Electronic state modification in laser deposited amorphous carbon films by the inclusion of nitrogen

Abstract

In this study, we investigate the effect of the inclusion of nitrogen in amorphous carbon thin films deposited by pulsed laser deposition, which results in stress induced modifications to the band structure and the concomitant changes to the electronic transport properties. The microstructural changes due to nitrogen incorporation were examined using electron energy-loss spectroscopy and Raman scattering. The band structure was investigated using spectroscopic ellipsometry data in the range of 1.5–5 eV, which was fitted to the Tauc Lorentz model parametrization and optical transmittance measurements. The dielectric constant evaluated using optical techniques was compared to that obtained with electrical measurements, assuming a Poole-Frenkel type conduction process based on the best fits to data. The electrical conduction mechanism is discussed for both low and high electric fields, in the context of the shape of the band density of states. By relating a wide range of measurement techniques, a detailed relationship between the microstructure, and the optical and the electrical structures of a -CN _x films is obtained. From these measurements, it was found that, primarily, the change in density of the film, with increasing nitrogen pressure, affects the band structure of the amorphous carbon nitride. This is due to the fact that the density affects the stress in the film, which also impacts the localized states in the band gap. These results are supported by density of states measurements using scanning tunneling spectroscopy.

Full text

Page 1

Electronic state modification in laser deposited amorphous carbon films
by the inclusion of nitrogen
Y. Miyajima,1,a?G. Adamopoulos,1S. J. Henley,1V. Stolojan,1Y. Tison,1E. Garcia-Caurel,2
B. Drévillon,2J. M. Shannon,1and S. R. P. Silva1,b?
1Nano-Electronic Centre, Advanced Technology Institute, University of Surrey, Guildford, GU2 7XH,
United Kingdom
2Laboratoire de Physique des Interfaces et des Couches Minces, UMR 7647 CNRS, Ecole Polytechnique,
91128 Palaiseau Cedex, France
?Received 18 February 2008; accepted 10 July 2008; published online 17 September 2008?
In this study, we investigate the effect of the inclusion of nitrogen in amorphous carbon thin films
deposited by pulsed laser deposition, which results in stress induced modifications to the band
structure and the concomitant changes to the electronic transport properties. The microstructural
changes due to nitrogen incorporation were examined using electron energy-loss spectroscopy and
Raman scattering. The band structure was investigated using spectroscopic ellipsometry data in the
range of 1.5–5 eV, which was fitted to the Tauc Lorentz model parametrization and optical
transmittance measurements. The dielectric constant evaluated using optical techniques was
compared to that obtained with electrical measurements, assuming a Poole-Frenkel type conduction
process based on the best fits to data. The electrical conduction mechanism is discussed for both low
and high electric fields, in the context of the shape of the band density of states. By relating a wide
range of measurement techniques, a detailed relationship between the microstructure, and the optical
and the electrical structures of a-CNxfilms is obtained. From these measurements, it was found that,
primarily, the change in density of the film, with increasing nitrogen pressure, affects the band
structure of the amorphous carbon nitride. This is due to the fact that the density affects the stress
in the film, which also impacts the localized states in the band gap. These results are supported by
density of states measurements using scanning tunneling spectroscopy. © 2008 American Institute
of Physics. ?DOI: 10.1063/1.2977718?
I. INTRODUCTION
During the last few years, carbon nitride and its counter-
part, amorphous carbon nitride ?a-CNx?, has emerged as a
material of high technological and scientific interest.1–6The
interest in carbon nitride has been renewed due to its poten-
tial as an electronic thin film for both cold cathode displays7
and electrode materials in electrochemical studies of water
treatment.8Electronic properties have been reported for both
hydrogenated and nonhydrogenated amorphous carbon ni-
tride. In particular, it is found that nitrogen incorporation has
a beneficial effect on electric field emission at low micro-
scopically applied fields by reducing the threshold electric
field.9,10The efficient inhibition of redox decomposition re-
actions of water using a-CN or a-CN:H electrodes over a
3–4 V potential window, as compared to only about 2 V for
metal electrodes, has also introduced many more potential
applications.11,12Additionally, the deposition of nitrogen-
containing amorphous carbon has received particular atten-
tion following the theoretical predictions of an ultrahard
silicon-nitride-like phase, i.e., ?-C3N4.13,14According to
these predictions, this phase would have insulating proper-
ties, hardness, and thermal conductivity comparable to that
of diamond.
Carbon nitride thin films can be deposited by a variety of
deposition techniques, namely sputtering,15plasma enhanced
chemical vapor deposition,16,17filtered cathodic vacuum arc
?FVCA?,18,19electron cyclotron wave resonance ?ECWR?,20
mass selected ion beam deposition,21and integrated distrib-
uted electron cyclotron resonance.22Among the deposition
techniques, pulsed laser deposition ?PLD? ?Refs. 23 and 24?
is a well-established technique for a-CNxthin film deposition
that allows for the tuning of the structural, optical, and elec-
tronic properties under varying growth conditions. Under-
standing the structure and the physics of amorphous carbon,
as well as its alloys, is essential to improve electronic prop-
erties of carbon-based amorphous semiconductors such as
carrier mobility, dielectric properties of low k carbon-based
films,25–27as well as electron emission from carbon cold
cathodes.7
This study aims to investigate a series of PLD grown
carbon nitride films with the aim of correlating the micro-
structure with the optical and the electronic transport prop-
erties for varying laser fluences and N2pressure.
II. EXPERIMENTAL DETAILS
A. Film deposition
PLD of nonhydrogenated carbon nitride films have been
investigated quite extensively over the past years. Most films
have been produced by deposition onto a suitable substrate
following nanosecond ultraviolet pulsed laser ablation ?PLA?
a?Electronic mail: y.miyajima@surrey.ac.uk.
b?Author to whom correspondence should be addressed. Electronic mail:
s.silva@surrey.ac.uk. FAX: ? 44 1483 68 6081.
JOURNAL OF APPLIED PHYSICS 104, 063701 ?2008?
0021-8979/2008/104?6?/063701/9/$23.00 © 2008 American Institute of Physics
104, 063701-1
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

Page 2

of a graphite target in a low-pressure nitrogen gas
atmosphere.28–30The ablation apparatus used in the present
study consisted of a high-vacuum stainless-steel chamber
that was evacuated using a turbo molecular pump, backed by
a mechanical pump, to a base pressure of 10−6Torr. The
chamber was back filled with N2during thin film deposition.
The incident laser beam was directed through one of the
chamber side arms that was sealed with a quartz window and
focused onto the target at 45° to the surface. The target, a 10
cm in diameter disk of highly oriented pyrolytic graphite
?Kurt J. Lesker, 99.99% purity?, was rotated to minimize
repeated ablation of the same spot. A KrF Lambda Physik
excimer laser ?LPX 210i?, operating at 248 nm with 25 ns
full width at half-maximum ?FWHM? pulse duration, was
used as the UV pulsed laser source. The surface of the target
was cleaned using laser ablation with a laser fluence of
4 J/cm2before depositions. The substrates were placed 6
cm from the graphite target. A series of a-C and a-CNxfilms
were deposited onto quartz, c-Si, and mica substrates at laser
density power between 4 and 12 J/cm2for various nitrogen
pressures with the repetition rate of 10 Hz. Quartz and c-Si
substrates were cleaned using a series of organic solvents,
rinsed with de-ionized ?DI? water, and dried using N2gas.
Deposition on mica was carried out immediately after cleav-
ing. The typical thicknesses of the PLD deposited films were
around 30 nm.
B. Electron energy loss spectroscopy
The sp2content of our films was estimated using elec-
tron energy-loss spectroscopy ?EELS? technique.19,31–33The
sp2fraction can be obtained from EELS using the peak cor-
responding to a transition from the occupied 1s to empty ??
states in the carbon K edge spectra. The area of the peak was
normalized within the region up to 294 eV and the area of
the peak was compared with graphite that is 100% sp2
bonded carbon.34Additionally, the nitrogen content can be
evaluated by comparing the carbon K edge and a nitrogen K
edge.35The valence electron density of the film can be evalu-
ated from plasmon energy in the low loss spectrum. It must
be noted that the density of the film can be obtained assum-
ing that the number of valence electrons is four for carbon
and five for nitrogen.36
A CM200 Philips Supertwin transmission electron mi-
croscope ?TEM? operating at 200 kV with a LaB6filament
was used for measurements with a Gatan Imaging Filter
GIF2000 spectrometer for EELS. Films with a thickness of
30–70 nm were floated off from the mica substrate by im-
mersing in DI water and then placed on copper TEM grids.
The obtained EELS spectra and results derived for the films’
density, sp2and nitrogen content for varying nitrogen pres-
sures, and laser power densities are shown in Figs. 1 and 2,
respectively.
C. Raman spectroscopy
Raman spectroscopy is one of the most popular and
powerful techniques for the nondestructive analysis of amor-
phous carbon and its alloys that provides important structural
information.37–40Unpolarized visible Raman spectra were
obtained using the 514.5 nm line of an Ar+laser. A Renishaw
micro Raman system 1000 spectrometer was used with a
laser output of 15 mW, which resulted in an incident power
at the sample of about 1 mW.
The recorded spectra between 800 and 1800 cm−1are
shown in Fig. 3. As a consequence of resonance effects, Ra-
man is much more sensitive to ? bonds than to ? bonds. In
addition to the second order TO peak of the Si substrate at
about 960 cm−1, the spectra show the usual amorphous car-
bonvibrationalfeatures,
1500–1580 cm−1and a weaker D peak around 1350 cm−1.
The G mode is the bond stretching mode between two
sp2-hybridized C sites ?olefinic or aromatic? while the D
mode is the breathing motion of sp2-aromatic rings usually
activated with disordered carbon where the graphene net-
work has been disrupted.
TheRamanspectrawas
Breit-Wigner-Fano41,42line shape to the G peak and Lorent-
zian line shape to the D peak.37The fitting results for the
integrated D and G intensities ratio, i.e., the G line width and
the G line position ??max?Ref. 37??, are shown in Fig. 4.
namelya
G
peakat
curvefittedusing the
D. Spectroscopic ellipsometry
The a-C films were studied with ex situ UV visible spec-
troscopic ellipsometry ?SE? in the range of photon energies
FIG. 1. The EELS spectra of a-CNxfilms for different background nitrogen
pressures and laser fluences of ?a? 4 and ?b? 12 J/cm2.
063701-2Miyajima et al. J. Appl. Phys. 104, 063701 ?2008?
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

Page 3

from 1.5 to 5 eV using the UVISEL system by Jobin Yvon.
For the analysis, a Tauc-Lorentz ?TL? analytical expression
for the dielectric function43,44was assumed. The SE provides
the complex refractive index and allows the energies Eg?for
the onset of the optical transitions? and E04?where the ab-
sorption ?=104cm−1? to be obtained. A four-parameter
model ?A, C, Eg, and ?inf? is sufficient to describe the optical
functions of the thin film to the accuracy of the ellipsometer.
Jellison and Modine44developed this model using the
Tauc joint density of states ?DOS? and the Lorentz oscillator.
The imaginary part ?iof the dielectric function is given by
the product of the imaginary part of Tauc’s dielectric func-
tion with the Lorentz one. In the approximation of parabolic
bands, Tauc’s dielectric function describes interband transi-
tions above the band edge. The new expression for ?iis set
up as
?i?E? =?
?E2− Eo
AEoC?E − Eg?2
2?2+ C2E2·1
E? for E ? Eg,
?1?
?i?E? = 0 for E ? Eg,
?2?
where Egis the optical band gap, A is the strength of the ?i
peak, C is the broadening term of the peak, and E0is the
peak central energy; thus E0is always larger than Eg. The
real part ?rof the dielectric function is derived from the
expression of ?iusing the Kramers-Kronig integration.
The TL expression is consistent with known physical
phenomena within the limitations of the model. At large E,
the ?i?E? of the TL model→0. This is consistent with ob-
served behavior in the x-ray and ?-ray regime, where it is
known that the absorption coefficient is very small. Further-
more, ?i?E?=0 below Eg. The only mechanisms that give a
nonzero value of ?i?E? below the band gap are those that are
explicitly ignored in the TL model, such as Urbach tail ab-
sorption and vibrational absorption in the infrared. Finally,
the TL expression is consistent with the Kramers-Kronig
derivation in that ?r?E? is determined by Kramers-Kronig
FIG. 2. The nitrogen gas pressure against ?a? sp2fraction, ?b? N content, and
?c? density for laser fluences of 4 and 12 J/cm2.
FIG. 3. The visible Raman spectra of a-CNxfilms for different nitrogen
pressures and laser fluences of ?a? 4 and ?b? 12 J/cm2.
FIG. 4. ?a? I?D?/I?G? peak intensity ratio, ?b? the FWHM of the G peak, and
?c? the G peak position for laser fluences of 4 and 12 J/cm2.
063701-3Miyajima et al. J. Appl. Phys. 104, 063701 ?2008?
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

Page 4

integration. The high frequency dielectric constant ??inf? is an
additional fitting parameter that prevents ?rfrom converging
to zero for energies below the band gap.
The fitted parameters of the TL model, as well as the
refractive index ?n? and the extinction coefficient ?k? disper-
sions, are shown in Figs. 5 and 6, respectively. Here it should
be pointed out that the model for inhomogeneous films,
which is usually expressed by the volume fraction of void,
was used.
E. Ultraviolet-visible-near infrared optical
transmittance
A Cary 5000, Varian ultraviolet-visible-near infrared
?UV-VIS-NIR? spectrometer was used for the optical trans-
mittance measurements in the range between 190 nm
??6.5 eV? and 3000 nm ?0.4 eV? in the dual beam mode.
Films deposited on quartz substrates were used for these
measurements.
The optical transmittance spectra is provided with
complementary optical characteristics, namely, E04?corre-
sponds to the absorption coefficient of 104cm−1? and the
Tauc optical band gap. The optical characteristics of the films
for various deposition conditions derived from SE and UV-
VIS-NIR are summarized in Fig. 7.
F. Electrical characterization
Electrical characterization was carried out using sand-
wich structures. A highly doped c-Si wafer was used as a
FIG. 5. SE TL model parameters. ?a? Tauc optical band gap Eg, ?b? the
broadening factor C, ?c? the transition peak central energy E0, and ?d? the
strength of the transition peak A.
FIG. 6. n and k dispersions for different background nitrogen pressures and
laser fluences of 4 and 12 J/cm2.
FIG. 7. The optical band gap ?E04and Tauc from UV-VIS-NIR and SE? and
the activation energy measured at low electric fields against the nitrogen gas
pressure for laser fluences of ?a? 4 and ?b? 12 J/cm2.
063701-4 Miyajima et al. J. Appl. Phys. 104, 063701 ?2008?
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

Page 5

substrate for the films and 100 nm of sputtered aluminum
was used to form a back contact. A circular top electrode
with an area of 1.5?10−4mm2was fabricated using a
shadow mask. The top electrodes were dc sputtered and com-
prised of 50 nm of Cr coated with 50 nm of Au. The thick-
ness of a-CNxfilms was about 30 nm.
Electrical measurements were carried out using a
Keithley 236 source meter with a temperature controlled
stage. The electrical characteristics at high electric fields
were measured at room temperature. In order to evaluate the
activation energy at low electric fields, the temperature was
varied between room temperature and 80 °C. The Current
Density-Electric Field characteristics are shown in Fig. 8.
The activation energy Eais plotted out in Fig. 7, which was
derived by assuming an Arrhenius relationship between cur-
rent and inverse temperature.
G. Scanning tunneling spectroscopy
Scanning tunneling spectroscopy ?STS? experiments
were performed at room temperature in a commercial ultra-
high vacuum scanning tunneling microscope in the ultra high
vacuum
?UHV-STM??Omicron
equipped with a scanning electron microscopic column using
electrochemically etched tungsten tips. During the experi-
ment, the base pressure in the UHV chamber was 2
?10−11mbar. The STS current-voltage curves were mea-
sured in the range of −1.5 to ?1.5 V for at least 16 positions
on a 100?100 nm2image, and are averaged in order to
obtain the DOS corresponding to a large area. Two samples
were measured, and both a-C and a-CNxfilms were depos-
ited with the laser fluence of 4 J/cm2.
VTmultiscanSTM?
III. RESULTS AND DISCUSSION
PLA of graphite results in the deposition of quite differ-
ent carbonaceous structures depending on the pressure of the
ambient gas in the chamber. The energy of the species in-
volved with the growth can also be changed by applying an
external bias to the chamber. A comparison of the PLA de-
posited material to that deposited using the conventional ca-
thodic arc process can be found in the literature.45Henley et
al.28investigated the PLA mechanism of a graphite target in
a chamber that contains Ar. The energy of the ablated carbon
ions exceed 100 eV, the optimal energy for sp3bond forma-
tion, in the vacuum when a laser fluence of 10 J/cm2is
used. When the Ar pressure is increased due to the additional
collisions of the gas ions with its concomitant decrease in
mean free path, the plume speed decreases. When the Ar
background pressure exceeds 20 mTorr, a shockwavelike
plume movement is observed and the speed reduces to
around 8 km/s, 400 ns after the ablation. The speed reduces
to 2.6 km/s after 800 ns with the Ar background pressure at
154 mTorr. Additionally, Fuge et al.46investigated the plume
dynamics during nanosecond and femtosecond ablations of
graphite in nitrogen. These studies must be carefully consid-
ered in order to understand the microstructural differences
initially revealed by EELS and Raman.
Following Fuge et al.,46a nitrogen background pressure
between 20 and 80 mTorr was chosen. This is because the
plume shows a shockwavelike propagation for pressures
above 20 mTorr, meaning that the transport of the carbon
atoms/clusters shows diffusive rather than ballistic character
at this pressure range. The most important factor that con-
trols the deposition process as well as the nitrogen incorpo-
ration ?and consequently the films structure? is the number of
the collisions that occur between the nitrogen molecules and
the laser ablated carbon atoms/ions/clusters in the chamber
once the nitrogen gas is introduced. The mean free path of
the nitrogen gas at the pressure of 0.75 mTorr ?1 mbar? has
been reported to be about 5 cm.47Taking into account that
this distance is shorter than the distance between the sub-
strate and the target ?6 cm in our experimental setup?, and the
nitrogen pressure is much higher ?20–80 mTorr?, we can
safely assume that at least one collision of the ablated carbon
species with a nitrogen molecule is expected before the car-
bon atom/cluster reaches the substrate. In such a diffusive
transport regime, the nitrogen is expected to be introduced
into the film microstructure. It is clear then that the increase
in the nitrogen pressure results in an increase in the multiple
scattering and reduces the speed of the plume ?or equiva-
lently the carbon species?. The small but notable decrease of
the film’s density for increased nitrogen pressure ?Fig. 2?c??
can be attributed to the reduced energy of the carbon species
incident on the substrate. Similar results have been reported
for PLA of graphite in an Ar atmosphere.28Figure 2?b? dem-
onstrates that the nitrogen pressure seems to have almost no
effect on the nitrogen content. It has been shown that the
peak of the emission of the CN radicals appears at about 8
mm from the target at a nitrogen background pressure of 10
mTorr.46Additionally, the mean free path of nitrogen at this
pressure is estimated around this distance, assuming that the
FIG. 8. J-E characteristics of devices deposited at various nitrogen pres-
sures and laser fluences ?a? 4 and ?b? 12 J/cm2.
063701-5Miyajima et al. J. Appl. Phys. 104, 063701 ?2008?
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

Page 6

CN radicals are formed at the initial collisions. Higher laser
fluences ablate more energetic carbon species.48These spe-
cies have a higher probability of reacting with nitrogen and
forming CN radicals, resulting in higher nitrogen incorpora-
tion, as shown in Fig. 2?b?. One of the reasons for an in-
crease in the sp2content with increasing nitrogen pressure
?Fig. 2?a?? is due to the decrease in the plume speed as a
result of collisions between the carbon species and the nitro-
gen molecules. The pressure range is sufficient to let these
carbon species experience diffusive transport before reaching
the substrate, and the incident speed is not high enough to
create high sp3fraction films. Furthermore, N, which prefers
to bond in a planer sp2configuration, dictates bond forma-
tion in the plume, which then gets transported to the sub-
strate.
The above trends are confirmed by the Raman spectra
characteristics that are shown in Fig. 3 and the analysis that
follows in Fig. 4. Indeed, the G peak remains almost constant
with increasing nitrogen pressure ?i.e., the nitrogen content?.
This indicates that N addition has replaced the C=C olefinic
groups with aromatic groups. Figure 4?c? shows the G-line
width variation as a function of the nitrogen pressure. Just to
note that the G peak width indicates the bond angle distor-
tions in the excited configurations. It has been reported49that
the G-line width decreases with increasing N content, indi-
cating loss of disorder of small clusters for higher N content.
In the nitrogen pressure range of 20–80 mTorr, where the
results in a nitrogen content is about 25 at % in our films, no
G-line variation has been observed. The I?D?/I?G? ratio ?Fig.
4?a?? shows the same trend. A major effect of N incorpora-
tion is the increase in the clustering of the sp2phase, which
is indicated by the D peak. The I?D?/I?G? ratio variation
with increasing nitrogen content confirms that the fundamen-
tal dependence of I?D?/I?G? is on sp2clustering.
The TL SE model fitted parameters are shown in Fig. 5.
If the valence and the conduction bands are treated as single
peaks, the peak position of a-CNxshould shift toward the
Fermi level, compared with a-C, since the ? and ??bands
become smaller in comparison to the ? and ??bands, which
become larger. This explanation can be supported by the
EELS K edge data ?see Fig. 1?, which can be treated as a
modification to the shape of the conduction band. This ex-
plains the drop observed in the mean peak position E0and
the peak strength A, and the constant value of the broadening
factor C. What is remarkable is the small increase in the
optical band gap considering the large increase in nitrogen
background pressure.
Figure 7 shows the optical band gap ?both Tauc and E04?
that was derived using different techniques. These results are
different to those reported in Ref. 49 where nitrogen incor-
poration decreases the optical band gap due to an increase in
the sp2content of the film and its consequences. Similar
results were reported on ta-CNxdeposited using FVCA and
ECWR.50On the other hand, dc sputtered low sp3fraction
a-CNxshows an increase in optical band gap with increasing
N/C ratio.50However, in the latter cases, the optical band
gaps are larger than those of the present study; thus there
must be another mechanism that reduces the optical band gap
in PLD a-C and a-CNxfilms.
The increasing optical band gap may be attributed to the
decreased defect density in the band gap since the films get
less dense. The obvious difference in the optical properties of
the nitrogen free films grown at different laser fluences is due
to the fact that the higher laser fluence forms higher quanti-
ties of sp3bonded material. Higher sp3fraction films show
higher E0and A values for the nitrogen free a-C, especially
those grown at 12 J/cm2. The small optical band gap a-C
films with high sp3fraction and dense film properties have
been reported elsewhere,51and the reduced band gap is at-
tributed to the high internal stress. The internal stress in the
a-C films deposited at more than 12 J/cm2is considered to
be high due to the films being able to delaminate easily,
whereas a-CNxfilms have good adhesion to the substrates.
Thus, the high sp3fraction films have high internal stress
that may create localized states in the band gap and thus
contribute to a low optical band gap.
The J-E characteristics are shown in Fig. 8. It appears
that the films become more resistive with increasing nitrogen
pressure for both laser fluences used. This is consistent with
the SE analysis, which concluded a lower defect density with
increasing nitrogen pressure, as well as the optical analysis,
which showed an increased band gap that would increase the
resistivity of the films. The free carrier concentration in a
semiconductor has an inverse exponential relationship to the
optical band gap. It was found that some samples were under
stress and failed at high electric fields. In most cases, electric
fields of less than 2.0?105V/cm should be applied for re-
producible results for these films due to their high stress. In
the case of a-CNxdeposited at 80 mTorr and a laser fluence
of 4 J/cm2, the field was limited to 1.0?105V/cm.
Figure 7 also show the activation energy Eaat low elec-
tric fields, which has comparable values to those of the op-
tical band gap evaluated using different techniques.
A possible conduction mechanism for such low electric
fields was reported to be band tail hopping,52via defects
?band tail? above the Fermi level. The defect states are
thought to lie above the Fermi level and the electrical acti-
vation energy indicates the energy between the Fermi level
and the defects. If the Fermi level lies in the middle of the
band gap, band tail hopping seems to occur at the bottom of
the empty ??band. It is consistent with the similar values
measured for both of Eaand optical band gaps.
Figure 9?a? shows the dielectric constant that was ob-
tained from SE. In general, there is a good correlation be-
tween the dielectric constant obtained from SE and that ob-
tained from conductivity measurements. Such a correlation is
found for a-C and a-CNxgrown at 20 mTorr of nitrogen gas
?see Fig. 9?.
The conductivity of a-CNxappears to be independent of
the laser fluence and it decreases with increasing nitrogen
pressure. This may also be related to the decrease in the
number of defects in the band gap due to the decrease in the
internal stress. It is believed that the decrease in the defect
states at the midband gap is reflected by the decrease in the
activation energy at low electric fields. In this region, the
conduction mechanism appears to be hopping within a defect
063701-6 Miyajima et al.J. Appl. Phys. 104, 063701 ?2008?
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

Page 7

band, and the conductivity decreases with increasing activa-
tion energy and a decrease in the defect density within the
band gap.
The conduction mechanism at high electric fields is nor-
mally different to that at low fields.53Figure 10 shows the
log of the current density versus the square root of the elec-
tric field at room temperature as a function of the nitrogen
pressure for laser fluences of 4 and 12 J/cm2. At electric
fields higher than 104V/cm, there is a linear dependence,
revealing that the main conduction mechanism appears to be
either Poole-Frenkel ?PF? bulk limited or Schottky barrier
contact limited conduction.
For these samples, two types of contacts were used. One
was the n+-Si substrate while the other was a Cr top contact.
However, the J-E characteristics seem to be symmetric, in-
dicative that the conduction mechanism is not contact domi-
nated. Thus, we come to the conclusion that the conduction
mechanism for these films could be a bulk limited process
such as PF conduction.
The current density J due to PF conduction through the
film is given by
J = eN??E exp?− ?? − ??PF/???E?
kT
?,
?3?
where
?PF=?
e3
??r?0
,
?4?
? is the trap depth, E the electric field across the film, N the
density of the neutral trapping centers, ? the effective mo-
bility, and ? is a coefficient that changes from one to two,
depending on the number of the charged defects and the
degree of compensation. When ?=1 is satisfied, the formula
is similar to the classical PF conduction model.53Here, it
should be noted that the neutral trapping centers are not nec-
essarily the same as the defects detected by electron spin
resonance ?ESR?, which detects unpaired spins.
The application of the classical PF ??=1? conduction
model of a-C and a-CNxcan provide us with an effective
dielectric constant. These values, as well as those for the
square of the refractive index n ?at 2 eV? assuming a low
imaginary component, were derived from the TL model dis-
persion curves ?Fig. 6? and are shown in Fig. 11 along with
those calculated from single wavelength ellipsometry. It is
demonstrated that the approximations that were used for both
optical and electrical measurements are very consistent. Ad-
ditionally, the decrease in the refractive index ?at 2 eV? for
increasing nitrogen gas pressure is consistent with the de-
crease in the films density,54as discussed earlier.
If the PF conduction mechanism dominates at high
fields, then the activation energy should be proportional to
the square root of the electric field. A a-CNxfilm that was
grown at a laser fluence of 4 J/cm2in 20 mTorr nitrogen gas
was measured to check for the change in activation energy
with electric field, as shown in Fig. 12. The estimated trap-
ping center depth was found to be about 0.16 eV. In addition,
the product N??can also be evaluated using the above equa-
tion. For the device used in Fig. 12, N??was estimated to be
about 3?1015?cm V s?−1. Investigation using ESR provides
a defect density of less than 1021cm−3for nonhydrogenated
a-C ?Ref. 55? and 1018cm−3for sputtered a-CNx.56Assum-
ing that the density of the neutral trapping centers is less than
this value, the mobility is higher than 10−4cm2/Vs when
FIG. 9. The N2pressure dependence of ?a? the dielectric constant at high
energy derived from SE and ?b? the conductivity at low electric fields less
than 104V/cm at room temperature.
FIG. 10. PF plots against N2pressure for laser fluences of ?a? 4 and ?b?
12 J/cm2.
063701-7Miyajima et al. J. Appl. Phys. 104, 063701 ?2008?
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

Page 8

N?1018cm−3for a-CNxfilms. This is consistent with the
reported field effect mobility of a-CNxbeing up to about
10−4cm2/V s.57,58The field effect mobility is estimated
from low electric fields ??103V/cm? whereas the effective
mobility in the conduction band used the PF plot at high
electric fields ??104V/cm?. At low electric fields, the con-
ductive mechanism is thought to be a hopping process
through the defect band, whereas the PF effect occurs at the
bottom of the conduction band. Thus, the mobility estimated
from PF plots can be higher than that extracted using field
effect transistor measurements. It should also be pointed out
that band tail hopping has been reported in a-CNxfilms and
that the PF effect might occur through defects in the band
tail22,52instead of the edge of the conduction band.
Figure 13 shows the normalized conductance of a-C and
a-CNx. The normalized conductance is proportional to the
DOS, and a negative sample bias shows the valence band of
the films, whereas a positive sample bias shows the conduc-
tion band of the films. From these DOS measured using STS
under UHV conditions, the gap of a-C ??0.9 eV? measured
appears to be larger than that of a-CNx??0.7 eV?, which is
in agreement with previous reports and literature with re-
gards to ta-CNxbut in disagreement with those results re-
ported for dc sputtered a-CNx.49,50However, the optical band
gap decreases when nitrogen is introduced. A remarkable
feature of Fig. 13 is the shallow slope of the DOS in the
conduction band. This is thought to be the reason why the
optical band gap is smaller than that reported in previous
reports.49,50The long band tail narrows the optical band gap
compared with the effective band gap for both a-C and
a-CNxfilms. This feature may be attributed to the high stress
in these films.
Therefore, the band diagram of PLD a-C and a-CNxcan
be illustrated, as shown in Fig. 14. The DOS of carbon films
consist of ? and ??bands attributed to sp2hybridization, and
? and ??bands associated with sp3hybridization. Moreover,
the band tail exists between ? and ??bands. The optical
band gap attributed to the transitions from the occupied ? to
the unoccupied ??bands, which are nonsymmetrical bands,
was confirmed with the DOS obtained by STS analysis ?see
Fig. 7 and 12?.
IV. CONCLUSION
This study demonstrates the effect of nitrogen inclusion
on the electronic states and the physical properties, including
electrical properties, in a-C films deposited using PLD. The
pressure range of nitrogen was chosen such that diffusive
FIG. 11. The dielectric constant obtained from PF plots, and n2at 2 eV
extracted from n and k dispersions with monochromatic ellipsometry.
FIG. 12. The activation energy versus the square root of the electric field for
a-CNxfilm deposited in 20 mTorr N2gas with a laser fluence of 12 J/cm2.
FIG. 13. The DOS derived from STS for PLD a-C and a-CNxfilms.
FIG. 14. The band diagram of PLD ?a? a-C and ?b? a-CNx.
063701-8 Miyajima et al.J. Appl. Phys. 104, 063701 ?2008?
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp

Page 9

carbon transport occurred during deposition and incorpora-
tion of nitrogen into the films. A series of techniques were
applied for structural, optical, and electrical characterization,
namely, EELS, Raman spectroscopy, SE, and UV-VIS-NIR
spectroscopy along with I-V measurements. It has been
shown that nitrogen incorporation changes the sp3fraction of
the films. Once nitrogen was incorporated, no strong depen-
dence between the nitrogen gas pressure, the laser fluence,
and the microstructure was found. However, the background
pressure during the deposition did affect the density, which
can in turn affect the stress in the films, and hence the local-
ized states in the band gap. Therefore, an increase in the
nitrogen pressure decreases conductivity and increases the
optical band gap. The presence of the long band tail is sup-
ported by the DOS close to the Fermi level obtained by STS
measurements and is in agreement with the broadening fac-
tor C obtained from SE. Thus, the presence of a long band
tail should be carefully considered when using a-C and
a-CNxfilms deposited by PLD in electronic devices.
Two types of conduction mechanism were found for low
and high electric fields. Band tail hopping appears to be the
dominant conduction mechanism at low electric fields while
PF conduction was confirmed to dominate conduction at
high fields. The fact that transport is dominated by hopping
between localized neutral defects suggests that shallow nitro-
gen donors are not present in large concentrations.
ACKNOWLEDGMENTS
The authors are grateful to the EPSRC Swindon, U.K.
for funding this program via a Portfolio Partnership award.
1M. Moseler, P. Gumbsch, C. Casiraghi, A. C. Ferrari, and J. Robertson,
Science 309, 1545 ?2005?.
2S. Bhattacharyya, S. J. Henley, E. Mendoza, L. Gomez-Rojas, J. Allam,
and S. R. P. Silva, Nat. Mater. 5, 19 ?2006?.
3Properties of Amorphous Carbon, edited by S. R. P. Silva ?INSPEC, Lon-
don, 2003?.
4Amorphous Carbon: State of the Art, edited by S. R. P. Silva, J. Robertson,
W. I. Milne, and G. A. J. Amaratunga ?World Scientific, Singapore, 1997?.
5J. Robertson, Mater. Sci. Eng., R. 37, 129 ?2002?.
6J. Windeln, C. Bram, H. L. Eckes, D. Hammel, J. Huth, J. Marien, H.
Rohl, C. Schug, M. Wahl, and A. Wienss, Appl. Surf. Sci. 179, 167
?2001?.
7S. R. P. Silva, J. D. Carey, G. Y. Chen, D. C. Cox, R. D. Forrest, C. H. P.
Poa, R. C. Smith, Y. F. Tang, and J. M. Shannon, IEE Proc.: Circuits
Devices Syst. 151, 489 ?2004?.
8H. Cachet, C. Debiemme-Chouvy, C. Deslouis, A. Lagrini, and V. Vivier,
Surf. Interface Anal. 38, 719 ?2006?.
9G. A. J. Amaratunga and S. R. P. Silva, Appl. Phys. Lett. 68, 2529 ?1996?.
10B. S. Satyanarayana, A. Hart, W. I. Milne, and J. Robertson, Appl. Phys.
Lett. 71, 1430 ?1997?.
11G. Adamopoulos, C. Godet, C. Deslouis, H. Cachet, A. Lagrini, and B.
Saidani, Diamond Relat. Mater. 12, 613 ?2003?.
12K. S. Yoo, B. Miller, R. Kalish, and X. Shi, Electrochem. Solid-State Lett.
2, 233 ?1999?.
13A. Y. Liu and M. L. Cohen, Science 245, 841 ?1989?.
14A. Y. Liu, R. M. Wentzcovitch, and M. L. Cohen, Phys. Rev. B 39, 1760
?1989?.
15R. Ohr, B. Jacoby, M. von Gradowski, C. Schug, and H. Hilgers, Surf.
Coat. Technol. 173, 111 ?2003?.
16S. R. P. Silva and G. A. J. Amaratunga, Thin Solid Films 253, 146 ?1994?.
17S. R. P. Silva, B. Rafferty, G. A. J. Amaratunga, J. Schwan, D. F. France-
schini, and L. M. Brown, Diamond Relat. Mater. 5, 401 ?1996?.
18G. A. J. Amaratunga, V. S. Veerasamy, W. I. Milne, C. A. Davis, S. R. P.
Silva, and H. S. Mackenzie, Appl. Phys. Lett. 63, 370 ?1993?.
19P. J. Fallon, V. S. Veerasamy, C. A. Davis, J. Robertson, G. A. J. Amara-
tunga, W. I. Milne, and J. Koskinen, Phys. Rev. B 48, 4777 ?1993?.
20S. Rodil, N. A. Morrison, W. I. Milne, J. Robertson, V. Stolojan, and D. N.
Jayawardane, Diamond Relat. Mater. 9, 524 ?2000?.
21Y. Lifshitz, S. R. Kasi, and J. W. Rabalais, Phys. Rev. Lett. 62, 1290
?1989?.
22C. Godet, N. M. J. Conway, J. E. Bouree, K. Bouamra, A. Grosman, and
C. Ortega, J. Appl. Phys. 91, 4154 ?2002?.
23M. P. Siegal, J. C. Barbour, P. N. Provencio, D. R. Tallant, and T. A.
Friedmann, Appl. Phys. Lett. 73, 759 ?1998?.
24S. R. P. Silva, S. Xu, B. K. Tay, H. S. Tan, H. J. Scheibe, M. Chhowalla,
and W. I. Milne, Thin Solid Films 290–291, 317 ?1996?.
25A. Grill, Diamond Relat. Mater. 10, 234 ?2001?.
26M. Aono and S. Nitta, Diamond Relat. Mater. 11, 1219 ?2002?.
27M. Aono, S. Nitta, T. Katsuno, T. Itoh, and S. Nonomura, Appl. Surf. Sci.
159–160, 341 ?2000?.
28S. J. Henley, J. D. Carey, S. R. P. Silva, G. M. Fuge, M. N. R. Ashfold,
and D. Anglos, Phys. Rev. B 72, 205413 ?2005?.
29A. A. Voevodin, J. G. Jones, J. S. Zabinski, Z. Czigany, and L. Hultman,
J. Appl. Phys. 92, 4980 ?2002?.
30T. Sato, A. Narazaki, Y. Kawaguchi, and H. Niino, Appl. Phys. A: Mater.
Sci. Process. 79, 1477 ?2004?.
31S. D. Berger, D. R. Mckenzie, and P. J. Martin, Philos. Mag. Lett. 57, 285
?1988?.
32J. Fink, T. Mullerheinzerling, J. Pfluger, A. Bubenzer, P. Koidl, and G.
Crecelius, Solid State Commun. 47, 687 ?1983?.
33S. R. P. Silva and V. Stolojan, Thin Solid Films 488, 283 ?2005?.
34A. J. Papworth, C. J. Kiely, A. P. Burden, S. R. P. Silva, and G. A. J.
Amaratunga, Phys. Rev. B 62, 12628 ?2000?.
35S. E. Rodil, W. I. Milne, J. Robertson, and L. M. Brown, Appl. Phys. Lett.
77, 1458 ?2000?.
36A. C. Ferrari, A. Libassi, B. K. Tanner, V. Stolojan, J. Yuan, L. M. Brown,
S. E. Rodil, B. Kleinsorge, and J. Robertson, Phys. Rev. B 62, 11089
?2000?.
37A. C. Ferrari and J. Robertson, Phys. Rev. B 61, 14095 ?2000?.
38K. W. R. Gilkes, H. S. Sands, D. N. Batchelder, W. I. Milne, and J.
Robertson, J. Non-Cryst. Solids 227–230, 612 ?1998?.
39G. Adamopoulos, J. Robertson, N. A. Morrison, and C. Godet, J. Appl.
Phys. 96, 6348 ?2004?.
40V. I. Merkulov, J. S. Lannin, C. H. Munro, S. A. Asher, V. S. Veerasamy,
and W. I. Milne, Phys. Rev. Lett. 78, 4869 ?1997?.
41S. Prawer, K. W. Nugent, Y. Lifshitz, G. D. Lempert, E. Grossman, J.
Kulik, I. Avigal, and R. Kalish, Diamond Relat. Mater. 5, 433 ?1996?.
42M. Yoshikawa, Mater. Sci. Forum 52–53, 365 ?1090?.
43J. Tauc, R. Grigorov, and A. Vancu, Phys. Status Solidi 15, 627 ?1966?.
44G. E. Jellison and F. A. Modine, Appl. Phys. Lett. 69, 371 ?1996?.
45S. Xu, B. K. Tay, H. S. Tan, L. Zhong, Y. Q. Tu, S. R. P. Silva, and W. I.
Milne, J. Appl. Phys. 79, 7234 ?1996?.
46G. M. Fuge, M. N. R. Ashfold, and S. J. Henley, J. Appl. Phys. 99, 014309
?2006?.
47A. Chambers, Mordan Vacuum Physics ?CRC, Boca Raton, FL, 2005?.
48F. Claeyssens, R. J. Lade, K. N. Rosser, and M. N. R. Ashfold, J. Appl.
Phys. 89, 697 ?2001?.
49A. C. Ferrari, S. E. Rodil, and J. Robertson, Phys. Rev. B 67, 155306
?2003?.
50S. E. Rodil, S. Muhl, S. Maca, and A. C. Ferrari, Thin Solid Films 433,
119 ?2003?.
51J. Schwan, S. Ulrich, T. Theel, H. Roth, H. Ehrhardt, P. Becker, and S. R.
P. Silva, J. Appl. Phys. 82, 6024 ?1997?.
52G. Lazar, K. Zellama, M. Clin, and C. Godet, Appl. Phys. Lett. 85, 6176
?2004?.
53Y. Miyajima, J. M. Shannon, S. J. Henley, V. Stolojan, D. C. Cox, and S.
R. P. Silva, Thin Solid Films 516, 257 ?2007?.
54T. Schwarz-Selinger, A. von Keudell, and W. Jacob, J. Appl. Phys. 86,
3988 ?1999?.
55R. C. Barklie, Diamond Relat. Mater. 10, 174 ?2001?.
56M. A. Monclus, D. C. Cameron, A. K. M. S. Chowdhury, R. Barkley, and
M. Collins, Thin Solid Films 355–356, 79 ?1999?.
57F. J. Clough, W. I. Milne, B. Kleinsorge, J. Robertson, G. A. J. Amara-
tunga, and B. N. Roy, Electron. Lett. 32, 498 ?1996?.
58Y. Hayashi, N. Kamada, T. Soga, and T. Jimbo, Diamond Relat. Mater. 15,
1015 ?2006?.
063701-9 Miyajima et al.J. Appl. Phys. 104, 063701 ?2008?
Downloaded 31 Mar 2009 to 131.227.178.132. Redistribution subject to AIP license or copyright; see http://jap.aip.org/jap/copyright.jsp