See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/222124526

Chemical vapor deposition of thin graphite films of nanometer thickness

Article  in  Carbon · September 2007

DOI: 10.1016/j.carbon.2007.05.028

CITATIONS READS

354 1,309

4 authors:

Alexander Obraztsov Ekaterina Obraztsova

Lomonosov Moscow State University Russian Academy of Sciences
241 PUBLICATIONS   3,164 CITATIONS    13 PUBLICATIONS   556 CITATIONS   

SEE PROFILE SEE PROFILE

Anastasia. Tyurnina Aleksey Zolotukhin

Brunel University London Picodeon Ltd Oy, Oulu, Finland
32 PUBLICATIONS   1,229 CITATIONS    29 PUBLICATIONS   705 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

CVD graphene growth View project

EcoUltra2D View project

All content following this page was uploaded by Anastasia. Tyurnina on 08 March 2018.

The user has requested enhancement of the downloaded file.

 Carbon 45 (2007) 2017–2021
www.elsevier.com/locate/carbon

Chemical vapor deposition of thin graphite

films of nanometer thickness
a,b,*
A.N. Obraztsov , E.A. Obraztsova b, A.V. Tyurnina b, A.A. Zolotukhin b

a
University of Joensuu, Joensuu 801101, Finland
b
Moscow State University, Moscow 119992, Russia

Received 9 March 2007; accepted 31 May 2007

Available online 24 June 2007

Abstract

Well-ordered graphite films with a thickness of a few graphene layers have been grown on Ni substrates by chemical vapor deposition
(CVD) from a mixture of hydrogen and methane activated by a DC discharge. According to Auger, Raman and scanning tunneling
microscopy (STM) data the CVD graphite film thickness is about 1.5 ± 0.5 nm. The graphene layers were perfectly adhered to the sub-
strate surface except for upthrusted ridges of a few tens of nanometers in height. STM has revealed an atomically smooth surface with the
atomic arrangement typical of graphite between the ridges. A difference in the thermal expansion coefficients of nickel and graphite is
considered as a reason for the ridge formation.
 2007 Elsevier Ltd. All rights reserved.

1. Introduction Graphene can be produced by mechanical exfoliation of

Graphite is a widely used material with well-developed
synthesis methods and properties investigated in detail
[1]. A peculiarity of graphite is its layered structure formed
by parallel two-dimensional graphene sheets weakly cou-
pled by van der Waals interaction. Each graphene sheet
looks like a hexagonal network of carbon atoms connected
by strong covalent ‘‘in-plane’’ r r bonds. The delocalized
electrons appear only due to an additional p p bonding of
electronic orbitals oriented perpendicularly to the graphene
plane. They are responsible for the graphite electrical con-
ductivity which is highly anisotropic. Recently, the uncon-
ventional electric field and quantum Hall effects have been
demonstrated experimentally for a single sheet of graphite–
graphene [2,3]. These experiments have triggered a great
interest to the graphite films containing one or a very few
graphene sheets [4–7], because the unusual electron behav-
ior is considered as a result of two-dimensional confine-
ment of charge carriers.
*
Corresponding author. Address: Moscow State University, Moscow
119992, Russia. Fax: +7 495 939 2988.
E-mail address: obraz@polly.phys.msu.ru (A.N. Obraztsov).
individual layers from the surface of highly oriented pyro-
lytic graphite (HOPG) crystal [2–5]. Thin heteroepitaxial
graphite films consisting of a few graphene layers can be
produced by graphitization of silicon carbide surface [6,8]
or dissociation of ethylene gas on Ni(1 1 1) surface in
ultra-high vacuum [9]. However, such ‘‘hand-made’’ graph-
ene or heteroepitaxial thin films have relatively small lateral
sizes and are suitable only for pure scientific studies. For
deposition of extended films appropriate for practical
applications a chemical vapor deposition (CVD) from acti-
vated gaseous phase could be efficient. But up to now only
rather thick graphite films with a high number of structural
defects were deposited by CVD technique [1].
In this paper, we describe a high yield technique of
deposition of the large area graphitic films with thickness
of a few graphene layers. The results of structural charac-
terization of such films are also reported.
2. Experimental
The graphite films were grown by chemical vapor deposition technique
from a hydrogen–methane gas mixture activated by DC discharge. A
detailed description of the deposition facility and the growth method

0008-6223/$ - see front matter  2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.carbon.2007.05.028
2018 A.N. Obraztsov et al. / Carbon 45 (2007) 2017–2021
has been done elsewhere [10]. Our CVD system allows production of dif-
ferent types of carbon films ranging from polycrystalline diamond to car-
bon nanotubes. A type of the material deposited depends essentially on the
methane concentration and on the substrate temperature. In this work, the
process parameters corresponding to graphite phase deposition have been
chosen [10]: the gas mixture composition H2:CH4 = 92:8; the total gas
pressure P = 80 Torr; the substrate temperature of 950 C. In these condi-
tions the nano-graphite meso-porous layers were deposited after 1–1.5 h
[10]. The deposition procedure includes pumping out the reactor chamber
by a rotary pump to the base pressure value of 10 3 Torr followed by
introduction of a pure hydrogen and ignition of DC discharge at pressure
of about 40 Torr. Then the hydrogen pressure was increased up to 80 Torr
simultaneously with increasing the discharge current. This procedure took
about 10 min. It was followed by an introduction of the methane gas into
the reactor and an increase of its content up to 8 vol%. The discharge cur-
rent was fixed at value about 0.5 A/cm2 being typical for the graphite
phase formation [10].
Optical emission spectra (OES) measured in DC discharge plasma have
revealed an important role of carbon dimers C2 generated in the plasma
near the substrate surface. The dimers may be considered as building
blocks for the graphite deposit [10]. In present study the deposition time
has been reduced to 5–10 min to obtain the thinner films. In our CVD sys-
tem the main parameters are stabilized usually not earlier than 15 min
after the process starts. Thus, the process is not completely under control
during first minutes of deposition. For instance, the intensities of OES
peaks (including those corresponding to C2 dimers) are varied with time
in contrast to the fixed values observed in case of the standard deposition
conditions [10]. However, the deposition procedure has shown surprisingly
good repeatability of the results despite of the incontrollable character of
its initial stage.
In this work, the silicon and nickel square plates with thickness of
about 0.5 mm and lateral dimensions of 10 · 10 mm or 25 · 25 mm were
used as substrates. The single crystalline silicon substrates have been pre-
pared by cutting the standard polished Si(1 0 0) wafers. The nickel sub-
strates were made by cutting of the polycrystalline Ni sheets with
thickness of 0.5 mm produced by rolling. The substrates were treated in
pure hydrogen DC plasma to remove the surface oxide layer before the
interaction with methane started.
The CVD-deposited graphite films were studied by Raman spectros-
copy using a Fourier-transform Raman apparatus Bruker FR 1 106 with
excitation by a YAG:Nd laser operating at wavelength 1064 nm. A special
care has been taken to avoid the material heating by laser radiation in
course of measurements. For this purpose the Raman spectra were first
measured with different laser power densities. The final level of the laser
power density was chosen in the middle of the range where no temporal
changes in the line positions and shapes were observed. Raman spectra
of HOPG sample (HOPG SPI-3 GRADE supplied by SPI, West Chestr,
PA 19381, USA) were recorded at the same conditions to obtain the ref-
erence data.
The surface morphology was studied with a scanning electron micros-
copy (SEM) using LEO Supra 50 PV instrument and a scanning tunneling
microscopy (STM). An Auger electron spectroscopy (AES) was used for
elemental analysis of the surface. The AES and STM measurements were
carried out at room temperature under ultra-high vacuum (UHV) condi-
Fig. 1. The SEM images films grown in identical CVD conditions on Si (a) and Ni (b and c) substrates. The image (c) has been obtained with the sample
tilt of 75.
tions in a multi-chamber setup equipped with a cylindrical mirror analyzer
Riber OPS-200 (for AES), Sigma Scan GPI-300 [11] (for STM) and other
instruments for the surface analysis and treatments. A basic pressure in the
setup did not exceed 2 · 10 10 Torr. A preparation of W-tips for STM
measurements was made by an electrochemical etching outside and a sub-
sequent sharpening inside the vacuum chamber with Ar+-bombardment
[12]. The samples were investigated by AES and STM in ‘‘as grown’’ form
(without any UHV treatments).
3. Results and discussion
Our previous studies have revealed no significant differ-
ence in the structure of graphite-like materials deposited on
Ni and Si substrates in CVD-process lasted 45 min or
longer [10]. For the shorter deposition times (5–10 min)
substantially different carbon materials have been grown
onto substrates made from these two materials. To provide
the identical conditions of CVD process for both types of
substrates the sets of Ni and Si wafers have been located
side by side on a holder in the deposition chamber. Each
set consisted of four samples (10 · 10 mm). The films
deposited onto both types of substrates looked very
smooth and continuous, but their morphologies were dif-
ferent. The SEM images of films grown on both substrates
are shown in Fig. 1.
The film grown on Si substrate consists of numerous
tiny grains (more clearly seen under the higher resolution)
and looks more or less homogeneous over the whole sam-
ple. The film deposited on Ni has a specific morphology
with the smooth surface areas separated from each other
by out-of-plane extrusions. A typical size of the smooth
surface regions is about 1 lm. The ridges between such
regions form their own net on the film surface. The SEM
image obtained for the tilted sample (Fig. 1c) shows a for-
mation of the net of ridges at the boundaries of the smooth
micron-size areas. Some of the surface regions look like
swells formed by the film partially exfoliated from the sub-
strate. An example of such swelling is marked by arrow in
Fig. 1c. The films grown onto nickel contain also the ball-
like structures on the surface. An origin of these structures
is unclear. Taking in account an absence of any non-carbon
inclusions shown by AES we suppose that these balls are
the islands of amorphous carbon.
The films grown have been analyzed by Raman scatter-
ing spectroscopy. Fig. 2 shows the typical Raman spectra
of films grown on Si (Fig. 2(1)) and Ni (Fig. 2(2) – solid
 A.N. Obraztsov et al. / Carbon 45 (2007) 2017–2021
Fig. 2. The Raman spectra for CVD films deposited onto Si (1) and Ni (2
– solid curve) substrates and for HOPG sample (2 – dotted curve). The PL
background subtraction has been performed for spectrum 1. The
excitation wavelength was 1064 nm.
curve) substrates. For comparison the Raman spectrum of
HOPG sample has been registered (Fig. 2(2) –dotted
curve). All peaks with Raman shifts below 1000 cm 1
observed in the spectrum 1 correspond to one- and two-
phonon Raman scattering of silicon. These lines are rather
intensive because the excitation wavelength (1064 nm) pro-
vides resonance conditions for the Raman scattering in Si.
The main fundamental Raman mode of silicon at 520 cm 1
is so intensive that it is partially cut by the graph frame
(Fig. 2). Due to the resonance conditions an intensive pho-
toluminescence (PL) band of the silicon substrate (with
maximum around 1.08 eV) is also observed. The PL back-
ground has been subtracted for spectrum 1 (Fig. 2).
The other bands in the spectrum 1 correspond to the
first- and second-order Raman scattering in the carbon
film. The shapes and widths of these bands are typical of
disordered carbon. They are similar, for instance, to the
bands of amorphous carbon (a-C) films deposited by a
magnetron sputtering [13].
Fig. 3. The STM images of the surface of graphite film grown on Ni coinciding with the atomic arrangement of an ideal single crystal of graphite (a) and
with the graphene layers stacking fault producing the moiré (b).
2019
On the contrary, the Raman spectrum observed for the
film deposited onto Ni substrate (Fig. 2(2) – solid line) is
very similar to that of HOPG (Fig. 2(2) – dotted line).
The most pronounced peaks in these spectra are centered
around 1582 cm 1 and 2600 cm 1. The line at 1582 cm 1
(the so-called ‘‘G’’ – ‘‘graphite’’ band) is assigned to the
one-phonon zone-center Raman scattering in graphite
[14]. Another Raman line, usually observed for graphitic
materials in the vicinity of 1350 cm 1, is labeled as ‘‘D’’-
‘‘disorder’’-induced mode. The light scattering on phonons
with non-zero k-vectors contributes to this ‘‘D’’-signal. Its
intensity increases with the number of defects in graphite
lattice. The frequency position of ‘‘D’’-band depends on
excitation wavelength. In case of excitation at 1064 nm
the band is centered on 1300 cm 1 [15,16]. For the CVD
films grown on Ni and for the HOPG sample the intensity
of this band was found in our measurements to be just
slightly above the noise level. In the scale of Fig. 2, it is
not observable at all. While the low intensities of ‘‘D’’-
bands show a high structural perfection of both materials,
a difference in the widths of ‘‘G’’-bands may be attributed
to a relatively less ordering of the CVD film in comparison
with that of HOPG due to an existence of inhomogeneous
mechanical stress in the films.
For two-phonon Raman scattering on the same pho-
nons with non-zero k-vectors the structural defect presence
is not needed. As a result a corresponding Raman line
(labeled in the literature as ‘‘D*’’ or ‘‘2D’’ [7]) is more
intensive in the spectrum of perfect graphite than the
one-phonon ‘‘D’’-line. The intensity of ‘‘2D’’ band is
increased with excitation wavelength due to a selective real-
ization of resonance conditions for different phonons. For
HOPG this band could demonstrate a double-peak struc-
ture with the excitation in a visible spectral range [7]. But
with the excitation at 1064 nm, used in our experiments,
‘‘2D’’-band of HOPG looks like a single Lorentzian cen-
tered at 2604 cm 1 (Fig. 2(2) – dotted curve). At the same
time the CVD film grown on Ni demonstrates an asymmet-
ric ‘‘2D’’ Raman band, which exhibits two features: a
shoulder centered at 2604 cm 1 (at the same frequency as
2020 A.N. Obraztsov et al. / Carbon 45 (2007) 2017–2021

that of HOPG) and a down-shifted main peak centered at

2560 cm 1. A similar downshift and splitting of ‘‘2D’’
Raman band have been observed and considered as a
‘‘fingerprint’’ for graphite sheets with a number of layers
from 1 to 6 [7].
The Raman spectroscopy data are in well agreement
with the direct structural studies by scanning tunneling
microscopy (STM). The low-resolution STM images of
the CVD films grown on Ni substrates are very similar to
those obtained with SEM (see Fig. 1b and c). They exhibit
a presence of the smooth regions bordered by ridges with a
typical height of 10–50 nm. The STM images with the
higher spatial resolution exhibit an atomic-scale smooth-
ness of a significant part of the smooth film surface. In
some areas we have detected STM images typical for the
atomic arrangement of a single crystal of graphite (see
Fig. 3a). In other areas we have obtained various types
of the moiré images (see Fig. 3b). The latter type of STM
images (moiré) is resulted usually from a distortion in
stacking of the graphite atomic layers [17].
The STM technique has been used also to estimate the
thickness of CVD films. For this purpose a part of the film
was scraped off the Ni substrate by STM tip in course of
scanning in the contact mode. The complete removal of
the carbon material was controlled by consequent STM
imaging of the scanned area and the surface electrical con-
ductivity measurements [18]. STM profile measurement of
the border of the scrapped region has allowed estimation
of the CVD film thickness as 1.5 ± 0.5 nm.
Additional information about the thickness of CVD film
has been obtained with AES analysis. The typical graphite-
type Auger spectra (C KLL peak) have been found for C-
covered Ni substrate. No other elements (including Ni)
have been found on the surface. Taking into account the
depth of Auger analysis for C KLL (275 eV) the minimal
thickness of the continuous graphite film can be estimated
as 1 nm [19].
Thus, AES, STM and Raman data have shown that the
films deposited by CVD method on Ni substrates have a
well-ordered crystalline graphite structure. The film thick-
ness is about 1.5 ± 0.5 nm. Taking into account a polycrys-
talline metal type of Ni substrates we can propose a
mechanism of the graphite structure formation which
includes three stages (see Fig. 4). During Stage I the nickel Fig. 4. A schematic view of three stages of the deposition process of
oxides are removed from the substrate surface due to its graphite film on Ni substrate (see comments in the text).
contact with the hydrogen plasma. The treated surface of
Ni substrate kept some roughness and a polycrystalline
structure typical for metals produced by rolling. At Stage ordered graphite structure is possible neither on Si surface
II gaseous methane is introduced into the reactor chamber nor on Ni with the surface orientation different from (1 1 1)
and forms C2 dimers in the plasma. These dimers are due to a larger lattice mismatch between the graphitic film
absorbed on the substrate surface from the activated gas growing and the substrate. In the latter case only the for-
and form different carbon structures. We assume that some mation of amorphous carbon may occur. In the former
parts of polycrystalline Ni substrate have (1 1 1) Ni atomic case the heteroepitaxial islands of graphite on Ni (1 1 1)
arrangement which is suitable for a heteroepitaxial graph- are growing in lateral directions with the time and covering
ite formation due to a very good matching of the lattice the neighboring areas. Then the separate islands meet and
parameters of nickel and graphene [9,20,21]. At the same join each other for formation of a final smooth graphite
synthesis parameters a heteroepitaxial growth of the film with a little amount of structural defects.
 A.N. Obraztsov et al. / Carbon 45 (2007) 2017–2021
The Stage III takes place after switching off the plasma
and substrate cooling down to the room temperature. The
nickel substrate and the graphite film should change sizes
in accordance with their thermal expansion coefficients.
This coefficient for Ni is changed from 12.89 to 21.0 ·
10 6 K 1 for the temperature range 300–1273 K [22], while
for graphite this value changes from 1.22 to 0.7 · 10 6 K 1
[1] in the same temperature range. Such strong difference in
the thermal expansion coefficients should lead to a much
stronger reduction of the lateral sizes of Ni substrate than
those of graphite film. As a result, a mechanical stress
should appear in the film during the sample cooling. A
release of this stress could lead to formation of the ridges
or/and swells in the areas with the weakest adhesion
between the film and the substrate.
Using the thermal expansion coefficient data [1,21], we
have estimated the difference in the reduction of linear sizes
for the graphite film and the Ni substrate after their cooling
from the deposition temperature (about 1200 K) down to
the room temperature. It appeared to be about 31 ± 8 nm
per length of 2 lm, which is an average distance between
the ridges in SEM images (see Fig. 1). This value coincides
well with the difference between the total length measured
along the typical ridge profile and along its base. According
to STM data this difference for the typical ridges on the
graphite film surface was about 35 ± 15 nm.
4. Conclusions
The well-ordered graphite films of nanometer thickness
have been grown on Ni substrates by chemical vapor depo-
sition from hydrogen–methane gas mixture activated by
DC discharge. These films have the atomically smooth
micron-size regions separated from each other by ridges.
The ridges have been formed due to the difference in ther-
mal expansion coefficients for graphite and nickel. The film
thickness has been estimated as 1.5 ± 0.5 nm. The pro-
posed mechanism of the film formation assumes a nucle-
ation process as a heteroepitaxial growth of graphene
sheets on (1 1 1) surface of Ni crystallites composing the
substrate. The films grown under the same conditions on
Si substrates have a highly disordered (amorphous) struc-
ture due to a larger lattice mismatch between Si and
graphite.
Acknowledgements
The authors are grateful to Drs. K.N. Eltsov and B.V.
Andryushechkin from General Physics Institute of Russian
Academy of Sciences (Moscow, Russia) for collaboration
in STM studies and registration of the STM images.
This work was supported by EU FP6 Project No. 12881
FERROCARBON: ‘‘Room Temperature Ferromagnetism
in Graphite and Fullerenes’’ and by INTAS (Grant nos.
#04-84-297 and #05-109-4966).
View publication stats
2021
References
[1] Pierson HO. Handbook of carbon, diamond and fullerenes. Park
Ridge (NJ): Noyes; 1993.
[2] Novoselov KS, Geim AK, Morozov SV, Jiang D, Zhang Y, Dubonos
SV, et al. Electric field effect in atomically thin carbon films. Science
2004;306(5696):666–9.
[3] Novoselov KS, Geim AK, Morozov SV, Jing D, Katsnelson MI,
Grigorieva IV, et al. Two-dimensional gas of massless Dirac fermions
in graphene. Nature 2005;438:197–201.
[4] Zhang YB, Tan Y-W, Stormer HL, Kim Ph. Experimental observa-
tion of the quantum Hall effect and Berry’s phase in graphene. Nature
2005;438:201–5.
[5] Novoselov KS, McCann E, Morozov SV, Fal’ko VI, Katsnelson MI,
Zeitler U, et al. Unconventional quantum Hall effect and Berry’s
phase of 2 in bilayer graphene. Nature Phys 2006;2(3):177–80.
[6] Ohta T, Bostwick A, Seyller Th, Horn K, Rotenberg E. Controlling
the electronic structure of bilayer graphene. Science 2006;313(5789):
951–4.
[7] Ferrari AC, Meyer JC, Scardaci V, Casiraghi C, Lazzeri M, Mauri F,
et al. Raman spectrum of graphene and graphene layers. Phys Rev
Lett 2006;97:187401–4.
[8] Forbeaux I, Themlon J-M, Debever J-M. Heteroepitaxial graphite on
6H–SiC(0 0 0 1): interface formation through conduction-band elec-
tronic structure. Phys Rev B 1998;58(24):16396–406.
[9] Gamo Y, Nagoshima A, Wakabayashi M, Terai M, Oshima C.
Atomic structure of monolayer graphite formed on Ni(1 1 1). Surf Sci
1997;374(1):61–4.
[10] Obraztsov AN, Zolotukhin AA, Ustinov AO, Volkov AP, Svirko
YuP. Chemical vapor deposition of carbon films: in-situ plasma
diagnostics. Carbon 2003;41(4):836–9.
[11] Eltsov KN, Klimov AN, Priadkin SL, Shevlyuga VM, Yurov VYu.
Ultra high vacuum scanning tunneling microscope for chemical
surface reactions study and nanotechnology. Phys Low Dimen Struct
1996;7–8:115–21.
[12] Eltsov KN, Shevlyuga VM, Yurov VYu, Kvit AV, Kogan MS. Sharp
tungsten tips prepared for STM study of deep nanostructures in
UHV. Phys Low Dimen Struct 1996;9–10:7–12.
[13] Ferrari AC, Robertson J. Resonant Raman spectroscopy of disor-
dered, amorphous, and diamond-like carbon. Phys Rev B 2001;64(7):
075414–27.
[14] Nemanich RJ, Solin SA. First- and second-order Raman scattering
from finite-size crystals of graphite. Phys Rev B 1979;20(2):392–401.
[15] Pocsik I, Hundhausen M, Koos M, Ley L. Origin of the D peak in
Raman spectrum of microcrystalline graphite. J Non-Cryst Solids
1998;227–230(2):1083–6.
[16] Thomsen C, Reich S. Double resonant Raman scattering in graphite.
Phys Rev Lett 2000;85(24):5214–7.
[17] Pong W-T, Durkan C. A review and outlook for an anomaly of
scanning tunneling microscopy (STM): superlattices on graphite. J
Phys D: Appl Phys 2005;38(21):R329–55.
[18] These measurements were made in collaboration with Prof. A.N.
Titkov from Ioffe Physico-Technical Inst. (S. Petersburg, Russia). The
detailed description of these experiments is under preparation for
publication now.,
[19] Seah MP, Dench WA. Quantitative electron spectroscopy of surfaces:
a standard data base for electron inelastic mean free paths in solids.
Surf Inter Anal 1979;1(1):2–11.
[20] Tontegode AYa. Carbon on transition metal surfaces. Prog Surf Sci
1991;38(3–4):201–429.
[21] Bertoni G, Calmels L, Altibelli A, Serin V. First-principles calculation
of the electronic structure and EELS spectra at the graphene/Ni(1 1 1)
interface. Phys Rev B 2004;71(7):075402–8.
[22] Kollie TG. Measurements of the thermal-expansion coefficient of
nickel from 300 to 1000 K and determination of the power-law
constants near the Curie temperature. Phys Rev B 1977;16(11):
4872–81.