JOURNAL OF APPLIED PHYSICS
VOLUME 86, NUMBER 11
1 DECEMBER 1999
Micro-Raman spectroscopic analysis of tetrahedral amorphous carbon
films deposited under varying conditions
E. Liua)
School of Mechanical and Production Engineering, Nanyang Technological University, 50 Nanyang
Avenue, Singapore 639798, Singapore
X. Shi, B. K. Tay, L. K. Cheah, H. S. Tan, J. R. Shi, and Z. Sun
School of Electrical and Electronic Engineering, Nanyang Technological University, 50 Nanyang Avenue,
Singapore 639798, Singapore
共Received 17 December 1998; accepted for publication 26 August 1999兲
The structure of tetrahedral amorphous carbon 共ta-C兲 films deposited by a filtered cathodic vacuum
arc has been studied using micro-Raman spectroscopy in terms of substrate bias, nitrogen gas partial
pressure 共ta-C:N films兲 or aluminum content in a mixed aluminum/carbon target 共ta-C:Al films兲
during deposition. The first-order Raman spectra generally show a broad feature overlaid by a
disordered 共D兲 peak and a graphitic 共G兲 peak. The contribution of sp 3 bonding to the Raman
spectrum is not explicit, since the Raman phonon line is more sensitive to the sp 2 carbon bonding
due to its larger Raman scattering cross section. However by comparing the ratios of the intensities,
the full widths at half maximum 共FWHM兲, and the peak areas between the D and G peaks, the sp 3
contribution may indirectly be reflected by the complex Raman features. The G peak position for the
ta-C and ta-C:N films appears to not change significantly with the change of substrate bias voltage
or N2 partial pressure, whereas the shift of the D peak is more appreciable. On the contrary, the G
peak position for the ta-C:Al films shows a continuous decrease with increasing Al content. For the
undoped ta-C films, the minimum intensity, area, and FWHM ratios between the D peak and the G
peak are obtained at a bias around ⫺100 V, which corresponds to the maximum sp 3 content in the
ta-C films. These ratios for the ta-C:N and ta-C:Al films, however, generally increase with increased
N or Al content, which indicates the increase of sp 2 bonded clusters. © 1999 American Institute
of Physics. 关S0021-8979共99兲01223-2兴
I. INTRODUCTION
Ionized carbon atoms with varying energies can form
different types of carbon bonding in the film. The energy can
be controlled by applying a variable bias to the substrate. It
has been reported that the maximum sp 3 content can be obtained at a certain carbon ion impinging energy, for instance,
around 100 eV.9
The undoped ta-C is p type with the Fermi level lying
about 0.25 eV above the valence band mobility edge.10 By
incorporation of nitrogen, the Fermi level can be shifted toward the conduction band. Thus, ta-C can be doped or alloyed by N depending on the amount of N in the film,11,12
which is also related to the nitrogen doping efficiency.13
However, excess nitrogen atoms in the film may create C/N
compound or promote the formation of sp 2 clusters.
Metal-containing DLC materials are interesting since
they may satisfy some special requirements of certain properties such as the electrical and mechanical properties, wear
resistance and friction.14 It was reported that hydrogenated
amorphous carbon 共a-C:H兲 films containing certain metals at
the few percent level can have wear resistance and friction
coefficients comparable to those of unmetallized a-C:H
films.15 However, the structure and properties of metalcontaining ta-C films have seldom been reported.
The Raman active vibration numbering for natural single
4
and Z⫽4 per unit cell is
crystal graphite with symmetry D 6h
E 2g mode which exhibits a single Raman peak at around
Diamondlike carbon 共DLC兲 materials have many potential applications due to their superior thermal, electronic, optical, mechanical and tribological properties.1–7 The properties of these materials are closely related to their bonding
structure, which is in turn determined by the deposition conditions. However, the preparation of these materials by
chemical vapor deposition 共CVD兲 or physical vapor deposition 共PVD兲 processes has not yet been standardized.
DLC films can be deposited by a diversity of
techniques.8 A filtered cathodic vacuum arc can be used to
deposit tetrahedral amorphous carbon 共ta-C兲 films containing
sp 3 carbon bonding up to 85%.9 However, the s p 2 carbon
bonding, impurities and defects in the films can certainly
degrade these materials. For the production of DLC materials
with CVD, it is believed that the deposition conditions, such
as the substrate temperature, bias and gas flow rate, are crucial in controlling the quality and morphological characteristics of the films. However, with PVD processes at lower
temperatures or at room temperature, the thermal effect diminishes and the substrate bias becomes the most important
parameter.
a兲
Author to whom correspondence should be addressed; electronic mail:
[email protected]
0021-8979/99/86(11)/6078/6/$15.00
6078
© 1999 American Institute of Physics
Liu et al.
J. Appl. Phys., Vol. 86, No. 11, 1 December 1999
1580 cm⫺1, namely the G peak. This peak is associated with
the in-plane C–C stretching mode of the s p 2 hybridized carbon atoms. If graphite becomes disordered in the carbon layers, the G peak broadens.16,17 For polycrystalline graphite,
depending on the size of the crystallites, the Raman selection
rule k⫽0 is no longer valid and a second peak at 1350 cm⫺1
appears, namely the disordered or D peak. If the long-range
order of the crystalline material is lost and the carbon phase
becomes glassy, both the G and the D peaks broaden. The
bond angle distortions may shift the D peak to a lower wave
number.18
Raman spectroscopy has been widely used to study the
structure of DLC materials.19,20 However, unlike resonant
Raman scattering,21 visible laser light such as 514.5 nm light
is not sensitive to s p 3 carbon bonding in DLC materials.
Therefore, fitting D and G peaks has commonly been used to
interpret the Raman spectra acquired from DLC materials.
Using a single Breit-Wigner-Fano 共BWF兲 line shape, as was
proposed by Prawer and co-workers,22 the emergence of the
D peak can be used to predict the appearance of s p 2 clusters
in the film for films containing a high s p 3 content.
In this article the structure of ta-C films deposited at
different substrate biases or containing varying N or Al contents is studied using micro-Raman spectroscopy with 514.5
nm laser light in terms of the D peak and G peak shifts or the
ratios of the Raman peak intensities, full widths at half maximum 共FWHM兲 and integrated peak areas between D and G
peaks.
II. EXPERIMENTAL DETAILS
The films used in this study were deposited by a filtered
cathodic vacuum arc 共FCVA兲 process. Details of the FCVA
system are given elsewhere.9 In the FCVA process, the ionized atoms produced from the target in the vacuum chamber
(10⫺4 – 10⫺7 Torr兲 were accelerated through a mechanicalelectrical-magnetic filtering bend towards the substrate and
further deposited on the substrate. Undoped n-type 具100典 silicon wafers were used as the substrate. The wafers were first
cleaned in detergent liquid and then in de-ionized water using an ultrasonic machine before being placed in the vacuum
chamber. The substrate surface was cleaned further by Ar ion
bombardment in the vacuum chamber prior to deposition.
The wafer surface was then coated with a layer of thin film.
For the ta-C films, a bias voltage from ⫺20 to ⫺160 V was
applied to the substrate during deposition. All the depositions
were carried out at room temperature 共22 °C兲. For deposition
of the ta-C:N films, nitrogen gas was fed into the deposition
chamber through a mass flow controller. The nitrogen partial
pressure was varied between 5⫻10⫺4 and 7⫻10⫺7 Torr. For
the ta-C:Al films, mixed aluminum/graphite 共Al/C兲 targets
were used in place of a pure graphite target during deposition. The Al/C targets, with varying Al contents were made
of Al and graphite powder under pressure of 0.6 GPa. The
deposition procedure was the same as that for the pure graphite target. Both the ta-C:N and the ta-C:Al films were deposited at a bias of ⫺100 V, since this is the energy known to
give the highest s p 3 content for ta-C films.
6079
A micro-Raman spectroscope 共Ramascope, Renishaw
1000兲 using 514.5 nm Ar laser light was used to characterize
the film structure. The laser output power was 10 mW. Filtering of 50% power was also used to reduce the power
density at the sample. The laser beam was focused onto the
sample surface using an optical microscope with a magnification of 50⫻ 共laser spot size ⬃1 ␮m兲. The Raman spectra
were acquired in the range of 1100–1900 cm⫺1 to evaluate
differences in the structures of the films in terms of the firstorder D and G peak positions and the ratios of the peak
amplitudes, FWHMs, and integrated peak areas between the
corresponding D and G peaks.
Atomic force microscopy 共AFM兲 共S-3000, Digital
Instruments兲 was used to measure the surface morphological
characteristics of the films.
III. RESULTS AND DISCUSSION
Figure 1 shows AFM images of the surface morphologies of ta-C, ta-C:N and ta-C:Al films, where the ta-C film
was deposited at a bias of ⫺100 V, the ta-C:N film was
deposited at a N2 partial pressure of 1⫻10⫺5 Torr and the
ta-C:Al film was grown using a C/Al target containing 5
at. % Al. It can be seen that the ta-C film contains fine asperities. The ta-C:N film shows larger clusters. The cluster
size of the ta-C:Al film is between that of the ta-C and the
ta-C:N films. It is hypothesized that the fine clusters in the
ta-C film are due to a large percentage of sp 3 carbon bonding formed by the high impinging carbon energies.
Typically in the ta-C films, the predominant component
is sp 3 bonded carbon together with certain sp 2 bonded carbon clusters. The Raman phonon lines are more sensitive to
the sp 2 carbon bonding due to its larger Raman scattering
cross section. The Raman spectra for the films in this study
give a broad band overlaid by G and D peaks as shown in
Figs. 2–4. The contribution from the sp 3 component to the
Raman spectra is not explicit. However, by comparing the
intensities, full widths at half maximum and integrated peak
areas between the D and the G peaks, and by referring to the
information acquire by some other techniques, e.g., electron
energy loss spectroscopy 共EELS兲 and x-ray photoelectron
spectroscopy 共XPS兲 etc., it can be deduced that the sp 2
bonded clusters change in structure concomitantly with a
change in the sp 3 content. Figures 2 and 3 show that the G
peak position appears to not change significantly with the
change of deposition parameters such as the substrate bias
and N partial pressure, while the D peak position has
changed significantly with the deposition conditions. There
is a tendency that the ratios of the intensities, FWHMs and
peak areas between the D and G peaks first decrease and then
increase with an increase of substrate bias during deposition.
The ratios I D /I G and FWHMD /FWHMG both show a minimum at a bias of around ⫺100 V. This trend is in agreement
with the results previously measured by EELS,9 from which
the maximum sp 3 content in the ta-C films was obtained at a
bias of around ⫺100 V. The minimum of these ratios indicates that the amount of sp 2 content has reduced, which may
indirectly indicate that the sp 3 content has increased.
6080
J. Appl. Phys., Vol. 86, No. 11, 1 December 1999
Liu et al.
FIG. 2. Raman spectra of ta-C films deposited at different substrate biases:
共a兲 Raman spectra overlaid with fitted curves and 共b兲 D and G peak positions
and ratio of the peak intensities, FWHMs and integrated peak areas between
the D and the G bands.
FIG. 1. Surface morphology of films: 共a兲 ta-C film, 共b兲 ta-C:N film deposited at a N2 partial pressure of 1⫻10⫺5 Torr and 共c兲 ta-C:Al film deposited
using a C/Al target containing 5 at. % Al.
However, the D peak of the Raman spectra of the ta-C
and ta-C:N films shows more evidence of dependence on the
substrate bias and the N2 partial pressure. Figure 2 shows
that the D peak first shifts to a lower frequency with an
increase of bias and then shifts to a higher frequency with a
further increase of bias. The minimum Raman shift for the D
peak occurs at a bias of ⫺100 V, which corresponds to the
maximum sp 3 content in the ta-C films.
It is well known that the G band of bulk graphite is
located at 1580 cm⫺1 of the E 2g mode which represents the
C–C stretching vibration in the graphite layer.23 In DLC materials, the decrease in the C–C stretching vibration corresponds to lengthening of the C–C bond and some carbon
atoms leave the layer. Under these conditions interlayer
bonds may appear. DLC materials therefore usually show a
lower G peak position. However, in this study, it is worth
noting that the G band of both ta-C and ta-C:N films does
not deviate considerably from that of graphite. Furthermore,
the G peak position is relatively insensitive to the substrate
bias or N2 partial pressure during deposition.
J. Appl. Phys., Vol. 86, No. 11, 1 December 1999
FIG. 3. Raman spectra of ta-C:N films deposited at different N2 partial
pressures: 共a兲 Raman spectra overlaid with fitted curves and 共b兲 D and G
peak positions and ratios of the peak intensities, FWHMs and integrated
peak areas between the D and the G bands.
Silva et al.24 have demonstrated that s p 2 bonded carbon
atoms are localized in graphitic clusters. The s p 2 clusters are
connected by the sp 3 bonded matrix. Since the films normally have a high compressive internal stress, the C–C
bonding is strained. Some carbon bonds may be shortened
and others are lengthened, depending on the bonding orientation relative to the stress direction. Ager et al.25 measured
the compressive stress of ta-C films using Raman spectroscopy. However, their measurement was based only on the G
peak shift to higher frequency and not on that of the D peak.
McCulloch et al. have suggested that the change in the intensity ratio between the D and G peaks reflect the change in
the size of the s p 2 bonded clusters.26,27 As the size of the
Liu et al.
6081
FIG. 4. Raman spectra of ta-C:Al films deposited using C/Al targets with
varying Al contents: 共a兲 Raman spectra overlaid with fitted curves and 共b兲 D
and G peak positions and ratios of the peak intensities, FWHMs and integrated peak areas between the D and the G bands.
clusters increases, the D peak intensity increases. The D
band shift is therefore not related to the stress effect, which
may also indicate the effect of the sp 2 cluster size.
At low N doping levels, the dominant tetrahedral structure of ta-C is not expected to alter substantially and the
band gap remains nearly constant. The N atoms form a dopant configuration (N⫹
4 ). However with a high content of N in
the film, the graphitization around the N atoms can lead to
graphitic chains throughout the material.
By comparing Fig. 2共b兲 with Fig. 3共b兲, it is interesting to
note that varying the substrate bias voltage over a fairly wide
range does not give rise to large sp 2 bonded clusters,
whereas adding N above a certain level leads to graphitization. In Fig. 2共b兲, the ratios can be used as a sensitive mea-
6082
Liu et al.
J. Appl. Phys., Vol. 86, No. 11, 1 December 1999
sure of sp 2 bonded carbon clusters in the ta-C films. Then in
Fig. 3, the film structure may have changed from doped
ta-C:N to alloyed ta-C:N when the N partial pressure is
higher than 5⫻10⫺4 Torr.
For the Al-containing films, both the G and D positions
shift to lower frequencies with an increase in the Al content.
This trend may be caused by the Al induced stress release or
by the high Al content. Another study28 has shown that the
internal stress has been reduced from about 12 GPa for the
ta-C films to about 1 GPa for the film deposited with an Al/C
target containing 15 at. % Al. Ager et al.25 have reported a G
peak shift of about 20 cm⫺1 on a ta-C film deposited at ⫺100
V bias compared to a free standing film. However, the G
peak shift is up to about 90 cm⫺1 for the Al-containing ta-C
films in this study. It is therefore inferred that the G peak
position shift is induced by not only the internal stress release.
Figure 4共b兲 shows the position of the D and G peaks as
the aluminum content is varied in the films. The graph shows
that both peaks change position as the aluminum content in
the target is increased. The D peak shifts from 1406 to 1327
cm⫺1 when the Al content is increased from 0.7 to 15 at. %.
The G peak shifts from 1569 to 1487 cm⫺1.
For the peak intensity and area ratios, the measurements
indicate that both increase as the content of aluminum increases. To explain this phenomenon, we propose to classify
the clusters into two categories, one ranging from atoms to
nanocluster and the other from nanoclusters to macroparticles.
The shift of the G peak position to a lower frequency can
be partly related to the reduction of compressive stress as
aluminum is introduced into the films.29 The hypothesis for
this is that the vibrational frequency in a material is proportional to its interatomic forces between the atoms. If the material is strained the spacings between the atoms change correspondingly, which further results in a change in the
vibrational frequency. However, it may not be realistic to
talk about strain relief to explain shifts of the order of 100
cm⫺1. With a high content of Al, the material may have been
modified dramatically. It is well known that Al is much
heavier than C in terms of atomic weight. The vibrational
modes of Al–C will tend to occur at much lower frequencies
than those of C–C bonds. In case the bonding between Al
and C forms during deposition as it does in Al–C compounds, large downward shifts of the Raman peak positions
are expected.
The shift of the D peak position to lower frequency may
be due to the further increase of the degree of disorder of the
sp 2 bonded clusters. The bond angles of the bonds in the
clusters also become further distorted30 and the bond lengths
increase, which weakens the bond strength. This gives rise to
lowering of the D peak position. That the intensity of the D
peak increases and the intensity of the G peak decreases
means that the size of the clusters is increasing.26,27 This
implies that the ratio of the peak intensities of the D and G
peaks should increase as the amount of aluminum increases
关Fig. 4共a兲兴.
The increase in the D peak intensity may also be related
to the relative increase in the amount of s p 2 bonded clusters
embedded in the sp 3 bonded matrix in the ta-C:Al films. The
difference in the ratio of the FWHMs is relatively small compared to the other two ratios.
The normal ta-C films are composed of the sp 3 bonded
carbon matrix with segregated sp 2 bonded carbon clusters.31
By introducing aluminum, it is expected that more doubly
bonded carbon clusters or even Al clusters may be created in
the film at the expense of the number of tetrahedral bonds.
The ratios of the D peak intensity, FWHM and peak area
to those of the G peak increase slightly with an increase of
Al content. With a small Al content, an increase in the
amount of sp 2 bonding appears to not be appreciable. However with the relatively high Al content in the film, the D
peak develops faster than the G peak. When the Al content is
small the amount of tetrahedral bonding may change less,
and the Al atoms may also partially act as electron acceptors,
forming doped structures. When excess Al is introduced into
the film, the Al atoms may react with C to form Al/C
compounds,32 or may exist in Al clusters, which is now being investigated.
IV. CONCLUSIONS
The structure of ta-C, ta-C:N, and ta-C:Al films was
studied using micro-Raman spectroscopy with 514.5 nm laser light.
It was shown that the G peak position appeared to not
change significantly with substrate bias for the ta-C films or
with N2 partial pressure for the ta-C:N films during deposition, while the D peak position significantly shifted.
For the ta-C films, there was a tendency for the ratios of
the intensities, FWHMs and peak areas between D and G
peaks to first decrease and to then increase with an increase
of substrate bias during deposition. The minimum ratios
were obtained at a bias of around ⫺100 V, which corresponds to the maximum sp 3 content in the ta-C films.
For the ta-C:N films, the ratios first slowly increased
with an increase of N2 partial pressure when the N2 partial
pressure was lower than 1⫻10⫺4 Torr, and then increased
relatively fast with an increase of N2 partial pressure when
the N2 partial pressure was higher than 1⫻10⫺4 Torr. The
sp 2 component may grow much faster when the N2 partial
pressure is higher than 1⫻10⫺4 Torr.
Compared to the ta-C films deposited in a fairly wide
range of bias voltages, the ta-C:N film graphitization was
more significant when the N content was above a certain
concentration.
For the Al-containing films, both the G and the D peak
positions shifted to lower frequencies with an increase of Al
content. The peak intensity and area ratios increased as the
content of aluminum increased. The difference in the FWHM
ratio was relatively small compared to that of the other two
ratios. More doubly bonded carbon clusters or even Al clusters may be created at the expense of the number of tetrahedral bonds. It is proposed that the amount of tetrahedral
bonding in the films may be less changed, when the Al content was small. The Al atoms in the film may react with C to
from Al/C compounds or may exist in Al clusters when excess Al is introduced into the film.
Liu et al.
J. Appl. Phys., Vol. 86, No. 11, 1 December 1999
A. Grill, V. Patel, and B. Meyerson, Surf. Coat. Technol. 49, 530 共1991兲.
N. Missert, T. A. Friedmann, J. P. Sullivan, and R. G. Copeland, Appl.
Phys. Lett. 70, 1995 共1997兲.
3
J. Jaskie, MRS Bull. 21, 59 共1996兲.
4
E. Liu, B. Blanpain, E. Dekempeneer, F. Löffler, J.-P. Celis, and J. R.
Roos, Diamond Films Technol. 4, 37 共1994兲.
5
E. Liu, B. Blanpain, X. Shi, J.-P. Celis, H.-S. Tan, B.-K. Tay, and J. R.
Roos, Surf. Coat. Technol. 106, 72 共1998兲.
6
L. K. Cheah, X. Shi, E. Liu, and J. R. Shi, Appl. Phys. Lett. 73, 2473
共1998兲.
7
E. Liu, X. Shi, L. K. Cheah, Y. H. Hu, H. S. Tan, J. R. Shi, and B. K. Tay,
Solid–State Electron. 43, 427 共1998兲.
8
Y. Catherine, Diamond and Diamond-Like Films and Coatings, edited by
R. E. Clausing et al. 共Plenum, New York, 1991兲, p. 193.
9
Shi Xu, B. K. Tay, H. S. Tan, Li Zhong, Y. Q. Tu, S. R. P. Silva, and W.
I. Milne, J. Appl. Phys. 79, 7239 共1996兲.
10
V. S. Veerasamy, G. A. J. Amaratunga, C. A. Davis, W. I. Milne, P.
Hewitt, and M. Weiler, Solid-State Electron. 37, 319 共1994兲.
11
A. Y. Liu and M. L. Cohen, Science 245, 841 共1989兲.
12
C. Niu, Y. Z. Lu, and C. M. Lieber, Science 261, 334 共1993兲.
13
V. S. Veerasamy, J. Yuan, G. A. J. Amaratunga, W. I. Milne, K. W. R.
Gilkes, M. Weiler, and L. M. Brown, Phys. Rev. B 48, 17954 共1994兲.
14
H. Dimigen, H. Hübsch, and R. Memming, Appl. Phys. Lett. 50, 1056
共1987兲.
15
M. Grischke, K. Bewilogua, K. Trojan, and H. Dimigen, Surf. Coat. Technol. 74–75, 739 共1995兲.
16
R. O. Dillon, J. A. Woollam, and V. Katkanant, Phys. Rev. B 29, 3482
共1984兲.
6083
H. Tsai and D. B. Bogy, J. Vac. Sci. Technol. A 5, 3287 共1987兲.
D. S. Knight and W. B. White, J. Mater. Res. 4, 385 共1989兲.
19
N. Wada, P. J. Gaczi, and S. A. Solin, J. Non.-Cryst. Solids 35&36, 543
共1980兲.
20
N. Rouzaud, O. Oberlin, and C. Beny-Bassez, Thin Solid Films 105, 75
共1983兲.
21
K. W. R. Gilkes, H. S. Sands, D. N. Batchelder, W. I. Milne, and J.
Robertson, J. Non-Cryst. Solids 227–230, 612 共1998兲.
22
S. Prawer, K. W. Nugent, Y. Lifshitz, G. D. Lempert, E. Grossman, J.
Kulik, I. Avigal, and R. Kalish, Diamond Relat. Mater. 5, 433 共1996兲.
23
P. V. Huong, Diamond Relat. Mater. 1, 33 共1991兲.
24
S. R. P. Silva, J. Robertson, G. A. J. Amaratunga, B. Rafferty, L. M.
Brown, J. Schwan, D. F. Franceschini, and G. M. Mariotto, J. Appl. Phys.
81, 2626 共1997兲.
25
J. W. Ager III, S. Anders, A. Anders, and L. G. Brown, Appl. Phys. Lett.
66, 3444 共1995兲.
26
D. G. McCulloch, S. Prawer, and A. Hoffman, Phys. Rev. B 50, 5905
共1994兲.
27
D. G. McCulloch and S. Prawer, J. Appl. Phys. 78, 3040 共1995兲.
28
E. Liu, X. Shi, and B. K. Tay, Technical Report, Nanyang University,
Singapore 共1998兲.
29
Q. Wei, R. J. Narayan, J. Narayan, J. Sankar, and A. K. Sharma, Mater.
Sci. Eng., B 53, 262 共1998兲.
30
P. K. Bachmann and D. U. Wiechert, Diamond Relat. Mater. 1, 422
共1992兲.
31
J. Robertson, Diamond Relat. Mater. 1, 397 共1992兲.
32
B. Maruyama, F. S. Ohuchi, and L. Rabenberg, J. Mater. Sci. Lett. 9, 864
共1990兲.
1
17
2
18