the Cu and Ni
catalysts, since addition of Cu to a Ni surface increases the
activation energy barrier, Eact, of the reaction CH4 - CH3 + H
by 1.3 times, and Eact for the reaction CH - C + H by 1.8 times
than that of pure Ni(111).18
Therefore, carbon atoms coexist with hydrogen atoms on
the surface of Ni and Cu catalysts, and consequently their
mutual interaction may change the adsorption and diffusion
properties of carbon atoms. Thus, a potential role of hydrogen
in determining the graphene growth mechanism and kinetics
on Cu and Ni has also to be clarified to gain useful information
on the optimal CH4–H2 ratio to improve graphene deposition.
Here we report on fundamental insight into the role of
hydrogen in graphene CVD growth kinetics on Ni and Cu
monitored for the first time in real-time during graphene
deposition. We demonstrate the competition between carbon
diffusion and hydrogen diffusion, occurring differently for Ni
and Cu, and the competitive role of H2 dissociative chemi-
sorption and CH4 dehydrogenating chemisorption, which
affect the carbon uptake and, hence, the graphene growth
and thickness on the two catalytic substrates. The different
interaction and recombination of hydrogen on Cu and Ni
determining the different active carbon precursor for graphene
on the two metals is highlighted. We also report Raman mea-
surements of how the hydrogen dilution affects the graphene
quality. Understanding better and providing other elements to
the kinetics of graphene growth is helpful to define the optimal
CH4/H2 ratio, which ultimately can contribute to improve
graphene layer thickness uniformity even on polycrystalline
substrates.
Experimental
Graphene deposition
Graphene was grown by chemical vapor deposition (CVD)
from mixtures of CH4 : H2 = 100 : 0–50 sccm gases at a
temperature of 900 1C and at a total pressure of 4 Torr in a
stainless-steel CVD reactor. Ar was also used as diluents to
keep the CH4 partial pressure constant, when varying the H2
This journal is c the Owner Societies 2011
View Online
flux. The samples were then cooled at a rate of B2 1C min1 in
1 Torr of H2.
300 nm Ni/300 nm SiO2/Si and 300 nm Cu/300 nm SiO2/Si
obtained by sputtering were used as substrates. They were
pre-annealed at 400 1C in UHV for nickel and copper oxide
desorption and then heated to 900 1C in 1 Torr of H2 for
recrystallization.
Real time kinetic characterization: spectroscopic ellipsometry
The advantage of the non-destructive and non-intrusive real
time monitoring exploiting spectroscopic ellipsometry (SE)19 is
that it can be applied to detect a few monolayers of graphene
on any substrate, transparent (like SiC) as well as opaque (like
metals). Recently Kravets et al.20 have reported the optical
constant, including the dielectric function, of a monolayer of
exfoliated graphene supported on a SiO2/Si wafer. Nelson
et al.21 also reported the optical properties for CVD graphene
after transferring it to a glass substrate. These previous studies
have set reference optical properties for graphene monolayers,
which were exploited in the present study to monitor, for the
first time, the graphene growth kinetics.
Spectroscopic ellipsometry19 monitored in real-time the
growth by directly recording the pseudodielectric function
hei = he1i + ihe2i of the metal catalyst as well as of graphene
layers, which is related to the extinction coefficient hki and
refractive index hni of materials by the following equation
" #
ð1  rÞ2
hei ¼ he1 i þ ihe2 i ¼ sin2 f 1 þ tan2 f ¼ hðn þ ikÞ2 i
ð1 þ rÞ2
where f is the angle of incidence fixed at 701 and r is the
complex reflection coefficient for the parallel, p, and perpendi-
cular, s, polarizations, defined as
rp
r¼ ¼ tan CeiD
rs
where tan C represents the change of amplitude of the reflected
polarized light beam with respect to the linearly polarized
incident beam, while the phase change between the two polari-
zations is related to cos D. rp and rs are the Fresnel reflection
coefficients. In the kinetic mode, ellipsometric spectra were
acquired every 1 s using a phase-modulated spectroscopic
ellipsometer (UVISEL, Horiba Jobin Yvon) in the 0.75–6.5 eV
spectral range with a 0.01 eV resolution.
Here, the kinetic data are shown at the probing photon
energy of 4.2 eV, because this energy is close to the interband
transition of nickel, which has a main absorption peak above
4 eV due to transitions from the lowest d-band to a free-
electron-like band along all the three directions (L, D and S in
the Brillouin zone22) of copper, which also has two main
interband transitions above 4 eV,23 and graphene, which
shows an absorption peak at 4.6 eV due to a van Hove
singularity in the graphene density of states.20,21
Graphene characterization: Raman spectroscopy
Raman spectroscopy has shown to be a powerful tool to assess
thickness and quality of graphene layers.24 Raman spectra
were then collected using a LabRAM HR Horiba-Jobin Yvon
spectrometer with the 532 nm excitation under ambient conditions
Phys. Chem. Chem. Phys., 2011, 13, 20836–20843 20837

Published on 17 October 2011 on http://pubs.rsc.org | doi:10.1039/C1CP22347J
Downloaded by Brown University on 13 July 2012
Fig. 1 Real-time evolution of he1i monitored at the photon energy of
4.2 eV during exposure of the (a) 300 nm Cu/300 nm SiO2/Si at various
temperatures and (b) 300 nm Ni/300 nm SiO2/Si substrate at 900 1C to
hydrogen. The H2 pressure is 0.1 Torr. The time when hydrogen is
introduced into and removed from the reactor is also shown. The inset
in (a) shows the like-Arrhenius plot for the activation energy, Ea, of
hydrogen diffusion into polycrystalline Cu.
at low laser power (o1 mW) to avoid laser-induced damage
using a Horiba Jobin-Yvon LabRAM HR spectrometer.
Raman mapping with a 1 mm resolution was also run.
Results and discussion
At the typical growth temperature of graphene by CVD of
approximately 900 1C, the diffusion coefficient of hydrogen in
Cu is approximately one order of magnitude higher than for
Ni, i.e., the diffusion coefficient of H2 in Cu is 2  104 cm2 s1,25
while it is lower than 5  105 cm2 s1 for Ni,25 implying a lower
hydrogen solubility in Ni than in Cu.26
20838 Phys. Chem. Chem. Phys., 2011, 13, 20836–20843
View Online
The different response of Ni and Cu to hydrogen already
occurs during the substrate annealing/cleaning step preceding
the graphene growth. The hydrogen interaction with Ni and
Cu has been monitored in real time by recording variation of
the Cu and Ni dielectric function during exposure to hydrogen.
Fig. 1 compares the variation of the real part of the dielectric
function (he1i = n2  k2, where n is the refractive index and k is
the extinction coefficient, being sensitive to variation of Cu and
Ni density by hydrogen or carbon in-diffusion and incorpora-
tion) during exposure of cleaned and annealed Ni and Cu to
hydrogen.
The observed reversible phenomenon with respect to hydrogen
for Cu indicates that hydrogen readily diffuses into Cu, and it
out-diffuses when hydrogen is turned off and pressure decreased.
At the probing wavelength reported in Fig. 1, the sampling depth
is approximately 15 nm and 12 nm for Cu and Ni, respectively.
The slope of the he1i variation is indicative of the diffusion rate,
which is plotted in the like-Arrhenius plot shown in the inset to
yield the activation energy for the hydrogen in-diffusion in
polycrystalline Cu of 0.20  0.01 eV. Conversely, no reversible
response is measured for Ni, consistently with a lower diffusivity
of hydrogen in Ni, indicating that mostly hydrogen recombines
on and desorbs from a Ni surface, according to reaction (1),
where (s) indicates a Ni surface site. This is also consistent with
the fact that for Ni very low temperatures (e.g. 350 K) are
needed to desorb hydrogen from a Ni(111) surface because of
the unique recombination reaction of subsurface with surface
hydrogen on Ni27
H(s) + H(s) - H2m + 2s (highly favored for Ni, not for Cu)
(1)
This different behavior of hydrogen is responsible for the
different trends shown in Fig. 1b: as soon as hydrogen is turned
on the slight increase in he1i can be ascribed to fast removal of
residual oxygen/contaminants and of hydrogen (according to
reaction (1)) from the Ni surface, but after that no reversible
in-diffusion/out-diffusion is detected.
The next step to understand is: how does the hydrogen–
catalyst interaction play when in the presence of CH4 and how
does it affect the graphene growth?
Fig. 2 compares the kinetics monitored during the growth of
graphene on polycrystalline Ni and Cu by CH4–H2. The time
when CH4 is in and off is also indicated, since typically the
cooling down of the samples occurs in H2. Different phenomena
can be read in Fig. 2. Specifically, as soon as the CH4 flows into
the reactor, a transient region (AB) followed by a fast carbon
in-diffusion-controlled kinetics is observed for Ni. In the tran-
sient regime (AB) CH4 is chemisorbed and catalytic dissocia-
tion to carbon on the surface occurs, and both the carbon and
hydrogen at the surface can contribute to the removal of
residual NiO from the surface. The duration of this transient
regime depends on the status, i.e., cleaning and crystallinity, of
the surface of the Ni catalyst; the better the Ni surface, the
more sites are available for CH4 chemisorption and dissocia-
tion and, therefore, the shorter in time this transient regime
(see more in the below for discussion of Fig. 5). Interestingly,
even when CH4 is turned off and the Ni sample cooled down,
the same trend is still observed although with a different slope,
This journal is c the Owner Societies 2011
 View Online
Published on 17 October 2011 on http://pubs.rsc.org | doi:10.1039/C1CP22347J
Downloaded by Brown University on 13 July 2012

Fig. 2 (a) Real-time kinetic profiles of the real part, he1i, of the pseudodielectric function recorded at the photon energy of 4.2 eV during graphene
growth on polycrystalline 300 nm Ni/300 nm SiO2/Si (black curve) and on polycrystalline 300 nm Cu/300 nm SiO2/Si (red curve) (CH4 : H2 =
100 : 3). At A-point CH4 is injected into the reactor, at C-point CH4 is stopped and sample cooled down, at D-point temperature reached 400 1C
and the final point is at room temperature (RT). (b) Typical Raman spectra of graphene grown on Cu and Ni with the corresponding photographs
as inset. (c) Sketch of graphene formation by both direct chemisorptions/deposition on Cu and precipitation/segregation on Ni.

until a temperature of approximately 400 1C is reached. Below The first step to be considered is the competitive dissociative
400 1C the he1i trend reverses and a decrease of he1i is observed. chemisorptions of H2 and physisorption of CH4 on surface
400 1C is the critical temperature above which carbon diffusion sites (s) of Ni and Cu, according to reactions (2) and (3),
in/out of Ni bulk occurs, and below which carbon diffusion is respectively
kinetically inhibited.28
Conversely, in the case of copper, as soon as CH4 is flowing a H2 + 2(s) - H(s) + H(s) (2)
slight linear variation is seen with time until saturation is reached
CH4 + (s) - CH4(s) (3)
(no diffusion-like profiles), and when the CH4 is stopped and the
sample cooled down, no further variation is observed. While we have previously discussed that the sticking coefficient
Furthermore, in order to better understand the role of of hydrogen is higher for Cu than for Ni, CH4 easily physi-
hydrogen and carbon on the Ni and Cu surfaces, we also sorbs on Ni, since the negligible physisorption activation
analyzed the kinetics recorded when after growing graphene energy of 0.01 eV for reaction (3).29 Specifically, the initial
the CH4 is still let to flow in and hydrogen is stopped. In the sticking coefficient of CH4 on Cu is four times lower than Ni,
case of Ni, no appreciable hydrogen reversible phenomenon because of the higher activation energy of 201 kJ mol1 (B2 eV)
was observed (see Fig. 3a). Conversely, for graphene growth for CH4 physisorption on Cu.30
on Ni a variation of he1i occurs when CH4 is off (e.g. see also As for the next steps of CH4 dehydrogenation according to
Fig. 2), whereas for graphene on Cu a variation of he1i occurs the following reactions
when H2 is off, as shown in Fig. 3b.
These different observed profiles can be rationalized in light CH4(s) + (s) - CH3(s) + H(s) (4)
of the different mechanisms for graphene growth on Ni and
CH3(s) + (s) - CH2(s) + H(s) (5)
Cu, which have to take into account the different catalytic
activity of Cu and Ni in the dehydrogenation of CH4 and the CH2(s) + (s) - CH(s) + H(s) (6)
different interaction with hydrogen in addition to the different
C-solubility. CH(s) - C(s) + H(s) (7)

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 20836–20843 20839

Published on 17 October 2011 on http://pubs.rsc.org | doi:10.1039/C1CP22347J
Downloaded by Brown University on 13 July 2012
Fig. 3 Real-time evolution of he1i monitored at the photon energy of
4.2 eV during exposure of (a) Ni and (b) Cu at 900 1C to CH4–H2 =
100 : 5 (T = 900 1C; P = 4 Torr) for graphene growth. The time when
hydrogen is introduced into and removed from the reactor is also shown.
previous studies have already demonstrated that the dissocia-
tive chemisorption of CH4 resulting in the chemisorption of
CH3 and H on Ni according to reaction (4) is the rate limiting
step,31,32 and that at temperatures above 625 K CH3 is
subsequently stepwise dehydrogenated according to reactions
(5)–(7),31–36 and finally at T > 625 K the surface C atoms
diffuse into the Ni bulk where C-nucleation and segregation
take place. Comparing the effectiveness of the CH4 dissocia-
tive chemisorption on the Ni and Cu catalysts according to
reaction (4), previous studies30 determined that the CH4 dissoci-
ative chemisorption probability at 800 K on clean Cu(100) is
4.1  1011, which is five orders of magnitude lower than that for
clean Ni(100), which results to be 3.7  106. The lower activity
of Cu compared to Ni in the catalytic dissociative chemisorptions
of CH4 can be rationalized considering that it occurs by the
electron transfer from the C–H bonds to the 3d orbitals of the
catalysts, with Ni having two 3d unpaired electrons and Cu
having only one unpaired electron available for the interaction
(Cu has electron configuration [Ar]3d104s1, since an electron
passed from the 4d-orbital to 3d to generate a filled 3d electron
shell, which is the most stable configuration).
Therefore, the two competitive dissociative chemisorption
reactions of H2 (reaction (2)) and of CH4 (reaction (4)) compete
with a different effectiveness on Ni and Cu, the former being
favored on Cu, the latter on Ni, determining a different initial
carbon uptake on Ni and Cu surfaces. Furthermore, for the
sequence of dehydrogenation steps (5)–(7), the chemisorbed
H(s) originating by reaction (2) can have an inhibitor role by
blocking/reducing the number of surface sites available for CH4
dissociative chemisorptions and subsequent dehydrogenation.37,38
This hydrogen inhibitor role can be clearly seen in the growth
20840 Phys. Chem. Chem. Phys., 2011, 13, 20836–20843
View Online
Fig. 4 Real-time evolution of he1i monitored at the photon energy of
4.2 eV during graphene growth on 300 nm Cu/300 nm SiO2/Si at
various CH4/H2 ratios. The Raman spectra for 2 samples are also
shown with an explicit I2D/IG ratio, indicative of the graphene thick-
ness, and IG/ID ratio, indicative of defects.
kinetics on Cu in Fig. 4, which clearly shows the decrease of
the graphene deposition rate on Cu with the increase of the H2
concentration, since the CHx (x = 1–4) dissociative chemi-
sorption probability is proportional to the free Cu surface sites
and, hence, a higher partial pressure of H2 would reduce
surface sites, further inhibiting reactions (4)–(7). Furthermore,
a possible increase of the etching rate of graphene by the
higher hydrogen content would also contribute to slow down
the overall growth rate.
Conversely, in the case of Ni, the opposite phenomenon
occurs. It is very improbable to find subsurface hydrogen in
Ni at the growth temperatures of graphene because of the
so-called phenomenon of hydrogen ‘‘resurfacing’’,39 i.e., hydrogen
in a subsurface site migrates to the surface and quickly
recombines with another surface hydrogen, desorbing from
the Ni surface according to reaction (1) experimentally demon-
strated by data in Fig. 1. The fast H2 desorption makes
available more surface sites where CHx can further dissocia-
tively chemisorb according to reactions (4)–(7). Furthermore,
in the Ni case, the hydrogen desorption can even aid the
dehydrogenation of CHx (x = 1–3) according to the additional
processes
CHx(s) + H(s) - CHx1(s) + s + H2m (8)
Therefore, in the case of Ni, the fast recombination and
desorption of hydrogen makes available sites for the dissocia-
tion of CH4, which is followed by a fast in-diffusion of C,
favoring all dehydrogenation steps down to C in-diffusion into
Ni. Furthermore, reaction (8) puts in evidence the role of
hydrogen in the non-self-limited graphene growth on Ni,
i.e., the surface catalyst is never poisoned by carbon and there
are always sites available at the Ni surface for the CH4
dehydrogenation.
This model agrees with and rationalizes the observed pro-
files in Fig. 2 for Ni, i.e., there is no accumulation of H in Ni
because of reaction (1), so that no reversible response to H- is
seen, on the other hand, because of reaction (7), C-diffusion
profiles are seen in Fig. 2.
Conversely, reaction (7), i.e., dehydrogenation of CH is
more difficult to complete on Cu because it has been reported
This journal is c the Owner Societies 2011
 View Online

that on both Cu(100) and Cu(111) the C + H state i.e.,

C-atoms on the Cu surface, has a higher energy than CH4
of 2.75 and 3.6 eV (for the two Cu orientations). In fact,
differently from all other metals like Ni, Pd and Ru where
dehydrogenation of CH4 is exothermic,40,41 Cu is the only
metal where the same reaction is endothermic.42
According to first-principles calculations within density
functional theory,43 C–C dimers are more stable on all sites
of a Cu surface. Therefore, in the case of Cu, instead of
reaction (7) the following reaction has to be considered in
carbon deposition

CH(s) + CH(s) - (s)CQC(s) + H2m (9)

Published on 17 October 2011 on http://pubs.rsc.org | doi:10.1039/C1CP22347J

consistently with the experimental observation of H2 out-

diffusion shown in Fig. 3. The importance of reaction (9) is
the formation of C–C bonds with sp2 hybridization. The
transient product from reaction (9) further aromatizes to the
benzene ring on the Cu surface sites.
Downloaded by Brown University on 13 July 2012

The different role of hydrogen in the graphene kinetics on

Ni can be inferred from Fig. 5, which compares the kinetics of
graphene growth on Ni at various CH4/H2 ratios: for each
growth run reported in the figure the CH4 flow is kept constant
at 100 sccm while the H2 flux is increased from 5 to 50 sccm.
Interestingly, different slopes of the he1i variation are observed
depending on the hydrogen content, the slope i.e., Dhe1i/Dt
being a measure of the process rate.
Here two kinetic regions can be distinguished, a transient
region whose rate decreases with the increase of the H2 content,
and a C-diffusion regime whose rate slightly increases with the
increase of the hydrogen content, indicating that hydrogen also
changes the adsorption and diffusion of C in Ni. The decrease in
the rate by the H2 in the initial regime can be rationalized by
reaction (1), i.e., assuming that the dissociative chemisorption Fig. 5 (a) Real-time evolution of he1i monitored at the photon energy
of CH4 (reaction (4)) is the rate limiting step, the graphene of 4.2 eV during graphene growth on 300 nm Ni/300 nm SiO2/Si
deposition rate for the first layers depends on the number of Ni at various CH4/H2 ratios. (b) The kinetic normalized profile for
sites available for the CH4 dissociative chemisorption and CH4 : H2 = 100 : 10 shows fitting quality (green and red lines) accord-
catalytic dehydrogenation, and surface sites can be hindered ing to the 2nd Fick’s diffusion law.
by competitive hydrogen dissociative chemisorptions—reaction
(1). Conversely, the C-diffusion region is independent of the
hydrogen, since it depends only on the diffusivity, D, of C in Ni kinetics and avoid defects due to the transferring to SiO2)
and by the C concentration at the surface (for the in-diffusion) shows that by decreasing the H2 content, more than one
and in the Ni (for the out-diffusion and segregation steps). This monolayer can be achieved on Cu also, consistently with
is demonstrated by Fig. 5b, which shows that the two segments previous findings.16 Furthermore, taking the IG/ID Raman
in the region of C-diffusion well fit the solution of Fick’s second peak intensity ratio as indicative of defects, the increase of
law (x is the diffusion depth and t is the time) hydrogen also results in a higher defect density. Here, Fig. 6
shows, for the first time, that, among defects, point defects
 
x due to sp3 C–H bonds have to be considered as demonstrated
cðx; tÞ ¼ c0 erfc pffiffiffiffiffiffi
2 Dt by the IR reflection measurements taken on the graphene/Cu
using the BESSY synchrotron light, showing the C–H stretch-
The fit gives the same D value for the two segments, indicating ing peak at 2924 cm1. The fact that the C–H band increases
that it is always the C-atoms in-diffusing (when CH4 is flowing) with the angle of incidence indicates that the C–H stretching
and out-diffusing (when CH4 is off); the two segments only vibration is out-of-plane, as schematized in the inset (a similar
differ in the c0 values being in the first segment the surface band has not been observed for graphene grown on Ni).
concentration coming from reactions (4)–(7) and in the second Fig. 7 shows that there is an effect of the H2 concentration
case the value saturated in the Ni bulk. on the graphene thickness and quality evaluated by the Raman
Finally, an effect of H2 on the graphene quality is also spectra, also for graphene grown on Ni, as a consequence of
found. The Raman analysis of the graphene grown on Cu with the initial impact of hydrogen on the CH4 dissociative chemi-
various H2 contents reported in Fig. 4 (spectra are on the sorptions (reaction (1)) and, therefore, of the solid concentration
as-grown samples on Cu to compare directly the effect of growth of carbon established at the Ni surface. Noteworthily, the D peak

This journal is c the Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 20836–20843 20841

Published on 17 October 2011 on http://pubs.rsc.org | doi:10.1039/C1CP22347J
Fig. 6 IR reflection spectra at various angles of incidence (551–751)
taken on the graphene/Cu using the BESSY synchrotron light, show-
ing a stretching peak at 2924 cm1. The inset shows a sketch of the
out-of-plane C–H defect.
Downloaded by Brown University on 13 July 2012
Fig. 7 (a) Typical Raman spectrum for graphene grown at CH4 : H2 =
100 : 5; the spectrum is directly on the Ni substrate; no D-peak is
observed. (b) Dependence of the 2D peak intensity and full width at
half maximum, FWHM, as a function of increasing H2. The inset also
shows an optical microscope photograph of the graphene on Ni, where
the dark-gray spots indicate thicker graphene.
at 1350 cm1 associated to defects is absent for the as-grown
graphene on Ni, consistently with the absence of the C–H
band in IR reflection measurements. The optical microscopy
photograph shows graphene regions of different thickness. The
lightest gray regions show a full width at half maximum (FWHM)
20842 Phys. Chem. Chem. Phys., 2011, 13, 20836–20843
View Online
(39 cm1) and a single symmetric Lorentzian line shape profile
peaked at 2716 cm1 (on Ni) of the 2D band, with a 2D-to-G
intensity ratio I2D/IG E 2.6 corresponding to a monolayer or
bilayer graphene, covering up to 80% over 22 500 mm2, for a
very low H2 concentration. Dark gray regions are few-layer
(L) graphene with L > 3 as indicated by the Raman spectrum
showing an I2D/IG E 0.9, and a 2D FWHM of 43 cm1. A
decrease of the I2D/IG ratio as well as an increase of the 2D
FWHM are observed with the increase in hydrogen, indicating
also in the case of Ni a deleterious effect on the graphene
quality.
Conclusions
In summary, graphene has been grown by low pressure CVD
using a range of CH4–H2 gas compositions on polycrystalline
Cu and Ni catalysts aiming at understanding the role of
hydrogen in differentiating graphene growth kinetics on Cu
and Ni. The peculiarity of this work has been the real-time
monitoring of the CVD kinetics of graphene on Ni and Cu
substrates. In addition to the well reported different solubility
of C in Cu and Ni, hydrogen also plays an important role
in graphene growth kinetics and quality as summarized in the
following points:
- During the annealing/cleaning of Cu and Ni prior to
initiating the graphene growth, typically run in H2, as well
as during graphene deposition from CH4–H2 mixtures, mole-
cular and atomic hydrogen readily diffuse in Cu, while it
recombines on the Ni surface, in agreement with the different
hydrogen diffusion coefficients (the diffusion of H2 in Cu is
2  104 cm2 s1, while it is 5  105 cm2 s1 for Ni). An
activation energy of 0.20  0.01 eV has been determined for
the diffusion of H2 into polycrystalline Cu.
- Hydrogen dissociative chemisorption competes with CH4
dissociative dehydrogenation on surface sites. This H2 com-
petition plays a kinetic inhibitor role mainly for graphene
growth on Cu.
- Hydrogen slows down the deposition kinetics of graphene
on Cu blocking sites on the Cu surface. Hydrogen has also a
negative effect on the quality of the graphene grown on Cu,
and contributes to create point defects consisting of hybridized
sp3 C–H bonds.
- Hydrogen resurfacing and surface combination aid in
keeping sites on the Ni surface free for the CHx (x = 1–4)
dehydrogenation, yielding diffusion/segregation of carbon.
By optimizing and minimizing the hydrogen content, graphene
without a D-peak in Raman spectra can be grown on poly-
crystalline Ni.
Thus, a better understanding of the role of hydrogen in the
CVD kinetics and graphene properties provides an additional
element to optimize growth parameters, which ultimately can
contribute to improve graphene layer thickness uniformity
even on polycrystalline substrates.
Acknowledgements
The authors thank Mr Alberto Sacchetti at IMIP-CNR for the
technical assistance in performing growth experiments. We
also acknowledge Dr Tom Oates and Dr Karsten Hinrichs at
This journal is c the Owner Societies 2011

ISAS-Berlin for the Infrared ellipsometry measurements at
BESSY. The financial contribution of the FP7 European
project NIM-NIL (GA 228637) is also acknowledged.
Notes and references
1 A. K. Geim and K. S. Novoselov, Nat. Mater., 2007, 6, 183.
2 A. K. Geim, Science, 2009, 324, 1530.
3 C. Berger, Z. Song, X. Li, X. Wu, N. Brown, C. Naud, D. Mayou,
T. Li, J. Hass, A. N. Marchenkov, E. H. Conrad, P. N. First and
W. A. de Heer, Science, 2006, 312, 1191–1195.
4 X. S. Li, C. W. Magnuson, A. Venugopal, R. M. Tromp,
J. B. Hannon, E. M. Vogel, L. Colombo and R. S. Ruoff, J. Am.
Chem. Soc., 2011, 9, 2816.
Published on 17 October 2011 on http://pubs.rsc.org | doi:10.1039/C1CP22347J
5 C. Mattevi, H. Kim and M. Chhowalla, J. Mater. Chem., 2011,
21, 3324.
6 J. Cho, L. Gao, J. Tian, H. Cao, W. Wu, Q. Yu, E. N. Yitamben,
B. Fisher, J. R. Guest, Y. P. Chen and N. P. Guisinger, ACS Nano,
2011, 5, 3607.
7 X. Li, W. Cai, J. An, S. Kim, J. Nah, D. Yang, R. Piner,
A. Valemakanni, I. Jung, E. Tutuc, S. K. Banerjee, L. Colombo
and R. S. Ruoff, Large-area synthesis of high-quality and uniform
Downloaded by Brown University on 13 July 2012
graphene films on copper foils, Science, 2009, 324, 1312.
8 N. Liu, L. Fu, B. Dai, K. Yan, X. Liu, R. Zhao, Y. Zhang and
Z. Liu, Nano Lett., 2011, 11, 297.
9 Y. Lee, S. Bae, H. Jang, S. Jang, S. E. Zhu, S. H. Sim, Y. I. Song,
B. H. Hong and J. H. Ahn, Nano Lett., 2010, 10, 490.
10 S. Bae, H. Kim, Y. Lee, X. F. Xu, J. S. Park, Y. Zheng,
J. Balakrishnan, T. Lei, H. R. Kim and Y. I. Song, Nat. Nanotechnol.,
2010, 5, 574.
11 Q. Yu, L. A. Jauregui, W. Wu, R. Colby, J. Tian, Z. Su, H. Cao,
Z. Liu, D. Pandey, D. Wei, T. F. Chung, P. Peng, N. P. Guisinger,
E. A. Stach, J. Bao, S. S. Pei and Y. P. Chen, Nat. Mater., 2011,
10, 443.
12 X. Li, W. Cai, L. Colombo and R. S. Ruoff, Nano Lett., 2009,
9, 4268.
13 Z. Luo, Y. Lu, D. W. Singer, M. E. Berck, L. A. Somers,
B. R. Goldsmith and A. T. Charlie Johnson, Chem. Mater.,
2011, 23, 1441.
14 H. Chen, W. Zhu and Z. Zhang, Phys. Rev. Lett., 2010,
104, 186101.
15 I. Vlassiouk, M. Regmi, P. Fulvio, S. Dai, P. Datskos, G. Eres and
S. Smirnov, ACS Nano, 2011, 5, 6069.
16 S. Bhaviripudi, X. Jia, M. S. Dresselhaus and J. Kong, Nano Lett.,
2010, 10, 4128.
17 L. Gao, W. Ren, J. Zhao, L. P. Ma, Z. Chen and H. M. Cheng,
Appl. Phys. Lett., 2010, 97, 183109.
18 W. An, X. C. Zeng and C. Heath Turner, J. Chem. Phys., 2009,
131, 174702.
This journal is c the Owner Societies 2011
View Online
19 M. Losurdo, M. Bergmair, G. Bruno, D. Cattelan, C. Cobet, A. de
Martino, K. Fleischer, Z. Dohcevic-Mitrovic, N. Esser, M. Galliet,
R. Gajic, D. Hemzal, K. Hingerl, J. Humlicek, R. Ossikovski,
Z. V. Popovic and O. Saxl, J. Nanopart. Res., 2009, 11, 1521.
20 V. G. Kravets, A. N. Grigorenko, P. R. Nair, P. Blake,
S. Anissimova, K. S. Novoselov and A. K. Geim, Phys. Rev. B:
Condens. Matter Mater. Phys., 2010, 81, 155413.
21 F. J. Nelson, V. K. Kamineni, T. Zhang, E. S. Comfort, J. U. Lee
and A. C. Diebold, Appl. Phys. Lett., 2010, 97, 253110.
22 P. B. Johnson and R. W. Christy, Phys. Rev. B: Solid State, 1974,
9, 5056.
23 K. Stahrenberg, Th. Herrmann, K. Wilmers, N. Esser and
W. Richter, Phys. Rev. B: Condens. Matter, 2001, 64, 115111.
24 A. C. Ferrari, J. C. Meyer, V. Scardaci, C. Casiragji, M. Lazzeri,
F. Mauri, C. Piscanec, D. Jiang, K. S. Novoselov and S. Roth,
Phys. Rev. Lett., 2006, 97, 187401.
25 L. Katz, M. Guinan and R. J. Borg, Phys. Rev. B: Solid State,
1971, 4, 330.
26 E. M. Sacris and N. A. D. Parlee, Metall. Mater. Trans. B, 1970,
1, 3377.
27 R. Baer, Y. Zeiri and R. Kosloff, Phys. Rev. B: Condens. Matter,
1997, 55, 10952.
28 F. C. Schouten, E. Te Brake, O. L. J. Gijzeman and G. A. Bootsma,
Surf. Sci., 1978, 74, 1.
29 W. An, X. C. Zeng and C. H. Turner, J. Chem. Phys., 2009,
131, 174702.
30 I. Alstrup, I. Chorkendorff and S. Ullmann, Surf. Sci., 1992,
264, 95.
31 M. B. Lee, Q. Y. Yang and S. T. Ceyer, J. Chem. Phys., 1987,
87, 2724.
32 M. B. Lee, Q. Y. Yang, S. L. Tang and S. T. Ceyer, J. Chem. Phys.,
1986, 85, 1693.
33 P. M. Holmblad, J. Wambach and I. Chorkendorff, J. Chem.
Phys., 1995, 102, 8255.
34 T. P. Beebe, Jr., D. W. Goodman, B. D. Kay and J. T. Yates, Jr.,
J. Chem. Phys., 1987, 87, 2305.
35 I. Chorkendorff, I. Alstrup and S. Ullmann, Surf. Sci., 1990,
227, 291.
36 L. Hanley, Z. Xu and J. T. Yates, Jr., Surf. Sci., 1991, 248, L265.
37 J.-W. Snoeck, G. F. Froment and M. Fowles, J. Catal., 1997,
169, 240.
38 A. Becker, Z. Hu and K. J. Hüttinger, Fuel, 2000, 79, 1573.
39 G. Henkelman, A. Arnaldsson and H. Jónsson, J. Chem. Phys.,
2006, 124, 044706.
40 C. J. Zhang and P. Hu, J. Chem. Phys., 2002, 116, 322.
41 I. M. Ciobica, F. Frechard, R. A. van Senten, A. W. Kleyn and
J. Hafner, J. Phys. Chem. B, 2000, 104, 3364.
42 W. Zhang, P. Wu, Z. Li and J. Yang, arXiv:1101.3851v1 [cond-mat.
mtrl-sci.].
43 H. Chen, W. Zhu and Z. Zhang, Phys. Rev. Lett., 2010,
104, 186101.
Phys. Chem. Chem. Phys., 2011, 13, 20836–20843 20843