FULLERENE-LIKE CARBON NITRIDE –
A NEW THIN FILM MATERIAL
Jörg Neidhardt1,2, Esteban Broitman3 and Lars Hultman2
1)
Christian Doppler Laboratory for Advanced Hard Coatings, Department of
Physical Metallurgy and Materials Testing, University of Leoben, 8700 Leoben,
Austria.
2)
Thin Film Physics Division, Department of Physics (IFM), Linköping University,
58183 Linköping, Sweden.
3)
Department of Chemical Engineering, Carnegie Mellon University, Pittsburgh, PA
15213, USA
1
INTRODUCTION
It is remarkable that a coating material finds large scale application as a new
industrial coating material just a few years after its discovery. That is the very case for
carbon nitride (CNx). As such it has, e.g., replaced Diamond-like Carbon (DLC) as
protective coating on magnetic storage media and the corresponding read and write
heads [1], which is a clear indication for its advantageous properties [2][3].
First depositions of CNx by rf magnetron sputtering were already reported in
the late 70’s by Cuomo et al. [4], but it was not until the early 90’s that wide scientific
attention was drawn to this material by theoretical prediction of the super hard -C3N4
phase by Liu and Cohen [5]. Studies on CNx as hard coating material were therefore
pioneered by researches at Northwestern University [6][7]. Many deposition methods
and techniques have been employed in the quest for the elusive super-hard crystalline
phase of -C3N4 [8]. Despite numerous publications, not much convincing evidence
1
has been shown so far for its synthesis. A few groups, however, claim the synthesis of
nanocrystallites of -C3N4 [9][10] and other crystalline polymorphs [11]. Difficulties
arise primarily from the nitrogen deficiency in the material due to the inherent drive
of nitrogen to form strongly covalent bonded N2, which readily desorbs [12]. In order
to incorporate the required 57% of nitrogen the focus was mainly on high energetic
deposition techniques to provide for pronounced non-equilibrium growth conditions.
These high energies, however, suppressed any structure evolution by displacing film
atoms via knock on collisions. Despite these efforts, the nitrogen content in the
resulting amorphous CNx material rarely surpasses 30 at% (see section 2.3 “Structure
Evolution Mechanisms”) and any nitrogen concentration exceeding this limit seems to
be due to N2 gas trapped in a porous film structure.
If, on the other hand, alternative deposition techniques such as reactive
magnetron sputtering of carbon in a nitrogen-containing atmosphere are employed,
the energy per deposited particle is rather low. This is mainly due to the low ion to
neutral arrival rate ratio inherent to the technique. Therefore, most film-forming
particles carry thermal energies, which allows for the evolution of structured films.
Reactive magnetron sputtering with rather low ion assist energies was employed by
Sjöström et al., which led to the discovery of the fullerene-like allotrope of carbon
nitride (CNx) in Linköping in ’94 [13]. This nano structured CNx (0<x<0.25) thin film
material was found to be much more interesting from a materials research and
applications point of view as its amorphous counterpart. By using nitrogen as a
reactive gas a film structure consisting of bent and frequently intersecting planes
evolves at elevated temperatures, whereas pure carbon films sputtered under the same
conditions result in a porous under-dense soot like material [14].
2
2
FULLERENE-LIKE MATERIALS AND MICROSTRUCTURE
Fullerene-like materials are characterised as compound materials with, by the
effect of the structural incorporation of odd-membered rings, heavily bent and
intersecting hexagonal basal planes. These structures are besides CNx observed in,
e.g., C [15], BN:C [16], BN [17], MoS2 [18], TiO [19], NbS2 [20] and WS2 [21]. Out
of these materials systems only C, CNx and BC:N exhibit cross-linking, which enables
the material to extend the strength of the covalent bonded two-dimensional hexagonal
network into three dimensions.
Figure 1: Plan view high-resolution transmission electron
micrograph of the curved and intersecting fullerene-like sheets in
CNx for a film grown at 450 C with an N2 fraction of 0.5 and a
bias voltage of -25 V. The inset illustrates the buckling of the
graphene sheets by pentagon-incorporation obtained from abinitio calculations [22].
Fullerene-like CNx consists of, upon substitution of nitrogen for carbon, bent,
intersecting and cross-linked graphite sheets, as seen in the high-resolution
transmission electron microscopy plan-view image in Figure 1. This implies that the
majority of carbon is incorporated in a three-coordinated planar sp2-hybridised state.
The double bond between sp2-hybridised carbon as in graphite, is with a bond energy
3
of 6.3 eV compared to 3.6 eV, actually stronger than the single bond sp3-hybride bond
of diamond [23]. However, the overall strength of graphite is limited by its 2dimensional structure of hexagonal planes and, thus, the rather weak inter-planar vander-Waals-type bonding. Hence, graphite deforms easily by gliding of basal planes. If
on the other hand nitrogen is incorporated during growth the energy required to form
pentagons is considerably lowered and thus the graphitic sheets curve and buckle
(inset Figure 1) [24]. This enables the material to extend the extraordinary strength of
a planar sp2-coordinated carbon network in three dimensions.
The nitrogen-induced alleged cross-linkage between the sheets contributes
considerably to the strength of fullerene-like CNx by preventing inter-planar slip (see
next section 2.2 "The Role of Nitrogen” for a more detailed description). This results
in an extremely fracture tough, elastic and compliant material, which deforms by
reversible bond rotation and bond angle deflection rather than slip and bond breaking
[3]. Diamond-like carbon films for instance exhibit also properties beneficial for
many applications, such as high hardness due to a high fraction of sp3 carbon bonding
and favourable tribological behaviour. However, contrary to fullerene-like CNx, the
application of DLC films suffers mostly from the inherently higher brittleness of the
material.
Depending on the growth conditions, fullerene-like CNx thin films exhibit a
large variation in basal plane curvature and, hence, extension and alignment of sheets.
Thus, due to the variation in microstructure it denotes a material with a large spread in
properties, e.g., elastic modulus, elasticity and electrical resistance [14][25][26], all
having in common an extraordinary mechanical strength. This provides for tailoring
of the properties to given applications as, e.g., wear protective coatings.
4
2.1
Bonding Configuration
Fullerene-like CNx is a compound material where nitrogen is incorporated in a
carbon matrix. The effect of impurities such as hydrogen might also be of importance,
as it is known to improve the properties of, e.g., DLC films [27]. However, the very
low concentrations of impurities such as hydrogen and oxygen in the magnetron
sputtered CNx films should not have any significant effect upon the film structure. On
the other hand Hellgren et al. studied the effect of intentional hydrogen addition to the
discharge during CNx deposition [28]. It was shown that hydrogen terminates
potential bonding sites for CN precursors and, thus, the formation of a fullerene-like
structure is hindered. Additionally, hydrogen is enhancing the etching process by
formation of, not only volatile C2N2 molecules (see next section 2.2 "The Role of
Nitrogen”), but also NH3 and volatile CxHy species.
Carbon is an excellent constituent in the synthesis of hard compounds, since it
fulfils both requirements of a small atomic size and the ability to form strong bonds.
The four valence electrons of carbon ([He] 2s2 2p2) are able to form three
hybridisation states enabling three major bond configurations; the sp3 tetrahedral
bonds like in diamond, the sp2 planar graphite-like bonds and the sp bonds like in
alkynes. In Figure 2 it is depicted that for sp3-hybridised carbon, all four electrons are
able to form -bonds pointing towards the corners of a tetrahedron enabling for a
strong 3-dimensional network. In a sp2-bonded network, only three electrons form
three planar bonds, while the fourth forms a -orbital perpendicular to the plane of the
three -bonds. The electron in the -orbital can be delocalised to strengthen the bonds
by forming a resonance structure of single and double bonds in an appropriate matrix.
Sp-hybridised carbon, for instance, forms two linear, but very strong bonds due to the
5
possible overlap of two orthogonal -orbitals. The strength of carbon compounds is
easily envisioned by the example of diamond, still the hardest material known, due to
the strong carbon-carbon bond, the high three-dimensional coordination and the small
atomic size. However, the overlap of the -orbitals for the sp and sp2 carbon-carbon
bonds, commonly referred to as double and triple bonds, results in bond energies of
6.3 and 8.7 eV, which are much higher bond energy compared to 3.6 eV for the single
bond as in the sp3 hybrid. This gives graphite and polymers their extraordinary inplane/chain strength.
sp3
sp2
sp
Figure 2: Schematic of the three different valence electron
hybridisation states of carbon; sp3 as in diamond, sp2 as in
graphite and alkenes and sp as for alkynes. The elongated orbitals
represent the hybrids able to form -type molecular bonds, while
the smaller round orbitals are the remaining 2p electron(s) that
can form -type bonds.
Nitrogen, like carbon, predominantly forms covalent bonds. Its valence
electrons ([He] 2s2 2p3) exhibit, despite the additional electron, bond hybridisations
similar to carbon. However, the extra valence electron implies some differences. In
case of sp3-hybridised nitrogen, four hybrid orbitals are formed just like in carbon,
while the extra electron forms a non-bonding lone pair with an electron in a hybrid
orbital leaving only three -bonding orbitals (Figure 3). The bond angle is slightly
deflected to 109° instead of 107° compared to diamond due to the effect of the more
6
spacious lone pair. There are two possibilities for the sp2-hybridised bond
configuration. First, the extra electron is able to form a non-bonding lone pair with a
sp-hybrid orbital leaving the atom with two non-linear -bonds and the possibility to
participate in -resonance structures with the electron in the remaining 2p orbital. For
the second case, the extra electron forms the lone pair with the remaining 2p orbital
leaving the atom with three planar -bonds and no possibility to contribute to a resonance structure, due to the filled, non-bonding 2p orbital. Sp-hybridisation of
nitrogen is also common, whereas the atom can form one  and two -bonds, while
the 5th valence electron forms a lone-pair with the second sp-hybrid orbital.
sp3
sp2
sp2
..
N
..
N
N
H H H
Figure 3: Schematic of different electron hybridisation states for
nitrogen; sp3 like in ammonia, sp2 two-fold coordinated like in
pyridine and 3-fold coordinated as in a substitutional graphite site.
The difference in coordination is due to the location of the nonbonding electron pair.
Due to the inherently disordered structure of CNx compounds, it has been
difficult to determine the exact bonding structure of carbon and nitrogen in the
material. Thus, complementary spectroscopic techniques have to be used to
investigate the local bonding environment of fullerene-like CNx. It has been shown
that the bonding structure can be readily analysed by X-ray photoelectron
7
spectroscopy (XPS) [29][30]. The N1s core level peak has typically two strong
components (Figure 4) P1 at 400.7 eV and P2 at 398 – 399 eV, and at least two
smaller ones P3 positioned at 402 eV and P4 at 399.5 eV. The assignment of these
peaks has been widely divergent in literature [31]. However, strong support can be
found for the following peak interpretation by comparing the XPS results of a wide
range of films with other spectroscopic techniques, such as x-ray absorption/emission
spectroscopy (XANES, XES) [32], nuclear magnetic resonance spectroscopy (NMR)
[33], electron energy loss spectroscopy (EELS) [34] and theoretical calculations [32].

P1 (400.7 eV) corresponds to 3-fold coordinated sp2-hybridised nitrogen bonded
to three carbon atoms, substitutional in a predominantly sp2-coordinated graphite
network (see inset Figure 4), while a sp3-hybridised carbon as neighbour cannot be
ruled out. N-H bonding originating from reactions with ambient air might also
have contributions in this region.

P2 (398.0 – 399.0 eV) has initially been assigned to nitrogen bonded to an at least
partially sp3-hybridised carbon matrix, but has rather to be seen as nitrogen in a
pyridine-like structure (see inset Figure 4), where a 2-fold coordinated sp2hybridised nitrogen forms two bonds to an otherwise graphitic structure. This
bonding configuration may occur along edges of planes or next to defects and is
essential for eventual cross-linking.

P3 (402.5 eV) is normally assigned to N-O bonds formed after air exposure [31].
However, recent results from in-situ XPS in a UHV environment show no such
correlation. Instead P3 scales with the intensity of P1, suggesting that it originates
from a similar substitutional graphite structure, where nitrogen donates an electron
8
to the structure leaving it with a virtual positive charge and thus increasing the
binding energy [35].

P4 (399.5 eV) is a minor contribution and most likely corresponds to sphybridised nitrogen in a nitrile structure. Raman spectroscopy is a more suited
technique for the detection of residual nitriles [36] and has indicated that a minor
fraction is present in fullerene-like CNx films. N-H bonding configurations due to
contamination or surface reactions may also contribute in this region [35].
Normalized Intensity (arb. units)
600 ºC
P1
P3
P2
P4
150 ºC
P1
P3
450 ºC
300 ºC
P2
0 ºC
1
0.5
0.16
0
-100 ºC
N
N
••
404 402 400
398 396
404 402 400
398 396
Binding Energy (eV)
Figure 4: XPS N1s photoelectron spectra of CNx films synthesised
at various N2 fractions, a bias voltage of –25 V and substrate
temperatures from 600 down to –100 C. P1-P4 indicate the 3 to 4
spectral contributions. The insets illustrate the associated major
bonding configurations (figure taken from reference [30]).
In reference [30] it is shown by combining XPS and HRTEM (SAED) that the
N1s core level P1/P2 intensity ratio reflects the structural arrangement of CNx films.
A decreasing extension of the fullerene-like sheets implies a decreasing number of
9
nitrogen in a substitutional site (P1) and in turn more nitrogen bonded along edges
and defects (P2). Hence, a pronounced fullerene-like microstructure consisting of
extended planar domains is characterised by a large P1/P2 ratio, where the peaks are
well separated (2 eV) and P1 is dominating the spectra. If, on the other hand, the
fullerene-like structure is disrupted the P1/P2 ratio is decreasing.
Detailed analysis based on the XPS C1s photoelectron peak is impractical,
since it is typically broad and its contributions cannot by extracted separately. Still,
the C1s peak is observed to broaden towards higher binding energies as soon as
nitrogen is incorporated. This is consistent with the above-mentioned distribution in
bonding environments and the broadening is correspondingly more pronounced at
lower growth temperatures [34].
The information extracted from XPS and XANES (NEXAFS) are in many
aspects complementary, while the latter assesses the electronic structure of a
compound at a much superior resolution. Using this technique, the bonding structure
of CNx - in particular the fullerene-like allotrope - was analysed in detail by Gago et
al. [37]. As seen in Figure 5 the C1s spectra consists of two broad major peaks around
284 and 290 eV corresponding to 1s  * and 1s  * transitions, respectively. The
intensity of the * peak and the position of the * absorption edge indicate the
dominance of sp2 hybrids, as compared to the reference spectra of graphite and
diamond. The states below the graphite peak (G) are attributed to defects in the
graphitic network [38], while the formation of various CN bonds is revealed by peaks
in the 290-294 eV region. The small (C2) peak indicates the existence of a minor
fraction of nitrile bonding configurations regardless of the substrate temperature [39].
This is not surprising given the large number of preformed CN species in the
deposition flux (see next chapter). The graphitisation occurring at higher temperatures
10
can be seen by an increasing (G) peak. A complete deconvolution and peak
assignment for the C1s spectra is, as for XPS, very complex due to the multitude of
XANES Intensity (arb. units)
bonding configurations.
*
 states
*
N2 N3
N1
 states
C2
723 K
723 K
573 K
573 K
423 K
423 K
273 K
273 K
Diamond
cBN
143 K
143 K
Graphite G
280
N4
285
hBN
C1s
290
295
395
400
N1s
405
410
Photon Energy (eV)
Figure 5: XANES spectra at the C1s and N1s edge for CNx films
grown at a N2 fraction of 0.16, -25 V bias and various substrate
temperatures. The spectra of graphite and diamond as well as
hexagonal and cubic BN are shown as a reference for sp2 and sp3
hybridisations (figure taken from reference Error! Reference
source not found.).
Similar to XPS, more direct information can be extracted from the N1s
spectra, where four distinctively different peaks in the * can be resolved (Figure 5).
Nitrogen also seems to be predominantly sp2-hybridised as indicated by the
pronounced * region. The peaks N1, N2 and N4 are assigned to nitrogen bonded in a
pyridine, nitrile and graphite-like configuration (comparable to P2, P4 and P1 in the
XPS spectra), whereas N3 is still under discussion.
The development of the peak N2 as a function of the deposition temperature as
analyzed by XANES (see Figure 5) indicates that the nitrile configuration CN is
most pronounced at lower temperatures. This is due to the fact that the triple bond,
11
present in most of the preformed CN species in the flux, stays intact upon partial nonaligned incorporation due to a decreased reactivity at lower temperatures (see section
2.3 “Structure Evolution Mechanisms”). At higher temperature, however, these
species are incorporated along edges of sheets and thus the electron of the triple bond
forms a third bond or a lone pair, as it is the case for a graphite or pyridine-like
configuration. In addition, the remaining nitrile groups react to form C2N2, which
desorbs by the chemical desorption process described above.
In congruence with XPS, the incorporation of nitrogen in a graphite-like
configuration (N4) is preferred at elevated temperatures and lower N2 fractions. This
alongside with lower number of nitrogen incorporated in a pyridine (N1) or nitrile
(N2) configuration is the fingerprint of an evolving fullerene-like microstructure,
which is even more pronounced at lower N2 fractions. However, these bonding states
are still present, while especially the pyridine-like configuration might play an
important role for inter-planar cross-linking.
2.2
The Role of Nitrogen
The substitution of nitrogen for carbon in the graphite basal planes affects both
curvature of planes and their reactivity, which is one of the key processes for the
formation of fullerene-like CNx solid films at elevated temperatures. For comparison,
pure carbon films grown under the same conditions result in an amorphous underdense structure consisting of aromatic disordered fragments. It is only upon
introducing of even small fractions of nitrogen in the discharge that dense and
structured films form. Three major roles were assigned to nitrogen and will be
described in the following sections.
12
Evolution of Curvature
The first role is that nitrogen incorporation induces curvature. This is evident
from the fingerprint-like appearance of basal planes of a fullerene-like CNx film in
Figure 1. The highly corrugated and curved basal planes imply a radius of curvature in
the nanometer range. This curvature is induced for all nitrogen fractions of a N2/Ar
magnetron sputter discharge, while its degree can be controlled by the N2 fraction
[14]. Thus, nitrogen plays an essential role for inducing spherical curvature. For pure
carbon structures curvature is known to be due to the incorporation of odd-member
rings in an otherwise hexagonal structure. As, e.g., 12 pentagons provide for the
spherical appearance of closed caged carbon fullerenes, such as C60. Experimental
results as well as calculations actually suggest that iso-structures to C60 might exist in
fullerene-like CNx thin solid films [40]. These so called nano-onions (Figure 6) form
under certain deposition conditions, such as elevated temperatures (Ts = 450 °C) and
low N2 fractions (0.1 – 0.16).
5nm
Figure 6: Spatially resolved electron energy loss spectroscopy
showing the CCD counts and relative nitrogen concentration
scanned across a CNx nano onion as indicated in the respective
high-resolution electron transmission microscopy planview image.
The inset shows the structure of C48N12 believed to exist in the
centre of the nano onion [40].
13
These experimental findings were, furthermore, verified by total energy
calculations, which show that nitrogen stabilizes pentagons in aromatic carbon
structures [24]. However, the structures are shown to be less stable if more than one
carbon is substituted in a ring and/or nitrogen is incorporated next to each other,
which from geometrical conditions explains the apparent saturation in maximum
nitrogen concentration of approximately 20-25 at%.
In summary, both experiments and theoretical calculations suggest that
nitrogen triggers the evolution of curvature by stabilizing the required pentagons at
much lower energies as compared to pure carbon structures, while the incorporation
of more than one nitrogen atom in a carbon ring structure seems to be energetically
less favourable [24].
Formation of Inter-planar Cross-linking
Another role of nitrogen is the promotion of inter-planar cross-linking. The
combination of mechanical properties, such as elasticity and hardness of fullerene-like
carbon nitride is superior to most carbon-only materials (see section 3.1 “Indentation
Hardness and Modulus”). The mechanical strength of graphite, for instance, is limited
by its easily activated slip system along the graphite basal planes. On the other hand,
this weakening inter-planar slip also holds the key for the extremely low friction.
Carbon fullerene cages, instead, are known to be extremely resilient and to withstand
severe deformation without any plastic yield via broken carbon-carbon bonds [41].
However, the mechanical properties of crystalline compounds formed from fullerenes
- so called fullerites - still suffer from the rather weak inter-molecular bonding. It has
also been shown that temperature treatment can initiate inter-molecular covalent
cross-linking or polymerisation, which improves the mechanical strength [42][43].
14
The other extreme of hard carbon-only materials is represented by diamond, where the
highest degree of three-dimensional cross-linking is provided by a 100 % sp3
hybridisation. Also diamond-like carbon thin films synthesised by various techniques
exhibit a significant fraction of sp3-hybridised carbon that give the material an
outstanding hardness, while still retaining favourable low friction properties.
However, these super hard materials have only a limited tolerance to deformation and
overloading results in imminent brittle fracture and, thus, mechanical failure of the
substrate-coating system.
Fullerene-like CNx compared to common super-hard thin film materials (cBN, DLC) has a low to moderate resistance to penetration (hardness), as probed by
nano-indentation [3]. Still, the indentation response surpasses crystalline graphite by
far, which implies the existence of strong inter-planar cross-links or interlocking of
sheets. The most striking property, however, is its outstanding resistance to plastic
deformation (see section 3.1 “Indentation Hardness and Modulus” for details). In fact,
fullerene-like CNx deforms predominantly elastically, which combined with its
compliance results in a resilient and fracture tough material [3][14]. Nitrogen plays a
crucial role, since pure carbon materials grown under similar conditions have a sootlike appearance, which is in conjunction with a deficient cohesion and strength.
The precise nature of the cross-linking mechanisms is, however, still under
discussion. A first explanation for cross-linking mechanisms was established by total
energy calculations performed in reference [24]. The presented results suggest that,
upon nitrogen-incorporation in a hexagon, it is for the adjacent carbon atom
energetically more favourable to change from an sp2 to an sp3-hybridised state, which
could provide for the required out of plane bond. For that reason, nitrogen-doped
hetero fullerenes also known as aza-fullerenes, e.g. C59N instantaneously dimerise
15
upon synthesis via a carbon-carbon bond next to the substituted nitrogen atom [44].
Experimental results, however, indicate that fullerene-like CNx is a predominantly sp2
bonded [45][46] and the existence of significant numbers of sp3-hybridised carbon
remains to be affirmed. On the other hand only a few inter planar cross-links are
needed to increase the stiffness significantly and any low fraction of sp3-hybridised
carbon might pass undetected due to experimental limitations of, e.g., NMR.
Cross-linking at the nitrogen site on the other hand is rather unlikely. Firstly, it
was shown by total energy calculations in reference [24] that the structure is more
stable if two incorporated nitrogen atoms in a graphite fragment are separated by at
least two carbon atoms. Secondly, chemical synthesis of, e.g., aza-fullerene dimmers
indicates that the cross-linking is initiated at the carbon next to an incorporated
nitrogen atom and not at the nitrogen site itself [44].
The high stiffness of fullerene-like carbon nitride, in contrast to planar
graphite, can also be explained by the 3-D interlocking of curved meandering
fullerene-like sheets. In addition, purely sp2-based cross-linking mechanisms might
also be present, such as branching of sheets and/or carbon-carbon bond rotation as
suggested in references [15][33][45][47]. For the latter, the electron lone pair in the 2p
orbital of substitutional incorporated nitrogen or the formation of pentagons disturbs
locally the -resonance structure. Therefore, the double bond resonance structure
between two neighbouring carbon atoms is disrupted and, thus, the carbon-carbon
bond is free to rotate due to a missing 2p-2p orbital coupling. This might result in an
out-of-plane rotation of the subsequently growing structures and hence enables crosslinking leading to a multitude of possible configurations. This is most likely if one
considers also the disruption of the continuous sheet-structure by the incorporation of
16
two-coordinated, pyridine-like bonded nitrogen that is present in significant numbers
as suggested by spectroscopy-based techniques.
Figure 7: Stable cross-linked structures with cross-linking bridge
(enlarged in the inset).
Most recent computational studies corroborated the existence of out-of-plane
bond deflection and rotation in the presence of multi-atomic precursors [48]. Figure 7
exemplifies two of the energetically stable structures exhibiting cross-linking obtained
by simulated (synthetic) growth.
(a)
(b)
(c)
(d)
2
Figure 8: Alternative sp -based cross-linking mechanisms via a
nitrogen-induced vacancy (a) and pentagon formation (b), which
results in out-of-plane bending (c) and finally provides a
nucleation site for out-of-plane growth (d) [33]. The grey shaded
atoms represent carbon and the black nitrogen; the hydrogen
(white) atoms are terminating the fullerene-like fragment.
17
Another potential sp2-based cross-linking mechanism is branching of
fullerene-like sheets. It was shown in section 2.1 that nitrogen is bonded primarily in
two different configurations; substitutional at carbon sites, which does not affect
continuous planar growth or in a pyridine configuration, where one of the three bonds is terminated by the electron lone-pair. The latter implies the existence of
vacancy defects as depicted in Figure 8 (a). Additionally, nitrogen induces pentagon
formation, which provides for spherical curvature at the vacancy site (b). This
presented configuration results in out-of-plane nucleation sites (c), where threedimensional growth can be initiated (d). These ideas were confirmed by initial
calculations [33].
Formation of CxNy (x,y  2) species
Carbon Nitride thin films are commonly grown by plasma-activated deposition
techniques, where carbon is prone to react with activated nitrogen originating from the
discharge. This reactivity of nitrogen is considered to be the dominant compound
forming reaction on surfaces for most PVD nitride films, due to the rather low
probability of gas phase reactions under the sub-atmospheric pressures. However, the
experimentally established existence of a low-mass gaseous compound in the carbonnitride system, namely C2N2 [49], adds numerous degrees of freedom to the formation
mechanisms of fullerene-like CNx films as compared to other nitrides forming from
single-atomic precursors.
As a consequence the deposition flux from a carbon target sputtered in a N2containing atmosphere consists to a large extent of multi atomic species. They form
via reaction between the carbon atoms in the target and the vast number of nitrogen
ions implanted into the top surface layers during sputtering [49]. In turn C2N2 and
18
fragments thereof are emitted whereas their number scales with the nitrogen fraction
in the discharge [49].
The presence of these nitrogen-containing species in the deposition flux leads
to various effects also on the substrate site. One important process is that they
partially react upon adsorption to form volatile gases such as N2 and C2N2, which
subsequently desorb. This results in a film-surface etching commonly referred to as
chemical sputtering [50]. This process is observed for various CNx deposition
techniques especially those utilizing hyper-thermal species to assist the growth
[51][52][53][54]. It was first established for ion beam assisted deposition, where a
deficiency of carbon and nitrogen in the film was detected compared to the calibrated
fluxes of both elements [55]. The extent of the chemical desorption is determined by
the reaction probability to form a volatile gas and, thus, depends on type and mobility
of adsorbed species. The overall process can therefore be seen as a preferential
etching of less favourably bonded adsorbates and, thus, results in dense and smooth
films. Prevailing CN precursors bonded in an aligned manner along edges of sheets
are more resilient to concurrent etching and therefore play an important role for the
evolution of fullerene-like planes (see next chapter). The chemical desorption process
also accounts for the apparent maximum of incorporated nitrogen at 25at%, due to
its higher desorption probability.
2.3
Structure Evolution Mechanisms
For magnetron sputtering, the presence of nitrogen in the discharge gas leads
to complex chemical reactions at the target and substrate surface [45]. The intense
high-energy ion bombardment at the target results in sputtered target constituents,
whereas most of the nitrogen ions are implanted during the process. The implanted
19
nitrogen radicals react readily with the carbon atoms in the target. Thus, the
composition of the target surface is altered substantially, especially if one considers
the initially rather low sputter yield of pure C. Hence, predominantly CxNy (x,y  2)
molecules are emitted from such a dynamically self-modified target surface, due to
the relatively high strength of the CN bond, alongside with desorbing gases like N2
and C2N2. The extent of these processes scale with the availability of N2 in the
discharge and therefore lead to a strong increase in flux of preformed CN species at
higher N2 fractions, while the carbon flux only plays a minor role [49].
Although low ion assist energies (20-50 eV) are generally beneficial for film
growth by enhancing ad-atom mobility at relatively low substrate temperatures, it has
been shown that they are not the dominant structure-determining factor in case of
fullerene-like CNx films. Instead the flux and type of preformed CN species carrying
only thermal energy is crucial for the evolution of fullerene-like structures. These
molecules might be stabilized against desorption in an evolving sheet by multiple
bonding on the C and N side upon aligned incorporation and, thus, promoting the
evolution of planar structures. Gueorguiev et al. has corroborated the energetic
advantage of this process using step-by-step ab-initio approaches in reference [56].
Furthermore, the formation of C2N2 as volatile gas is also active on the substrate site
and provides for selective etching of misaligned molecules by reaction with adsorbed
CN and their subsequent desorption. This process accounts for the apparent saturation
in nitrogen content at approximately 25 at% nitrogen incorporated into the films,
which is in congruence with the theoretical established most stable structures
[40][48]. The incorporated nitrogen seems to originate predominantly from the
already bonded nitrogen in the preformed molecules arriving at the substrate, while
the extensive low energetic nitrogen ion bombardment from the discharge plasma
20
readily reacts upon adsorption to form volatile N2. Thus, the evolution of the
fullerene-like microstructure has to be seen as a delicate interplay of adsorbing
preformed molecules, which are either incorporated in a stabilized and, thus, aligned
manner or desorb via the formation of C2N2 if they are less favorably bonded [45].
These requirements of low to moderate energy input during growth via ion
bombardment or substrate heating and in particular the existence of a predominant
number of preformed molecules in the film forming flux are not only inherent to
reactive magnetron sputtering, but can also be found for pulsed laser deposition [54].
Both techniques result in fullerene-like CNx films, whereas the structured domains are
much larger for magnetron sputtered films. On the other hand, if no preformed species
are present in the deposition flux no structure evolution is observed as it is the case for
low-energy ion beam assisted deposition [57]. For magnetron sputtering the growth
parameters such as substrate temperature, N2 fraction and bias voltage strongly affect
the fullerene-like structure evolution in terms of extension, alignment and crosslinking. All of the mentioned parameters primarily concern the magnitude of the
chemical desorption and, thus, the selectivity incorporation sites either by influencing
the ad-atom (molecule) mobility (bias, temperature) or by changing the number and
type of film forming species (N2 fraction).
The general trends in terms of deposition parameters and resulting structures
are summarized in Figure 9. According to HRTEM [30] and XPS (Figure 4), films
grown at higher temperatures (Ts > 300°C) and intermediate bias (Vs  -25 V) are all
fullerene-like, since both parameters provide sufficient selectivity to incorporate most
of the arriving species in an aligned fashion, regardless the N2 fraction. The
availability of N2, however, determines the extension and alignment of the sheets,
since the dominance of preformed species at higher N2 fractions promotes the growth
21
of extended, well-aligned sheets resulting in an almost graphite like appearance. At
lower N2 fractions the flux of carbon atoms becomes relevant instead and the
complexity and total number of the preformed species is decreasing, which in turn
leads to shorter heavily bent fullerene-like fragments. This is an effect of a lower
selectivity of incorporation sites via more complex reaction pathways for the
formation of C2N2.
Figure 9: Structure zone diagrams of structures and properties
observed for magnetron sputtered fullerene-like CNx thin films
depending on the substrate temperature, the degree of ion
bombardment, comprising the ion to neutral ratio and ion energy
and the N2 fraction in the discharge. The HRTEM plan-view
inserts should be seen as representatives only (taken from
reference [45]).
22
If the temperature is decreased (T < 300 °C) the ad-atom (molecule) mobility
is reduced and an increasing number of preformed molecules are incorporated in an
out-of-plane orientation due to a decreased efficiency of the chemical desorption
process. This results in the amorphisation of the otherwise graphitic nature of the
films grown at higher N2 fractions. The smaller number of species arriving at lower
N2 fractions is, however, still effectively incorporated in an aligned fashion into short
fragments down to a deposition temperature of 150 °C.
A higher bias or ion energy, exceeding the displacement energy of carbon (E >
30 eV), on the other hand induces an increasing number of defects into the growing
sheets and thus decreases the fullerene-like fragment size. This is most pronounced at
higher N2 fractions, most likely due to a high sensitivity of the aligned incorporation
process towards energy input (Figure 9).
3
3.1
MECHANICAL AND TRIBOLOGICAL PROPERTIES
Indentation Hardness and Modulus
The mechanical properties, such as hardness and modulus, of macroscopic
materials are easily determined by, e.g., tensile testing and purely plastic indentation
methods [58]. These properties are also of vital importance for thin film materials.
Their evaluation is, however, not straight forward mainly due to limitations in lateral
dimensions and the rather small volume of the analysed thin film material. The
problem is partially overcome by depth sensing indentation techniques, where load
and displacement of a sharp indenter penetrating a thin film on a substrate can be
recorded on µN and nm scale [59]. In the past, however, the extreme elasticity of
fullerene-like CNx made the straightforward application of this technique rather
23
ambiguous in terms of quantification [60][61]. In order to overcome these problems,
the elastic-plastic behaviour of the material was investigated in detail in reference [3].
The so gained understanding allowed for the numerical evaluation of fullerene-like
CNx by novel adaptation of existing techniques [14]. However, these methods could
not be applied to the extremely thin films and, thus, purely elastic indents
accompanied by Hertzian modelling had to be employed [26].
Load (  N)
500
TiN
CN0.18
E=350 GPa
H=25 GPa
E=120 GPa
H=18 GPa
-SiO2
CN0.13
E=70 GPa E=45 GPa
H=10 GPa H= 8 GPa
400
300
200
100
0
0
20
40
60
Displacement (nm)
80
Figure 10: Load-displacement curves of two CNx films grown at –
25 V bias (CN0.13), –40 V bias (CN0.18) and at a N2 fraction of 0.5
in comparison to TiN and fused silica (a-SiO2) as reference
materials (taken from reference [35]).
Fullerene-like CNx denotes a new class of super elastic materials, due to its
unique microstructure of bent and cross-linked basal planes, which extends the
extraordinary strength of the a sp2-hybridised planar network in three dimensions. The
detailed studies mentioned above provided for the analytical tool to study the
correlation between the microstructure and mechanical properties in reference [14].
Here, the influence of the growth conditions, such as N2 fraction and ion energy, were
investigated and it was shown that the resistance to penetration (hardness) can be
24
varied over a wide range by increasing the number of cross-linking sites between the
fullerene-like sheets by, e.g., increasing the ion energy.
Figure 10 shows two typical examples of load-displacement curves for CNx
films grown at a N2 fraction of 0.5, but for two different bias voltages. CN0.18 grown
at -40 V bias has a higher degree of cross-linkage and therefore exhibits a higher
resistance to penetration (smaller maximum displacement) while the more graphitic
like appearance of CN0.14 with its extended, well-aligned sheets (Figure 9) has a fairly
low resistance to penetration, due to the weak inter-planar interactions via cross-links.
These differences also manifest itself in lower modulus and hardness numbers.
Remarkably, both films, despite their different compliance, show no or only very little
tendency to plastic deformation and the indentation energy is exclusively stored
elastically and released upon unload. This high elasticity is readily appreciated by
comparing the residual displacement of the load displacement curves of fullerene-like
CNx to depicted reference materials such as TiN and fused silica (a-SiO2).
Ts (C)
150
450
N2
fraction
Hardness (GPa)
-10 V
-25 V
-40 V
Modulus (GPa)
-10 V
-25 V
0
7
115
0.16
15
166
1
10
91
0
0.3
11
0.16
10
18
6
12
1
5
43
-40 V
75
120
50
135
Table 1: Hardness and Modulus values determined by elasticplastic indentations with a diamond indenter of cube corner
geometry for various growth conditions. Films considered
fullerene-like are underlined.
Similar tendencies as for the dependence of H and E on the ion energy are
observed for a change in substrate temperature (Table 1). Both growth parameters
basically influence the number of defects and, thus, the number of potential cross-
25
linking sites. A lower substrate temperature increases the H and E significantly;
presumably by increasing the number of cross-linking sites via incorporation of out-of
plane aligned preformed species (see previous section). However, this increase in
cross-linking sites goes hand in hand with a decrease in extension of fullerene-like
planes. This implies restrictions as for elastic buckling and compression of inter
planar spacing. Thus, the more resistant films grown at higher bias and lower
temperatures commonly exhibit a reduced relative elastic recovery [14].
Furthermore, the microstructure and, thus, the properties of fullerene-like CNx
are also influenced by the N2 fraction, see Figure 9 and reference [14]. A pronounced
fullerene-like structure as synthesized at lower N2 fraction (0.16) and a moderate bias
(-25 V) consists of heavily bent sheets (Figure 1), which by curling restrict their size
and extension as compared to more elongated sheet structures grown at higher N2
fractions. This implies a higher cross-linking density and thus a higher resistance to
indentation, but lower relative elastic recovery.
The detailed analysis of the mechanical response of fullerene-like CNx
characterizes a new class of extremely resilient materials, where deformation energy
is predominantly stored elastically. This can be understood by the cross-linked planar
structure of fullerene-like CNx, where the sheet-like domains wrap and buckle under
load, while the presumed strong C-C cross-links prevent gliding. Thus, plastic
deformation by mechanisms similar to graphite is inhibited. Additionally, the interplanar strength of a predominantly sp2-hybridised material is considerable, which in
turn makes plastic yielding by cracking or bond breaking less likely. On the other
hand, the rather weak intra-planar van-der-Waals forces and the large lattice spacing
provides for a high compressibility, which explains the observed low to moderate
resistance to penetration. Therefore fullerene-like CNx deforms by bond angle
26
deflection rather than breaking providing for the compression of the intra-planar
lattice spacing, much like in a “super hard rubber”. A similar behaviour is also
inherent to C60 Fullerenes, which are known to withstand extreme deformation down
to 20% of their original size without yielding [41]. However, the strength in fullerene
solids (fullerites) is limited due to the weak cross-linking between the molecules. This
is not the case for fullerene-like CNx where the curved sheets extent throughout the
material providing for its elasticity on a macroscopic scale.
General Comments on the Concept of Hardness
According to reference [62] “Hardness is not difficult to measure, it is difficult
to define”. In a more general sense, hardness is defined as resistance of a material to
deformation under external load, which permanently affects the surface by plastic
deformation. Thus, the conventional hardness concept has to be applied with great
care for extremely elastic materials, such as fullerene-like CNx. For instance, TiN is
perceived as a hard material with a low tendency to plastic deformation, due to its
much higher hardness number compared to CNx (Figure 10). Contrarily to TiN, CNx
leaves no residual imprint and can therefore be considered much tougher, since it
exhibits a lower tendency to plastic deformation, despite its higher compliance. This
leads to ambiguities in the common understanding of “hardness” and, therefore, the
indentation response of fullerene-like CNx is best seen, as the one of a super hard
elastomer, whereas the hardness number becomes less meaningful [3].
3.2
Tribological Properties
Tribology can be understood as “the science and technology of interacting
surfaces in motion” in respect to friction and wear [63]. A more recent tribology-
27
related engineering application is to coat surfaces with thin films to improve their
tribological behaviour by combining, e.g., the toughness of a bulk steel substrate with
the low friction and wear properties of, e.g., carbon-based coating materials.
Only a few theories are applicable to predict and interpret the performance of a
tribological pairing. One is elastic-plastic theory, which concerns the corresponding
deformation of surfaces in contact as to their load bearing capabilities [64]. The other
is the wide, but not yet fully understood, field of tribo-chemistry; where the chemical
interactions between two surfaces in a tribological contact are considered. These
reactions might lead to chemically modified counterparts and wear debris. Also the
formation of chemically altered transfer films adherent to at least one of the
counterparts falls in this category [65]. However, the engineering science of tribology
in terms characterisation of, e.g., a substrate coating system remains complex and is
often performed on model systems to mimic a “real life” contact.
Hard and Brittle
Plastic Elastic
Super Elastic
TiN
DLC
*)
Diamond Al2O3
HSS
CN0.18 CN0.13
thin film
thin film
a)
b)
a)
b)
b)
b)
b)
H
(GPa)
80
23
25
9
10
-20
18
7
E
(GPa)
1050
320
350
250
100
-200
120
37
H/E
0.076
0.072
0.071
0.036
 0.1
0.15
0.19
a) taken from reference [66]
b) measured data
*) high speed steel
Table 2: Moduli and hardness numbers of some reference
materials compared to two CNx films grown at 25 eV (CN0.13) and
40 eV (CN0.18) alongside with their H/E ratio. A ratio exceeding
0.15 indicates a super elastic behaviour.
The outstanding mechanical properties, in particular the high compliance,
predict fullerene-like CNx to be a promising candidate as a tribological coating. Its
high hardness-to-modulus ratio is descriptive for a super elasticity (Table 2) and is as
28
such the delimiting factor in the so-called plasticity index describing the compliance
of a material [67]. The extremely high H/E ratio suggests that asperities in a
tribological contact are likely to behave elastic [3][68]. This is of crucial importance;
since microscopic asperities are present for all technological surfaces and as such bear
the load in a tribological contact. Their total area, however, is only a fraction of the
nominal contact area, which results locally in extreme pressures. Under these
conditions, commonly used hard thin films materials are likely to behave in a brittle
manner, which might result in abrasive wear debris. Fullerene-like CNx on the other
hand is highly compliant (see previous section) and, thus, the contact pressure is
dissipated.
Furthermore, the rather low modulus (E < 200 GPa) of fullerene-like CNx
suggests that the macroscopic contact stresses are dissipated over larger volumes
compared to hard coatings with higher moduli (250 GPa < E < 450 GPa). This in turn
reduces stress discontinuities along the film-substrate interface and, thus, prevents
substrate deformation. Additionally, the shear stresses along the interface plane for a
strained substrate-coating system are also reduced, which might be beneficial for the
adhesion of the film under load [69].
Friction and Wear of fullerene-like CNx
Fullerene-like CNx was tested in reference [70] and benchmarked to the
behaviour of amorphous CNx as well as graphitic and diamond like carbon films.
Reciprocal sliding was chosen as contact geometry, where a coated substrate moves
relative to a sapphire counterpart. The transparent sapphire sphere allowed for in-situ
visualization and Raman spectroscopy in order to elucidate the eventual formation of
transfer layers. The tribo-chemistry between two counterparts in respect to the
formation of debris and transfer layers depends strongly on the composition of the
29
surrounding gas and the dissipated energy in the contact [71][72]. The friction
coefficient, defined as the tangential/normal force ratio, was recorded by a lateral
force sensor.
0.4
0.3
ambient
dry
0.3
0.2

ambient
0.2
dry
0.1
0.1
FL 16_150_25
FL 100_450_10
0.0
0
0.4
1000
2000
3000
4000
0.0
0
dry
3000
4000
0.3
ambient
0.2
0.1
ambient
0.2
0.1
FL 100_450_25
0.0
2000
dry
0.3

1000
0.4
0
500
1000
100_150_25
0.0
1500
0
200
400
0.3
0.3
ambient
dry
0_450_25Graphitic
0.2
0.2

600
ambient
0.1
0.1
0_150_25DLC
FL 100_450_40
0.0
0.0
0
1000
2000
3000
Cycles
0
1000
2000
3000
4000
Cycles
Figure 11: Friction curves of fullerene-like (FL) and amorphous
CNx compared to pure carbon films tested in dry and ambient air.
The
nomenclature
as
in
e.g.
100_450_25
reads
[N2(%)_Ts(C)_VB(V)] [70].
The coefficient of friction for fullerene-like CNx for the above-described setup
and under an average contact pressure of 0.7 GPa is with approximately 0.2  0.05
30
slightly higher than the values for pure carbon films (0.15  0.02), while no difference
is observed for amorphous CNx (Figure 11) [70]. These values can be considered low
in comparison to the friction coefficient of, e.g., Si (0.6) and TiN (0.7) under similar
conditions. A low coefficient of friction is desired for a tribological contact not only
because it increases the system effectiveness, but most importantly it also reduces the
thermal energy dissipated in the asperities, which prevents them from, e.g., oxidising
[63]. Furthermore, a dependence on the humidity of the surrounding atmosphere is
observed (Figure 11), while no correlation in between the friction coefficient and
hardness (resistance to penetration) can be detected [70]. This alongside with the
unsteady friction traces as seen in Figure 11 indicates the formation of third bodies in
-1
Wear Rate (10 mm N m )
the tribological contact [73].
-1
carbon (ambient)
a-CNX (ambient)
100
FL-CNX (dry)
-7
3
FL-CNX(ambient)
Dry Air
Ambient Air
10
FL 100_450_40
FL 100_450_25
FL 100_450_10
0_450_25
100_150_25
FL 16_150_25
0_150_25
1
0
2
4
6
8
10 12 14 16
Hardness (GPa)
Figure 12: Wear rates of various fullerene-like (FL) and
amorphous CNx as well as carbon films for sliding in dry and
ambient air. In the right graph the relation to the hardness is
plotted for the three groups of films. The hardness independent
wear rate for ambient sliding of fullerene-like CNx is indicated by
the horizontal line (guide to the eye only) (taken from reference
[45]).
31
The wear rate of fullerene-like CNx is significantly lower compared to
amorphous CNx (100_150_25) and carbon films (0_150_25, 0_450_25) (Figure 12).
Most remarkably the wear rates of fullerene-like CNx in ambient air are almost
constant despite the dramatic difference in hardness ranging from 5 to 15 GPa
depending on the synthesis conditions [70]. For dry conditions, the wear rates are
increasing and scale with the hardness of the films (Figure 12). This indicates that for
ambient sliding either thin transfer layers might have formed or that the reactive
surface sites of CNx were passivated by reactions with ambient moisture. Both effects
result in the separation of the two bulk counterparts and, thus, the wear rate is not
solely based on the mechanical properties of the film anymore. During dry sliding,
however, eventually formed transfer layers break up and the initially low coefficient
of friction is increasing and wear accelerates as seen by the unsteady friction traces
for dry sliding in Figure 11.
4
ELECTRICAL PROPERTIES
We have shown in the precedent section the growing interest in the use of
carbon nitride films for applications requiring hard protective coatings. The potential
for optical and electronic application, however, remains largely unexplored [74].
The study of the electrical properties of carbon nitride films started,
paradoxically, before Liu and Cohen [5] predicted their super-hard properties. The
first studies were motivated by the possible electronic doping of amorphous carbon
films. Jones [75] observed that nitrogen was able to increase the conductivity of aC:H films. However, it was unclear whether N increased the conductivity by doping –
raising the Fermi level towards the conduction band, or by graphitization of the
bonding- narrowing the band gap.
32
For the use of carbon nitride films in electrical devices such as diodes
[76][77], field effect transistors [78], field emitters [79] microsliding switches [80],
and planar waveguides [81], control of the electrical pr operties is very important.
Several experiments have been performed on controlling the electrical resistivity, , in
hydrogenated carbon nitride films (CNxHy) and hydrogen-free carbon nitride films
(CNx) by the variation in the content of nitrogen. However, the nitrogen doping
efficiency is very low and the observed changes in the electrical conductivity can also
been explained from structural changes in the films [74] [82] [83] or by the amount of
H in the films [25] [82]-[84].
The difficulty to correlate the properties of carbon nitride films with their
electrical properties, as determined by the preparation process, stems from the
generally poor definition of all deposition conditions of a specific film, the difficulty
of structural characterization of the materials, and the lack of standardization of
electrical characterization. Films with various nitrogen and/or hydrogen contents
exhibit different physical properties, such as carbon hybridization, refractive index,
hardness, stress, and electrical properties in controlled conditions.
4.1
Electrical properties of hydrogen-free carbon nitride
Although many studies on CNx have reported changes in the resistivity (),
there is only limited knowledge on their electrical properties [76] [85]-[89].
Furthermore, apparently contradictory results have been published regarding the
decrease [74] [76] [78] [85] [87] [91]-[93] or increase [85] [87]-[89] [94]-[96] of 
with the incorporation of nitrogen in the films. Moreover, the reported values for the
resistivity of CNx films span over a wide range, from semimetals with  ~ 10-3 cm
[74] [97] to dielectrics with  ~ 1013 cm [85] [98].
33
The most important factor in determining the electrical properties of
hydrogen-free amorphous carbon films is the sp3/sp2 bonding fraction;  decreases as
the sp3/sp2 ratio increases due to the reduced number of mobile  electrons [98].
Recently, Broitman et al [74] have shown a contradictory result. They have
studied hydrogen-free CNx films with different N content (x). In their films, the
sp3/sp2 C bonding fraction in the CNx films increases slightly with increasing nitrogen
content. However, increasing x from 0 to 0.15 gave an increase of the conductivity
from about 25 to 250 -1 cm-1, and no further change in  was observed for higher x
values (Figure 12).
Figure 12: Electrical conductivity  at room temperature of
hydrogen-free CNx films as a function of nitrogen concentration X.
The line is a guide for the eye only (taken from reference [74]) .
In order to explain this result, Broitman studied the main trends for how the
morphology changes depending on nitrogen content (Figure 13). The film deposited
in pure Ar (Fig. 13a), have a rough granular fracture surface with macroscopic voids
and a RMS roughness of 15 nm. The sample deposited with 5 % of N2 in the gas
mixture (x = 0.15, Fig. 13b) shows a columnar structure with a high degree of
34
alignment and a RMS roughness of 2.5 nm. For the film deposited in pure N2 (x =
0.25, Fig. 13c), a dense, closely packed morphology with a low roughness (RMS =
0.4 nm) is observed. This trend can be explained by a “chemical etching” process that
takes place in the presence of nitrogen (for details see 2.2 “The Role of Nitrogen”).
Atoms not strongly bonded to the growth surface are likely to be desorbed through the
formation of volatile N2 or C2N2 molecules, forcing the formation of a denser
structure.
Figure 13: Cross-sectional SEM micrographs of CNxHy films
deposited on Si at 350 C: (a) C (grown in pure Ar); (b) CN0.15
(95% Ar + 5% N2); (c) CN0.25 (pure N2); and (d) CH0.19 (99% Ar +
1% H2) (taken from reference [25]).
A correlation between electrical properties and morphology can be stated.
From the electrical point of view, (a) voids and (b) surface roughness can be seen as
defects where the electrons suffer scattering processes, thus decreasing the
conductivity of the film. Semimetals can be considered as metals with an abnormally
small number of carriers (electrons and holes), such that it is appropriate to use the
35
Matthiessen’s rule for estimating the film conduction properties [100]. That is, the
total conductivity T of the films will be the result of the individual contributions
from the intrinsic conductivity i and the conductivities a and b from both defects
(a) and (b) in the form
1/T = 1/i + 1/a + 1/b.
(1)
When N is incorporated in the film, both a and b increase due to a lower
surface roughness and a higher film density, screening a possible decrease of i
produced by an increasing number of sp3 bonds.
Figure 14: Temperature dependence of dark conductivity  for
CN0.25 film (taken from reference [74]).
Insight into the conduction mechanism could be obtained from the variation of
the electrical conductivity with sample temperature [74]. Figure 14 shows the
logarithm of the dark conductivity versus reciprocal temperature for a film with x =
0.25. The data points lie on two straight lines of different slopes at low and high
temperature, respectively. The temperature dependence of the conductivity for films
36
containing different N concentrations exhibits rather similar increases with increasing
temperature.
The electrical conductivity dependence with the temperature in Fig. 14 can be
explained through the two-channel model of conduction proposed by Nagels et al.
[101]. According to this model, if two different mechanisms of conduction are present
at low and high temperature, a kink in the conductivity curves will be observed. This
behavior has been reported before in a-C:H:N films [102] [103], but in hydrogen-free
amorphous carbon-nitride films the electrical characterization has been done in most
publications in a low [94] or high temperature region [85] [86] [88] [97]. Only
Monclus et al. [89] have reported the electrical properties of CNx films between 90
and 310 K. However, they did not analyze the conductivity at low temperatures.
The data belonging to samples prepared with 0%, 5%, 20%, and 100% of N2
in the sputtering gas were fit in the high- and low-temperature regions by the
relationship [103]
 = j exp(-Ej/KT),
(2)
where the preexponential j is a temperature-independent term, Ej is the
activation energy for conduction (that is, the difference in energy between the bottom
of the conduction band and the Fermi level) and j= H and L corresponds to the high
and low-temperature regions, respectively. In the high-temperature range, the fits
show an increase of EH from 1.7 meV to approximately 8 meV when the N
concentration increases from 0% to 20%, while in the low-temperature range EL
increases from 0.4 to 2 meV. The values of activation energy are consistent with the
results of Ricci et al. [103].
37
The linearity at high and low temperatures of log[(T)] vs T-1 is indicative of a
thermally activated process [103]. However, a nearly equally good linear fit of the
low-temperature data when plotted log[(T)] vs T-4 (indicative of a variable hopping
conduction mechanism [94] [103]) could be obtained.
A photoconductive effect, giving a ~5% increase in the nominal conductivity,
was observed by Broitman [74] during electrical measurements under light
illumination of 60 W m-2 intensity, however, no correlation with the amount of N in
the films could be established. The presence of photoconductivity in CNx films has
been reported before by Takada et al. [85].
4.2
Electrical properties of hydrogenated carbon nitride
It is well known that the electrical properties of carbon films depend strongly
on their hydrogen content [98]. Incorporation of hydrogen results in terminating sp2bonded clusters, as well as promoting the formation of sp3-bonded networks, both
resulting in a decreased electron mobility and an increase in the number and extent of
voids in the films [82][99]. In the case of CNxHy films, no systematic studies on the
electrical properties have been reported because most films have been deposited by
CVD techniques, where it is difficult to make an independent control of N and H
sources [74]. As a consequence, contradictory results have been published regarding
the possible [75][84][91] or non-existent [92][93] electronic doping of CHY films by
nitrogen incorporation to explain the variation of resistivity. Furthermore, the lack of
control in the background pressure, which often contains H2O and H2, probably gives
unintentionally added hydrogen in many films reported as grown apparently “free” of
hydrogen [105] [106].
38
30
0% N
2
5% N
[H] (at.%)
20
2
100% N
2
10
a)
0
0
5
10
15
20
H fraction (%)
2
 (-cm)
10 3
0% N
10 2
5% N
10 1
100% N
2
2
2
10 0
10 -1
10 -2
b)
10 -3
0
5
10
15
20
H fraction (% )
2
Figure 15: (a) H concentration in the films, and (b) Electrical
resistivity  of CNxHy films as a function of the H2 fraction in the
discharge gas for films grown in Ar/H2, Ar/H2(5%)/H2 (resulting
N/C ratio ~ 0.08-0.15) and N2/H2 (N/C ~ 0.22-0.25) at
temperatures of 350 °C. The lines are for guiding the eye only
(taken from reference [25]).
Broitman [25] correlated the H content in their films, as measured by Nuclear
Reaction Analysis, as a function of the H2 fraction in the gas phase for films grown in
Ar/H2, N2/H2 and Ar/N2(5%)/H2 discharges (Fig 15a). With increasing H2 fraction in
the gas, the film hydrogen concentration initially increases, but saturated when more
than ~5% H2 is added to the discharge gas. The electrical resistivity of the films
follows the same trend, as shown in Fig. 15b. The film that exhibits the largest change
in  (~ four orders of magnitude) is the one grown in Ar/H2 mixtures, which is also
the one that accommodates the highest H content. In general, a good correlation is
39
found between the resistivity and the H content in the film. However,  does not seem
to saturate with increasing H2 fraction in the gas phase as fast as the actual content in
the film. This can probably not be explained by variations in the nitrogen
concentration. The N/C ratio in the films grown in N2/H2 varied in the range 0.220.25, while for the films grown in Ar/H2(5%)/H2, N/C decreased from ~0.15 to ~0.08
as the H2 fraction was increased from 0 to 15%. Instead, this could be explained by a
continuous change of the microstructure to shorter basal planes that become
terminated by H, as previously observed by high-resolution transmission electron
microscopy [28].
Another factor that may play an important role for the CNxHy films, as for the
non-hydrogenated CNx, is the film morphology and surface roughness. The presence
of hydrogen has a similar effect as N, but here the formation of hydrocarbons, and, if
N is also present, CNH and NH3, is responsible for the densification of the structure,
as seen in Fig. 13d (RMS = 1 nm) [28].
As discussed above, the resistivity of non-hydrogenated CNx films as a
function of N concentration changes mainly due to the changed morphology induced
by the N. Hydrogen has a similar effect on the film morphology, but the electrical
response is the opposite. Thus, in this case, the reduced surface roughness and number
of voids is not affecting  as much as the increased degree of bond saturation by
hydrogen.
4.3
Environmental effects
It can be noted from Fig. 15a that the carbon film grown in pure Ar contains
substantial amounts of H, although not intentionally added. This indicates that H
40
diffuses into porous films when exposed to atmosphere (c.f. Fig. 13a). Thus,
depending on the porosity of the films, the electrical properties can also be expected
to be influenced by ambient humidity.
1.8
0% N
2
/
o
1.6
1.4
1.2
5% N
1.0
100% N
0
50
100
150
200
2
2
250
Time (hours)
Figure 16: Change of relative electric resistance /0 of CNx films
after exposure to atmosphere. 0 is the initial resistance of the asdeposited sample. The lines are for guiding the eye only (taken
from reference [25]).
Broitman [25] have recently shown how the humidity present in the
environment during the electrical characterization affects the measurements. Figure
16 shows the change of relative resistivity /0 of CNx films (grown without H2) after
exposure to atmosphere, o being the initial resistivity of the as-deposited sample. The
resistivity of the film sputtered with pure Ar increased dramatically up to 70 % after
10 days, while for the films sputtered in the presence of nitrogen the increase in /0
was less than 10 %. Several experiments showed that  changes only when the films
were exposed to the atmosphere, i.e., the resistivity did not change significantly when
the samples were stored in vacuum. This has to be explained by diffusion, presumably
of water molecules or water-related species, into the structure [25] [28] [107].
41
A corresponding change of the electric resistance of carbon films under
exposure to air, due to the adsorption of water vapour has been reported in literature
[108] [109]. Also CNx films have been reported to absorb water, as determined by the
IR absorption in the range of the OH stretching mode around 3500 cm -1 [107].
Absorption by as much as 5 wt% has been observed by weight increase measurements
[110]. Absorption of water vapor is a reasonable explanation to the increase of  over
time, since the magnitude of the change shows a strong correlation with the degree of
porosity in the films (c.f. Fig 13). The other active component in air, oxygen, can be
excluded since XPS [28] [77] and ERDA [14] analysis of the samples have shown no
presence of oxygen in the films.
In an attempt to analyze the observed time dependence in terms of a diffusion
phenomenon, Broitman [25] have shown the variation of relative resistance (R-R0)/R0
versus time was plotted in log-log scales (Fig. 17). In a normal diffusion process
through isotropic material the diffusion front is Gaussian, and a log-log plot of the
quantity of diffusing species versus diffusion time gives a linear relationship with a
slope of 0.5 [111].
1
0
(R-R )/R
0
0% N
2
0.1
5% N
100% N
2
2
0.01
10
100
Time (hours)
Figure 17: Variation of relative resistance (R-R0)/R0 versus time.
Ro is the initial resistance of the as-deposited sample (taken from
reference [25]).
42
The plot in Fig. 17 exhibits two separate quasilinear regimes. The first one,
during the period 0-100 hours, has a slope around 0.60 for the sample done at 0 % N2,
~ 0.53 for the sample done at 5 % N2, and ~0.50 for the sample deposited in 100 %
N2. Processes with slopes higher than 0.5 are dominated by an enhanced diffusion commonly observed in porous systems [111]. In our case, the water molecules diffuse
in samples that exhibit different porosity scale, which might explain the observed
variation.
The second regime beyond 100 hours is slower, with a slope around 0.15 for
all samples. The propagation is dominated by a “sub-diffusive” behavior [111]. This
abnormal diffusion is more frequent and could be explained by a trapping mechanism
associated with the adsorption of a water molecule on the surface and/or inside the
pores.
4.4
Electronic Structure
The relatively high electrical conductivity, the optical data, as well as the
observed photoconductivity, suggests an electronic structure having a large number of
-type band-tail states, indicating a structure containing sp2-hybridized bonds.
Previous experimental results, obtained by x-ray photoelectron spectroscopy [30]
[112] [113] and in situ scanning tunneling microscopy [112] support this
interpretation.
Ricci et al. [103] suggested an electronic-band model for  electrons in
disordered carbons, with the presence of localized states situated at the tails of the
energy bands between EC (conduction) and EV (valence), and the Fermi level EF lies
somewhere within this mobility gap (Fig. 18). When the separation between the 
bands closes with the Fermi level position around the mobility shoulder value EC or
43
just above, an increase in the electrical conductivity and a decrease in the band gap up
to a semimetallic behavior should be observed. The temperature-dependent
conductivity, the photoconductivity, and the optical data obtained in our experiments
[25][74] can be examined within this model. At higher temperatures conduction is due
to carriers excited beyond the mobility shoulders into nonlocalized or extended states.
An increased excitation can also be expected from the extra energy carried by
photons. Furthermore, in the lower temperature range, conduction could arise from
carriers being excited into the localized states at the band edges, or hopping or
tunneling between localized states near the Fermi level. In order to distinguish a
possible hopping mechanism, a more extensive study at lower temperatures (< 130 K)
is necessary.
Figure 18: Schematic electronic-band structure for  electrons in
an amorphous carbon film for overlapping bands.
5
Concluding Remarks
Fullerene-like CNx is shown to be an inherently nanostructured material
consisting of bent and cross-linked predominantly sp2-hybridised graphene planes.
44
The incorporation of sp2-hybridised nitrogen, either substitutional for carbon in a
graphene network or in a pyridine-like manner, induces bending by formation of
pentagons and initiates cross-linking, while a cross-linking mechanism on the C sp3
site still cannot be ruled out. Although low ion assist energies are generally beneficial
for film growth by enhancing ad-atom mobility at relatively low substrate
temperatures, it has been shown that it is not the structure-determining factor in case
of fullerene-like CNx films. Instead the flux of preformed CxNy (x,y  2) species,
originating from a self-modified C target surface that acts as a virtual CVD source
under N-ion bombardment, is a crucial aspect for understanding the evolution of
fullerene-like structures. That is because these species are stabilized by multiple
bonding on the C and N side upon aligned incorporation during growth and, thus,
support the evolution of sheet structures. Furthermore, the formation of volatile CN
species namely C2N2 is also active on the substrate side and provides for selective
etching of misaligned molecules by reaction with adsorbed CN and subsequent
desorption. Thus, the evolution of the fullerene-like microstructure has to be seen as a
delicate interplay of adsorbing preformed molecules, which are either incorporated in
a stabilized and, thus, aligned manner or desorb via the formation of C2N2 if they are
less favourably bonded.
The mechanical deformation of fullerene-like CNx thin films is in general
determined by the rather weak van-der-Waals interactions between the planes and
inter-planar cross-links. The fairly large (0.35 nm) inter-planar spacing is easily
compressed under stress, whereas inter-planar slip, as it is limiting the strength of
graphite, is hindered due to the strong alleged C-C cross-linking and mechanical
interlocking. Hence, fullerene-like CNx is an extremely resilient material where
deformational stresses are dissipated over large volumes. The likeliness of bond
45
breaking and slip is reduced enabling the structure to recover after unload. Therefore,
one can imagine the deformation mechanism like the reaction of a nano-structured
polymeric sponge of great intra-planar strength upon compression, where the walls
buckle without rupturing or breaking. It is concluded that on a macroscopic scale,
fullerene-like CNx exhibits a fairly low resistance to indentation (high compliance)
combined with a comparatively low modulus. It also shows very little tendency to
plastic deformation. This superior resiliency to deformation makes asperities in a
tribological contact more likely to deform elastic. Hence, brittle fracture is hampered
giving rise to possible applications in especially abrasive environments. Furthermore,
the low modulus might be also of importance for thin film applications, since it
dissipates the contact stresses over large volumes making adhesive failures of the
film-substrate system less likely, due to reduced stress gradients at the interface. The
above-mentioned facts lead to a reduced wear compared to amorphous CNx or pure
carbon. This in combination with the rather low coefficient of friction makes
fullerene-like CNx a promising candidate for tribological applications, while the
process control provides for tailoring the mechanical properties.
The electrical properties of hydrogen-free CNx films is mainly determined by
the morphology of the film (i.e., surface roughness and degree of porosity), while the
sp3/sp2 bonding ratio or actual N content cannot by itself explain the variations in .
However, the presence of nitrogen during growth has a large indirect effect, as it
promotes the formation of smoother and more dense films. Adding H2 during
deposition also results in an increased density and reduced surface roughness. In this
case, however, an increased resistivity was observed due to the termination saturation
of bonds by hydrogen. A similar effect was observed when exposing CNx films to the
atmosphere, due to water vapor diffusing into the films. Thus, the presence of
46
hydrogen during deposition, the film morphology, as well as the ambient conditions
during measurement must be considered in the analysis of the electrical properties.
ACKNOWLEDGEMENTS
The authors acknowledge J.-E. Sundgren, H. Sjöström, N. Hellgren, R. Gago, B.
Holloway, S.J. Bull as well as I. Singer for valuable discussions. J. Neidhardt was
supported by SKF/ERC and E. Broitman by the National Science Foundation, Award
Nr. 0408574, “Vapor Lubrication of Magnetic Data Storage”.
REFERENCES
[1]
D.B. Bogy, W. Fong, B.H. Thornton, Z. Hong, H.M. Gross and C.S. Bathia,
IEEE Transactions on Magnetics 38, 1879 (2002).
[2]
A. Wienss, M. Neuhauser, H. Schneider, G. Persch-Schuy, J. Windeln, T.
Witke, U. Hartmann, Diamond Relat. Mater. 10, 1024 (2001).
[3]
I.A. Garcia, E.G. Berasategui, J. Neidhardt, N. Hellgren, S.J. Bull, T.F. Page
and L. Hultman, Philos. Mag. A 82, 2133 (2002).
[4]
J.J. Cuomo, P.A. Leary, D. Yu, W. Reiter and M. Frish, J. Vac. Sci. Technol.
16, 299 (1979).
[5]
A.Y. Liu and M.L. Cohen, Phys. Rev. B 41, 10727 (1990).
[6]
M.Y. Chen, X. Lin, V.P. Dravid, Y.W. Chung, V.W. Wong, and W.D. Sproul,
Surf. Coat. Technol. 54–55, 360 (1992).
[7]
D. Li, S. Lopez, Y.-W. Chung, M.S. Wong and W.D. Sproul, J. Vac. Sci.
Technol. A 13, 1063 (1995).
[8]
E.G. Wang, Prog. Mater. Sci. 41, 241 (1997).
[9]
S.K. Mishra and L.C. Pathak, Materials Letters 59, 3481 (2005).
[10]
Qi Zhang, Zhi Cheng Qin, Jing Zhu, Phil.Magn. Lett. 80, 585 (2000).
[11]
Li Chao, Chuan-Bao Cao and He-Sun Zhu, Materials Letters 58, 1903 (2004).
47
[12]
A. Badzian, T. Badzian, R. Roy and W. Drawl, Thin Solid Films 354, 148
(1999).
[13]
H. Sjöström, S.Stafström, M. Boman and J.-E. Sundgren, Phys. Rev. Lett. 75,
1336 (1995).
[14]
J. Neidhardt, Zs. Czigány, I.F. Brunell and L. Hultman, J. Appl. Phys. 93,
3002 (2003).
[15]
I. Alexandrou, H.-J. Scheibe, H.-J. Kiely, C.J. Papworth, G.A.J. Amaratunga
and B. Schultrich, Phys. Rev. B 60, 10903 (1999).
[16]
M. Johansson, H. Sjöström and L. Hultman, Vacuum 53, 451 (1999).
[17]
D.L. Strout, J. Phys. Chem. A 104, 3364 (2000).
[18]
M. Chhowalla and G.A.J. Amaratunga, Nature 407, 164 (2000).
[19]
S. Avivi, Y. Mastai and A.J. Gedanken, J. Am. Chem. Soc. 122, 4331 (2000).
[20]
G. Seifert, H. Terrones, M. Terrones and T. Frauenheim, Sol. State. Commun.
115, 635 (2000).
[21]
A. Rothschild, J. Sloan, R. Tenne, J. Am. Chem. Soc. 122, 5169 (2000).
[22]
Å. Johansson, Charge Dynamics and Electronic Structure of -Conjugated
Systems (Linköping Studies in Science and Technology, Doctoral Thesis no.
783, 2002).
[23]
W.W. Porterfield, Inorganic Chemistry – A Unified Approach (Academic
Press, London, 1993).
[24]
S. Stafström, Appl. Phys. Lett. 77, 3941 (2000).
[25]
E. Broitman, N. Hellgren, J. Neidhardt, I. F. Brunell and L. Hultman, Journal
Electron. Mater. 31, 11 (2002).
[26]
J. Neidhardt, H. Högberg, and L. Hultman, Thin Solid Films 478, 34-41
(2004).
[27]
J. Robertson, Mater. Sci. Eng. R 37, 129 (2002).
[28]
N. Hellgren, M.P. Johansson, B. Hjörvarsson, E. Broitman, M. Östblom, B.
Liedberg, L. Hultman and J.-E. Sundgren, J. Vac. Sci. Technol. A 18, 2349
(2000).
[29]
W. J. Gammon, O. Kraft, A. C. Reilly and B. C. Holloway, Carbon 10, 1917
(2003).
[30]
J. Neidhardt, Zs. Czigány, and L. Hultman, Carbon 42, 2729 (2004).
[31]
S.E. Rodil, Recent Devel. Appl. Phys. 6, 391 (2003).
48
[32]
N. Hellgren, J. Guo, C. Såthe, A. Agui, J. Nordgren, Y. Luo, H. Ågren and J.E. Sundgren, Appl. Phys. Lett. 76, 4348 (2001).
[33]
W.J. Gammon, (Doctoral Thesis, College of William and Mary, Williamsburg,
Virginia, USA, 2003).
[34]
N. Hellgren, N. Lin, E. Broitman, V. Serin, S.E. Grillo, R. Twesten, I. Petrov,
C. Colliex, L. Hultman and J.-E. Sundgren, J. Mater. Res. 16, 3188 (2001).
[35]
L. Hultman, J. Neidhardt, N. Hellgren, H. Sjöström and J.-E. Sundgren, Mater.
Res. Bull. 28, 194 (2003).
[36]
A.C. Ferrari, S.E. Rodil and J. Robertson, Phys. Rev. B 67, 155306 (2003).
[37]
R. Gago, I. Jiménez, J. Neidhardt, B. Abendroth, I. Caretti, L. Hultman and W.
Möller, Phys. Rev. B 71, 125414 (2005).
[38]
R. Gago, I. Jiménez, J.M. Albella, A. Climent-Font, D. Cáceres, I. Vergara,
J.C. Banks, B.L. Doyle and T.J. Terminello, J. Appl. Phys. 87, 8174 (2000).
[39]
I. Jiménez, R. Gago, J.M. Albella and T.J. Terminello, Diamond Rel. Mater.
10, 1166 (2001).
[40]
L. Hultman, S. Stafström, Zs. Czigány, J. Neidhardt, N. Hellgren, I. F. Brunell,
K. Suenaga and C. Colliex, Phys. Rev. Lett. 87, 225503-1 (2001).
[41]
R.P. Webb and M. Kerford, Nucl. Instrum. Methods Phys. Res., Sect. B 180,
32 (2001).
[42]
Ya.M. Soifer, N.P. Kobelev and V.M. Levin, J. Alloys Compd. 310, 292
(2000).
[43]
A.V. Talyzin, L.S. Dubrovinsky, M. Oden and U. Jansson, Diamond Relat.
Mater. 10, 2044 (2001).
[44]
C.M. Brown, L. Cristofolini, K. Kordatos and K. Prassides, J. Am. Chem. Soc.
118, 8715 (1996).
[45]
J. Neidhardt, Fullerene-Like Carbon Nitride Thin Solid Films, PhD Thesis No.
877, and the references therein, Institute of Technology, Linköping University,
Linköping, Sweden, 2004.
[46]
W.J. Gammon, D. I. Malyarenko, O. Kraft, G. L. Hoatson, A. C. Reilly and B.
C. Holloway, Phys. Rev B 66, 153402 (2002).
[47]
S.J. Townsend, T.J. Lenosky, D.A. Müller, C.S. Nichols and V. Elser, Phys.
Rev. Lett. 69, 921 (1992).
[48]
G.K. Gueorguiev, J. Neidhardt, S. Stafström and L. Hultman, Chem. Phys.
Lett. 410, 228 (2005).
49
[49]
J. Neidhardt, B. Abendroth, R. Gago, W. Möller and L. Hultman, J. Appl.
Phys. 94, 7059 (2003).
[50]
K. J. Boyd, D. Marton, S. S. Todorov, A. H. Al-Bayati, J. Kulick, R. A. Zuhr
and J. W. Rabalais, J. Vac. Sci. Technol. A 13, 2110 (1995).
[51]
J. Vlček, K. Rusnak, V. Hejek and L. Martinu, J. Appl. Phys. 86, 3646 (1999).
[52]
N. Hellgren, M. P. Johansson, E. Broitman, P. Sandström, L. Hultman and J.E. Sundgren, Thin Solid Films 382, 146 (2001).
[53]
R. Kaltofen, T. Sebald and G. Weise, Thin Solid Films 290-291, 112 (1996).
[54]
A. A. Voevodin, J. G. Jones, Zs. Czigány, L. Hultman and J. S. Zabinski, J.
Appl. Phys. 92, 4980 (2002).
[55]
S. S. Todorov, D. Marton, K. J. Boyd, A.H. Al-Bayati and J. W. Rabalais, J.
Vac. Sci. Technol. A 12, 3192 (1994).
[56]
G.K. Gueorguiev, J. Neidhardt, S. Stafström and L. Hultman, Chemical
Physics Letters 401, 288-295 (2005).
[57]
R. Gago, M. Vinnichenko, J. Neidhardt, U. Kreissig, Zs. Czigany, L. Hultman,
A. Kolitsch and W. Möller, Thin Solid Films 483, 89-94 (2005).
[58]
J.F. Shackelford, Introduction to Materials Science for Engineers (Prentice
Hall, New Jersey, USA, 2000)
[59]
R.F. Bunshah, Handbook of Hard Coatings (Noyes Publications, New Jersey,
USA, 2001)
[60]
H. Sjöström, L. Hultman, J.-E. Sundgren, S.V. Hainsworth, T.F. Page and
G.S.A.M. Theunissen, J. Vac. Sci. Technol. A 14, 56 (1996).
[61]
N. Hellgren, M.P. Johansson, E. Broitman, L. Hultman and J.-E. Sundgren,
Phys. Rev. B 59, 5162 (1999).
[62]
D.M. Teter, MRS Bull. 23, 22 (1998).
[63]
I.M. Hutchings, Friction and Wear of Engineering Materials (Edward Arnold,
London, UK, 1992).
[64]
E. Rabinowicz, Friction and Wear of Materials, 2nd edition (John Wiley &
Sons, New York, USA, 1995).
[65]
I.L. Singer, S.D. Dvorak, K.J. Wahl and T.W. Scharf, Third Body Processes
and Friction of Solid Lubricants Studied by In Situ Optical and Raman
Tribometry, Proceedings of the 28th Leeds-Lyon Symposium on Tribology,
Edited by D. Dowson, M. Priest, G. Dalmaz, A.A. Lubrecht, Tribology series.
40 (Elsevier Scientific Publ. Co. London UK, 2002).
50
[66]
J.M. Molina-Aldereguia, S.J. Llloyd, M. Oden, T.Joelson, L. Hultman and
W.J. Clegg, Philosophical Magazine A 82, 1983 (2002).
[67]
J.A. Greenwood and J.B.P. Williamson, Proc. Roy. Soc. Lond. A 295, 300
(1966).
[68]
J.F. Palacio, S.J. Bull, J. Neidhardt and L. Hultman, Thin Solid Films,
accepted for publication (2005).
[69]
N. Schwarzer, Surf. Coat. Technol. 133, 397 (2000).
[70]
J. Neidhardt, L. Hultman, E. Broitman, T.W. Scharf and I.L. Singer, Diamond
Relat.Mat. 13, 1882 (2004).
[71]
J.C. Sánchez-López, M. Belin, C. Donnet and E. Elizalde, Surf. Coat. Technol.
160, 138 (2002).
[72]
I. Singer, MRS Bull. 23, 37 (1998)
[73]
I.L. Singer, S.D. Dvorak, K.J. Wahl and T.W. Scharf, J. Vac. Sci. Technol. A
21, 232 (2003).
[74]
E. Broitman, N. Hellgren, K. Järrendahl, M.P. Johanson, S. Olafsson, J.-E.
Sundgren, and L. Hultman. J. Appl. Phys. 89, 1184 (2001).
[75]
D.I.Jones and A.D.Stewart, Philos. Mag. B 46, 423 (1982)
[76]
Z.J. Zhang, S. Fan, J. Huang, and C.M. Lieber, Appl. Phys. Lett. 68, 2639
(1996).
[77]
C. Running, U. Griesmeier, M. Gross, H.C. Hofsäss, R.Downing, and
G.P.Lamaze, Diamond Relat. Mater 4, 666 (1995).
[78]
D.R. McKenzie, Y. Yim, N.A. Marks, C.A. Davis, B.A. Pailthorpe, G.
Amaratunga, and V.S. Veerasamy, Diam. Relat. Mat. 3, 353 (1994).
[79]
A.Modinos, and J.P.Xanthakis, Appl.Phys.Lett. 731874 (1998).
[80]
H.Kinoshita, and A. Yamauchi, J.Vac.Sci.Technol. A14, 1933 (1996).
[81]
I. Huttel, J.Gurovic, F. Cerny, and J.Pospisil, Diamond Relat. Mater 8, 628
(1999).
[82]
N.-H. Cho, K.M. Krishman, D.K. Veirs, M.D. Rubin, C.B. Hopper,
B.Bhushan, and D.B. Bogy, J. Mater. Res. 5, 2543 (1990).
[83]
M. Zhang, P. Wanga, and Y. Nakayama, Jpn. J. Appl. Phys., Part 1 36, 4893
(1997).
[84]
A. Grill, V. Patel, and S. Cohen, Diamond Relat. Mater 3, 281 (1994).
[85]
N. Takada, K. Arai, S. Nitta, and S. Nonomura, Appl. Surf. Sci. 113-114, 274
(1997).
51
[86]
K. B. Lee, D. J. Lee, and Y. W. Shin, Mater. Res. Soc. Symp. Proc. 446, 401
(1997).
[87]
N. Laidani, A. Miotello, and J. Perrière, Appl. Surf. Sci. 99, 273 (1996).
[88]
A. Wei, D. Chen, N. Ke, W. Y. Cheung, S. Peng, and S. P. Wong, J. Phys. D
31, 1522 (1998)
[89]
M.A. Monclus, D.C. Cameron, and A.K.M.S. Chowdhurry, Thin Solid Films
341, 94 (1999).
[90]
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 (1993)
[91]
S.R.P. Silva, B. Rafferty, G.A.J. Amaratunga, J. Schwan, D.F. Franceschini,
and L.M. Brown, Diamond Relat. Mater. 5, 401 (1996).
[92]
O. Amir, and R. Kalish, J. Appl. Phys. 70, 4958 (1991).
[93]
J. Schwan, W. Dworschak, K. Jung, and H. Ehrhardt, Diamond Relat. Mater.
3, 1034 (1994).
[94]
Y.Aoi, K.Sakurada, and E.Kamijo, Appl.Phys. A 79 (2004) 1533-1536.
[95]
F. Falk, J. Meinschien, G. Mollekopf, K. Schuster, and H. Stafast, Mater. Sci.
Eng., B 46, 89 (1997).
[96]
J. Bulí , M. Jelínek, V. Vorlí ek, J. Zemek, and V. Pe ina, Thin Solid Films
292, 318 (1997).
[97]
G.Lazar, M.Clin, S.Charvet, M.Therasse, C.Godet, K.Zellama, Diamond
Relat. Matter 12, 201 (2003).
[98]
N.Konofaos, E.K.Evangelou, S.Logothetidis, and M. Gioti, J. Appl. Phys. 91,
9915 (2002).
[99]
J. Robertson, and E.P. O’Reilly, Phys. Rev. B 35, 2946 (1987).
[100] E. Broitman, P. Alonso, and R. Zimmerman, Thin Solid Films 277, 192
(1996).
[101] P. Nagels, R. Callaerts, M. Denayer, and R. DeConinck, in Proceedings of the
5th International Conference on Amorphous and Liquid Semiconductors,
edited by J. Stuke and W. Brenig (Taylor & Francis, London, 1973), p. 867.
[102] G. L. Doll, J. P. Heremans, T. A. Perry, and J. V. Mantese, J. Mater. Res. 9, 85
(1994).
[103] M. Ricci, M. Trinquecoste, F. Auguste, R. Canet, and P. Delhaes, J. Mater.
Res. 8, 480 (1993).
[104] E. A. Davis and N. F. Mott, Philos. Mag. 22, 903 (1970)
52
[105] J.M. Schneider, A. Anders, B. Hjörvarsson, I. Petrov, K. Macák, U.
Helmersson and J.-E. Sundgren, Appl. Phys. Lett. 74, 200 (1999).
[106] V. Hajek, K. Rusnak, and J. Vicek, J. Vac. Sci. Technol A 17, 899 (1999).
[107] F.Alvarez, N.M. Victoria, P. Hammer, F.L. Freire, Jr., and M.C. dos Santos,
Appl. Phys. Lett. 73, 1065 (1998).
[108] C.L. Jensen, and R.A. Hoffman, J.Vac. Sci. Technol A 9, 116 (1991).
[109] J.P. Łukaszewicz, Sens. Actuators B 60, 184 (1999).
[110] S. Vepřek, J. Weidmann, and F. Glatz, J. Vac. Sci. Technol. A 13, 2914
(1995).
[111] P. Shewmon, “Diffusion in solids”, Minerals, Metals & Materials Society,
Warrendale (1989).
[112] W.T. Zheng, H. Sjöström, I.Ivanov, K.Z.Xing, E.Broitman, W.R.Salaneck,
and J.-E.Sundgren, J.Vac.Sci.Technol A 14 2696 (1996)
[113] N.Lin, H.Hellgren, M.P.Johansson, L.Hultman, R. Erlandsson, and J.-E.
Sundgren, Phys. Rev. B 61, 4898 (2000).
53