/ www.sciencexpress.org / 7 May 2009 / Page 1 / 10.1126/science.1171245
Graphene has been attracting great interest because of its
distinctive band structure and physical properties. Today,
graphene is limited to small sizes since it is produced
mostly by exfoliating graphite. We grew large area
graphene films of the order of centimeters on copper
substrates by chemical vapor deposition using methane.
The films are predominantly single layer graphene with a
small percentage, <5%, of the area having few layers, and
are continuous across copper surface steps and grain
boundaries. The low solubility of carbon in copper
appears to help make this growth process self-limiting.
Graphene film transfer processes to arbitrary substrates
were also developed, and top gated field effect transistors
fabricated on Si/SiO
2
substrates showed electron
mobilities as high as 4300 cm
2
V
–1
s
–1
at room temperature.
Graphene, a monolayer of sp
2
-bonded carbon atoms, is a
quasi-two-dimensional (2D) material. Graphene has been
attracting great interest because of its distinctive band
structure and physical properties (1). Today, the size of
graphene films produced is limited to small sizes, usually less
than 1000 μm
2
, because it is produced mostly by exfoliating
graphite, which is not a scalable technique. Graphene has also
been synthesized by desorption of Si from SiC single crystal
surfaces which yields a multilayered graphene structure that
behaves like graphene (2, 3), and by a surface precipitation
process of carbon in some transition metals (4–8).
Electronic application will require high-quality large area
graphene that can be manipulated to make complex devices
and integrated in silicon device flows. Field effect transistors
(FETs), fabricated with exfoliated graphite have shown
promising electrical properties (9, 10), but these devices will
not meet the silicon device scaling requirements, principally
power reduction and performance. One proposed device that
could meet the silicon roadmap requirements beyond the 15-
nm node by S. K. Banerjee et al. (11) The device is a
“BisFET” (bilayer pseudospin FET) device which is made up
of two graphene layers separated by a thin dielectric. The
ability to create this device can be facilitated by the
availability of large area graphene. Making transparent
electrode, another promising application of graphene, also
requires large films (6, 12–14).
At this time, there is no pathway for the formation of a
graphene layer that can be exfoliated from or transferred from
the graphene synthesized on SiC but there is a way to grow
and transfer graphene grown on metal substrates (5–7).
Although graphene has been grown on a number of metals
now, we still have the challenge of growing large area
graphene. For example, graphene grown on Ni seems to be
limited by its small grain size, presence of multilayers at the
grain boundaries, and the high solubility of carbon (6, 7). We
have developed a graphene chemical vapor deposition (CVD)
growth process on copper foils (here, 25 μm thick). The films
grow directly on the surface by a surface catalyzed process
and the film is predominantly graphene with less than 5% of
the area having two- and three-layer graphene flakes. Under
our processing conditions, the two- and three-layer flakes do
not grow larger with time. One of the major benefits of our
process is that it can be used to grow graphene on 300-mm
copper films on Si substrates (a standard process in Si
technology). It is also well known that annealing of Cu can
lead to very large grains.
As described in (15), we grew graphene on copper foils at
temperatures up to 1000ºC by CVD of carbon using a mixture
of methane and hydrogen. Figure 1A shows a scanning
electron microscope (SEM) image of graphene on a copper
substrate where the Cu grains are clearly visible. A higher
resolution image of graphene on Cu (Fig. 1B) shows the
presence of Cu surface steps, graphene “wrinkles,” and the
presence of non-uniform dark flakes. The wrinkles associated
with the thermal expansion coefficient difference between Cu
and graphene are also found to cross Cu grain boundaries,
indicating that the graphene film is continuous. The inset in
Fig.1B shows transmission electron microscope (TEM)
images of graphene and bilayer graphene. Using a process
similar to that described in (16), the as-grown graphene can
be easily transferred to alternative substrates such as SiO
2
/Si
Large-Area Synthesis of High-Quality and Uniform Graphene Films on Copper Foils
Xuesong Li,
1
Weiwei Cai,
1
Jinho An,
1
Seyoung Kim,
2
Junghyo Nah,
2
Dongxing Yang,
1
Richard Piner,
1
Aruna Velamakanni,
1
Inhwa Jung,
1
Emanuel Tutuc,
2
Sanjay K. Banerjee,
2
Luigi Colombo,
3
* Rodney S.
Ruoff
1
*
1
Department of Mechanical Engineering and the Texas Materials Institute, 1 University Station C2200, The University of Texas
at Austin, Austin, TX 78712–0292, USA.
2
Department of Electrical and Computer Engineering, Microelectronics Research
Center, The University of Texas at Austin, Austin, TX 78758, USA.
3
Texas Instruments Incorporated, Dallas, TX 75243, USA.
*To whom correspondence should be addressed. E-mail: colombo@ti.com (L.C.); r.ruoff@mail.utexas.edu (R.S.R.)
/ www.sciencexpress.org / 7 May 2009 / Page 2 / 10.1126/science.1171245
or glass (Fig. 1, C and D) for further evaluation and for
various applications; a detailed transfer process is described
in the supplemental section. The process and method used to
transfer graphene from Cu was the same for the SiO
2
/Si
substrate and the glass substrate. Although it is difficult to see
the graphene on the SiO
2
/Si substrate, a similar graphene film
from another Cu substrate transferred on glass clearly shows
that it is optically uniform.
We used Raman spectroscopy to evaluate the quality and
uniformity of graphene on SiO
2
/Si substrate. Figure 2 shows
SEM and optical images with the corresponding Raman
spectra and maps of the D, G, and 2D bands providing
information on the defect density and film thickness. The
Raman spectra are from the spots marked with the
corresponding colored circles shown in the other panels (in
Fig. 2, A and B, green arrows are used instead of circles so as
to show the trilayer region more clearly). The thickness and
uniformity of the graphene films were evaluated via color
contrast under optical microscope (17) and Raman spectra (7,
18, 19). The Raman spectrum from the lightest pink
background in Fig. 2B shows typical features of monolayer
graphene; e.g., ~0.5 G-to-2D intensity ratio, and a symmetric
2D band centered at ~2680 cm
–1
with a full width at half
maximum (FWHM) of ~33 cm
–1
. The second lightest pink
“flakes” (blue circle) correspond to bilayer graphene and the
darkest one (green arrow) is trilayer graphene. This thickness
variation is more clearly shown in the SEM image in Fig. 2A.
The D map, which has been associated with defects in
graphene, in Fig. 2D is rather uniform and near the
background level, except for regions where wrinkles are
present and close to few-layer regions. The G and the 2D
maps clearly show the presence of more than one layer in the
flakes. In the wrinkled regions, there are peak height
variations in both the G- and 2D-bands, and a broadening of
the 2D band. An analysis of the intensity of the optical image
over the whole sample (1 cm by 1 cm) showed that the area
with the lightest pink color is more than 95%, and all 40
Raman spectra randomly collected from such area shows
monolayer graphene. There is only a small fraction of trilayer
or few-layer (<10) graphene, <1%, and the rest is bilayer
graphene, ~3 to 4%.
We grew films on Cu as a function of time and Cu foil
thickness under isothermal and isobaric conditions. By using
the process flow described in (15) we found that graphene
growth on Cu is self-limited; growth that proceeded for more
than 60 min yielded a similar structure to growth runs
performed for about 10 min. For times much less than 10
min, the Cu surface is usually not fully covered [SEM images
of graphene on Cu with different growth time are shown in
fig. S3 (15)]. The growth of graphene on Cu foils of varying
thickness (12.5, 25, and 50 μm) also yielded similar graphene
structure with regions of double and triple flakes but neither
discontinuous monolayer graphene for thinner Cu foils nor
continuous multilayer graphene for thicker Cu foils, as we
would have expected based on the precipitation mechanism.
According to these observations, we concluded that graphene
is growing by a surface-catalyzed process rather than a
precipitation process as reported by others for Ni (5–7).
Monolayer graphene formation caused by surface segregation
or surface adsorption of carbon has also been observed on
transition metals such as Ni and Co at elevated temperatures
by Blakely and coauthors (20–22). However, when the metal
substrates were cooled down to room temperature, thick
graphite films were obtained because of precipitation of
excess C from these metals, in which the solubility of C is
relatively high.
In recent work, thin Ni films and a “fast cooling” process
were used to suppress the amount of precipitated C. However,
this process still yields films with a wide range of graphene
layer thicknesses, from 1 to a few tens of layers, and with
defects associated with fast cooling (5–7). Our results suggest
that the graphene growth process is not one of C precipitation
but rather a CVD process. The precise mechanism will
require additional experiments to understand, but very low C
solubility in Cu (23–25), and poor C saturation as a result of
graphene surface coverage may be playing a role in limiting
or preventing the precipitation process altogether at high
temperature, similar to the case of impeding of carburization
of Ni (26). This provides a pathway for growing self-limited
graphene films.
In order to evaluate the electrical quality of the synthesized
graphene, dual-gated field effect devices (FETs) using Al
2
O
3
as the gate dielectric were fabricated and measured at room
temperature. The data along with a device model which
incorporates a finite density at the Dirac point, the dielectric
as well as the quantum capacitances (9), are shown in Fig. 3.
The extracted carrier mobility is ~4300 cm
2
V
–1
s
–1
, with the
residual carrier concentration at the Dirac point of n
0
= 3.2 ×
10
11
cm
–2
. These data suggest that the films are of reasonable
quality, at least sufficient to continue improving the growth
process to achieve a material quality equivalent to the
exfoliated natural graphite.
References and Notes
1. A. K. Geim, K. S. Novoselov, Nat. Mater. 6, 183 (2007).
2. C. Berger et al., Science 312, 1991 (2006).
3. K. V. Emtsev et al., Nat. Mater. 8, 203 (2009).
4. P. W. Sutter, J.-I. Flege, E. A. Sutter, Nat. Mater. 7, 406
(2008).
5. Q. Yu et al., Appl. Phys. Lett. 93, 113103 (2008).
6. K. S. Kim et al., Nature 457, 706 (2009).
7. A. Reina et al., Nano Lett. 9, 30 (2009).
8. J. Coraux, A. T. N’Diaye, C. Busse, T. Michely, Nano Lett.
8, 565 (2008).
9. S. Kim et al., Appl. Phys. Lett. 94, 062107 (2009).
/ www.sciencexpress.org / 7 May 2009 / Page 3 / 10.1126/science.1171245
10. M. C. Lemme et al., Solid-State Electron. 52, 514 (2008).
11. S. K. Banerjee, L. F. Register, E. Tutuc, D. Reddy, A. H.
MacDonald, Electron Device Letters, IEEE 30, 158
(2009).
12. P. Blake et al., Nano Lett. 8, 1704 (2008).
13. R. R. Nair et al., Science 320, 1308 (2008).
14. X. Wang, L. Zhi, K. Müllen, Nano Lett. 8, 323 (2008).
15. See supporting material on Science Online.
16. A. Reina et al., J. Phys. Chem. C 112, 17741 (2008).
17. Z. H. Ni et al., Nano Lett. 7, 2758 (2007).
18. A. C. Ferrari et al., Phys. Rev. Lett. 97, 187401 (2006).
19. A. Das et al., Nat. Nanotechnol. 3, 210 (2008).
20. M. Eizenberg, J. M. Blakely, Surf. Sci. 82, 228 (1979).
21. M. Eizenberg, J. M. Blakely, J. Chem. Phys. 71, 3467
(1979).
22. J. C. Hamilton, J. M. Blakely, Surf. Sci. 91, 199 (1980).
23. R. B. McLellan, Scripta Metall. 3, 389 (1969).
24. G. Mathieu, S. Guiot, J. Carbané, Scripta Metall. 7, 421
(1973).
25. G. A. López, E. J. Mittemeijer, Scripta Materialia 51, 1
(2004).
26. R. Kikowatz, K. Flad, G. Horz, J. Vac. Sci. Technol. A 5,
(1987).
27. We would like to thank the Nanoelectronic Research
Initiative (NRI-SWAN; #2006-NE-1464), the DARPA
CERA Center, and The University of Texas at Austin for
support.
Supporting Online Material
www.sciencemag.org/cgi/content/full/1171245/DC1
Materials and Methods
Figs. S1 to S3
22 January 2009; accepted 9 April 2009
Published online 7 May 2009; 10.1126/science.1171245
Include this information when citing this paper.
Fig. 1. (A) An SEM image of graphene on a copper foil with
a growth time of 30 min and (B) a high-resolution SEM
image showing a Cu grain boundary and steps, 2- and 3-layer
graphene flakes, and graphene wrinkles. Inset in (B) shows
TEM images of folded graphene edges. (C and D) Graphene
films transferred onto a SiO
2
/Si substrate and a glass plate,
respectively.
Fig. 2. (A) SEM image of graphene transferred on SiO
2
/Si
(285-nm-thick oxide layer) showing wrinkles, and 2- and 3-
layer regions. (B) Optical microscope image of the same
regions as (A). (C) Raman spectra from the marked spots
with corresponding colored circles or arrows showing the
presence of graphene, 2 layers of graphene and 3 layers of
graphene; (D to F) Raman maps of the D (1300 to 1400 cm
–
1
), G (1560 to 1620 cm
–1
), and 2D (2660 to 2700 cm
–1
) bands,
respectively (WITec alpha300, λ
laser
= 532 nm, ~500-nm spot
size, 100× objector). Scale bars are 5 μm.
Fig. 3. (A) Optical microscope image of a graphene FET. (B)
Device resistance vs top-gate voltage (V
TG
) with different
back-gate (V
BG
) biases and versus V
TG
-V
Dirac,TG
(V
TG
at the
Dirac point), with a model fit (solid line).
