Regulating
Top
-
surface
G
raphene
G
rowth
by
“G
ettering”
C
arbon
Diffusion
at
B
ackside of the
Copper F
oil
Irfan H.
Abidi,
a
Yuanyue Liu,
b,c
Jie Pan,
d
Abhishek
Tyagi,
a
Minghao Zhuang,
a
Qicheng
Zhang,
a
Aldrine
A.
Cagang,
a
Lu
-
Tao
Weng,
a,e
Ping
Sheng,
d
William A. Goddard
b
and Zhengtang
Luo
*,a
a
Department of Chemical and Biomolecular
Engineering,
Hong Kong University of Science and
Technology, Clear Water Bay, Kowloon, Hong Kong
b
Materials and Process Simulation Center,
c
The Resnick Sustainability Ins
titute, California
Institute of Technology, Pasadena, California 91125, United States
d
Department of Physics and William Mong Institute of Nano Science and Technology, Hong
Kong University of Science and Technology, Kowloon, Hong Kong
e
Materials
Characterization and Preparation Facility, The Hong Kong University of Science and
Technology, Clear Water Bay, Hong Kong
Abstract
We
reported a
simple and practical
strategy
that enables us to control the nucleation density
and growth kinetics
for graphen
e grown
on the
top
-
surface
of metal substrate
through
getter
ing
the carbon source on
the
backside
of the
flat
Cu foil
,
during chemical
vapor
deposition
(CVD).
H
itherto,
for
CVD graphene grown
on
a flat Cu foil, merely top
-
surface
-
based growth mechanism has been emphasized, while
the effects from
diffused
carbon and graphene layer
formed
on
the
backside
of the
Cu foil
is normally overlooked
.
O
ur
systematic
experimental findings indicated
that
graph
ene
grown on the backside
,
governs the carbon diffusion through the bulk Cu
, thus strongly dominate
nucleation
process on the top surface
.
This understanding steers us to
devise a strategy of
regulating
carbo
n diffusion to the top surface
by using a
“getter”
substrate
such as
nickel
for carbon
at
backside
of the
Cu foil.
D
epth profiling of the nickel substrate,
along with Density
Functional Theory (DFT) calculation,
verifies
the
gettering role
of
nickel
support
.
Implementing, backside carbon gettering
(BCG) approach to single
-
crystal graphene
growth
resulted in lowering of the nucleation density by two orders of magnitude, enabling
growth of
single
-
crystal domains of
⁓
6 mm lateral size on untreated Cu foil.
Finally,
we
demonstrate
the growth of large area
polycrystalline
SLG
, free of unwanted MLG domains,
with significantly improved
the
field
-
effect
mobility of ~ 6800 cm
2
V
-
1
s
-
1
and
97.
7
%
optical
transmittance
.
Our approach provides an unusual methodology for control in chemical
vapo
r deposition
of 2D materials
.
Introduction
Graphene
brings the aspiration for the worldwide researchers to achieve highly efficient
and flexible electronic and optoelectronic devices, ow
ing to its exceptional optical and
elect
ronic mobility
characteristics.
1
Beyond the mechanical exfoliation
method, which
is
limit
ed in terms of size and uniformity, chemical vapor deposition (CVD)
on transition
metals
emerges as
a practical
and scalable method to meet the ever increasing demand of
uniform and large area graphene.
2
,
3
Among all
the
substrates,
copper (Cu) provides better
control for uniformity and number of layers of graphene
through
self
-
limiting
surface
mechanism
.
4
Although, considerable success have been
made
to gro
w
high
-
quality
CVD
graphene, yet
efforts required to accomplish the
goal of
obtaining
large, uniform, and
defect free
flat single layer graphene (SLG)
.
The major
challenges
are to
deal with defects
such as
g
rain boundaries and unwanted
multilayer
graphene (
MLG)
patches
beneath the
monolayer graphene,
acting as electron scattering barriers that
severely affect the electrical
as well as optical
performances.
5
-
7
In
the
last few years, numerous
efforts
underwent
to
develop
large
,
single crystal graphene
as a strategy
to
circumvent the
adverse effect of grain
boundaries
,
hitherto
limited to smaller size within practical time and growth conditions.
8
-
12
Moreover, t
he existence of
MLG patches
persists in single crystal SLG
growth, affecting
the essence of this technique.
7
,
10
Although, various approaches
have been adopted
to
restrict
the presence of
MLG patches, particularly
manipulating surface
treatment
,
13
-
15
re
-
solidifying
16
and
varying
composition of Cu foil,
17
or modifying
the
CVD process by
introducing pulsed growth.
18
Nevertheless, these
strategies are either unsucce
ssful to
completely limit t
he
MLG existence or worked
only
within constrained growth conditions
and requiring extensive pretreatments
. Therefore,
devising
a
robust
method to
synthesize
large
-
area
uniform
SLG
,
irrespective of growth conditions and surface treatment of Cu foil
,
is
indispensable
for
scaling up its growth at
the
industrial
level
.
On the other hand
,
the role
of carbon diffusion from interior surface of Cu
has been discussed for a few interesting
configuration
, such as
pocket or
deliberated gaps
,
during bi
/multi
-
layer
graphene (B
/M
LG)
growth,
19
-
22
but seldom explored a flat Cu foil, notwithstanding
,
more
practical and
industrially viable configuration.
Despite,
the top surface of flat Cu foil has been
considered solely responsible for bi
-
or multilayer graphene growth governing by two
mechanisms; 1) adsorption
-
diffusion and 2) gas phase penetration
.
23
-
26
Although
most of
the time graphene growth was observed on both sides of Cu foil,
7
,
21
,
27
see Table
-
1
supporting information.
Moreover,
enormous
top surface based strategies are employed to
contr
ol the nucleation density of graphene domains for large single
-
crystal graphene
growth,
9
,
10
,
12
,
28
ignoring the backside
effect, which indeed has a w
ide spectrum of
implementation.
Therefore, exploring the backside carbon diffusion and its impression on
the
kinetics
of graphene growth on
the
top
side of flat Cu foil is indispe
nsable
for
a
full
understanding
of CVD process
to obtain
desired graphene structures
.
In this work,
we
systematically studied the
role of carbon diffusion on the growth
mechanism of MLG domains as well as single crystal
graphene,
and redirected the growth
using “gettering” process at the backside by replacing the
support substrate for flat Cu foil
.
Our
studies demonstrated the
direct
correlation
betwe
en growth of
graphene
on
the
top
surface
and
the
graphene c
overage on
the
backside of Cu
foil that
governs
by
the C species
diffusion through bulk Cu
,
consistent
to Cu pocket configuration
.
19
,
22
Later,
we found this
C diffusion can be restricted completely
by “gettering” the accumulated carbon at the
backside
using a
nickel
support
substrate
.
Finally, density function theor
y (DFT)
calcu
lations validate
the role of nickel
substrate that provides preferential sites for C
species to
bind
as compare
d
to
the
Cu surface.
We implemented this unique backside
carbon gettering (BCG) approach to
alleviate
the nucleation density of graphene domains
as low as
6.11 nuclei/cm
2
,
which eventually lead
s
to the growth of
large size single crystal
graphene
domain
for diameter
reaches to
⁓
6 mm
.
More importantly,
with this method,
high
-
quality
single layer
graphene was produced even without surface treatment
,
a major
step towards industrial scale production
. O
ur
study provides
new understanding of
carbon
gettering
to
synthesize
uniform
high
-
quality
large area
single crystal graphene
for next
generation
electronic and
optoelectronic devices.
Results and
discussion
Figure 1a illustrates
the
configuration
of Cu foil
supported by a
solid
substrate
in our CVD
system.
Herein, w
e
intend to explore
the influence of
support
substrate
on
the
kinetics
of
graphene growth on
both sides of
Cu foil
,
as recent work has shown that the
support
substrate plays a crucial role for graphene growth
on only backside
.
29
,
30
W
e carried out
CVD growth on Cu foil resting on three different substrates (i) quartz (ii) Cu coated
quartz
(hereafter refer as “
quartz
(Cu)
”)
and (iii) nickel, shown inset of Figure 1(b).
Figure 1b
illust
rates the XRD spectra of
quartz
(Cu)
, showing the
successful
fusion of Cu
on quartz
during several CVD cycles,
31
and Ni substrate.
Figure1c
-
h depict
the top and back side of
the Cu foil after 5 minutes of CVD growth
using quartz
(
c,
f
)
, quartz(Cu)
(
d
,
g
)
and nickel
(
e
,h
)
, as
support
ing
substrate
,
using a reported visualization method
.
32
The
obtained
samples, are designated as Cu/quartz, Cu/quartz(Cu) and Cu/nickel, respectively
,
henceforth
.
On th
e top surface, t
he graphene coverage
s
are similar
at present
growth
conditions
and
reach
~
80
%
coverage
within
5 minute
s of CVD growth
,
shown
in F
igure
1c
-
e.
Nevertheless,
the growth
behavior
s
on the backside
distinct
, as shown in
Figure 1f
-
h
and
supplementary Figure
S1
. For instance,
Figure
1
f reveals
the graphene coverage
on
the
backside
was
<
20% for Cu/
quartz
after
5
minutes
of growth
,
significantly
slower compared
to
the
top surface (
~
80%).
While
,
Figure
1
g illustrates the
graphene coverage
reaches to
above 40% for Cu/quartz(Cu),
relatively faster than Cu/quartz. On the contrary,
we did
not observe any gr
aphene growth for Cu/nickel
as
shown
in
Figure
1
h, and it
persist
s
even
though
we prolonged the growth for
3
hours (
supplementary
Figure S
2
).
These
observations reflect the indispensable role of
the
support
substrate in CVD growth kinetics,
hitherto overlooked.
F
or Cu/quartz and Cu/quartz(Cu)
, t
he
differences
in growth rate
can
be explained
with respect to the limited
diffusion of
ga
s
ses
through the micro/nano gaps
present
between Cu foil and
the
underlying
substrate.
Such limited
diffusion is evident
from graphene
distribution
on
the
backside
of Cu
foil,
showing high density at
edges and
low concentration at the
middle part
(supplementary Figure S
3
).
The
asymmetric
growth
rate
on top and back
side
surface of Cu foil is similar to that observed
for
two surfaces of
Cu pocket configuration.
19
,
20
,
22
Furthermore, t
he amplified growth
rate
via
quartz
(Cu)
substrate can be explained considering the additional supply of catalyst (Cu evaporated on
quartz) for
accelerated
dehydrogenation of methane
gas, which eventually increases
graphene growth rate.
30
,
33
Figure 2a
-
c
illustrates t
he
optical images
of graphene
film grown on
the
top
surface of Cu
foil
after 60 minutes of CVD growth and transfer
red
to SiO
2
/Si substrate
.
The contrast
under
an
optical
microscope
allows
us
to
differentiate between
single layer graphene (
SLG
)
and
multilayer graphene (M
LG
)
domains
,
34
as pointed out with arrows.
W
e did not find
noticeable
MLG domains for Cu/nickel growth
, rather the film consists of uniform SLG
,
as shown in
Figure 2c
, in contrast to the presence of MLG for other two
Cu/quartz and
Cu/quartz(Cu)
samples
.
This is further confirmed
by
the
Raman spectra
acquired from
multiple spots, presenting
the presence
of MLG patches for
graphene growth on
Cu/quartz
and Cu/quartz(Cu), whereas
only
SLG film
were
observed for
Cu/nickel
,
evaluated
based
on
typical
intensity ratio of
G and 2D bands
for SLG and MLG
(inset of Figure 2a
-
c)
.
35
Figure 2d
-
f
plot
ted
M
LG domain size
present on
the
top
surface
of Cu foil
against CVD
growth time, along with the percentage of
Cu surface exposed (100
-
graphene coverage
(
%
)
) at back side o
f the Cu foil
.
Previously,
surface exposed at
the
top
surface of
Cu foil
is
considered as the main origin for continual growth of bi
-
multilayer graphene
under the
SLG
,
23
,
25
and
growth
ceased once
after the top surface is completely covered with
graphene
,
due to the self
-
limiting effect of Cu.
19
However,
Figure
2
d shows
M
LG
domains
continue
to
grow larger
even after
the
complete coverage of the
top surface
, also evident from optical
images shown in
supplementary
Figure S
4
.
Instead
, we
found that
growth kinetics of M
LG
domain
cor
relates
closely with the percentage coverage on the backside of Cu foil
, as
indicated in Figure 2e
where
the
growth
of
MLG
domains almost ceased completely after
20
minutes
of CVD
growth,
concurring
with complete coverage of graphene on
the
backside
.
Such
correlation is also demonstrate
d
by
Cu/nickel
sample
, as shown in Figure
2f.
This correl
ation
is consistent with previous observation that
carbon diffusion
from back
to
the top surface,
previously
reported
for BLG growth for Cu enclosure (pocket)
configuration.
19
,
20
In our case,
we proposed that carbon precursors diffuses through the
inevit
able
gaps between Cu foil and support substrate
, subsequently
dehydrogenate
on
backside of Cu foil to give C
active
species, which eventually diffuse through the bulk Cu
to nucleate MLG domains underneath the SLG layer on top side of Cu foil.
These active C
continues to diffuse through the
Cu
until a threshold concentration reached necessary to
start graphene nucleation on the backside.
36
Accordingly
,
once the graphene
completely
covers the
backside of Cu foil (
Figure 2e
), it passivates the catalytic surface
for further
dehydrogenation of carbon precursors
and also acts as a diffusion barrier,
23
resulting in
termination of carbon diffusion
through the bulk Cu,
eventually
leading to the cessation of
the
continual growth
of
MLG
domains even for prolonged CVD growth
.
On the other
hand,
for the Cu/nickel configuration,
the
absence
of graphene on
the
backside of Cu
foil
indicates the
lack
of carbon species available for graphene nucleation at
the
subsequent
surface
.
This
is consistent with the absence of MLG domains
on the top surface
(Figure
2c), due to the unavailability of C species for bulk diffusion
.
We believe nickel
plays an
important role
in
creating carbon deficiency in
the
vicinity of
back
Cu surface
.
Convincingly,
we found that
MLG growth on
the
top
surface
of Cu foil originates from the
bottom (backside), which can be suppressed by
terminating
the
carbon
supply to
the
top
surface.
As a proof of concept
, we extended our investigations
to CVD growth for large size MLG
using quartz and nickel as
a
support
substrate
.
Figure 3a
displays the o
ptical image
s
of
transferred
top
-
surface
graphene, showing
the presence of large size MLG domains for
Cu/quartz, after 120 minutes of CVD growth.
Although
the
top surface
is
covered
completely by SLG within 20 minutes
,
along
with very few small MLG domains
(supplementary Figure S
5
)
,
the
MLG domains
continue
to grow larger with
extended
CVD
growth
for
2 hours.
The MLG domains
continued to grow
even after the complete coverage
of top surface, once again rules out the
possibility that top surface is the only origin for
nucleation of MLG
, rather involves the other sources
.
Nevertheles
s,
Figure 3b reveals
th
at
graphene
film persists to be
MLG
-
free
uniform
monolayer
for Cu/nickel configuration
even though
for extended
growth to 120 minutes.
Figure 3c,d
illustrate t
he Raman spectra
acquired from marked areas indicate the features of MLG for Cu/quartz and uniform SLG
for Cu/nickel configuration
.
35
T
he Raman image
s
is
constructed by
using FWHM
(full
width at half maximum)
of
the
2D
band
,
evidencing
the uniform SLG growth
(FWHM
⁓
24
-
32 cm
-
1
)
35
,
37
on the
Cu/nickel
sample
whil
st
presence of MLG domains
(FWHM > 50
cm
-
1
)
for
the
Cu/quartz
system
,
as
evident
from Figure 3 e,f
.
In addition, the absence of
graphene on
the
backside
of
the
Cu foil
for Cu/nickel case
(supplementary Figure S
6
)
,
identifies
the role of
the
nickel
support
as
backside carbon gettering (BCG)
to limit the
bulk diffusion
through Cu
,
resulting in
a
lack
of MLG nucleation
on the
top
surface
.
To
further
confirm the role of carbon diffusion from
the
backside
of Cu foil for MLG growth,
we carried out CVD growth by wrapping/sealing the Cu foil around the quartz substrate to
close the possi
ble gaps for
leaking
-
in
the gas precursors to the back surface
, as a control
experiment
. The
growth resulted in uniform SLG on
the
top
side, while no growth on the
backside of Cu foil
, shown in
supplementary Figure S
7
, similar to the
aforementioned
BCG
method
.
Additionally, we
tune the gap size
to see the BCG effect by maintaining a gap of
0.5 mm and 2.0 mm between nickel support and Cu foil, shown in Figure S8. Interestingly,
by increasing the gap from 0.5 to 2 mm, the MLG patches start to appear again
indicating
the BCG effect is
diminishing.
These findings
cement
o
ur proposed mechanism
of
subsequent suppression of
MLG nucleation
through gettering the carbon
from
the
backside
of
the
Cu foil.
T
o
deeply
understand
the mechanism of
BCG
graphene growth
,
we
analyze
the chemical
nature of the substrates using
Time
-
of
-
Flight Secondary Ion Mass Spectrometry (ToF
-
SIMS)
.
In this experiment, the same Cu foil was partially supported (right half) with Ni
and the rest (left half) with
quartz substrate during CVD growth.
The Cu foil was
slightly
oxidized after CVD growth to visualize the
ir boundary
, followed with
ToF
-
SIMS ions
to
map
different possible
ions
.
Figure 4a
shows
ToF
-
SIMS mapping
which
illustrate
s
the lack
of C
2
-
ions on
Cu
surface
(right half)
supported with Ni substrate, indicating the absence
of C species or graphene,
38
in contrast to the surface supported with quartz
(left half)
.
Besides
, lack of
Cu
O
2
-
ions on the left half
specifies
the oxidation barrier
provided by
graphene film on subsequent Cu surface. It is worth noticing that we did not find any Ni
ions on
the
backside
of Cu foil exhibiting lack of any alloying/welding between Cu and Ni
substrate.
Furthermore, depth profiling of
nickel
support substrate before and after
CVD
growth reflects the penetration of C inside
the
nickel
support
during the
CVD
process
, as
shown in
Figure
4
b. The enhanced C concentration inside
the
nickel
support indicates
the
BCG
role
of nickel
to suppress the ca
rbon diffusion toward
the
top
side of Cu foil during
the
CVD
process
.
When
we replaced
the supposedly
inert support substrate (quartz) with
Ni, the
consumption of
dissociated C
predominantly
by
the
underlying
nickel support
resulted
in
a
lack
of
C
diffusion upwards to
the
top surface of Cu
.
Although, it is well known that
nickel has
a
higher
solubility of carbon as compared to that
of Cu,
39
but
it does not
explain the selective adsorption of carbon species on nickel surfaces
during CVD,
40
as we observed
the
complete
absence of graphene on the backside of Cu
foil (Figure 1h & 4a).
Therefore,
we perform
ed
d
ensity functional theory (DFT)
calculations
for our system
,
using the Vienna Ab
-
initio Simulation Package (VASP)
41
with
projector augmented wave (PAW) pseudopotentials
,
42
and the Perdew
-
Burke
-
Ernzerhof
(PBE) exchange
-
correlation functional
.
43
Ni or Cu is
modeled
by a slab with 6 layers and
5
×
5 surface supercell. 400 eV is used for the
plane
-
wave cut
-
off energy, and
the systems
are fully relaxed until the final force on each atom was < 0.01 eV/Å. 5
×
5
×
1 Monkhorst
-
Pack
k
-
points are sampled over the Brillouin zone.
W
e find that Ni provides
preferential
binding sites
(< 1eV)
for C compared with Cu
, as
shown
in
Figure 4c
, which offers a
thermodynamic driving force for C
to diffuse towards nickel rather
than
Cu
, in agreement
with
our
experimental findings
of BCG
outcome
. Interestingly, the most energetically
favorable si
te is located at
the sub
-
surface of Ni
(see Figure 4c for comparison of
the
energetics
of C at different sites of Ni and Cu)
.
This seems to be consistent with the SIMS
depth profile, which
reflects the penetration of C
near the surface, after
CVD
growth.
Moreover
,
we observe that the
BCG approach seems
to be
more effective
controlling
the
nucleation density during
the
oxygen assisted
graphene growth
.
9
,
29
Figure
5a shows the
schematic that how BCG works during the single crystal growth. The backside carbon
diffusion (white arrows
) for Cu/quartz
contributes to
increased
nucleation
density
at
the
top
surface
of Cu foil
,
due to
higher carbon content at the
Cu surface promotes the
nucleation of graphene domains.
44
In contrast, BCG mechanism limits the carbon diffusion
to the top surface (red arrow) for Cu/nickel, res
tricting the nucleation
density
at the top
.
Figure 5b depict
s
the plot for nucleation density of graphene domains
grown
on
the top
surface of Cu foil supported by quartz, nickel plate (Ni
-
P) and nickel foam (Ni
-
F).
The
nucleation density decreases two
orders
of magnitude
by just replacing quartz with
the
Ni
foam as a support, using same CVD conditions
, also
shown in Figure S
9
.
It is worth noting
that
no
chemical/
electropolishing
of Cu foil before CVD growth
is needed in our processes
.
Yet, t
he
nucleation
density reaches to
as low as 6.11 nuclei/cm
2
(0.06 nuclei/mm
2
)
for
Cu/Ni foam, which is among the lowest reported
so far
.
7
,
9
-
11
,
31
Interestingly
, oxygen
presence
accelerates the carbon diffusion through the bulk
Cu,
20
therefore,
suppressing the
carbon diffusion
thus favors the proces
s
.
This approach is
very
unique
and seldom
reported
to control the nucleation density on the top surface of
the
Cu foil while operating
from the backside, in contrast to the previously reported
merely
top
-
surface based
modifications.
8
-
10
,
12
,
28
,
31
Figure 5(c
-
e) shows the SEM images of
graphene domains grown on Cu foil using different
support substrates, the growth time was 10 minutes for (c) and 60 minutes for (d & e).
Interestingly, we also observed that using Ni foam instead of Ni plate also lowers down the
nucleation density from 44.
3 nuclei/cm
2
to 6.11 nuclei/cm
2
, at same conditions.
This
event
can be explained
by
the fact of amplified
carbon
gettering
capability
of Ni foam with more
surface,
and
creating
a
deficiency
of active carbon spe
cies in the local environment
,
with
higher
catalytic power for carbon precursor
s
competing with the subsequent Cu surface
,
as
verified with our DFT calculations discussed above
.
Figure 5f plot doma
in size
as a
function of
growth time, using three different support substrate. For Cu/quartz, the doma
in
size is limited to 361 μm on untreated Cu foil since high nucleation density resulted in
merging of domains within 30 minutes of growth.
Conversely
, Ni plate and
further
Ni foam
facilitate
growing
large millimeter
-
size
d
domains of single crystal graphen
e
assisted by
low nucleation density achieved through BCG effect. For Cu/Ni foam, graphene domains
with ⁓
4.5 mm lateral size
was grown within
5 hours
(
15 μm/min
) of CVD, which is
considered as the fast growth for millimeter
-
sized domains on Cu foil.
30
,
40
,
45
However,
decreasing the flow rate of methane produces single crystal domain of 6 mm in lateral size
on Cu/Ni foam (inset Figure 5f) with growth rate of 10 μm/min, faster than conventional
CVD meth
od.
7
-
11
,
31
Further, we verified the
single crystal nature of large graphene domains
by
performing selected area electron diffraction (SAED) on transferred domain to TEM
grid. SAED patterns were taken from more than 20 random spots over the 3 mm domain;
the same orientation of these patterns with rotation less than 1.5 degrees indicated the
single
crystallinity of the graphene, as shown in Figure S1
0
.
Figure 5g shows the photograph of
as grown isolated single crystal graphene domains on Cu/Ni foam, after
oxidation
for the
visualization. Figure 5h depicts the
domains
transferred to the 2 cm×2
cm SiO
2
/Si wafers
.
Furthermore, Figure 5i shows optical image revealing the benefit of BCG method to get
large single crystal graphene domains free of any bi
-
multilayer graphene patches, which
otherwise observed for Cu/quartz shown in Figure 5j and also in
most of the literature
reported previously.
7
,
10
,
11
,
31
Finally, we
implemented
our
BCG
approach
to
grow
MLG
-
free uniform
large area
continuous
SLG sheet
crucial
for efficient opto
electronic devices.
Here
in, we
demonstrate
the graphene
growth
on
“
untreated
”
Cu foil
using
quartz and nickel as
a
support
substrate,
respectively.
Figure
6
a
shows the optical images revealing
the
presence of aligned
MLG
patches underneath the SLG film
grown on
Cu/quartz
, analogous to
rolling features present
on raw Cu foil
(supplementary Figure S
11
)
. On the other hand, we get exclusively uniform
SLG
on Cu/nickel
within 30 minutes of CVD growth
using
the
BCG
method
,
without any
observable MLG patches
over the area of
2 cm×2 cm
, as shown in
Figure
6
b
and S
12
.
Furthermore, i
t is worth
noticing
that
our
BCG
method allow
s
the
total
elimination of the
complicated surface pretreatment step conventionally
required for graphene growth.
13
,
15
,
18
Achieving large area
MLG
-
free
uniform SLG on Cu foil
,
without any prior surface
treatment (chemical/electro
-
polishing) is
useful
particularly for industrial applications.
To evaluate the
electronic properties
,
we fabricate the FET devices
into Hall bar geometry
using electron beam lithography and oxygen plasma etching, as shown in F
igure
6
c.
T
he
resistivity (
s
)
is plotted as a function of gate voltage (V)
and
Hall
mobility
is measured
by fitting the data for
continuous
graphene synthesized using
conventional (
Cu/quartz
)
and
BCG
(
Cu/nickel
) CVD method
, as depicted in Figure
6
d.
The
field
-
effect
mobility of
electron and holes is represented
by
blue and red fitted curves, respectively.
For
graphene
synthesized by
conventional CVD
, the field effect curve
seems highly distorted and doped
as compared to that of
synthesized by
BCG method
. Moreover,
th
e
mobility
of
graphene
synthesized by
BCG method has
improved s
ubstan
tially and reaches to
68
00
cm
-
2
V
-
1
S
-
1
for
electrons, which is 2.6 times higher than graphene grown on Cu/quartz (2600 cm
-
2
V
-
1
S
-
1
)
.
The remarkable improvement in electrical transport properties
can be
attributed to the
complete removal of MLG patches
using
our novel method.
18
,
46
Further
improvement
s
in
mobility values
may also
be
achieved by optimizing
the transfer method
through reducing
the effect of contaminations
.
20
,
29
Nevertheless
,
our method provides an
additional
advantage to omit
the
extra
step of removing backside graphene
on Cu foil
,
7
which
resulted
in more efficient and cleaner transfer using
wet transfer
method
, as shown in
supplementary Figure S
1
3
.
Furthermore, we transferred graphene films on
the
glass
to
evaluate the transmittance of SLG grown by
the
conventional
and BCG method,
as shown
in F
igure
6
e
.
At 550 nm wavelength, t
he
measured
transmittance
was 9
7
.
7
% and 9
6
.
1
% for
SLG grown on Cu/nickel and Cu/quartz, respectively.
Lower attenuation of transmitted
light (
2
.
3
%) for graphene grown by BCG method, reflects the complete elimination of
scattering centers (MLG patches).
Therefore,
our
“gettering”
CVD
method
steers the
conventional CVD process towards new possibilities to synthesize
high
-
quality
exclusive
SLG
either continuous or
large
single crystal domains for
scaling up the potential graphene
-
based
transparent and
high
-
performance
opto
electronic devices.
Conclusion
In conclusion,
we
insight
deeply
into the origin of multilayer graphene nucleation by
adopting the strategy of using
a
“gettering”
substrate during
the
CVD
process
. Our results
indicate
graphene growth
on the top surface is
strong
ly
depen
ded
on the
back surface of
Cu foil that
pursue
s
with carbon diffusion through bulk Cu. By shutting down the carbon