of 31
Nanotechnology
ACCEPTED MANUSCRIPT
Single-step growth of graphene and graphene-based nanostructures by
plasma-enhanced chemical vapour deposition
To cite this article before publication: Nai-Chang Yeh
et al
2019
Nanotechnology
in press
https://doi.org/10.1088/1361-6528/aafdbf
Manuscript version: Accepted Manuscript
Accepted Manuscript is “the version of the article accepted for publication including all changes made as a result of the peer review process,
and which may also include the addition to the article by IOP Publishing of a header, an article ID, a cover sheet and/or an ‘Accepted
Manuscript’ watermark, but excluding any other editing, typesetting or other changes made by IOP Publishing and/or its licensors”
This Accepted Manuscript is
© 2019 IOP Publishing Ltd
.
During the embargo period (the 12 month period from the publication of the Version of Record of this article), the Accepted Manuscript is fully
protected by copyright and cannot be reused or reposted elsewhere.
As the Version of Record of this article is going to be / has been published on a subscription basis, this Accepted Manuscript is available for reuse
under a CC BY-NC-ND 3.0 licence after the 12 month embargo period.
After the embargo period, everyone is permitted to use copy and redistribute this article for non-commercial purposes only, provided that they
adhere to all the terms of the licence
https://creativecommons.org/licences/by-nc-nd/3.0
Although reasonable endeavours have been taken to obtain all necessary permissions from third parties to include their copyrighted content
within this article, their full citation and copyright line may not be present in this Accepted Manuscript version. Before using any content from this
article, please refer to the Version of Record on IOPscience once published for full citation and copyright details, as permissions will likely be
required. All third party content is fully copyright protected, unless specifically stated otherwise in the figure caption in the Version of Record.
View the
article online
for updates and enhancements.
This content was downloaded from IP address 131.215.71.192 on 22/01/2019 at 16:44
IOP
Publishing
Nanotechnology
Nanotechnology
XX
(XXXX) XXXXXX
https://doi.org/XXXX/XXXX
xxxx-xxxx/xx/xxxxxx
1
© xxxx IOP Publishing Ltd
Single-step growth of graphene and graphene-
based nanostructures by plasma-enhanced
chemical vapour deposition
Nai-Chang Yeh
1,2
, Chen-Chih Hsu
1
, Jacob Bagley
3
, and Wei-Shiuan Tseng
1
1
Department of Physics, California Institute of Technology, Pasadena, California 91125, USA
2
Kavli Nanoscience Institute, California Institute of Technology, Pasadena, California 91125, USA
3
Department of Chemistry, California Institute of Technology, Pasadena, California 91125, USA
E-mail: ncyeh@caltech.edu
Received xxxxxx
Accepted for publication xxxxxx
Published xxxxxx
Abstract
The realization of many promising technological applications of graphene and graphene-
based nanostructures depends on the availability of reliable, scalable, high-yield and low-cost
synthesis methods. Plasma enhanced chemical vapor deposition (PECVD) has been a
versatile technique for synthesizing many carbon-based materials, because PECVD provides
a rich chemical environment, including a mixture of radicals, molecules and ions from
hydrocarbon precursors, which enables graphene growth on a variety of material surfaces at
lower temperatures and faster growth than typical thermal chemical vapor deposition (T-
CVD). Here we review recent advances in the PECVD techniques for synthesis of various
graphene and graphene-based nanostructures, including horizontal growth of monolayer and
multilayer graphene sheets, vertical growth of graphene nanostructures (VG-GNs) such as
graphene nanostripes (GNSPs) with large aspect ratios, direct and selective deposition of
monolayer and multi-layer graphene on nanostructured substrates, and growth of multi-wall
carbon nanotubes (MWCNTs). By properly controlling the gas environment of the plasma, it
is found that no active heating is necessary for the PECVD growth processes, and that high-
yield growth can take place in a single step on a variety of surfaces, including metallic,
semiconducting and insulating materials. Phenomenological understanding of the growth
mechanisms are described. Finally, challenges and promising outlook for further development
in the PECVD techniques for graphene-based applications are discussed.
Keywords: graphene, vertically grown graphene nanostructures (VG-GNs), graphene nanostripes (GNSPs), carbon nanotubes
(CNTs), plasma enhanced chemical vapour deposition (PECVD), thermal chemical vapour deposition (T-CVD)
1. Introduction – A new era of graphene-based
technologies
Graphene, a single layer of carbon atoms forming a
honeycomb lattice structure with two-dimensional
sp
2
bonding, has simulated intense research activities since its
experimental isolation in 2005 because of its superior and
novel electronic, optical, thermal, magnetic and mechanical
properties that are promising for a wide range of
technological applications [1-4]. Despite abundant scientific
advances and prototype device demonstrations mostly based
on small pieces of mechanically exfoliated graphene, the full
potential of graphene-based technologies in such large-scale
Page 1 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
2
applications as in supercapacitors and lithium-ion batteries
for energy storage, in photovoltaic and fuel cells for energy
conversion and in large panels for display cannot be realized
unless high-quality graphene can be reliably synthesized in
mass quantities and at low costs. On the other hand,
compatibility of the graphene growth process with CMOS
technologies will be critical to the realization of graphene-
based nanoelectronic and optoelectronic applications.
Moreover, efficient and scalable graphene transfer
technologies from their growth substrates to desirable targets
without degrading the sample quality will be necessary for
many industrial applications. These technical challenges
place strong constraints on the methods of graphene
production. The purpose of this review is to provide an
update on recent advances in the graphene growth
technologies based on plasma-enhanced chemical vapour
deposition (PECVD), including the growth of graphene
sheets and graphene-based nanostructures on different
substrates, as well as the selective growth of graphene on
nanostructures in real industrial semiconductor wafers, and
then presents an outlook for the promises and challenges
ahead to advance the research and technology.
1.1. Current techniques used for synthesis of graphene
and graphene-based nanostructures
To date several methods have been developed for mass-
production of graphene, which include liquid-phase
exfoliation of graphite [5, 6], synthesis on SiC [7, 8], thermal
chemical vapour deposition (T-CVD) [9], and plasma-
enhanced chemical vapour deposition (PECVD) [10, 11].
Among these different synthesis methods, T-CVD has been
developed for growing large-area graphene with reasonably
high quality. However, T-CVD growth of graphene generally
requires multiple processing steps and relatively long time in
both substrate preparation and graphene growth [12-14].
Moreover, high-temperature processes (~ 1000
C) in the T-
CVD synthesis are incompatible with applications relevant to
semiconducting industry, and the high thermal budget adds
further constraints on mass production. Recently, oxygen and
oxygen containing species were also found to play an
important role in graphene synthesis, such as graphene
nucleation, graphene shape, and bilayer and multilayer
graphene formation [12, 15-17].
In contrast to the T-CVD growth method, PECVD has
proven to be a versatile approach that offers a number of
advantages [10, 11]. PECVD has been widely used for
synthesizing many carbon-based materials, such as
diamonds, graphene, vertically oriented graphene nano-walls
and nano-sheets, and carbon nanotubes (CNTs). The plasma
can provide a rich chemical environment, including a
mixture of free radicals, photons, energetic electrons, excited
molecules and active ions. This environment enables
graphene growth on different surfaces at relatively lower
temperatures and faster growth than T-CVD [10, 11].
Additionally, PECVD techniques can be employed for fast
and large-scale functionalization of graphene and related
materials, which is a versatile approach that further broadens
the scope of graphene-based applications. These advantages
make PECVD growth of graphene and graphene-related
nanostructures highly attractive, and have been considered as
a promising technique to improve the compatibility of
graphene growth with semiconducting manufacturing
processes.
1.2. Plasma sources for PECVD growth techniques
Before we proceed further with the PECVD conditions for
graphene synthesis, it is informative to briefly summarize
typical plasma sources used in PECVD processes. Different
plasma sources can be categorized by different power
frequencies, which include: direct current (DC) gas discharge
sources; radio frequencies (RF) sources with frequencies
ranging from 1 to 500 MHz; microwave (MW) sources with
frequencies ranging from 0.5 to 10 GHz; and combinations
of the aforementioned types.
Typical DC-PECVD synthesis involves a relatively simple
setup using the parallel-plate dc glow discharge. With a
sufficient potential applied between the planar cathode and
anode, Townsend breakdown will occur and plasma is
generated [10]. The RF sources have three main modes for
coupling the energy of an RF generator to the plasma: the
evanescent electromagnetic (H) mode, the propagating wave
(W) mode, and the electrostatic (E) mode [10]. The MW
sources typically operate with either the transverse magnetic
(TM) or transverse electric (TE) propagation modes [10].
The choice of the plasma source and reactor configuration
for PECVD growth of graphene depends on the specific
applications and the growth substrates. Among different
plasma sources and reactor configurations, the most efficient
delivery of power per unit volume is the TM-MW
frequencies. On the other hand, in the event of delicate
substrates or specific applications that are sensitive to ionic
bombardment and/or ultraviolet (UV) light exposure,
configurations with remote plasma arrangements may
become desirable. A recent review by Z. Bo
et al.
[10]
contains comprehensive accounts and references for various
plasma sources and reactor configurations that are employed
for PECVD growth of graphene nanostructures, and so we
shall not elaborate different techniques further here. In the
following we shall simply specify the PECVD method used
when we compare the growth conditions and the resulting
graphene characteristics by various research groups.
1.3. Outline of this review
The remaining part of this article is structured as follows.
Section 2 is devoted to the status of horizontal growth of
large-area graphene sheets and selective growth of graphene
Page 2 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
3
on nanostructures by PECVD, studies of the resulting
graphene characteristics, and discussions of their potential
applications. Section 3 describes high-yield vertical growth
of graphene nanostructures, including quasi-one dimensional
graphene nanostripes (GNSPs) by seeded PECVD methods,
and the studies of their properties. In Section 4 recent
progress in the vertical growth of multi-wall carbon
nanotubes (CNTs) on different substrates by PECVD is
reviewed. In Section 5 we discuss important issues
associated with further development of PECVD and related
techniques for graphene-based applications. Finally we
summarize in Section 6 the status of PECVD growth of
graphene and graphene based nanostructures, and present an
outlook for new directions and challenges.
2. PECVD growth of graphene on different substrates
In this section, we review recent progress in PECVD growth
of graphene sheets on different substrates and selective
horizontal growth of multilayer graphene on nanostructures
of real industrial semiconductor wafers.
2.1. Reviews of PECVD growth of graphene
2.1.1 PECVD growth of graphene sheets on transition
metal substrates
Generally speaking, graphene growth could be achieved at a
reduced temperature by PECVD on different transition metal
substrates such as Co [18, 19], Ni [20-23], and Cu [11, 24-
31]. Among the pioneering works, Woo
et al.
grew high-
quality and uniform graphene films at 850 °C in a remote
radio frequency PECVD (RF-PECVD) system [20]. SEM
images and electron backscattering diffraction map showed
highly crystallized graphene with few atomic defects and
well-ordered structure. The carrier mobility was ~ 4500
cm
2
V
-1
s
-1
at room temperature. Nandamuri
et al.
also
employed a RF-PECVD system to synthesize multilayer
graphene (MLG) on Ni (111) single crystals and
polycrystalline Ni foils in about one-minute growth time
[21]. The size of the graphene domains was found to be
consistent with the dimensions of the flat grain Ni surfaces,
which ranged from ~ 1 μm to ~ 20 μm, suggesting epitaxial
growth on the Ni polycrystalline substrate. Subsequently,
single layer graphene (SLG) was successfully grown on Ni
foil by MW-PECVD with the growth temperature from 450
to 750 °C [22], and the number of graphene layers could also
be controlled by changing the gas mixture of H
2
/CH
4
ratio.
Peng
et al.
demonstrated the synthesis of few-layer graphene
sheets on an ultra-thin Ni film coated on SiO
2
/Si substrate
using low temperature RF-PECVD without introducing any
H
2
[23], where the number of graphene layers could be
controlled by the thickness of Ni film at 475 °C. In general, it
is found that the use of Ni substrates for graphene synthesis
by PECVD methods typically yields MLG due to the high
carbon solubility in Ni, similar to the findings in T-CVD
graphene growth.
On the other hand, Cu foils have been widely used for
graphene growth because of the low cost and commercial
availability. Moreover, Cu is an excellent substrate for
synthesizing high quality monolayer graphene due to the low
carbon solubility and its catalytic nature. With the help of
plasma, hydrocarbon precursors can break apart more easily
so that carbon atoms and radicals can directly assemble into
graphene on the Cu surface. For example, Kim
et al.
have
deposited large-area graphene-like films on Al and Cu foils
using a microwave assisted surface wave plasma CVD (MW-
SWP-PECVD) method at a substrate temperature ~ 400 °C
[24]. However, graphene films grown by MW-SWP-PECVD
required high power (3 to 4.5 kW), and their Raman
spectroscopy exhibited large
D
and
D
peaks, which
indicated lots of defects and boundaries. Similarly, Terasawa
et al.
investigated the growth mechanism of graphene in
varying the growth conditions by a RF-PECVD [25]. When
the substrate temperature was kept at 500 °C, carbon nano-
walls (CNWs) were found on the Cu surface. On the other
hand, as the substrate temperature was raised to 900 °C, SLG
was fabricated with a small
D
peak. It was found that the
growth of SLG was activated by the Cu catalytic surface at
high substrate temperature, whereas the growth of CNWs
was initiated by the hydrocarbon radicals in the plasma.
In the following years, the growth of graphene based on
PECVD methods faced two major challenges. One was to
fabricate continuous and large areas of graphene films, and
the other was to improve the quality of the PECVD-grown
graphene. Yamada
et al
. combined a PECVD process at a
low substrate temperature of ~ 380 °C and a roll-to-roll
process for mass production of graphene [26]. Although the
resulting graphene structures were defective, Raman spectra
along the width direction of the Cu foil was found to be
uniform. This result suggested that roll-to-roll growth by the
PECVD method appeared promising for realizing continuous
and large area graphene films in industrial production.
Meanwhile, many research groups have experimented
different PECVD systems and growth conditions to improve
the quality of graphene grown on Cu substrates. Kim
et al
.
demonstrated synthesis of SLG on polycrystalline Cu foils
under various Ar/CH
4
and H
2
/CH
4
gas ratios for substrate
temperatures ranging from 700 °C to 830 °C [27]. The grain
size of graphene was found to range from ~ 0.4 to ~ 3 μm
and the shape was arbitrary or rounded hexagonal. In
particular, they found out even without any H
2
flow, methane
alone could still provide enough hydrogen species for single-
layer graphene synthesis on Cu by PECVD, which implied
that methane could be the source for both hydrogen and
carbon. Similar graphene films were synthesized using
remote MW-PECVD at substrate temperature of 600 °C with
Page 3 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
4
various ratios of methane and hydrogen mixture [28]. The
remote plasma setup used in the work could reduce defects
incurred from ion bombardment in the plasma. Moreover, it
was found that with increasing H
2
flow rate, Raman spectra
showed higher-quality graphene films with larger 2
D
-to-
G
and smaller
D
-to-
G
intensity ratios as well as narrower
FWHMs of the 2
D
band. Nang
et al.
also demonstrated
graphene synthesis on Cu foils by means of inductively-
coupled plasma chemical vapour deposition (ICPCVD) [30].
Graphene quality was found to improve with increasing
growth time and plasma power. In 2015, Laan
et al.
used RF-
PECVD to grow SLG at substrate temperatures as low as 220
°C [31]. These graphene films could be easily removed from
Cu foils and transferred to other substrates by dipping the
sample into water. Recently, Boyd
et al.
[11] demonstrated a
single-step process to grow graphene by MW-PECVD
without any active heating of the substrates. The graphene
sheets were almost strain free and the mobility was as high as
>~ 60, 000 cm
2
/V-s at 300 K.
In Table 1 we summarize the aforementioned PECVD
growth conditions on metallic substrates and the resulting
graphene characteristics from measurements of Raman
spectroscopy, sheet resistance and electrical mobility. Here
we note that Raman spectroscopy has been a powerful
experimental tool that can reveal numerous important
characteristics of graphene samples [32]. As detailed in Refs.
[32,33], Raman spectroscopic studies of graphitic samples
can provide useful information about the number of layers,
stacking order, disorder, as well as the behaviour of electrons
and phonons in the samples. The most prominent features in
the Raman spectra of monolayer graphene include the so-
called
G
-band appearing at ~ 1582 cm
1
and the 2
D
(also
known as
G
) band at ~ 2700 cm
1
, using laser excitation at
2.41 eV. In the case of a disordered sample or at the edge of
a graphene sample, the so-called disorder-induced
D
-band at
around ~ 1350 cm
1
is also present. The
G
-band is a first-
order Raman process associated with the doubly degenerate
(in-plane transverse optical and longitudinal optical) phonon
mode (
E
2
g
symmetry) at the centre of the Brillouin zone. The
D
-band is a second-order Raman process involving one in-
plane transverse optical phonon and a defect mode due to
imperfections such as defects, edges and folds. The 2
D
-band
corresponds to a second-order Raman process that involves
two in-plane transverse phonon modes, which is a mode of
particular interest for analysing the number of graphene
layers and stacking order in multilayer graphene based on its
linewidth and peak position [32]. For instance, the 2
D
-band
of bilayer graphene with perfect Bernal A-B stacking should
be considered as the superposition of four Lorentzian peaks
that correspond to four different double resonance processes
associated with bilayer graphene, whereas that of trilayer
graphene with A-B-A stacking would involve six Lorentzian
peaks [32]. In the case of perfect 3D graphite, the 2
D
-band
consists of two Lorentzian peaks, whereas for randomly
oriented, turbostratic multilayer graphene layers, the 2
D
-
band appears as a single Lorentzian like that in monolayer
graphene but with a larger linewidth [32]. Additionally, the
biaxial strain (
ll
+
tt
) associated with monolayer graphene
can be estimated by considering the Raman frequency shifts
 
0
m
m
  
  
and the
Grȕneisen
parameter
bia
x
m
[33]:
 
bi ax
m
m
0
m
ll
tt
  
,
(1)
where m (=
G
, 2
D
) refers to the specific Raman mode, and
is the corresponding resonant frequency in the absence of
0
m
strain. Furthermore, for polycrystalline graphene samples,
the average in-plane
sp
2
crystallite size (
L
a
) of the samples
may be estimated by using the intensity ratio (
I
D
/
I
G
) [34]:
 
4
1
560
nm
D
a
G
L
E
I
L
I
 
 
 
,
(2)
where
E
L
denotes the excitation energy of the laser source.
Therefore, a variety of useful information about the quality
and structural characteristics of graphene samples can be
obtained by analyzing the peak frequencies, linewidths and
relative intensities of prominent Raman modes.
2.1.2 PECVD graphene on non-metal substrates
The rich radical environment provided by PECVD has also
been exploited to synthesize graphene on non-metallic
substrates. In particular, direct growth of graphene on
semiconducting substrates (such as Si and Ge) and dielectric
substrates (such as SiO
2
and HfO
2
) is technologically very
important for better integration of graphene into current
semiconductor industry [35, 36]. Moreover, direct growth of
graphene on desirable non-metallic substrates eliminates the
need of transferring samples from their growth substrates,
preventing potential degradation of sample quality during the
transfer process. By using precursors such as CH
4
, C
2
H
2
, H
2
,
and CO
2
, direct growth of graphene has been demonstrated
on semiconducting substrates of Ge [37-40] and Si [41-44],
and on dielectric substrates of GaN [45, 46], SiC [44, 47],
SiO
2
[48-51], hexagonal boron nitride (
h
-BN) [52], Al
2
O
3
,
sapphire, quartz, mica, and even 4-inch glass wafers [53, 54].
On Ge, growth of high-quality graphene with large and
uniform areas (up to centimeters) had been achieved [39]. On
the other hand, for graphene growth on Si or SiO
2
substrates,
the inert silicon surface makes the nucleation and deposition
of graphene very difficult to happen. To date, it is still
challenging to directly synthesize large area, flat graphene
sheets on silicon without using any metal catalysts. Given the
reactive radicals inside the plasma cavity, PECVD growth
could be a good solution that provides a better environment
Page 4 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
5
for anchoring carbon and also higher energy for reactions to
take place as compared with thermal CVD systems.
However, it was found that metal-assisted growth was often
needed in the PECVD growth, because without any transition
metal catalysts, graphene tended to stop growth quickly after
initial nucleation and the resulting sample often contained
many structural defects, forming networked nano-graphite
[41], island type nano-graphene [42], or graphene nano-walls
[43]. For instance, Kim
et al.
used a remote Cu catalyst
method by putting a Cu foil on top of a SiO
2
substrate [51].
On the other hand, continuous SLG sheets could be grown
between a thin Ni film and a SiO
2
interface. The carrier
mobility ranged from 43 to 580 cm
2
/V-s, which was almost
on the same order as that of the graphene directly fabricated
on the SiO
2
substrate by thermal CVD.
In contrast to catalysing graphene growth by transition
metals, Wei
et al.
found a critical equilibrium state between
H
2
plasma etching and CH
4
or C
2
H
4
plasma CVD [55], which
led to the growth of micrometer-scale graphene directly on
sapphire, HOPG, and SiO
2
substrates at temperatures as low
as 400 °C. However, the growth process described by Wei
et
al.
[55] required preparation of seeds on the substrates first,
which was followed by multiple synthesis steps, suggesting
the generic difficulties in direct growth of graphene on non-
metallic substrates. Similarly, Kato
et al.
developed a method
for growing graphene directly on a SiO
2
substrate by rapid-
heating plasma CVD so as to provide sufficient energy [56].
Hexagonal boron nitride (
h
-BN) is a wide bandgap
insulator known to be an ideal substrate for graphene because
of its atomically flat surface, small lattice mismatch to
graphene and the ability to tune graphene electronic structure
[57-60]. Therefore, it is highly desirable to experiment direct
PECVD growth of graphene on
h
-BN. In 2013, Yang
et al.
reported epitaxial growth of single-domain graphene on
h
-
BN by a plasma-assisted deposition method [52], and found
that the growth of continuous single-crystalline SLG or
bilayer graphene films was only limited by the size of the
h
-
BN substrates.
Gallium nitride (GaN) is a binary III/V direct bandgap
semiconductor commonly used for applications in various
optoelectronic, high-power and high-frequency devices. Sun
et al.
first attempted direct growth of graphene on the surface
of the GaN (0001)/sapphire substrate by T-CVD at 950°C
under a high flow of ammonia without metallic catalysts
[45]. They found that the synthesized carbon thin films were
largely
sp
2
bonded, macroscopically uniform, NS electrically
conducting, with optical transparencies comparable to that of
exfoliated graphene [45]. Subsequently, Kim
et al.
demonstrated plasma-assisted direct synthesis of graphene
films on GaN substrates [46], and found that the active layer
in the light emitting diode (LED) was not degraded by the
growth temperature of graphene so that the deposited
graphene could act as a transparent conducting electrode
without the need of any transfer process.
While progress has been made in the PECVD growth of
graphene on non-metallic substrates, it is clear that much
more effort is still needed to establish the growth conditions
for synthesizing high-quality large-area graphene on a wide
variety of dielectric and semiconducting surfaces.
Page 5 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
6
Figure 2.1 (a) A schematic of the PECVD setup for synthesis of graphene sheets and VG-GNs without active heating.
Adapted from Ref. [11]. (b) Schematic illustration of the single-step PECVD growth mechanism of graphene on
copper. (c-e) False-color SEM images of graphene grown for excessive time and transferred to single crystalline
sapphire, with increasing magnification from left to right, showing well-aligned, hexagonal adlayer graphene domains
(dark) on the bottom monolayer graphene (light), which illustrate how the hexagonal grains nucleated along parallel
lines of the copper foil coalesce into a single sheet of graphene.
Page 6 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
7
2.2. Single-step deposition of high-mobility graphene
at reduced temperatures
We have recently developed a new growth method based
on MW-PECVD under no active heating to the substrates
[11, 61]. This new method has been shown to reproducibly
achieve, in one step, high-mobility large-sheet graphene
samples that are nearly strain free [11].
The PECVD system is schematically illustrated in Figure
2.1(a). It consists an Evenson cavity and a power supply
(MPG-4, Opthos Instruments Inc.), which provides an
exciting MW frequency of 2.45 MHz to generate plasma. A
residual gas analyzer (RGA) is used to monitor the precursor
and by-products partial pressure. The gas delivery system
consists of mass flow controllers (MFCs) for H
2
, CH
4
and
Ar. The CH
4
gas flow is controlled by a leak valve placed
before the methane MFC. Other than these typical gases to
grow large-area graphene sheets, a quartz container stored
with some substituted aromatics may be attached to the
growth chamber via a leak valve and a quarter-turn, shut-off
valve. With the addition of substituted aromatics such as 1,2-
dicholorobenzene
(1,2-DCB),
1,2-dibromobenzene(1,2-
DBB), 1,8-dibromonaphthalene (1,8-DBN) and toluene as
the seeding molecules, we can choose to grow graphene
nanostripes (GNSPs) [61] and carbon nanotubes (CNTs)
vertically, which will be elaborated later in Sections 3 and 4.
Table 1.
Comparison of different synthesis conditions for PECVD-growth of monolayer graphene on metallic
substrates and the resulting characteristics of the graphene samples. Here RT refers to room temperature.
Page 7 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
8
In this new approach, cyano radicals play an important
role in a hydrogen-methane plasma to remove Cu native
oxide without active heating. After Cu is smoothly etched,
graphene growth is found to nucleate from arrays of well-
aligned hexagonal domains and eventually coalesced into a
large sheet of ~ 1 cm
2
, as schematically shown in Figure
2.1(b) and further exemplified in Figure 2.1(c)-(e) [11].
Detailed Raman spectroscopic studies of these SLG graphene
sheets generally revealed excellent spectral quality, as
exemplified by a typical point spectrum in Figure 2.2(a), and
maps as well as the corresponding histograms of the 2
D
-to-
G
and
D
-to-
G
intensity ratios over a (100 × 100) μm
2
area of a
graphene sample on Cu foil in Figures 2.2(b) and 2.2(c),
respectively. Clearly the samples are mostly SLG with few
defects, as further manifested in Figures 2.2(d)-(e) by the
histograms of the FWHM of the 2
D
mode, and the intensity
ratios of the 2
D
-to-
G
and
D
-to-
G
modes, (
I
2D
/
I
G
) and (
I
D
/
I
G
),
respectively.
Moreover, estimates of the magnitude of strain from both
Raman spectroscopy and scanning tunnelling microscopy
(STM) [11] revealed that the spatial strain distribution in the
PECVD-grown graphene, as exemplified in Figure 2.3(a) for
the biaxial strain determined from the Raman spectroscopic
studies using Equation (1) and in Figures 2.3(b) and 2.3(c)
for strain derived from atomically resolved STM topographic
studies [11], is nearly strain-free with an average of <~
0.07% strain. The average strain in PECVD-grown graphene
was consistently more than one order of magnitude smaller
than that of the SLG grown by the T-CVD method, as
exemplified by the strain maps shown in Figures 2.3(b) and
2.3(c). The finding of much reduced strain in our PECVD-
grown graphene samples is also consistent with their much
better electrical mobility, typically 30,000
70,000 cm
2
/V-s
at 300 K, which is comparable to the best values (40,000
60,000 cm
2
/V-s) reported in multi-step, thermal CVD-grown
single crystalline graphene at 1.7 K [12].
In addition to the high electrical mobility, we note that
large-area strain-free graphene may be applied to
strain
engineering
of novel nano-electronics by transferring strain-
free monolayer graphene to substrates with pre-designed
nanostructures to induce controlled spatial distributions of
strain [62]. This approach is achievable because the local
electronic properties of graphene are known to be highly
susceptible to nanoscale lattice distortions [62 - 65]. Thus,
proper design of the strain induced on graphene by its
underlying, nanoscale architected substrates can result in
desirable modifications to the local electronic properties of
graphene [62]. Such nanoscale strain engineering of
graphene for novel electronics [62] is only feasible with the
availability of sizable and nearly strain-free graphene
Figure 2.2 Raman spectroscopic characterizations of
monolayer graphene samples grown by the single-step
PECVD process [11]: (a) A representative point Raman
spectrum of the PECVD-grown graphene taken at laser
wavelength at 532 nm. (b) A spatial map for the intensity
ratio of the 2D-to-G modes over an area of (100
100)
m
2
with the spectra were taken at 2 mm per pixel steps.
(c) A spatial map for the intensity ratio of the D-to-G
modes over the same (100
100)
m
2
area as shown in
(b). (d) Histogram of the FWHM of the 2D mode taken
over the same (100
100)
m
2
area as shown in (b) and
(c). (e) Histogram of the intensity ratios of 2D-to-G
modes, (
I
2D
/
I
G
), as shown in (b). (f) Histogram of the
intensity ratios of D-to-G modes, (
I
D
/
I
G
), as shown in (c).
Figure 2.3 Comparison of the strain in as-grown
monolayer graphene samples on Cu by PECVD and T-
CVD methods [11]: (a) Strain map (upper panel) and the
corresponding histogram (lower panel) of PECVD-grown
graphene over a (100
100)
m
2
area as derived from
Raman spectroscopic studies. (b) Strain map (left panel)
and the corresponding histogram (right panel) of PECVD-
grown graphene over a (10
10) nm
2
area as derived
from STM topographic studies. (c) Strain map (left panel)
and the corresponding histogram (right panel) of T-CVD
grown graphene over a (10
10) nm
2
area as derived
from STM topographic studies.
Page 8 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
9
synthesized by the PECVD method described in Reference
[11].
2.3. Selective growth of PECVD graphene on Cu
nanostructures
The impermeability [66] and oxidation resistance [67] of
graphene makes it an ideal candidate as a protection barrier
material for Cu interconnects to prevent oxidation and
diffusion of Cu into the underlying low-k dielectrics. Kang
et
al.
[68] synthesized graphene on Cu conducting lines of 2
m width by the T-CVD method, and found that the
resistance was reduced by 2%–7% and the breakdown
current density was increased by 18% compared to pure Cu
wires. Mehta
et al.
[69, 70] investigated low temperature
(~650 °C) deposition of graphene around Cu nanowires, and
found that the graphene deposition enhanced both the
electrical and thermal conductivity. They also demonstrated
successful blockage of Cu ion diffusion by large area multi-
layer graphene membranes deposited directly on silicon
oxide using PECVD. However, these experimental studies all
involved high temperature synthesis of graphene (> 550 °C),
which was incompatible with typical CMOS processing
temperatures (< 450 °C). Further, none of the
aforementioned studies were carried out on realistic
industrial wafers with high-density nanostructures on delicate
low-k dielectrics.
We have recently demonstrated the feasibility of growing
graphene selectively on nanoscale Cu-interconnects in
realistic industrial semiconductor wafers by a single-step
PECVD technique and have further optimized the processes
to minimize plasma-induced damages to the underlying low-
k dielectrics. Our substrates were fabricated by an industrial
semiconducting company. As shown in Figure 2.4(a), SEM
images with increasing magnification show micro-sized Cu
pads (middle image) and Cu nanostructures with different
widths (right image). The bright areas in the SEM images
represented Cu coated device and interconnect structures,
and the dark areas represented a low k dielectric material.
Using proper growth conditions, we could reproducibly
deposit high-quality and continuous graphene layers onto the
Cu nanostructures in the industrial wafers. Moreover, the low
plasma power density (10 W for ~ 1 cm
3
volume) employed
in the PECVD growth corresponded a gas temperature < ~
100
C, which was fully compatible with the current
fabrication processes in semiconducting industry.
However, the PECVD process was found to induce some
damages to the low-k dielectrics, depending on the growth
parameters and schemes of experimental setups. The
damages induced by plasma to the low-k dielectrics were
mainly due to ion bombardment and UV light, as
schematically shown in the left panel of Figure 2.4(b). In
Figures 2.4(c) and 2.4(d) we show the AFM images of the
region indicated by the blue box in Figure 2.4(a) after
graphene growth. In Figure 2.4 (c) for graphene grown under
direct exposure to plasma, many clusters appeared on the
surface due to the low-k material etched by plasma and
redeposited on the Cu surface and the graphene structures
grown on Cu appeared inhomogeneous. To mitigate the
plasma induced damages to the low-k dielectrics, we
introduced a holed HOPG plate as a UV absorber, as
Figure 2.4 (a) SEM images of industrial wafers with
increasing magnification (by
10 from the left to the
middle and by
20 from the middle to the right). Right
image shows Cu nanostructures with different widths, and
the smallest features are <~ 20 nm. (b) Schematic
drawings of the side view for the direct PECVD growth
configuration (left) and that for the PECVD growth
configuration with a holed HOPG plate as the UV
absorber (right). (c) & (d) AFM images corresponding to
the region indicated by the blue box in (a), after direct
PECVD growth (left) and growth with HOPG absorber
(right), respectively. (e) & (f) Raman spectra of (c) and
(d), respectively.
Page 9 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
10
schematically shown in the right panel in the Figure 2.4 (b).
Using this configuration, we were able to block most of the
UV light and minimize damages to the low-k dielectrics.
AFM images taken after graphene growth with the HOPG
UV absorber as shown in Figure 2.4 (d) revealed significant
improvement on the surface and better Raman spectra
(Figures 2.4 (e) and 2.4 (f)).
3. Vertically grown graphene nanostructures (VG-
GNs)
In addition to the horizontal growth of graphene sheets and
nanostructures on substrates, there has been much interest
and development in the vertical growth of graphene
nanostructures. For instance, vertical graphene nanowalls,
also known as carbon nanowalls (CNWs) or vertically
oriented graphene, are multilayer graphene structures
oriented in such a way that the graphene honeycomb lattice is
perpendicular to the growth substrate, forming wall-like
structures. These
vertically grown graphene nanostructures
(VG-GNs) generally consist of varying shapes besides the
wall-like or sheet-like nanostructures, including graphene
nanostripes (GNSPs) with large length-to-height aspect ratios
[61] as exemplified in Figure 3.1(a), and graphene nano-
flowers [10, 61] that exhibit extensive branching behavior of
quasi-one-dimensional graphene nanostructures. VG-GNs
typically grow uniformly over a substrate in random
directions, forming a mesh-like graphene mat, as exemplified
in Figure 3.1(b) for GNSPs. VG-GNs can grow up to tens of
microns in height with interlayer spacings ranging 0.34 ~
0.37 nm [10], and can have either perfect AB stacking [71]
or turbostratic stacking [61]. Additionally, PECVD synthesis
of VG-GNs has been accomplished on a variety of substrates
such as Cu [61], Ni [72], Al [73], Si [74], graphite [75] and
polymers [76]. Although VG-GNs can also be synthesized
via thermal CVD [77], in-liquid plasma [78] and
electrophoretic deposition [79], PECVD has been the most
common method.
VG-GNs are an appealing material for four primary
reasons. First, as graphene-based nanostructures, VG-GNs
retain various favorable properties of graphene, such as the
high electron mobility [61]. Second, the mesh-like network
of VG-GNs creates a porous three dimensional architecture
of graphene nano-sheets, yielding extremely high surface
areas [80]. Third, VG-GN films have abundant exposed
edges, which facilitate active electrochemical properties [81]
and interesting chemistries [82, 83]. Finally, VG-GNs
represent a potential route for large scale production of
chemically pure and high quality graphene materials [61].
The favorable properties of VG-GNs have found ways
into a variety of applications. Among them, energy-related
applications are most common, primarily in supercapacitors
[73, 84–96], but also in batteries [97–107], solar energy
conversion [108–115] and fuel cells [116–120]. Field
emission [121–126] and molecule detection [127–136] are
also notable applications. Other applications include memory
devices [79], antibacterial coatings [137], semiconductors
[138], catalysis [139], magnetic sensors [140], tissue
engineering [141, 142], and heat dissipating [143] and
hydrophobic coatings [144]. The extensive study of VG-GNs
as an energy material and plethora of other proposed
applications suggest that VG-GNs may soon become
industrially relevant.
In this section we highlight recent progress made in the
PECVD fabrication of VG-GNs, including strides made in
controlling VG-GN material characteristics, novel synthetic
procedures, VG-GN heteroatomic doping, nanoparticle
inclusion, and VG-GN thin film coatings. We will also
discuss recent progress in the characterization of VG-GNs.
Finally, future prospects to advance the field of VG-GNs will
be discussed. For reviews describing preliminary aspects of
VG-GNs, such as the growth mechanism and the basic
synthesis procedures, the readers may refer to References
[10, 145]. Additionally, a review describing the use of VG-
GNs as field emitters can be found in Reference [146].
3.1. Fabrication progress
Typical PECVD growth of VG-GNs takes place in medium
vacuum (10 mTorr ~ 10 Torr), requires a carbon source (
e.g.
,
methane), an etchant (
e.g.
, hydrogen) and a radiation source
(
e.g.
, microwave) to induce the plasma. Most VG-GNs
growth also requires high substrate temperatures (
e.g.
, 800 ~
1300 K). The radiation induced plasma consists of carbon,
etchant radicals and ions. The seeding of VG-GN growth
may take place at graphitic defects that form during early
growth [10] or directly at the plasma-induced radical sites of
the substrate [61]. After a VG-GN is seeded, carbon radicals
may continue to deposit and contribute to the growth of VG-
GNs, while etchants remove defects and amorphous carbon.
More in-depth discussion of typical growth procedures and
mechanisms and be found in the review by Bo
et al
[10].
3.1.1 VG-GN material
characteristics.
Although VG-GNs are graphene-based nanomaterials, even
the highest quality VG-GNs do not demonstrate the ideal
properties of two-dimensional graphene [61]. Therefore,
research efforts that correlate the growth parameters with the
quality of VG-GNs are important to advance this field.
Additionally, modifications (
e.g.
, defect incorporation,
chemical doping,
etc.
) to VG-GNs can result in certain
desirable properties through functionalization at the expense
of other properties (
e.g.
, electron mobility). Significant
studies in these respects will be discussed in this section.
Page 10 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
11
Recently, we have demonstrated VG-GN materials with
large aspect ratios (>~ 100) by using aromatic molecules in
the PECVD growth as the seeding molecules [61]. These
VG-GNs are dubbed as
graphene nanostripes
(GNSPs) to
emphasize their large aspect ratios and also to differentiate
them from
graphene nanoribbons
(GNRs) that typically
referred to quasi-one-dimensional graphene nanostructures
that exhibit quantum confinement effects [61]. As
exemplified in Figure 3.1(c) for an GNSP that was
transferred and isolated on a silicon substrate, the GNSP
exhibited dimensions of <~ 1
μ
m in width and 125
μ
m in
length, which corresponded to a large aspect ratio of ~ 125:1.
These GNSPs with large aspect ratios are also high quality
materials that can be produced with high yield and at room
temperature. We further found that although the as-grown
GNSPs formed highly interconnected networks and the
distance between joint points was typically short (
e.g.
, 1
μ
m,
see Figure 3.1(b)), long GNSPs could be isolated from
samples because GNSPs network joint points did not
represent the end of a GNSP, but rather represented splitting
points where the multiple GNSP layers split in different
directions [61]. This notion was confirmed by studying the
SEM images of all transferred samples after sonication in
solvent, which always revealed combinations of long and
straight GNSPs together with interconnected webs of short
graphene pieces, as exemplified in Figure 3.1(c). Moreover,
the synthesis involved substituted benzene carbon sources
that led to C
6
radicals and C
6
H
6
, as shown by the
representative RGA data in Figure 3.2(b). The presence of C
6
radicals and C
6
H
6
molecules in addition to methane played
Figure 3.1: Characterization of DCB-catalysed PECVD-grown GNSPs [61]: (a) Tilted (45°) SEM image of GNSPs. (b)
Top-down SEM image of as-grown GNSPs. (c) – (f) SEM images of exfoliated GNSPs, demonstrating the variety of
graphene nanostructures that can be obtained by exfoliating as-grown GNSPs. The exfoliated nanostructures shown here
range in dimensions from 1
μ
m
125
μ
m (panel c) to 30 nm
1100 nm (panel f). In addition to completely separated
individual GNSPs and some interconnected, web-like graphene pieces (panel c), exfoliation also produced two (panel d)
or more (panels e and f) GNSPs fused together. (g) A representative Raman spectrum of GNSPs, showing characteristic
D-
,
G
-,
D
-
and 2
D
- peaks. (h) A high resolution TEM image of GNSPs demonstrating the atomic structure. (i) A
representative SAED pattern of GNSPs, which reveals two sets of incommensurate six-point patterns (highlighted by the
colored lines) overlapping each other, suggesting that this sample consists of multiple graphene layers that are rotated
with respect to each other.
Page 11 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
12
an important role in the rapid room-temperature growth and
the development of GNSPs with large aspect ratios, which
was in contrast to typical PECVD growth of VG-GNs that
primarily involved C
2
radicals from such carbon sources as
hydrocarbon (C
2
H
2
or CH
4
) and fluorocarbon (CF
4
, CHF
3
or
C
2
F
6
). In the latter case the resulting nanostructures were
mostly “wall-like” with typical length-to-width aspect ratios
<~ 10, whereas the growth rate may be increased with
increasing plasma power [10]. Additionally, we found that
DCB precursors further enhanced the presence of C
2
radicals
(Figure 3.2(c)), which also contributed to the high-yield
growth of GNSPs [61].
Hydrophobicity/hydrophilicity is generally an important
characteristic of nanomaterials. In the case of graphene,
while it is intrinsically hydrophilic, airborne contaminants
readily render graphene hydrophobic [147, 148]. Given that
many electrochemical applications require hydrophilic
electrode materials to be compatible with aqueous solvents
and the strong interest in using VG-GNs as electrode
materials, several investigations have focused on achieving
hydrophilic
VG-GNs
for
various
electrochemical
applications. In these studies, hydrophilicity (also known as
wettability) is quantitively determined by placing a water
droplet on a VG-GN surface and then measuring its contact
angle with the surface; a decreased contact angle signifies
increased hydrophilicity. In the following we discuss two
recent studies regarding hydrophilicity of VG-GNs.
Zhang
et al
[74] performed post fabrication argon
sputtering of VG-GNs to increase the surface defect content
as determined by Raman and x-ray photoelectron
spectroscopies. These argon-sputtered VG-GNs with higher
defect concentrations were found to be more hydrophilic than
as grown VG-GNs, suggesting that the hydrophilicity of VG-
GNs can be manipulated via defect engineering.
Subsequently, Shaui
et al
[72] studied the dependence of
VG-GN hydrophilicity on the VG-GN areal density. Here the
VG-GN areal density is determined by the average distance
between individual vertical graphene structures. Shaui
et al
produced VG-GNs with interstructure distances ranging from
15 nm to 306 nm on metallic substrates and demonstrated
that hydrophilicity increased with increasing VG-GN areal
density. They further showed that the Raman spectra of all
samples were nearly identical, suggesting that the defect
contents were similar in all samples so that the increased
hydrophilicity may be primarily attributed to the increasing
VG-GN areal density. Shaui
et al
also conducted
supercapacitor measurements and showed that supercapacitor
properties in terms of the specific capacitance and specific
Figure 3.2: PECVD growth mechanism of GNSPs catalysed by DCB [61]: (a) Schematic illustration of a proposed model
for DCB-catalysed growth of GNSPs on copper. (b) RGA data of different molecules as a function of time, where the
shaded area represents the time interval when plasma is turned on. (c) OES data during the growth of GNSPs for different
ratios of DCB to methane.
C
2
C
6
C
6
H
6
Page 12 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
13
energy improved with increasing VG-GN areal density. On
the other hand, it was not clear whether the improved
properties were due to increased hydrophilicity or increased
edge content or both, as the graphene edge states generally
have a higher capacitance than the graphene basal plane [81].
In addition to providing the necessary radicals for VG-
GNs growth, plasma in the PECVD synthesis contains high
energy electrons and ions [149] that can contribute to defects
in VG-GNs. The kinetic energy of such charged species is
greatest near the plasma source. Cuxart
et al
[75]
demonstrated a remote growth of VG-GNs where the growth
substrate was not placed directly in the plasma generating
medium. Although the VG-GNs thus synthesized were
compromised in quality as determined by their Raman
spectroscopic characteristics, the demonstration of remote
plasma growth was significant because it suggested the
feasibility of PECVD growth of VG-GNs on delicate
substrates that may be susceptible to high-energy ion
bombardment and/or UV-induced damages.
3.1.2 Precursors for the PECVD
synthesis.
An important consideration in optimizing the PECVD
synthesis of VG-GNs is to achieve reproducible mass
production at both reduced costs and minimum
environmental impact. A feasible approach is to explore
alternative carbon precursors and/or innovative substrates,
which may be better suited for particular applications. We
review below recent progress in the exploration of different
carbon precursors and substrates.
As mentioned previously, we have recently achieved a
high yield and single-step deposition of high quality GNSPs
by including substituted benzenes in the PECVD growth
[61]. Specifically, trace amounts of 1,2-dichlorobenzene
were included in a hydrogen/methane plasma, and, despite its
low content, the 1,2-dichlorobenzene had significant effects
on the growth. Additionally, other substituted aromatics such
as 1,2-dibromobenzene, 1,8-dibromonapthalene and toluene
were used as precursor molecules for the growth of GNSPs,
although 1,2-dichlorobenzene was found to be the most
effective [61]. Similarly, Lehmann
et al
[150] reported using
paraxylene (a substituted aromatic) in the synthesis VG-GNs,
although in their work paraxylene was the sole carbon source
and the synthesis was carried under argon gas flow.
A favorable choice of precursors should not only take into
consideration of the waste items and natural products but
also the cost and accessibility. In this context, evaporated
essential oil had been used as a carbon precursor and
combined with a hydrogen/argon plasma to produce VG-
GNs [151]. Similarly, butter [152], lard oil [153] and
sugarcane [154] had been placed on metal substrates and
exposed to hydrogen/argon plasmas to produce VG-GNs.
Finally, Zhou
et al
. [155] used graphene oxide as a precursor
to synthesize VG-GNs by first converting the graphene oxide
into a graphene aerogel, then exposing the graphene aerogel
to a methane/hydrogen plasma at 1050 K. Although this
approach involved several steps and produced VG-GNs with
excess oxygen impurities, its novelty is noteworthy.
In addition to novel precursors, different growth substrates
have also been demonstrated in recent years. Most notably,
VG-GNs had been grown on silicon nanowire meshes with
nanowire diameters as small as 37 nm [156]. The resulting
composite of VG-GNs on silicon nano-mesh had extremely
high surface areas. Besides silicon nanowires, synthesis of
VG-GNs on a polymer [76] and aluminum foils [73] has also
been demonstrated recently. These results suggest the
versatility of PECVD growth so that a variety of precursors
and substrates can be tailored with proper growth conditions
to specific applications.
3.1.3 Heteroatom
doping.
Despite many desirable properties (
e.g.
, high electron
mobility), the absence of a bandgap and the relative inert
chemical properties of graphene limit its applications to
photovoltaics and catalysts, including in fuel cells and
electrochemical energy storage. Heteroatom doping,
i.e.
,
substitution of non-carbon atoms into the graphene lattice,
has been employed to overcome this limitation, and the
selection and combination of specific heteroatom dopants
determines the performance of the resulting material in
applications [157]. Among the heteroatoms that have been
incorporated into horizontally grown graphene sheets include
boron, nitrogen, sulfur, iodine, bromine, phosphorus,
selenium, chlorine, and fluorine [158]. In contrast, fewer
heteroatoms have been attempted for doping into VG-GNs,
which include nitrogen [93,126,159], boron [121,160,161],
and fluorine [162–164]. Interestingly, co-doping of any two
heteroatoms to VG-GNs has not been reported. In the
following we review the successful doping procedures that
have been investigated to date for VG-GNs.
Nitrogen doping of VG-GNs has been achieved during the
PECVD growth and via post-growth modification.
Specifically, Soin
et al
achieved nitrogen doping by applying
a N
2
plasma to VG-GNs grown by a typical procedure [159].
Similarly, Wang
et al
. achieved nitrogen doping by adding
ammonia gas during their regular PECVD growth procedure
[93]. Interestingly, it was found that simply including N
2
gas
flow during the PECVD growth did not result in nitrogen
doping [159, 165]. Alternatively, Zhao
et al
. applied an
ammonia plasma to readily synthesized VG-GNs and were
able to achieve nitrogen doping to VG-GNs [126]. However,
such a post fabrication plasma treatment tends to result in
inhomogeneous nitrogen dopings, with primary nitrogen
concentrations accumulated near the surface of VG-GNs.
Boron doping of VG-GNs has only been achieved during
the growth process by including B
2
H
6
gas during the PECVD
synthesis of VG-GNs [121, 160, 161]. Although the
achievement of boron doping is notable, B
2
H
6
is dangerous
and expensive. Therefore, further exploration of other safer
Page 13 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
14
and more practical means of boron doping would be
desirable.
To date surface fluorination of VG-GNs has been
achieved in several ways, all of which via post fabrication
modification. Satulu
et al
. fluorinated VG-GNs via sputtering
of polytetrafluoroethylene polymer and also by exposing
VG-GNs to C
2
H
2
F
4
or SF
6
plasmas [162]. They also reported
superhydrophobicity in the resulting material. Davami
et al
.
achieved fluorination by simple exposure of VG-GNs to
XeF
2
[163]. On the other hand, Lin
et al
. achieved
fluorination of VG-GNs by cycling gallium ion
bombardment and XeF
2
exposure [164]. These studies all
reported a significant increase in the mechanical stiffness of
fluorinated VG-GNs.
3.1.4 Nanoparticle
anchoring.
Development of heterostructures of VG-GNs and various
nanoparticles has been explored in an effort to exploit the
attractive properties of each component. For instance,
electrochemically or catalytically active nanoparticles
anchored onto VG-GNs could enable higher surface areas
and benefit from the conductive VG-GN network. To date,
nanoparticle inclusion into VG-GNs has only been achieved
by post fabrication, and the methods used include sputtering
[166], atomic layer deposition [102], solvothermal synthesis
[167, 168], spin coating [169] and electrodeposition [98,
130].
Anchoring
lithium
titanate
[102],
hydrogen
molybdenum bronze [98] and molybdenum disulfide [156,
157] nanoparticles to VG-GNs have demonstrated enhanced
performance of the resulting heterostructures in lithium and
sodium ion batteries. On the other hand, tin oxide anchoring
to VG-GNs has shown improved formaldehyde detection
[130] and photoelectrochemical performance [169]. In
addition, silver, aluminum, cobalt, molybdenum, nickel,
tantalum, and silicon nanoparticles have been explored for
applications to surface-enhanced Raman spectroscopy [166].
3.1.5 Thin film
coatings.
Thin film coatings on VG-GNs have also been reported.
Davami
et al
[170] investigated the mechanical properties of
VG-GNs coated by thin-film Al
2
O
3
and demonstrated a
three-fold increase in the Young’s modulus of VG-GNs with
a 5 nm thick Al
2
O
3
coating. Park
et al
[171] sputtered
carbon, silicon, and silicon carbide onto VG-GNs and found
decreased electrical resistivity in each case. Thin film coating
on VG-GNs offers the possibility of creating arbitrary thin
films that exhibit the high surface area and significant
mechanical strength of the VG-GN networks, which may
lead to novel functional composite materials and so is worthy
of further exploration.
3.2. Characterization
Characterization of VG-GNs typically aims to demonstrate
the synthesis of high quality graphene and any properties that
may be useful for a specific application. The most definitive
characterization tools to demonstrate the growth of VG-GNs
is scanning electron microscopy (SEM) along with Raman
spectroscopy. SEM characterization is straightforward and
only involves identifying the VG-GN morphology. Detailed
analysis of a Raman spectrum can provide a variety of
information about the quality of the material, the number of
graphene layers, stacking order, defect type, and carrier type
and concentration [32, 33, 172]. Additionally, as described in
Section 2.1, the characteristic peaks (
G
, 2
D
,
D
and
D
) of the
Raman spectrum of graphene are easily identifiable and so
are convenient experimental signatures for confirming the
presence of graphitic materials. Exemplifying Raman spectra
for PECVD-grown monolayer graphene sheet and GNSPs
are shown in Figure 2.1(a) and Figure 3.1(g), respectively.
Generally the Raman spectra of pristine graphene monolayer
do not include the
D
-band, whereas the Raman spectra of
GNSPs and of all VG-GNs always include a
D
-band due to
many exposed edges in the nanostructures [61,173]. We
further note that the ratio of the 2
D-
band to
G
-band, (
I
2
D
/
I
G
),
is an important identifier of graphene quality. The value of
(
I
2
D
/
I
G
) typically ranges from 0.5 to 2, depending on the
number of layers, stacking order, carrier concentrations,
etc.
Besides SEM and Raman spectroscopy, the next most
common characterization methods include transmission
electron microscopy (TEM), selected area electron
diffraction (SAED), and x-ray photoelectron spectroscopy
(XPS). The extreme resolution of TEM allows for imaging
the edge structure of VG-GNs so that the number of layers
can be revealed [173] and even individual atoms of the
graphene lattice can be resolved [61]. SAED, typically
performed in the TEM instrument, has been employed to
determine the crystallinity of VG-GNs. From the SAED
studies, it is found that VG-GNs have shown either pristine
graphene stacking [71] or turbostratic (
i.e.
, misaligned)
stacking [61], although it is still unclear which growth
parameters may influence the stacking order of VG-GNs.
Examples of a TEM image and a SAED pattern of GNSPs
are shown in Figures 3.1(h) and 3.1(i), respectively. The
hexagonal graphene lattices can be seen in the TEM image,
and the SAED shows two overlapping six-point patterns,
implying that the sample has turbostratic stacking, whereas a
single six-point pattern would indicate regular stacking
[174].
XPS characterizations can provide critical information for
the chemical purity and/or heteroatom doping content of the
VG-GNs [93]. Additionally, high resolution spectra of the
carbon 1s peak can indicate the relative content of
sp
2
and
sp
3
hybridized carbon, which is an indicator of the defect
concentration and the quality of VG-GNs [74]. Ganeson
et al
[175] performed a detailed comparison of XPS and Raman
Page 14 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
15
spectroscopy measurements of VG-GNs and suggested that
the two techniques in tandem could provide information
about the type of defects present.
Other useful characterization methods include ultraviolet
photoelectron spectroscopy (UPS) for measuring the work
function of VG-GNs [176, 177], which provides useful
information about the doping level of the graphene
nanostructures; wettability studies, where the contact angle
of a water droplet on the VG-GN sample is determined [72,
74]; surface area measurements via nitrogen adsorption
Brunauer-Emmett-Teller (BET) method [178]; and electrical
transport measurements that determine the resistivity, carrier
concentration and electron mobility [61, 179, 180]. In
addition, certain application-specific characterizations, such
as the cyclic voltammetry measurements, are carried out to
evaluate the performance for energy storage applications.
While the research field of VG-GNs has largely focused
on applications and novel fabrication procedures, some
notable characterization of the novel properties of VG-GNs
have been made in recent years, which we summarize below.
3.2.1 Long term stability
measurements.
Ghosh
et al
[181] and Vizireanu
et al
[182] performed
stability measurements on VG-GNs but found contradictory
results. Ghosh
et al
exposed VG-GNs to ambient conditions
for six months and consistently performed Raman
spectroscopy, XPS, hydrophillicity and electrical transport
measurements. All of these measurements indicated stable
materials [181]. On the other hand, Vizireanu
et al
found
immediate changes in the hydrophillicity, and demonstrated
time-dependent chemical and morphological modifications
via Fourier-transform infrared spectroscopy, XPS and atomic
force microscopy [182]. Although the origin for these
incongruous findings is not clear, it may be the result of
differences
in
the
defect/edge/impurity
type
and
concentration afforded by the different synthetic procedures
used in each study. Further studies to correlate the growth
conditions and the stability of VG-GNs will be beneficial to
controlling the quality and optimizing the functionalization
of these nanomaterials .
3.2.2 Pressure dependence and thermal
conductivity.
Mishra
et al
[183] performed studies of the VG-GN pressure
dependence and thermal conductivity. Pressure dependent
shifting of the Raman modes revealed a reversible deviation
from the long-range structural order under pressures
exceeding 16 GPa. Mishra
et al
suggested that their findings
in the VG-GNs closely correlated with the pressure
dependent properties of seven-layer graphene sheets. They
further confirmed via TEM that the thickness of their VG-
GN samples was indeed consistent with seven layers of
stacked graphene. Hence, the pressure dependence of VG-
GNs behaved similarly to that of large area graphene sheets
of comparable thicknesses despite the differing geometries.
Mishra
et al
also determined the thermal conductivity of VG-
GNs via temperature dependent Raman spectroscopy [183].
3.3 Outlook for future development of VG-GNs
The most significant hindrance in the development of VG-
GN research stems from the complexity of plasma chemistry,
which is sensitive to the gas composition and flow rate,
plasma source and plasma power, precursor type, substrate
material, growth chamber dimension,
etc.
[149]. Given that
different research groups employ varying PECVD deposition
systems, it is generally difficult to exactly reproduce the
experimental conditions in different laboratories. In this
context, studies that investigate a single growth parameter
(
e.g.
, the flow rate) are not particularly effective in advancing
the development of VG-GNs for applications. On the other
hand, although exact experimental conditions may be
difficult to fully reproduce, VG-GNs of comparable
characteristics can still be developed in different systems.
Thus, the development of VG-GNs can benefit substantially
from high-standard and more thorough characterizations,
especially in application-focused studies.
In particular, more detailed Raman spectroscopic and XPS
studies could provide much better insights into the
microscopic properties of VG-GNs. For instance, most
studies simply use Raman spectroscopy to confirm the
presence of graphene by citing the characteristic
D
,
G
and
2
D
bands. On the other hand, simple peak fitting would
provider much more information, such as the
G
peak position
and the 2
D
peak symmetry that offer information about the
stacking order of graphene layers [48,61,172]; the intensity
ratio of the
D
to
G
peaks indicates the crystallite size
[61,172] according to the relation in Equation (2); the
G
and
2
D
peak positions indicate doping content [172]; and the
intensity ratio of the
D
to
D
peaks can indicate the type of
defects present,
i.e.
, edge, vacancy, graphitization,
etc.
[175].
XPS can reveal the relative content of
sp
2
and
sp
3
hybridized
carbon bonds, carbon vacancy defects, functionalization and
even the structural quality [175]. These exemplary studies
demonstrated by Ganesan
et al
[175] are highly informative
and can be standardized for meaningful comparison of the
physical properties of VG-GNs produced by different groups
under varying growth conditions.
In addition to standardizing the characterization of VG-
GNs, better methodology in the following prospects would
significantly advance the research of VG-GNs.
3.3.1 Defect characterizations.
Although Ganesan
et al
[175] have demonstrated XPS and
Raman spectroscopy characterization of VG-GNs whereby
the type and concentration of VG-GN defects was quantified,
more rigorous validation of this characterization method is
still needed. Specifically, the conclusions deduced by
Ganesan
et al
[175] depended on results associated with
Page 15 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
16
monolayer graphene, whereas VG-GNs are generally
multilayer structures. In order to validate the findings of
Ganesan
et al
, further studies that correlate XPS and Raman
spectroscopy measurements with direct microscopic
observation of the defects (
e.g.
, STM and TEM imaging)
would be particularly informative. Such correlation could
expedite the optimization of growth parameters for VG-GNs
to achieve desirable properties towards specific applications.
3.3.2 Computational
modelling.
In
addition
to
more
rigorous
and
standardized
characterization of VG-GNs, computation modelling for the
complex PECVD growth process can significantly advance
the field. For instance, although a phenomenological model
was proposed for the catalytic single-step growth process of
GNSPs [61], this model only considered the role of the
carbon source and the etchant in a very basic manner. On the
contrary, even a simple hydrogen/methane plasma involves
dozens of chemical species and radicals of different kinetic
energies [149], so that the complicated interplay of all
chemical species in the growth process of VG-GNs cannot be
well understood without computational modelling. Better
understanding of the influence of all plasma species on the
VG-GN growth process via computational modelling can
also help guide the synthesis procedure to achieve desirable
properties of VG-GNs.
3.3.3 Solution processing and transfer of
VG-GNs.
Although VG-GNs can be grown on a variety of substrates so
that pertinent substrates may be chosen for specific
applications, not all materials can withstand the PECVD
environment that involve energetic ions, electrons, radicals,
UV emission and at times relatively high temperatures.
Solution processing,
i.e.
, suspension of VG-GNs in solution
and removal of them from substrates compatible with the
PECVD process so that they can be transferred to other
substrates, would expand the potential applications of VG-
GNs. Furthermore, solution processing can enable additional
characterizations of VG-GNs by dispersing them, including
analysis of the aspect ratio and conductivity of individual
graphene nanostructures.
3.3.4 Heteroatom
doping
As discussed previously, heteroatom doping can expand the
properties of graphene by functionalization. However,
heteroatom doping has not been well explored in VG-GNs.
Further, the doping of VG-GNs that has been explored to
date often involves dangerous and expensive gases.
Expanding the palette of heteroatoms and establishing safe
and cost effective dopings of VG-GNs would enable further
application of VG-GNs.
VG-GNs research is a fast-growing field due to the myriad
of their potential applications. In particular, VG-GNs can
offer many of the attractive properties of graphene and
graphene-based, functionalized nanomaterials with the
possibility of scalable syntheses. Certain properties of VG-
GNs, such as the defect content and hydrophillicity, can also
be controlled. Additionally, nanoparticle inclusion and
heteroatom doping provide an even broader range of
possibilities. In recent years, major progress has been made
in the fabrication and modification of VG-GNs to tailor their
properties for specific applications. Advances in better
characterizations of VG-GNs have also been developed. By
holding a high standard of characterizations as well as
pursuing computational modelling and developing proper
transfer methods, we expect further progress in the
understanding and control of the VG-GN growth mechanism,
which can further enhance current applications and enable
new applications of VG-GNs.
4. Carbon nanotubes (CNTs)
Among
all
vertically
grown
graphene-based
nanomaterials, CNTs are one of the most promising
substances that have been widely used in many fields. Made
of rolled-up single or multiple graphene sheets, CNTs retain
most of the excellent properties of graphene while exhibit
additional quantum behaviour due to its one-dimensional
structure. In the case of single-wall carbon nanotubes
(SWCNTs), they can be either semiconducting or metallic,
depending on the chiral vector of the nanotube in real space
[184]. SWCNTs exhibit semiconducting property with
bandgap around 0-2 eV when they have zigzag structure.
Armchair SWCNTs are metallic with band degeneracy
between the highest
π
valance band and the lowest
π
conduction band at the K-point. The three-dimensional (3D)
structures of different helicities are illustrated in Figure 4.1.
On the other hand, multi-wall carbon nanotubes (MWCNTs)
are typically metallic [184]. The electrical conductivity of
CNTs can be up to ~ 10
5
S/cm [185], which implies high
charge transport capability. CNTs can be synthesized either
aligned or unaligned. Although CNTs with random
distribution are often sufficient for most applications, aligned
CNTs are highly desirable for vertically stacked, layer by
layer device structures.
The typical diameters of SWCNTs are around 0.8-2 nm,
whereas those of MWCNTs can be as large as hundreds of
nanometers. The lengths of CNTs range from tens of
nanometers to centimeters. The usual aspect ratio
(length/diameter) is in the range 10
2
–10
4
[186, 187]. Apart
from high electrical conductivity, CNTs also exhibits a high
thermal conductivity (~ 6000 W/m•K), which is five times
higher than copper. Moreover, CNTs exhibit high thermal
stability up to 2800 °C in vacuum. The mechanical property
of CNTs can be up to 45 billion Pascal (tensile strength),
which is 100 times stronger than stainless steel [188]. These
interesting properties make CNTs very attractive material for
many applications, including in organic solar cells [189-191],
Page 16 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
17
fuel cells [192, 193], batteries [194], supercapacitors [195,
196], water filters [197], and biosensors [198, 199].
4.1. Common synthesis techniques
CNTs are commonly synthesized by three techniques: the
arc-discharge method, laser-vaporization technique, and
chemical vapour deposition (CVD). The initial discovery of
CNTs involved the arc-discharge method, in which carbon
vapour was formed by an arc discharge between two carbon
electrodes under inert atmosphere [200]. A direct current
generated a very high temperature discharge (~ 2000–3000
°C) between the two electrodes and both SWCNTs and
MWCNTs could be synthesized. To date this method
remains the most common and easiest to grow CNTs. On the
other hand, the purity of resulting CNTs is generally poor
because metallic particles and amorphous carbons can
always be found along with CNTs, even though the CNTs
products from this method are usually well-crystallized,
In contrast, the laser-vaporization technique can easily
produce high-purity CNTs. The mechanism for the formation
of CNTs by laser-vaporization is similar to the arc discharge
method. Specifically, a high-purity graphite chunk is placed
into a high-temperature furnace with catalytic metals such as
Ni, Co, Pt. The graphite chunk is targeted and ablated by a
high-power laser; CNTs are thus formed and brought to a
copper collector [201]. Although laser-generated CNTs are
usually of high purity, their growth yield is usually very low,
making this technique inefficient for industrial applications.
In comparison with the aforementioned CNT synthesis
methods of arc discharge and laser vaporization, CVD has
been found to be a much better solution and is still the most
popular method these days. In particular, CVD synthesis of
CNTs is a practical and scalable technique suitable for mass
production. Recently, PECVD synthesis of CNTs has been
developed and is found to be even better than the thermal
CVD (T-CVD) method. In this section, we review methods
of CNT synthesis by T-CVD and PECVD, and also discuss
the latest progresses in this field.
4.1.1 Thermal CVD synthesis of
CNTs
A simple schematic diagram of a T-CVD setup is shown in
Figure 4.2. A typical T-CVD process involves a catalyst-
coated substrate, which is placed inside a quartz tube. The
entire growth system is evacuated down to around 10-300
Torr by a mechanical pump. The precursor gases are then
introduced through the tubular reactor while the tube is
heated up to a desired temperature, typically 600-1200
C.
The flow rate of hydrocarbon gases as well as the
temperature should be accurately controlled so the precursors
can be decomposed to provide reactive hydrocarbon for
CNTs formation. The presence of activated carbon molecules
or clusters in gaseous phase is critical for the growth process.
Commonly used gaseous carbon precursors include
acetylene, methane, ethylene, xylene and carbon monoxide.
These carbon sources can be gaseous, liquid, or even in solid
state form. In the case of liquid precursors, the liquid is
placed in a flask and usually an inert gas is pressured through
it to push the hydrocarbon vapour to the reaction zone.
Elevated temperatures can also be applied to increase the
evaporation rate of liquid precursors. In the case of solid
precursors, they may be directly kept in the upstream zone of
the reaction furnace. Some volatile solid materials such as
camphor and ferrocene can directly turn from solid to vapour
and then be introduced into the reactor as carbon sources.
After these hydrocarbons have been introduced into the
system, CNTs can then start to grow on the catalyst-coated
substrate in the tubular reactor. After the growth process is
completed, the furnace is slowly cooled down to room
temperature and CNTs are collected from the substrates.
Details of the catalysts and precursors will be further
discussed in the following subsection.
4.1.2 PECVD synthesis of
CNTs
In the case of PECVD synthesis of CNTs, the growth
conditions and mechanism are very different from those of
Figure 4.1 Schematic illustrations of 3D structures of
SWCNTs with different helicities and MWCNTs.
Figure 4.2 Schematic diagram of a thermal CVD setup
for CNTs synthesis.
Page 17 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript
Journal
XX
(XXXX) XXXXXX
Yeh N-C
et al
18
using thermal CVD reactors. While thermal CVD uses heat
to provide the necessary energy to decompose hydrocarbon
and to create reactive dangling bonds, PECVD processing
generates electromagnetic fields to ionize hydrocarbon
molecules and create radicals. A typical environment for
PECVD synthesis of CNTs is schematically shown in Figure
4.3. A catalyst-included substrate is placed on a sample
holder in the growth chamber. The catalyst-included
substrate can be a pure metal, such as copper or nickel, or
metal-coated materials. Various hydrocarbon gases of
different partial pressures are introduced into the growth tube
before the plasma is turned on. Additionally, carbon-based
precursors are widely used to promote CNTs formation by
providing extra carbon sources. Some of the commonly used
precursors as well as several novel precursors will be further
discussed in the following subsections. The plasma cavity is
placed in a position that covers the entire substrate surface
area. After lighting up the plasma, the resulting strong
electromagnetic field begins to ionize the hydrocarbon gases
inside the growth chamber and create reactive radicals
around the substrate. At the same time, the plasma can also
heat up the catalyst surface and provide more energy for
reactions. Certain large radicals or ions can even be
accelerated by the electromagnetic field and then bombard
the substrate surface to create nucleation sites for the growth
of CNTs. When hydrocarbon radicals reach the reactive
substrate surface, they start to form bonds and become the
bases of CNTs. If correct conditions are chosen, stable
plasma will continue to generate carbon sources so that
deposition of CNTs occurs on the substrate. The presence of
homogeneous plasma in a large region can deposit CNTs
over a large area, up to centimeters in scale. Generally
speaking, the catalyst-included substrate for CNTs deposition
is located in the region where plasma is applied, and the
growth of CNTs usually takes place following the direction
of the plasma field. At the same time, the vacuum pump of
the PECVD system removes the excess hydrocarbon
molecules and radicals, thereby preventing the formation of
clusters of amorphous carbon.
As discussed in Section 1, typical plasma sources include
direct-current (DC) discharge plasma, radio frequency (RF)
plasma, and microwave (MW) plasma, which provide
various energy densities in a local volume. Among them,
MW plasma is the most common choice for CNTs synthesis.
As compared to the thermal CVD, the use of PECVD makes
it possible to achieve deposits with higher yields of CNTs at
a relatively low temperature because the gas molecules can
be
decomposed
more
efficiently
by
the
strong
electromagnetic field.
4.2. Precursors and catalysts
The precursors and catalysts are the two primary factors that
induce the CNTs growth in both thermal CVD and PECVD
systems. We discuss some of the important precursors and
catalysts in the following subsections.
4.2.1 CNTs precursors
As mentioned before, carbon containing compounds have
been widely used as precursors to grow different kinds of
CNTs. The most commonly used precursors are methane
[202], ethanol [203], carbon monoxide [204], ethylene [205],
acetylene [206], benzene [207], and toluene [208]. These
precursors, especially acetylene and methane, are used to
synthesize CNTs in some early reports with various kinds of
catalysts [209]. In 2002, Maruyama
et al.
demonstrated the
use of alcohol as the carbon source to deposit high-purity
SWCNTs in a relatively low-temperature CVD system [210].
Owing to the etching effect of OH radicals attacking carbon
atoms with their dangling bonds, Maruyama
et al.
found that
side products such as amorphous carbon, multi-walled
carbon nanotubes, metal particles and carbon nanoparticles
are all significantly suppressed even at a relatively low
reaction temperature < 800 °C. Since then, alcohol has
become a popular precursor to grow low-cost, large-scale,
and high-purity SWCNTs in CVD systems. In 2004, Hata
et
al.
introduced an efficient CVD synthesis method for
SWCNTs where the activity and lifetime of the catalysts
were enhanced by water [211]. The catalytic activity
enhanced by water results in massive growth of very dense
and vertically aligned nanotube forests with heights up to 2.5
millimeters. Moreover, the CNT product can be easily
separated from the catalyst-included substrate, providing
nanotube material with purity above 99.98%. There have also
been a few kinds of new precursor explored. For example, in
2011, Zhao’s group used sesame seeds as a precursor to grow
CNTs [212]. By using this natural organic precursor
consisting of uniform microcells with a Fe-complex, the
authors found that the Fe-complex could release uniformly-
distributed Fe nanoparticles, which efficiently produced
CNTs arrays with lengths up to 100 μm. The aforementioned
results suggest that carbon-included precursors play crucial
Figure 4.3 Schematic illustration of a PECVD growth
environment (with carbon sources, precursors, hydrogen
plasma and various radicals) for CNTs synthesis on a
catalysed substrate.
Page 18 of 30
AUTHOR SUBMITTED MANUSCRIPT - NANO-119760.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript