MASARYK UNIVERSITY
Faculty of Science
Department of Physical Electronics
Preparation of nanostructured hard protective
coatings by magnetron sputtering
Habilitation thesis
Scientiﬁc ﬁeld: Plasma physics
Mgr. Pavel Souˇcek, Ph.D.
Brno 2017
I would like to thank all my colleagues present and the past, local as well as
from abroad. A special thanks are dedicated to my family.
2
Abstract:
The presented habilitation work called Preparation of nanostructured
hard protective coatings by magnetron sputtering summarises and comments
on the results acquired during my work at the Department of Physical Electronics
at the Faculty of Science of the Masaryk University and also on the
results gained in collaboration with colleagues from the University of Groningen,
PLATIT a.s. (formerly PIVOT a.s.), Institute of Physics of Materials
and Institute of Nuclear Physics of the ASCR, Institut des Mat´eriaux Jean
Rouxel Nantes and INP Greifswald. Ten papers providing a comprehensive
view of the preparation and analysis of magnetron sputtered hard titanium
and carbon-based nanocomposite coatings and next-generation prospective
molybdenum–boron–carbon coatings are chosen to be presented in this work.
Titanium and carbon-based coatings are shown as an example of presently
used coatings designed for maximum hardness. First, the deposition process
as such was studied in detail. The behaviour of the plasma parameters
on the set deposition conditions was understood. Similarities and diﬀerences
to classical reactive magnetron sputtering were identiﬁed. Large series
of nanocomposite nc-TiC/a-C:H coatings were deposited and studied from
diﬀerent angles. The relationships between the deposition parameters, the
structure of the coatings and their mechanical properties were described.
Superhard coatings were deposited. Molybdenum–boron–carbon-based coatings
are presented as an example of next-generation protective coatings still
in their early stages of development. These coatings are not superhard as
it was realised by the industry that the highest hardness does not necessarily
mean the best performance. The next generation of protective coatings
sacriﬁces some of the hardness but gains moderate ductility instead.
3
Contents
1 Introduction 6
2 Study of nanocomposite nc-TiC/a-C:H coatings prepared by
a hybrid PVD-PECVD process 8
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Process characterization . . . . . . . . . . . . . . . . . . . . . 9
2.2.1 Reactive magnetron sputtering or a hybrid PVD-PECVD
process . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2.2 The link between the process parameters and coating
properties . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.3 Deposition and analyses of nc-TiC/a-C:H coatings . . . . . . 13
2.3.1 Analysis of coatings with in-depth resolution . . . . . 13
2.3.2 Analyses of nc-TiC/a-C:H coatings prepared at low
ion bombardment . . . . . . . . . . . . . . . . . . . . . 15
2.3.3 Analyses of the sensitivity of the nc-TiC/a-C:H coatings
properties to the level ion bombardment controlled
by the applied magnetic ﬁeld . . . . . . . . . . 16
2.3.4 Investigation of the tribological properties of nc-TiC/aC:H
coatings in short-term repeating applications . . . 18
2.3.5 Properties of nc-TiC/a-C:H coatings prepared employing
High Power Impulse Magnetron Sputtering . . . . 20
2.3.6 The eﬀect of nickel doping on the properties of of ncTi(:Ni)C/a-C:H
coatings . . . . . . . . . . . . . . . . . 23
3 Nanostructured Mo-B-C coatings prepared by magnetron
co-sputtering 25
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
4
3.2 Synthesis of Mo-B-C coatings using pulsed-DC and their analysis
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3 Thermal stability of the Mo-B-C coatings . . . . . . . . . . . 30
Bibliography 32
Author’s publications not discussed in this work 34
Attached commented publications 36
5
Chapter 1
Introduction
The presented habilitation work called Preparation of nanostructured
hard protective coatings by magnetron sputtering summarizes and comments
on the results acquired during my work at the Department of Physical Electronics
at the Faculty of Science of the Masaryk University and also on the
results gained in collaboration with colleagues from other departments from
Czech Republic and abroad. I have participated in all the work presented
in this work with special focus on deposition and chemical and structural
analyses of nanocomposite and nanolaminate protective coatings.
The core of the presented habilitation work is divided into two mutually
independent sections. The ﬁrst section deals with nanocomposites composed
of nanocrystalline titanium carbide grains embedded in an amorphous hydrogenated
carbon matrix (nc-TiC/a-C:H). These coatings are relatively
well known in the world and publications in the literature can be found
since the 1980s (see Fig.1.1, where the works published till the early phases
of our study are shown). This type of coatings has been in the focus of researchers
due to high hardness reaching the superhard region, simultaneous
low coeﬃcient of friction and wear and also for its biocompatibility. The
second section is focussed on a diﬀerent type of coatings. During the last
several years, it was realized that optimizing coatings for maximum hardness
does not yield the best results for real life applications. This is because
superhard coatings generally exhibit one common downside — brittelnes.
If a crack is formed, it rapidly propagates. This to a premature failure of
the coating and of the coated tool as a whole. Recently, in 2009 a group of
unusually stiﬀ and ductile hard coatings was proposed based on theoretical
calculations by the group of prof. Schneider from Aachen University. Diﬀer-
6
1995 1996 1997 1998 1999 2000 2004 2005 201120102009200720032001 2002 2006 2008
Groningen (Pei, Galvan, Shaha)
Sevilla (Sánchez - López, Martínez-Martínez)
Nantes (Tessier, El Mel)
Koszalin (Czyzniewski, Gulbinski)
Kiev (Kulikovsky)
Ohio (Voevodin)
Uppsala (Wiklund, Lewin)
Duebendorf (Zehnder, Patscheider)
Linköping (Sundgren, Samuelsson, Sarakinos)
´
.
(early 80s)
Figure 1.1: The ﬁrst reports on preparation of nc-TiC/a-C:H coatings.
ent combinations of metal-, boron- and carbon-based materials in a Me2BC
structure were studied by ab initio calculations. These materials owe their
unique combination of high hardness and enhanced ductility to their inherently
nanolaminated microstructure. They exhibit high aspect ratio unit
cell that is similar to MAX phases with alternating planes containing stiﬀ
carbidic and boridic bonds with high degree of ionicity providing the high
stiﬀness and hardness, while metallic bonds provide moderate ductility. So
far only few papers numbering in single digits are available on this material.
Such coatings could serve as the next generation of protective coatings for
applications where current protective coatings fail and therefore were in the
focus of our latest research.
The individual chapters contain commentaries of published works that
are in the attachment. As all of the attached papers present a complete work
by themselves, I commented also on the facts that are not usually not widely
discussed in published papers like the motivation and encountered problems.
The most important results obtained during my work are naturally also
included. The list of references consists of only the works in the attachment
as all of them contain numerous references more than adequately describing
the studied ﬁeld.
7
Chapter 2
Study of nanocomposite
nc-TiC/a-C:H coatings
prepared by a hybrid
PVD-PECVD process
2.1 Introduction
The ﬁrst chapter describes the ﬁrst type of the studied coatings — ncTiC/a-C:H
coatings. It is divided into two sections. This ﬁrst section is primarily
focused on the study of the behaviour of the deposition process and
the plasma parameters. This section is built around the ﬁrst two attached
papers [1] and [2]. In many cases, a study focused on coating deposition is
focused only on the inﬂuence of a set deposition parameters such as the ﬂow
of the buﬀer and reactive gasses or power supplied to the targets on the
properties of the deposited coatings. Although this approach might be to a
certain limit valid, it omits the study of the plasma that is just taken as a
“black box” supplying the ﬁlm forming atoms. Such an approach led to an
inevitable confusion in the literature. While some call the process of sputtering
titanium target in argon/acetylene gas mixture reactive magnetron
sputtering, some call it hybrid PVD-PECVD process. Paper [1] is focused on
showing that these are not the same and on proving that only one of these
options is correct. Paper [2] is based on linking the deposition parameters
with the processes in the volume of the chamber and on the sputtered target.
8
These consequently inﬂuence the properties of the deposited coatings. The
ﬁndings published in this work will be later used for successful up-scaling of
the process to an industrial conﬁguration.
The second section is primarily focused on the deposited coatings. A
method allowing for analysis of the mechanical properties and composition
of the coatings with in-depth resolution with then available equipment was
presented ﬁrst and in-depth homogeneity of the coatings was studied [3].
Once uniformity of the coatings was established, the surface morphology,
adhesion, structure and mechanical properties of coatings prepared in relatively
mild conditions suitable for heat sensitive substrates such as steel was
described [4]. Next, the study was extended to analyses of coatings made
in harsher conditions with enhanced ion bombardment during the coating
growth [5]. Carbon-based coatings generally exhibit a low coeﬃcient of friction
and wear, however these parameters are evaluated from a steady state
of the measurement and the often unstable initial part of the measurement
is generally omitted. A study was undertaken to assess the applicability of
this method for short-term repeating applications such as drilling, milling,
etc. [6]. Then further concepts possibly extending the applicability of the ncTiC/a-C:H
coatings were followed. After recent acquirement of a HiPIMS
source, the eﬀects of the HiPIMS use were investigated. HiPIMS and DCMS
deposited coatings were compared and the key coating parameters leading
to signiﬁcant hardness increase inﬂuenced by HiPIMS were pinpointed [7].
Lastly, the concept of doping of the coatings with weak carbide forming
elements to alter the structure of the coatings was investigated [8].
2.2 Process characterization
2.2.1 Reactive magnetron sputtering or a hybrid PVD-PECVD
process
Magnetron sputtering is a classical physical vapour deposition (PVD)
method of preparation of coatings studied and used since the 1980s. The
basic physical principle of sputtering is knocking out of atoms from a cathode
(also called target) by incident energetic particles supplied by a plasma
ignited between the cathode and the grounded walls of the chamber. The
plasma is held in the vicinity of the target by a magnetic ﬁeld.
Reactive magnetron sputtering involves adding a reactive gas into the
9
chamber during magnetron sputtering. The present reactive gas can react
with the material of the target and, depending on the amount of the present
reactive gas, the process can progress in two diﬀerent modes. If the amount
of the reactive gas is low the target remains nearly clean, the pressure in
the chamber is generally low, the intensity of the optical emission lines corresponding
to target material and the deposition rate are high. This is the
so-called metallic mode. As the reactive gas ﬂow increases, the material of
the targets reacts with the target more and when a particular gas ﬂow is
reached, the racetrack i.e. the working area of the target, is nearly instantly
covered by products of reaction of the target material and the reactive gas.
For example, TiN is formed if titanium target is sputtered in argon/nitrogen
mixture and TiO2 is formed if oxygen is used instead of nitrogen. The process
is rapidly shifted into the so-called compound (also referred to as poisoned
mode). In this mode the pressure is signiﬁcantly higher as all the nitrogen
able to react with the material in the chamber has already reacted and the
excess gas is adding to the Ar background pressure. The intensity of the
target material lines in the optical emission spectra drops as the resulting
compound is sputtered instead of the target material. The compounds
formed on the target also typically show lower sputter yield resulting in
much-decreased deposition rate. The lower sputter yield, meaning that it
is more diﬃcult to sputter this compound than the original material, also
signiﬁes that if we want to shift back to the metallic mode, we need to decrease
the reactive gas ﬂow to a signiﬁcantly lower value than where the
transition from metallic mode to poisoned mode occurred. This behaviour
called hysteresis is a hallmark of the classical reactive magnetron sputtering1
Similarities and diﬀerences between processes where oxygen and acetylene
are used under otherwise identical conditions with a focus on the Ti
line intensities were studied in paper [1]. A sputtering system with a relatively
large 20 cm in diameter Ti target and moderate pumping speed of
0.04 m3s−1 was used to ensure the occurrence of hysteresis. The process with
oxygen exhibited a hysteresis behaviour as is typically observed. However,
the process with acetylene behaves diﬀerently. No hysteresis is observed in
our work. The Ti line intensity for every acetylene ﬂow during the ﬂow
increase matches the one measured during the acetylene ﬂow decrease.
1
The hysteresis can be suppressed by very high pumping speeds and/or by using very
small targets, but it is generally observed for standard deposition systems.
10
An eﬀort was made to model the experimentally observed lack of hysteresis.
The standard Berg’s model assuming that the reactive gas atoms
from dissociated from e.g. nitrogen or oxygen can react only with the surface
of the titanium target and forms a single layer of the reacted nitride or
oxide is modiﬁed. An assumption that carbon atoms can not only react with
titanium but also together and form a thicker carbon layer on the target and
the chamber walls is added. This assumption agrees with the experiments,
where we observe that if the process is stopped after the highest acetylene
ﬂow is reached and the chamber is opened, the target including the racetrack
is covered by thick black residue. It is found out that indeed this assumption
does lead to suppression of hysteresis. It is furthermore possible to estimate
the degree of dissociation of acetylene in the chamber. This is found to be
20 – 50 %.
Therefore, magnetron sputtering of titanium in argon/acetylene atmosphere
does not principally exhibit hysteresis. Thus, designating the studied
process reactive magnetron sputtering is not entirely justiﬁed and can lead
to certain misunderstanding. A more correct way would be to label the process
as a hybrid PVD-PECVD process as it combines elements typical for
PVD, i.e. sputtering of titanium target, and for PECVD, i.e. use of plasma
to dissociate a precursor gas in the volume and formation of thick ﬁlms
formed from the precursor elements.
2.2.2 The link between the process parameters and coating
properties
The eﬀort to link the processes taking place inside the chamber during
the deposition and the properties of the coatings follows [2]. The cathode
voltage and current (in constant power mode), the total pressure, argon, titanium
and hydrogen optical emission lines are measured for diﬀerent acetylene
ﬂows. Coatings are furthermore deposited at selected acetylene ﬂows.
All of the studied parameters are found to be evolving continuously, however,
in a non-monotonous manner. The behaviour of the observed parameters is
found out to be governed by the processes on the target. The target after
each deposition was imaged and areas with diﬀerent apparent colours were
assigned to their respective phases (Ti — metallic grey, TiC — dull grey,
C — black). The relative coverage of the racetrack by diﬀerent phases derived
from pixel counting of the target images was used to of understand
11
the process dynamics.
For low acetylene ﬂows the voltage increases due to the formation of
TiC layer on the surface of the racetrack on the target. The TiC layer is
formed at the expense of Ti, and the relative concentration of Ti in the
plasma decreases accordingly, while the H concentration increases due to
acetylene fragmentation in the plasma volume. The border between the low
and medium range of acetylene ﬂows is identiﬁed by a local maximum of
the cathode voltage and an according minimum of the cathode current. The
medium range of acetylene ﬂows is characterised by a slight decrease of the
cathode voltage. This is probably due to formation of sub-stoichiometric TiC
on the racetrack. The cathode voltage decreases with increasing stoichiometry
of the TiC compound with acetylene ﬂow increase. Ti concentration
in the plasma is still decreasing and hydrogen concentration continues to
increase. Stoichiometric TiC coatings are deposited in this zone. The border
between the medium and the high range of acetylene ﬂows is more distinct
and is characterised by a local minimum of the cathode voltage, an according
local maximum of the current and maxima in the hydrogen line and argon
line emissions. The hardest coatings are always deposited at this border. For
high acetylene ﬂows the target voltage rapidly increases due to rapid penetration
of the carbon layer into the target racetrack. The Ti concentration
in plasma further decreases but H concentration remains slightly decreasing.
The knowledge of the process behaviour later proves to be pivotal in our
work for successful up-scaling of the process on an industrial scale. This upsacling
is performed on a setup with a central cylindrical rotating cathode
supplied by PLATIT. The process analysis methods are limited to monitoring
the cathode voltage and current and the total pressure as these are
already a part of the industrial deposition apparatus. The behaviour of these
parameters is found to be similar to that observed in the laboratory scale
device with a planar target. The only diﬀerence is in the process changes
being more sudden. This is due to diﬀerent geometries. The magnetic ﬁeld
conﬁguration in our laboratory scale device with planar target allows for a
slow gradual growth of the TiC or C phases on the non-sputtered part of
the target and their gradual penetration into the target racetrack leading
to slower gradual evolution of the cathode voltage. The cylindrical target
and its magnetic ﬁeld conﬁguration in the industrial device are optimized
for much higher target utilization ( 85 % compared to ∼ 15 % for planar
12
target). This setup leads to the fact that virtually the whole cathode acts
as the racetrack. Therefore, the TiC or C layers can not gradually penetrate
from the non-sputtered parts of the target and the whole target undergoes
the change of the surface state nearly instantaneously. A robust automatic
algorithm for production of hard nc-TiC/a-C:H coatings is developed based
on the sudden evolution of the cathode voltage and the total pressure allowing
for a repeatable and reliable deposition of hard nc-TiC/a-C:H coatings
independently on the chamber and target history.
2.3 Deposition and analyses of nc-TiC/a-C:H coat-
ings
2.3.1 Analysis of coatings with in-depth resolution
One of the ﬁrst obstacles reached with analyses of the deposited coatings
was how to check the homogeneity of the coating with in-depth resolution [3].
The deposition process in our case is always reaching a steady state typically
for the ﬁrst ﬁfteen minutes. Then, for the rest of the process (typically 45
minutes), the process parameters are suﬃciently stable. It is desirable to
ﬁnd out if the evolution of the coating parameters follows the evolution of
the deposition process.
Figure 2.1a shows the standard way of measurement of mechanical properties
of coatings. If information from a deeper part of the coating is required,
a higher load is applied. This, however, increases the aﬀected coating volume
not only to the deeper parts of the coating, but also to the sides. Moreover,
another problem arises when a hard coating on a soft substrate is measured.
It is known that the soft substrate is already under plastic strain before the
lower part of the coating is aﬀected. Therefore, the data from the bottom
part of the coating cannot be reliably extracted.
An alternate method is proposed in our work (see Fig. 2.1b). An abrasion
of a spherical crater on a calotest machine is performed ﬁrst. Then
measurements with a constant load over the calotte with equidistant steps
follow. Determination of the indent distance from the calotte centre gives the
thickness of the ﬁlm at the given position2. This way it is possible to mea-
2
Measurement on a metallographic cut is also possible. It was, however, found out that
as shallow angle cuts are needed, machine polishing was not precise enough. Handmade
cuts seemed adequate, but careful measurements of their proﬁle revealed that they were
13
Figure 2.1: Comparison of a) standard and b) proposed measurements of
deeper parts of the coating.
Figure 2.2: Residual indents on a calotte after measurement of mechanical
properties.
sure mechanical properties from deeper parts of the coatings more directly,
eﬃciently and exactly. An example of a realised measurement of mechanical
properties on a calotte is presented in Fig. 2.2.
Furthermore, our calculations show that as the plastically inﬂuenced volume
of the coating is similar to that producing the X-ray signal in EDX, it
is possible to directly compare the mechanical properties and chemical composition
derived by EDX. We conclude that using the presented method, it
is possible to conﬁrm that the chemical composition, as well as the mechanical
properties, attain stable values during the early phase of the deposition
where the process is reaching the equilibrium. Then, the composition and
the hardness and Young’s modulus remain stable.
not linear. This made a recalculation of the distance to the depth imprecise. Machine
made calotes were found out to be the optimal solution as either analytical equation or
an imaging on a confocal microscope and approximating of the proﬁle with a parabola are
simple and give good results.
14
2.3.2 Analyses of nc-TiC/a-C:H coatings prepared at low ion
bombardment
Once the homogeneity of the coatings was established, the analysis of the
properties of the coatings as a function of the reactive gas ﬂow began [4].
The ﬁrst series of coatings is deposited using the so-called well-balanced
magnetic ﬁeld conﬁguration. It is known that this conﬁguration allows for
strong plasma conﬁnement in the vicinity of the target providing relatively
dense plasma close to the target. The shape of the magnetic ﬁeld lines does
not allow for a signiﬁcant release of ions from the plasma towards the growing
coating limiting its ion bombardment and energy inﬂux.
The deposition process in our work needed to be optimised at the start
of the series to ensure suﬃcient adhesion. The adhesion of the coating to the
selected substrate must be high enough for reliable nanoindentation measurements
as well as for future applications, where adhesion of the coating
to the coated tool is one of the critical parameters determining a successful
coating.
Argon ion etch cleaning of the substrates is performed ﬁrst for 35 min to
ensure a clean substrate surface. When the intended coating is deposited directly
onto the substrate, the adhesion is not high enough. This is illustrated
by signiﬁcant delamination around the residual crater after a Rockwell adhesion
test according to DIN CEN/TS 1071-18 standards with 1500 N load
shown in Fig. 2.3a. Therefore a pure titanium adhesion layer is deposited for
three minutes. This corresponds to the adhesion layer thickness of ∼ 700 nm.
The time of the interlayer deposition is selected as a minimum needed for
the setting of all the parameters used during the deposition. This includes
the setting of a manual matching box of the RF bias and manual logging of
the pressure, cathode voltage and current and arcing occurrence. Also, the
required acetylene ﬂow is slowly gradually set during the last minute as the
older mass ﬂow controller tends to open too much and to poison the process
if it is operated too quickly. The residual crater after the Rockwell test of a
coating with the adhesion interlayer is shown in Fig. 2.3b. The adhesion of
the coating with the adhesion layer is good in the HF0 – HF1 grade. Despite
the adhesion layer is relatively thick, its thickness is not further optimised
as the deposited coatings exhibit good adhesion.
After the deposition of the titanium interlayer, the deposition of the ncTiC/a-C:H
coating follows. The acetylene ﬂow during the deposition directly
15
Figure 2.3: Comparison of adhesion of coatings without (left) and with
(right) adhesion promoting titanium interlayer.
determining the carbon and titanium content is identiﬁed as the critical
parameter inﬂuencing the structure and the mechanical properties of the
deposited coatings in our work. XRD proves the presence of TiC grains of
sizes ranging from several tens of nm to several nm depending on the carbon
content in the coating. No Ti peaks are detected in the XRD patterns even
when the Ti/C ratio is > 1 indicating that substoichiometric TiC is formed
rather than Ti. The highest hardness and Young’s modulus of 46 GPa and
415 GPa, respectively, are reached for the Ti/C ratio of ∼ 3/4. This sample
has a dense columnar structure with elongated TiC grains ∼ 15 nm wide.
This sample also exhibits the most pronounced preferential (111) orientation
of the grains.
2.3.3 Analyses of the sensitivity of the nc-TiC/a-C:H coatings
properties to the level ion bombardment controlled
by the applied magnetic ﬁeld
This study was motivated by the drastically diﬀerent mechanical properties
of nc-TiC/a-C:H coatings with identical chemical composition reported
by various authors. These diﬀerences are illustrated in Fig. 2.4. The deposition
temperatures are typically 300 °C, which is signiﬁcantly lower
than the melting temperature of titanium (3067°C) and also lower than
the temperature, at which the carbon matrix starts to deteriorate (350 –
400 °C). The deposition temperature, therefore, isn’t a key factor for these
diﬀerences. This indicates that the diﬀerences in the ion ﬂux should be the
culprit. A comparison of the levels of the ion ﬂux between the published
16
Figure 2.4: Diﬀerent hardnesses of nc-TiC/a-C:H coatings reported by different
authors.
works is however nigh impossible. This is because not all the experimental
details are always presented in the published papers (e.g. substrate bias
voltage, magnetic ﬁeld conﬁguration, target size and target to substrate distance).
All of the presented parameters might suggest high levels of ion ﬂux,
but the only unreported parameter can completely alter this expectation.
For instance, the distance of 100 — 150 mm between the substrate and 2
or 3-inch magnetron targets quite common in research systems results in
very low plasma densities at the substrate even for highly unbalanced magnetrons.
Therefore, the paper aims to single out the eﬀect of the ion ﬂux is
a single apparatus [5].
Two series of coatings are deposited. One with a lower ion ﬂux and one
with a higher level of ion ﬂux. The lower ﬂux of ions is assured by using
a well balanced magnetic ﬁeld in the magnetron. Use of the well-balanced
conﬁguration results in saturated ion current on the substrate of 0.3 mA/cm2
corresponding to the ion ﬂux of 1.87·1015 cm2s−1. An unbalanced magnetic
ﬁeld3 is used for depositions with higher ion ﬂux. This conﬁguration results
in the saturated ion current of 2.1 mA/cm2 corresponding to the ion ﬂux of
1.31·1016 cm2s−1. Thus, the number of impinging ions was increased seven
3
Newly manufactured and retroﬁtted to the existing Alcatel system by Gencoa.
17
times.
The structure of the coatings is, somewhat surprisingly, not signiﬁcantly
aﬀected by the diﬀerent levels of the ion ﬂux. Also, the maximal values of the
hardness and elastic modulus are nearly identical. However, their evolutions
are slightly shifted to diﬀerent C/Ti ratios. Therefore, diﬀerent acetylene
ﬂuxes are needed in both cases to achieve the highest hardness.
2.3.4 Investigation of the tribological properties of nc-TiC/aC:H
coatings in short-term repeating applications
It has been previously shown by others as well as in our work that carbonbased
coatings can achieve a very low coeﬃcient of friction (CoF) of < 0.1
together with low wear rate. These works show that the CoF exhibits a
typical 3 zone behaviour — TiC, transition and C phase governed zones.
The ﬁrst zone is characterised by high CoF > 0.25 and high counterpart
wear. This is due to low carbon content, relatively high hardness and also
high roughness of the coating. This leads to signiﬁcant energy dissipation
in the wear of the ball resulting in high CoF. The transition zone already
exhibits lower CoF. However, the value of the CoF is not stable. This is
due to periodical formation and destruction of a carbon-rich transfer layer
eﬀectively insulating the coating and the counterpart leading to a three
body interaction. The periodical destruction of the transfer layer is caused
by a low amount of very hard sharp debris from the coating. The carbon
governed zone is characterised by low and stable CoF of ∼ 0.1. The carbonrich
transfer layer is fully formed, and the wear of the counterpart is low.
All of these tests focus on the long-term evolution of the wear and friction
of the sliding couple, and the so-called running-in stage i.e. the often unstable
initial phase of the sliding contact between fresh, unworn surfaces is rarely
studied and is typically omitted. The omitting of the running-in period is
justiﬁable if the tested sample is designed for applications where the same
parts are in constant contact. A car engine is such an example. On the
other hand, there are applications where the coating gets periodically in
contact with a new counterpart. The examples would be drilling, milling or
forming. If there is a transfer ﬁlm created between the contact surfaces of a
drill and one hole, this transfer ﬁlm does not play a role in drilling another
hole. What coeﬃcient of friction would better describe such applications —
the one measured in the running-in period or the one measured after? The
18
Figure 2.5: Examples of cracked pincushions.
running-in stage was studied in detail in both continuous and interrupted
tribotests in paper [6].
A nc-TiC/a-C:H coating prepared in an industrial apparatus with carbon
content of ∼ 90 at.% was chosen for this testing as the high carbon content
enables the creation of the transfer layer. These tests were carried out at
ambient temperature as we have already shown that the coatings oxidise
during high-temperature tests in air and various TiO2 phases are formed.
Also the growth defects in the coatings tend to crack at high temperatures
even without mechanical stresses (see Fig. 2.5). Such cracking leads to a
premature failure of the coatings.
In our work the usually long-term continuous tests are divided into
shorter runs. The wear track on the coating as well as the wear scar on
the counterpart are imaged after each run. The coeﬃcient of friction (CoF)
measurements reveal that the CoF is initially high for the ﬁrst several runs
but remains constant and low for later runs. The roughness of the coating is
found to be the main factor in determining the CoF behaviour. The initial
high CoF is caused by high initial roughness of the coating leading to slow
or no formation of the carbon rich transfer layer crucial for low CoF. After
the roughness of the hard coating decreases to a certain value, the transfer
layer can form fast and the resulting CoF is low and stable. Roughness of
the softer steel counterpart and any possible oxidation of the wear track are
found out to be of little inﬂuence. Therefore, although the standard CoF
analysis procedure involving quantiﬁcation of the value of the CoF measured
in the steady state region after the running-in period does not exactly
describe the initial use of the coated tool, it is relevant even for descrip-
19
tion of the most of the lifetime of coatings enduring short-term repeating
applications such as those found in machining.
2.3.5 Properties of nc-TiC/a-C:H coatings prepared employing
High Power Impulse Magnetron Sputtering
High Power Impulse Magnetron Sputtering, HiPIMS for short, is a variant
of magnetron sputtering, where the power to the cathode is supplied
in very short and very intense pulses with an extended oﬀ-time between
them. The power pulse is typically tens, or a few hundreds of microseconds
long and the duty cycle is usually around one or two percent. It has been
reported that this technique generally produces plasma two orders of magnitude
denser than in the case of DC magnetron sputtering, which leads to
signiﬁcant ionisation of the sputtered species. The ionisation of the sputtered
species routinely reaches more than 50 % in HiPIMS, whilst it is in the order
of few percent in the case of DCMS. It is also known that the ion current
to on the substrate in the pulse and directly after the pulse is signiﬁcantly
enhanced. This increased ion and energy ﬂux together with deposition from
ions instead of neutrals have been shown to alter the structure and overall
properties of diﬀerent types of coatings. Therefore, this method is used to
study the eﬀect of its use on nc-TiC/a-C:H coatings [7].
One of the main drawbacks of HiPIMS is a lower deposition rate than
for corresponding coatings prepared by DCMS. This is also observed in our
work. However, the deposition rate of coatings prepared by DCMS decreases
at higher acetylene ﬂows, while the deposition rate of coatings prepared by
HiPIMS increases. The reduction of the deposition rate in the DCMS case
can be explained by a lower ﬂux of titanium from the target due to the racetrack
being covered by carbon residue. The eﬀect of lower titanium sputtering
rate is probably not suﬃciently compensated by the increase of carbon
ﬂux on the coating. The increase in the deposition rate at high acetylene
ﬂuxes with HiPIMS can be explained by enhanced ﬂux of carbon because of
expected higher level of acetylene dissociation in the denser HiPIMS plasma,
as it is known that dissociation of molecular gasses such as oxygen and nitrogen
is considerably higher in HiPIMS compared to DCMS. Moreover, it has
been shown in the literature that acetylene gets more dissociated in denser
plasmas (RF compared to DC) again justifying the assumption of higher
level of acetylene dissociation in HiPIMS.
20
Diﬀerent dynamics on the target were observed in our work for both
plasma excitation methods. Targets imaged after the deposition revealed
that when DCMS deposition of coatings with high carbon content results
in the racetrack being nearly fully covered with a black residue, depositions
of corresponding coatings with HiPIMS lead to much cleaner target
probably due to Ti ion backattraction. This has a rather unexpected result.
The cleaner surface of the target even at processes leading to deposition of
coatings with carbon content higher than 90 at.% allow for a wider range of
reliably deposited coatings while using HiPIMS. In comparison, the DCMS
depositions do not allow for arc-free deposition of coatings with carbon content
70 at.% in our setup.
Two sets of coatings are deposited by DCMS and HiPIMS at the same
mean power, pressure and temperature. DCMS deposited coatings are prepared
with a self-bias of -100 V and HiPIMS coatings are made at ﬂoating
potential. HiPIMS deposited coatings exhibit structure reﬁnement caused
by the enhanced ion bombarbment — grains of similar size as in DCMS are
of better stoichiometry and are separated by a lower amount of the a-C:H
matrix leading to closer grain-to-grain distance. This is attributed to the
higher ion and energy ﬂux inherent to the HiPIMS process. This structure
reﬁnement leads to higher hardness, elastic modulus, as well as the H/E and
H3/E2 ratios of the HiPIMS deposited coatings.
The eﬀect of the bias for HiPIMS deposited coatings is also tested for
several samples to have direct comparison with the biased DCMS deposited
coatings prepared with -100 V bias. Use of HiPIMS together with RF bias
proved to be technically challenging. Although the substrate holder and the
power source are separated by two capacities in the standard matchbox, the
power source is not hard enough. A large number of positive ions generated
by HiPIMS imping the substrate holder and lower the voltage generated by
the power source. When the voltage is ∼ 120 V before the pulse, it starts to
decrease at the onset of the pulse. It steadily declines during the pulse, and
also for approximately 250 microseconds after the pulse as the remaining ions
generated in the plasma travel to the substrate. Then, the voltage begins
to rise to its pre-pulse value. This behaviour is illustrated by measurements
plotted in Fig. 2.6. Therefore, the use of the second HiPIMS channel or
addition of a large capacitor to the circuit in case of using the substrate bias
would be needed for the experiments to attain a stable bias value.
21
Figure 2.6: The evolution of the discharge voltage, discharge current and
bias voltage during the HiPIMS pulse.
Figure 2.7: Comparison of surface morphologies of coatings with similar
carbon content prepared using DCMS with bias (left), HIPIMS without
bias (centre) and HiPIMS with bias (right).
The comparison of the surface morphologies of coatings with similar
carbon content prepared using bias and DCMS (left), no bias and HIPIMS
(centre) and HiPIMS with bias (right) is shown in Fig. 2.7. While the shift
from the use of DCMS with the bias to HiPIMS without bias results in
some surface reﬁnement, the addition of bias to HiPIMS leads to a very
strong change of the surface structure. The surface exhibits occurrence of
local melting possibly due to local thermal spikes generated by the energetic
ion bombardment. Also, the TiC grain strain doubled, and the hardness
was reduced by half. Therefore, use of -100 V bias proved to be severely
detrimental to the properties of the deposited coatings. However, lower levels
of bias should be tried as well. This is planned for a future experiment.
22
2.3.6 The eﬀect of nickel doping on the properties of of ncTi(:Ni)C/a-C:H
coatings
The next part of our work is motivated by the works of prof. Ulf Jansson
and dr. Erik Lewin from the Uppsala University. They theoretically predicted
that the use of doping of the nc-TiC/a-C:H coatings by a weak carbide
forming element could, in particular conditions, lead to the formation
of a metastable TiXC phase, where the doping element X is incorporated
in the grain. The incorporation of the doping element into the grain should
change the coating microstructure such as the grain size, stoichiometry or
the average grain separation. Such changes in the microstructure could lead
to (favourable) changes in the mechanical properties. The metastable grains
might also favour release of carbon from the grain to the matrix under stress
and the coating could exhibit self-lubrication.
The same deposition conditions (i.e. power to the target, pressure, bias)
are used in our work to have direct comparison with previously deposited
undoped samples to isolate the eﬀect of the doping. Zinc, copper and nickel
are initially tested. The ﬁrst idea is to change the level of doping by placing
metal pellets on a pure titanium target. Zinc and copper are proved
to be unsuitable due to the high temperature of the target during the deposition.
The copper pellets are mostly thawed, zinc pellets ﬁrst melt and
then evaporate. Nickel as the doping element is thus chosen in our work [8].
The work with nickel pellets is complicated because of ferromagneticity of
nickel and placement of the pellets in a strong magnetic ﬁeld on the cathode.
Thus, in the end, the coatings are prepared by sputtering of an alloy TiNi
target in an atmosphere containing acetylene. The amount of the metals
supplied from the sputtering of the TiNi target steadily decreases with increased
acetylene ﬂow (and hence the carbon content of the coating) due to
the used deposition process. This results in a gradual decrease of the eﬀect
of the doping with increasing carbon content. It is concluded that nickel is
successfully incorporated into the grains as no diﬀractions corresponding to
Ti are observed and the cubic TiC lattice parameter of the doped coatings
is higher than that measured in coatings prepared without nickel. A change
in the microstructure of the coatings is observed as expected. Nickel addition
primarily leads to grain size decrease due to the suspected formation of
lattice defects induced by nickel incorporation into the grains. Lower mean
grain separation is also observed. The change in the microstructure leads to
23
somewhat enhanced hardness if the carbon content is 45 at.%. However,
it is strongly reduced if the carbon content is higher.
Therefore, Ni doping does not lead to deposition of coatings with higher
maximal hardness. However, nickel is incorporated to the grains, and selflubricity
of the coatings can still be achieved. This can be a subject of
subsequent tests.
24
Chapter 3
Nanostructured Mo-B-C
coatings prepared by
magnetron co-sputtering
3.1 Introduction
The second section is devoted to the deposition and analysis of a diﬀerent
type of nanostructured coatings containing molybdenum, boron and carbon.
The work on this type of coatings is inspired by the works of the group of
professor Jochen Schneider from Aachen University on the Mo2BC coatings.
Mo2BC in a bulk form was in the centre of focus of studies before, but it
was rather for its superconducting properties and not the mechanical properties.
Mo2BC in the form of a thin ﬁlm is ﬁrst synthesised in Aachen in
2009. They also calculated, that the elemental cell of crystalline Mo2BC
combining layers with high and low electron density intrinsically exhibits
high hardness and stiﬀness coupled with enhanced ductility. This would be
very advantageous for demanding applications in the industry, where the
current generation of although hard, but often brittle protective coatings
fails. The presented method of the synthesis on crystalline Mo2BC coatings
has one drawback. Substrate temperature 800 ° C is needed to obtain crystalline
coating when DC is used for the plasma excitation. This temperature
is decreased down to ∼ 400 ° C when HiPIMS is used. HiPIMS is, however,
still an expensive technology with many free parameters not wholly suitable
for an easy mass industrial use. Therefore, we use mid-frequency pulsed DC
25
plasma excitation to enhance the energy ﬂow to the substrate compared to
DC sputtering to lower the crystallisation temperature. We also studied if a
fully crystalline coating with a precise chemical composition is even needed
to achieve hard and crack resistant coatings. This is described in paper [9].
The second paper discussed in this chapter and the last in this work [10]
is focused on the changes of the structure and the mechanical properties of
the coatings with increasing annealing temperature.
3.2 Synthesis of Mo-B-C coatings using pulsedDC
and their analysis
The ﬁrst step in the deposition of Mo-B-C coatings is to establish what
pulsing frequency and duty cycle should be used to maximise the ion ﬂux
on the substrate to in turn maximise the energy ﬂow on the growing coating
to promote its crystallinity. This is tested by a simple planar Langmuir
probe placed on the substrate holder and DC-biased to -100 V to reach the
saturated ion current. The pulsing frequency is varied from 0 kHz (i.e. DC)
to 350 kHz. Duty cycles of 65 % and 85 % are tested for each frequency. The
dependency of the ion ﬂux on the pulsing frequency for both duty cycles
is plotted in Fig. 3.1. Increased pulsing frequency results in an increase of
the ion current. Also, the lower duty cycle shows a higher ion current. This
is understandable as the power was always the same, it is concentrated in
a shorter part of the pulse in case of the lower duty cycle. This results in
a denser plasma leading to higher ionisation of the argon as well as, to a
certain extent, of the sputtered species. Moreover, the literature search revealed
that not only the ion number increases with higher pulsing frequency.
Also, the ion energy distribution signiﬁcantly broadens, and three diﬀerent
energy populations are present. This results in yet higher energy ﬂow to the
substrate. Therefore, conditions with the highest ion current are chosen for
the depositions in our work — 350 kHz pulsing frequency with 65 % duty
cycle. Three targets are usually sputtered – molybdenum, carbon and B4C.
The pulsed DC output is connected to the carbon target as the vast majority
of the ions are argon ions and the highest power is always supplied to
the carbon target because of its low sputter yield. Several depositions are
carried out from a single compound Mo2BC target connected to the pulsed
DC power supply.
26
0 5 0 1 0 0 1 5 0 2 0 0 2 5 0 3 0 0 3 5 0 4 0 0
0 . 0
2 . 0 x 1 0
1 4
4 . 0 x 1 0
1 4
6 . 0 x 1 0
1 4
8 . 0 x 1 0
1 4
1 . 0 x 1 0
1 5
1 . 2 x 1 0
1 5
1 . 4 x 1 0
1 5
1 . 6 x 1 0
1 5
6 5 % d u t y c y c l e
8 5 % d u t y c y c l e
ionflux[cm
-2
s
-1
]
p u l s e f r e q u e n c y [ k H z ]
Figure 3.1: The ion ﬂux as a function of the pulsing frequency and the duty
cycle.
Once the pulsing conditions are established, the depositions begin. Coatings
with diﬀerent microstructures are prepared by varying the coating composition,
deposition temperature and bias. The microstructure of the coatings
is observed, and the crystallinity is ﬁrst evaluated by X-ray diﬀraction
and later by transmission electron microscopy with selected area electron
diﬀraction. Coatings with three diﬀerent types of microstructure are pre-
pared.
The ﬁrst type of coatings exhibits a small broad diﬀraction peak centred
at ∼ 38° (samples A1 and A2 in paper [9]). Dark ﬁeld transmission
electron microscope (TEM) micrograph reveals only short-range ordering
in the coating, which was conﬁrmed by SAED (see Fig. 3.2a). These nearamorphous
coatings exhibit the hardness of ∼ 20 GPa and elastic modulus of
∼ 280 GPa. The fracture resistance is tested by indentation with the load of
1 N. This load ensures that the Berkovich tip indents the sample to a maximum
depth exceeding the coating thickness. No crack formation is detected
on the load/displacement curves or the surface of the coating in the area
of the residual indent as is shown in Fig. 3.2b. This implied that obtaining
fully crystalline Mo2BC coating is not a strict condition for achieving hard
and fracture resistance coatings. Even the chemical composition of 50 at.%
27
Figure 3.2: Near amorphous Mo-B-C coatings. a) TEM b) residual indent.
of Mo, 25 at.% of B and 25 at.% C is not necessary as similar deformation
behaviour is observed even when the composition is varied.
Samples labelled as C in paper [9] are later divided into two subgroups.
Only sample C2 from this article falls into the ﬁrst subgroup. The ﬁrst
subgroup exhibits chemical composition approximately corresponding to
Mo2BC. The X-ray diﬀractograms have a more pronounced narrower asymmetrical
diﬀraction peak centred at ∼ 41°. This peak can be deconvoluted
into two partially overlapping peaks. The base broad peak resembled the one
seen in the near-amorphous coatings. The second one is narrower indicating
the presence of another phase with a higher level of crystallinity. The position
of the sharp peak is close to the reference value for (111) and (080) Mo2BC
diﬀractions. A bright ﬁeld TEM image with the corresponding SEAD pattern
presented in Fig. 3.3a shows a ﬁne columnar microstructure throughout
the whole thickness. The SAED pattern is composed of several diﬀuse rings
with a strong texture. These are indexed as corresponding to Mo2BC phase.
The individual crystallite size is estimated using the Scherrer equation from
the sharp peak of the XRD to be ∼ 4 nm. The small size of the crystallites
also explains the lack of other diﬀraction peaks in the X-ray diﬀractograms.
Due to the relatively large size of the elemental cell, each crystallite is composed
of only a few elemental cells. Therefore, only a very limited periodicity
makes the appearance of weaker diﬀraction peaks highly unlikely. In conclusion,
this subgroup of coatings is composed of Mo2BC nanocrystallites in a
Mo-B-C matrix in a nanocomposite fashion. The hardness of such samples
is higher ∼ 31 GPa probably owing to the nanocomposite structure. TEM
micrograph of a lamella cut from the area of the residual indentation im-
28
Figure 3.3: Nanocomposite Mo-B-C coating. a) TEM b) cross-section view
of a residual indent.
print in the coating prepared on the high-speed steel substrate is presented
in Fig. 3.3b. The high-speed steel substrate shears in several places under
the severe stresses induced by the high-load indentation tests (indicated by
the white arrows). The adhesion of the coating to the substrate proves to
be higher than the internal cohesion and cracks originate from the coating/substrate
interface, speciﬁcally from the places of the substrate failure.
However, the cracks stop propagating in the volume of the coating and do
not reach its surface. This strongly implies excellent fracture resistance of
the nanocomposite coatings.
The last subgroup is characterised by a strong excess of molybdenum in
the coating typically caused by application of bias during the deposition, as
the application of bias cause enhanced re-sputtering of the lighter elements.
Samples C1 and C3 belong to this group. The X-ray diﬀractograms exhibit
a similarly broad base peak as the ﬁrst two groups indicating the presence
of a amorphous Mo-B-C pahse. However, another yet harper peak is superimposed
and is shifted to the one seen in the nanocomposite group. The
presence of this peak sharp together with a small peak at ∼ 59° indicates the
presence of crystalline Mo phase. This is conﬁrmed by the electron diﬀraction
pattern presented in Fig. 3.4a. The TEM ﬁgure also revealed that the
coating is predominantly crystalline. The size of the Mo crystallites calculated
by the Scherrer equation was ∼ 6 nm. Therefore, these coatings are
prevalently nanocrystalline with lower amount of an amorphous phase. They
exhibit the hardness of ∼ 21 GPa, but the elastic modulus is higher than
that of the near-amorphous coatings. A lamella from the area under the
29
Figure 3.4: Nanocrystalline Mo-B-C coating. a) TEM b) cross-section view
of a residual indent.
residual indentation imprint is shown in Fig. 3.4b. The cermet substrate is
signiﬁcantly deformed, and the coating complies. Despite this large deformation,
the coating exhibits good adhesion and no delamination under the
imprint is observed. Furthermore, no cracks are detected in the bulk of the
coating. The fully plastic deformation response even under such severe deformation
without the presence of cracking in the coating proves high fracture
resistance of the prevalently nanocrystalline coatings.
In summary not only the predicted crystalline Mo2BC coatings exhibit
high hardness coupled with enhanced ductility. All of the prepared amorphous,
nanocomposite, as well as prevalently nanocrystalline coatings, show
hardness higher than 20 GPa with no cracks in the area of residual imprints
after high-load tests. State of the art protective coatings such as TiN, TiAlN
or TiB2 tested in the same manner fail and exhibit severe cracking. Cracks
in the bulk of deposited coatings are detected only if the substrate itself
shears. If the substrate is plastically deformed without shearing, the coating
complies, and no cracks in the bulk of the coating are detected. Such
material can pave the way to next generation of protective coatings for the
most demanding applications where deformation of the coated workpiece is
involved.
3.3 Thermal stability of the Mo-B-C coatings
The last paper covered in this work is focused on the study of the changes
in the structure and mechanical properties of the Mo-B-C coatings induced
by annealing of the sample. This is important to obtain the value of the
crystallization temperature, to identify the crytalline phase or phases and
30
to understand the changes of the mechanical properties once it is reached.
An amorphous sample in the as-deposited state with the composition close
to Mo2BC was chosen for this study as the changes in the structure should
be the most pronounced. Samples deposited on high-speed steel were tested
up to 600°C and samples deposited on tungsten carbide-cobalt cermet up to
1000°C.
The microstructure is evaluated by XRD. No signiﬁcant changes of the Xray
diﬀractogram up to 900°C are observed. Diﬀraction peaks corresponding
to Mo2C phase appear when the coating is annealed to 1000°C. These results
would indicate that the mechanical properties would not change up to 900°C.
This is not the case. The as-deposited coatings on both substrates exhibit
the hardness of ∼ 20 GPa. The annealing to 500°C and 600°C leads to
a distinct increase to ∼ 24 GPa in case of coatings on the high-speed steel
substrate and ∼ 23 GPa in case of coatings on the cermet substrate. The
small diﬀerence between the substrates can be explained by diﬀerent coefﬁcients
of thermal expansion of both substrates generating higher thermal
stress in case of the steel substrate. The same explanation cannot be used
to account for the diﬀerence between the as-deposited coating and the coating
annealed to 500°C. The load/displacement curves can shed some light
on this problem. Whereas the load/displacement curve in the case of the
as-deposited coating is smooth, the same curve in the case of the sample
annealed at 500°C exhibits several pop-ins typical for crystalline materials.
Therefore, even though the annealing temperature is relatively low, the areas
of the short-range ordering serve as seeds for crystallite growth. The size
of the crystallites is probably in the range of 1 – 2 nm, which is too small for
the X-ray diﬀraction to produce distinct peaks. The hardness of the coating
further increases to ∼ 31 GPa after annealing to 1000°C probably due to the
formation of Mo2C.
The fracture resistance is tested in the same manner as is described in
the previous section. The as-deposited coating do not exhibit any cracks in
or around the residual indentation imprint. Small picture-frame cracks in
the imprint are observed when the coating is annealed to 600°C. A distinct
cracks more than 5 µm long cracks from the corners of the imprint are
seen when the sample is annealed to 1000°C. Therefore, the hardness of the
deposited coating can be enhanced by post-deposition annealing, however,
this hardness increase is at the cost of fracture resistance decrease.
31
Bibliography
[1] T. Schmidtov´a, P. Souˇcek, P. Vaˇsina, Study of hybrid PVD–PECVD
process of Ti sputtering in argon and acetylene, Surf. Coat. Technol.
205 (2011) S299–S302
[2] T. Schmidtov´a, P. Souˇcek, V. Kudrle, P. Vaˇsina, Non-monotonous evolution
of hybrid PVD–PECVD process characteristics on hydrocarbon
supply, Surf. Coat. Technol. 232 (2013) 283–289
[3] P. Vaˇsina, P. Souˇcek, T. Schmidtov´a, M. Eli´aˇs, V. Burˇs´ıkov´a, M. J´ılek,
M. J´ılek jr., J. Sch¨afer, J. Burˇs´ık, Depth proﬁle analyses of nc-TiC/aC:H
coating prepared by balanced magnetron sputtering, Surf. Coat.
Technol. 205 (2011) S53–S56
[4] P. Souˇcek, T. Schmidtov´a, L. Z´abransk´y, V. Burˇs´ıkov´a, P. Vaˇsina,
O. Caha, M. J´ılek, A.A. El Mel, P.Y. Tessier, J. Sch¨afer, J. Burˇs´ık,
V. Peˇrina, R. Mikˇsov´a, Evaluation of composition, mechanical properties
and structure of nc-TiC/a-C:H coatings prepared by balanced
magnetron sputtering, Surf. Coat. Technol. 211 (2012) 111–116
[5] P. Souˇcek, T. Schmidtov´a, L. Z´abransk´y, V. Burˇs´ıkov´a, P. Vaˇsina,
O. Caha, J. Burˇs´ık, V. Peˇrina, R. Mikˇsov´a, Y.T. Pei, J.Th.M. De
Hosson, On the control of deposition process for enhanced mechanical
properties of nc-TiC/a-C:H coatings with DC magnetron sputtering at
low or high ion ﬂux, Surf. Coat. Technol. 255 (2014) 8–14
[6] R. ˇZemliˇcka, P. Souˇcek, P. Vogl, M. J´ılek, V. Burˇs´ıkov´a, P. Vaˇsina, Y.T.
Pei, On the signiﬁcance of running-in of hard nc-TiC/a-C:H coating
for short-term repeating machining, urf. Coat. Technol. 315 (2017) 17–
23
32
[7] P. Souˇcek, J. Daniel, J. Hnilica, K. Bern´atov´a, L. Z´abransk´y, V.
Burˇs´ıkov´a, M. Stupavsk´a, P. Vaˇsina, Superhard nanocomposite ncTiC/a-C:H
coatings: The eﬀect of HiPIMSon coating microstructure
and mechanical properties, Surf. Coat. Technol. 311 (2017) 257–267
[8] J. Daniel, P. Souˇcek, K. Bern´atov´a, L. Z´abransk´y, M. Stupavsk´a, V.
Burˇs´ıkov´a, P. Vaˇsina, Investigation of the Inﬂuence of Ni Doping on
the Structure and Hardness of Ti-Ni-C Coatings, J. Nanomater. 2017
(2017) 6368927
[9] L. Z´abransk´y, V. Burˇs´ıkov´a, P. Souˇcek, P. Vaˇsina, J. Burˇs´ık, On the
study of the mechanical properties of Mo-B-C coatings, Eur. Phys. J.
Appl. Phys. 75 (2016) 24716(7pp)
[10] L. Z´abransk´y, V. Burˇs´ıkov´a, P. Souˇcek, P. Vaˇsina, J. Dug´aˇcek, P.
Sˇtahel, J. Burˇs´ık, M. Svoboda, V. Peˇrina, Thermal stability of hard
nanocomposite Mo-B-C coatings, Vacuum 138 (2017) 199–204
33
Author’s publications not discussed in this work
[11] J. Daniel, P. Souˇcek, L. Z´abransk´y, V. Burˇs´ıkov´a, M. Stupavsk´a, P.
Vaˇsina, On the eﬀect of the substrate to target position on the properties
of titanium carbide/carbon coatings, Surf. Coat. Technol in press,
doi: 10.1016/j.surfcoat.2017.06.076
[12] M. Kotilainen, R. Krumpolec, D. Franta, P. Souˇcek, T. Homola, D.C.
Cameron, P. Vuoristo, Hafnium oxide thin ﬁlms as a barrier against
copper diﬀusion in solar absorbers, Sol. Energ. Mat. Sol. C. 166 (2017)
140–146
[13] R. ˇZemliˇcka, M. J´ılek, P. Vogl, J. ˇSr´amek, P. Souˇcek, V. Burˇs´ıkov´a,
P. Vaˇsina, Principles and practice of an automatic process control for
the deposition of hard nc-TiC/a-C:H coatings by hybrid PVD-PECVD
under industrial conditions, Surf. Coat. Technol. 304 (2016) 9–15
[14] J. Burˇs´ık, V. Burˇs´ıkov´a, P. Souˇcek, L. Z´abransk´y, P. Vaˇsina, Nanostructured
Mo-B-C coatings, Rom. Rep. Phys. 68 (2016) 1069–1075
[15] T. Homola, V. Burˇs´ıkov´a, T.V. Ivanova, P. Souˇcek, P.S. Maydannik,
D.C. Cameron, J.M. Lackner, Mechanical properties of atomic layer
deposited Al2O3/ZnO nanolaminates, Surf. Coat. Technol. 284 (2015)
198–205
[16] L. Z´abransk´y, V. Burˇs´ıkov´a, J. Daniel, P. Souˇcek, P. Vaˇsina, J.
Dug´aˇcek, P. Sˇtahel, O. Caha, J. Burˇs´ık, V. Peˇrina, Comparative analysis
of thermal stability of two diﬀerent nc-TiC/a-C:H coatings, Surf.
Coat. Technol. 267 (2015) 32–39
[17] P. Souˇcek, T. Schmidtov´a, V. Burˇs´ıkov´a, P. Vaˇsina, Y.T. Pei, J.Th.M.
De Hosson, O. Caha, V. Peˇrina, T. Mikˇsov´a, P. Malinsk´y, Tribological
properties of nc-TiC/a-C:H coatings prepared by magnetron sputtering
at low and high ion bombardment of the growing ﬁlm, Surf. Coat.
Technol. 241 (2014) 64–73
[18] A.A. El Mel, E- Gautron, F. Christien, B. Angleraud, A. Granier, P.
Souˇcek, P. Vaˇsina, V. Burˇs´ıkov´a, M. Takashima, N. Ohtake, H. Akasuka,
T. Suzuki, P.Y. Tessier, Titanium carbide/carbon nanocomposite
hard coatings: A comparative study between various chemical analysis
tools, Surf. Coat. Technol. 256 (2014) 41–46
34
[19] R. ˇZemliˇcka, M. J´ılek, P. Vogl, P. Souˇcek, V. Burˇs´ıkov´a, J. Burˇs´ık, P.
Vaˇsina, Understanding of hybrid PVD–PECVD process with the aim
of growing hard nc-TiC/a-C:H coatings using industrial devices with
a rotating cylindrical magnetron, Surf. Coat. Technol. 255 (2014) 118–
123
[20] L. Z´abransk´y, V. Burˇs´ıkov´a, P. Souˇcek, P. Vaˇsina, T. Gardelka, P.
Sˇtahel, O. Caha, V. Peˇcina, J. Burˇs´ık, Study of the thermal dependence
of mechanical properties, chemical composition and structure of
nanocomposite TiC/a-C:H coatings (Reprinted from Surface & Coatings
Technology, vol 242, pg 62-67, 2014), Surf. Coat. Technol. 255
(2014) 158–163
[21] J. ˇSperka, P. Souˇcek, J.J.W.A. Van Loon, A. Dowson, C. Schwartz,
J. Krause, Y. Butenko, G. Kroesen, V. Kudrle, Hypergravity synthesis
of graphitic carbon nanomaterial in glide arc plasma, Mater. Res. Bull
54 (2014) 61–65
[22] L. Z´abransk´y, V. Burˇs´ıkov´a, P. Souˇcek, P. Vaˇsina, T. Gardelka, P.
Sˇtahel, O. Caha, V. Peˇcina, J. Burˇs´ık, Study of the thermal dependence
of mechanical properties, chemical composition and structure of
nanocomposite TiC/a-C:H coatings, Surf. Coat. Technol. 242 (2014)
62–67
[23] J. ˇSperka, P. Souˇcek, J.J.W.A. Van Loon, A. Dowson, C. Schwartz, J.
Krause, Y. Butenko, G. Kroesen, V. Kudrle, Hypergravity eﬀects on
glide arc plasma, Eur. Phys. J. D. 67 (2013) 261
[24] F. Amato, N.R. Panyala, P. Vaˇsina, P. Souˇcek, J. Havel, Laser desorption
ionisation quadrupole ion trap time-of-ﬂight mass spectrometry of
titanium-carbon thin ﬁlms, Rapid Commun. Mass Spectrom. 27 (2013)
1196–1202
[25] L. Z´abransk´y, V. Burˇs´ıkov´a, J. Burˇs´ık, P. Vaˇsina, P. Souˇcek, O. Caha,
M. J´ılek, V. Peˇrina, Study of the indentation response og nanocomposite
n-TiC/a-C:H coatings, Chemick´e listy 106 (2012) S1508–S1511
[26] M. Eli´aˇs, P. Souˇcek, P. Vaˇsina, Inﬂuence of N2 and CH4 on deposition
rate of boron based thin ﬁlms prepared by magnetron sputtering,
Chemick´e listy 102 (2008) S1506-S1509
35
Attached commented publications
36
Study of hybrid PVD–PECVD process of Ti sputtering in argon and acetylene
Tereza Schmidtová a,
⁎, Pavel Souček a
, Petr Vašina a
, Jan Schäfer b
a
Department of Physical Electronics, Faculty of Science, Masaryk University, Kotlářská 2, CZ-61137, Brno, Czech Republic
b
INP Greifswald e.V., Felix-Haussdorff-Str.2, D-17489, Greifswald, Germany
a b s t r a c ta r t i c l e i n f o
Article history:
Received 10 September 2010
Accepted in revised form 16 February 2011
Available online 26 February 2011
Keywords:
Hybrid PVD–PECVD process
Reactive magnetron sputtering
Hysteresis
Modelling
Hybrid PVD–PECVD process of target sputtering in hydrocarbon containing atmosphere combines aspects of
both conventional reactive magnetron sputtering (PVD) and plasma enhanced chemical vapour deposition
(PECVD). Such process is being typically used for deposition of metal carbides embedded in hydrogenated
carbon matrix. Compared to the conventional co-sputtering of metal and carbon targets, in the hybrid
deposition process the source of the carbon is dissociated hydrocarbon vapour in plasma. The aim of this
paper is to study the extent of similarities or differences between this hybrid process and the conventional
reactive magnetron sputtering. We have chosen the sputtering of titanium target in acetylene containing
atmosphere as a representative of the hybrid processes. We focused on experimental measurements of the
hybrid PVD–PECVD process behaviour, the time necessary for the process to achieve steady-state conditions
and basic modelling of the process.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
Hybrid deposition processes [1–3] combine various techniques for
deposition of thin ﬁlms. Hybrid PVD–PECVD process of a target
sputtering in a hydrocarbon containing atmosphere combines aspects
of both conventional reactive magnetron sputtering (PVD) and
plasma enhanced chemical vapour deposition (PECVD) [4–6]. Such
hybrid processes are being used for synthesis of metal carbon
nanocomposite layered structures [7–10]. The source of the carbon
in the hybrid PVD–PECVD process is dissociated hydrocarbon vapour
and not the carbon target as in co-sputtering experiments. The main
research in this ﬁeld is focused on the study of properties of deposited
material and the process features are often neglected.
Adding acetylene in the hybrid process is very like adding oxygen or
nitrogen into the conventional reactive magnetron sputtering and
therefore there is a rightful question of similarity or difference between
these two processes. Reactive magnetron sputter deposition in oxygen/
nitrogen containing atmosphere is widely used for deposition of various
materials. In general, reactive magnetron sputtering controlled by a
supply ﬂow of the reactive gas suffers with undesirable hysteresis
behaviour. Ina certain range of reactive gas supply there exist two modes
of operation: the so-called metal mode and the compound mode [11].
The aim of this paper is an experimental study of the behaviour of
hybrid PVD–PECVD process and basic modiﬁcation of the current
model for the reactive magnetron sputtering [12] to involve aspects of
the studied hybrid PVD–PECVD process. We present the experimental
measurements of the hybrid process behaviour and discuss the time
necessary to achieve steady-state conditions during this process and
report on the hybrid process modelling using the approach of Berg
model [12].
2. Experimental conﬁguration
Measurements were done on the commercial magnetron sputtering
system Alcatel SCM 650 — the experimental set-up drawing can be found
at [13]. A titanium target (purity 99.99%) 20 cm in diameter was
sputtered downwards being mounted on a well balanced magnetron
head. The distance between the target and substrate was 7 cm. The argon
ﬂow was kept constant at 20 sccm for all experiments. For the study of
the hybrid PVD–PECVD process acetylene gas was used. For comparison
with the conventional reactive magnetron sputtering we have chosen
oxygen and nitrogen. The gas inlet position was located below the
substrate holder [13] however the pump entry covered both the region
bellow and above the substrate holder. Thus the amount of gas entering
the zone between the target and the substrate was lower than total
measured supply.Plasma was ignitedbya RFgenerator at 900 W.We had
chosen the RF discharge on purpose in order to eliminate the arcing
caused by a formation of dielectric carbon rich layer on the target surface.
The non-invasive widely used technique of optical emission
spectroscopy was used as the diagnostic tool. Jobin-Yvon Triax 320
and Avantes SD2000 spectrometers were used.
3. Results
Argon and titanium lines were always present in optical emission
spectra. For the reactive sputtering in oxygen also the oxygen line at
Surface & Coatings Technology 205 (2011) S299–S302
⁎ Corresponding author.
E-mail address: dorian@physics.muni.cz (T. Schmidtová).
0257-8972/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.surfcoat.2011.02.047
Contents lists available at ScienceDirect
Surface & Coatings Technology
journal homepage: www.elsevier.com/locate/surfcoat
777 nm was observed. In the case of hybrid sputtering with acetylene
only very weak lines of atomic carbon at 248 nm and molecular band
of CH radical at 431 nm were observed together with stronger atomic
hydrogen lines Hα, Hβ depending on the acetylene supply.
For both reactive sputtering and hybrid process we chose the Ti
line at 365 nm and studied its evolution on oxygen, nitrogen or
acetylene ﬂow. The measured data were normalised to reach 100% of
Ti line intensity for no supply of oxygen, nitrogen or acetylene and
clean target. In Fig. 1 are shown the measured dependencies of Ti line
intensity on oxygen and acetylene ﬂow. Each point corresponds to
steady-state conditions that were reached after setting of the desired
gas ﬂow. The starting point for increasing ﬂows was the clean target
and for decreasing ﬂows the target poisoned by a high amount of gas.
The hysteresis region for oxygen case (upper graph) is wide
because of a difference between titanium sputtering yield from the
cleaned part of the target and from the poisoned part of the target
(TiO2). The transition between the metal and the compound mode is
given by newly generated process conditions in the whole reactor
when a critical reactive gas ﬂow is reached. On the other hand the
hybrid process of titanium sputtering in argon and acetylene behave
differently than reactive magnetron sputtering (lower graph). Ti line
intensity is decreasing and increasing linearly on the amount of
acetylene ﬂow. Clearly no hysteresis is observed despite that the
sputtering yields of TiC [14] and C [15] are lower than of Ti [14,15].
In the hybrid PVD–PECVD process the addition of hydrocarbon gas
into argon sputtering of metal target results in formation of carbon
rich layers on the target surface as well as on all other reactor surfaces.
Depending on experimental conditions, we have observed black, thick
carbon rich rings surrounding the racetrack area of the target. Apart
from the racetrack the target was all covered by carbon rich layer.
In the reactive magnetron sputtering the only option how to getter
oxygen from the volume is a reaction at the target surface or with
sputtered titanium atoms at the substrate area. In the hybrid process
the acetylene dissociation takes place in magnetised plasma and
carbon and carbon fragments diffuse to reactor surfaces and attach.
These thick carbon rich layers on all reactor surfaces can store a large
amount of carbon to such an extent that in the volume the presence of
carbon containing radicals could be minimal, as was experimentally
observed. Therefore the hysteresis behaviour would be strongly
suppressed.
Another important issue during conventional reactive sputtering is
the ability of the process to readily respond to changes. New steadystate
conditions are achieved usually very fast after the change of the
gas ﬂow. This is a key problem for the process control in general
[16,17]. The ability of the hybrid process to absorb carbon in already
carbon rich layers can lead to signiﬁcant time prolongation of
achieving the steady-state conditions or can even cause a time drift
of the deposition process itself as carbon rich layers grow thicker and
thicker.
In Fig. 2 are shown the time evolutions of the Ti line intensity for
oxygen, nitrogen and acetylene. If 8 sccm of oxygen, nitrogen or
acetylene is added into the reactor, the steady-state conditions are
reached within 1 min. For the case of oxygen and nitrogen the
transition from the metal to the compound mode occurs in very short
time. If lower ﬂow than that corresponding to the transition is set, the
steady-state conditions are attained within several minutes. However
for lower ﬂow of acetylene the characteristic time is prolonged up to
tens of minutes as can be seen in lower graph of Fig. 2. These
characteristic times have to be taken into account during depositions
because until the steady-state conditions are reached the coating
composition can vary.
Since the hybrid PVD–PECVD process does not show the hysteresis
region, there is no reason to intentionally poison the target before the
deposition to operate it in the mode similar to the compound mode of
the reactive magnetron sputtering. Nevertheless, it would be useful to
know the characteristic times in this case. The target was at ﬁrst
intentionally fully poisoned by a high amount of oxygen, nitrogen or
acetylene applied for the period of 3 min and the desired ﬂow was set
after that. We observed that the reactive sputtering achieves the
steady-state in minutes, however the hybrid PVD–PECVD process
Fig. 1. Hysteresis behaviour of Ti line intensity on oxygen ﬂow for conventional reactive
magnetron sputtering (upper graph) and linear dependency of Ti line intensity on the
acetylene ﬂow for the hybrid process.
Fig. 2. Time evolution of Ti line intensity for oxygen, nitrogen and acetylene (upper
graph) and for different acetylene ﬂows (lower graph).
S300 T. Schmidtová et al. / Surface & Coatings Technology 205 (2011) S299–S302
reaches the steady-state in the order of tens of minutes. This is caused
by the formation of thin compound layers of TiN, TiO2 on the target
that are quickly out-sputtered compared to the thick carbon rich
layers formed during the 3 min of the target poisoning. If we let the
target to poison longer than 3 min, the times of achieving the steadystate
conditions of the PVD–PECVD process will be prolonged.
4. Hybrid process modelling
The hybrid process behaviour has been modelled for the steadystate
conditions using essentials of Berg model [12] which is widely
used for modelling of hysteresis behaviour of reactive magnetron
sputtering. Shortly, the Berg model is based on a description of the
target poisoning using an equation of balance between formation and
out-sputtering of the poisoned part of the target in steady-state
conditions. Another balance equation is written for the substrate.
Fluxes of metallic and compound species towards the substrate are
given by percentage size of poisoned part of the target multiplied by
corresponding sputtering yields. The reactive gas supply either
contributes to the formation of compounds at target and substrate
or leads to increase of reactive gas partial pressure. The surplus of
the reactive gas is pumped out from the system by a pump of given
pumping speed.
If we approach the hybrid process as conventional reactive
sputtering as in Berg model where the C can react with Ti exclusively,
forming TiC in a monolayer only, due to difference in sputtering yields
of Ti [14,15] and TiC [14] the hysteresis would occur as a result of
reduced supply of metal. However the observed behaviour of hybrid
process was different.
For the purpose of basic modelling of the hybrid PVD–PECVD
process the assumption of thick carbon layer forming on all surfaces
was included into Berg model. We assumed total C2H2 dissociation
only into C and H described by a dissociation degree. We then
presumed the total carbon consumption on all surfaces in contrast to
the Berg assumption of reactive gas consumption only on the clean
parts of the target and the substrate where titanium is exposed. The
assumption of the total carbon consumption at the target and the
substrate enabled the formation of thick carbon layers, in other words
the carbon fragments were allowed to attach also to already formed
carbon layers.
The equations expressing these ideas were compiled as in Berg
model. Our algorithm steps follow the equation order in given paper
[12]. Following model parameters were used. The target area 0.03 m2
and the pumping speed 0.04 m3
s−1
were experimentally measured.
The current corresponding to argon ion ﬂux towards the cathode 0.5 A
was deduced from reactive magnetron sputtering behaviour for the
same discharge power. The substrate area of 0.3 m2
was estimated
from reactor geometry. Sputtering yields of Ti 0.2 and C 0.03 were
taken from literature [15] for 200ceV of Ar+
that corresponds to RF
self-bias measured in our experiment. We assumed gas temperature
300 K. According to literature [18], the hydrogen content in a-C:H
matrix of n-TiC/a-C:H ﬁlm is typically ~10%. Low hydrogen content in
formed carbon rich layers facilitates the attachment of carbon
fragments and further growth [19,20]. In our model, the sticking
coefﬁcients of carbon were kept one for simplicity. Further, if the
sticking coefﬁcient of C on C was decreased by 20%, the overall
difference in model output was not more than 15% for higher ﬂows of
acetylene and even much lower for lower ﬂows.
In Fig. 3 the observed behaviour of the hybrid PVD–PECVD process
is plotted and it is compared to hybrid process simulations for
different dissociation degrees. We assumed that the modelled
sputtered Ti ﬂux corresponds to the measured Ti line intensity.
Clearly, no hysteresis was modelled. The shape of the modelled
dependency is mainly inﬂuenced by the dissociation degree, the
discharge current, the available consumption area and the pumping
speed. The best experimental ﬁt is for 25% dissociation degree. However,
the dissociation degree of C2H2 in the discharge zone is higher
due to the distant gas inlet that caused reduced amount of acetylene
actually entering the discharge region.
5. Conclusions
The hybrid PVD–PECVD process is unlike the conventional reactive
magnetron sputtering. We have experimentally observed differences
between the hysteresis behaviour of the conventional reactive
magnetron sputtering of titanium target in oxygen/nitrogen containing
atmosphere and the hybrid PVD–PECVD process of titanium target
sputtering in argon and acetylene. No hysteresis region for the hybrid
sputtering was found. This is an interesting fact from the industrial point
of view because the hybrid process does not require any process control
which is vital during conventional reactive magnetron deposition.
We presume forming of thick hydrogenated carbon rich layers
from hydrocarbon radicals on all surfaces close to the electrode where
the dissociation of acetylene takes place. Depositing coating is thus
created not only by sputtering of titanium and carbon from the target
surface but also by plasma induced polymerisation of hydrocarbon
radicals directly at the substrate.
We have discussed the characteristic time needed for the hybrid
process to achieve the steady-state conditions with ﬁndings that for
higher amount of added acetylene the faster the steady-state
conditions are reached. For higher ﬂows the steady-state conditions
are reached in within 1 min but for lower ﬂows the time prolongs
even to tens of minutes. The hybrid process behaviour has been
basically modelled for the steady-state conditions using essentials of
Berg model. The leading assumption of growing of thick carbon rich
layers on surfaces caused the hysteresis suppression in the model.
Acknowledgements
This work has been supported by GACR 104/09/H080, 202/08/
P038 and MSM contracts 0021622411 and OP R&DI CZ.1.05/2.1.00/
03.0086. Pavel Souček acknowledges as a holder of Brno PhD Talent
Financial Aid.
References
[1] J.P. Celis, D. Drees, M.Z. Huq, P.Q. Wu, M. De Bonte, Surf. Coat. Technol. 113 (1999)
165.
[2] A. Matthews, A. Leyland, Surf. Coat. Technol. 71 (1995) 88.
[3] A. Daniel, T. Duguet, T. Belmonte, Appl. Surf. Sci. 253 (2007) 9323.
[4] J.T. Harnack, F. Thieme, C. Benndorf, Surf. Coat. Technol. 39 (40) (1989) 285.
[5] J.T. Harnack, C. Benndorf, Mater. Sci. Eng. A140 (1991) 764.
[6] C. Benndorf, T.J. Harnack, U. Fell, Surf. Coat. Technol. 48 (1993) 345.
[7] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, A. Cavaleiro, Surf. Coat. Technol. 198 (2005)
44.
[8] P.Y. Tessier, R. Issaoui, E. Luais, M. Boujtita, A. Granier, B. Angleraud, Solid State Sci.
11 (2009) 1824.
Fig. 3. Experimental Ti line intensity and modelled ﬂux of sputtered Ti atoms on
acetylene ﬂow for different dissociation degrees of C2H2.
S301T. Schmidtová et al. / Surface & Coatings Technology 205 (2011) S299–S302
[9] W. Kulisch, P. Colpo, F. Rossi, D.V. Shtansky, E.A. Levashov, Surf. Coat. Technol.
118–119 (2004) 714.
[10] W. Gulbinski, S. Mathur, H. Shen, T. Suszko, A. Gilewicz, B. Warcholinski, Appl.
Surf. Sci. 239 (2005) 302.
[11] S. Berg, H.-O. Blom, T. Larsson, C. Nender, J. Vac. Sci. Technol. A 5 (1987) 202.
[12] S. Berg, T. Nyberg, Thin Solid Films 476 (2005) 215.
[13] M. Eliáš, P. Souček, P. Vašina, Chem. Listy 102 (16) (2008) 1506.
[14] O.A. Fouad, A.K. Rumaiz, S.I. Shah, Thin Solid Films 517 (2009) 5689.
[15] Y. Yamamura, H. Tawara, Acad. Press 62 (1996) 149.
[16] T. Kubart, O. Kappertz, T. Nyberg, S. Berg, Thin Solid Films 515 (2006) 421.
[17] B. Bellido-González, B. Daniel, J. Counsell, D. Monaghan, Thin Solid Films 502
(2006) 34.
[18] A. Czyżniewski, W. Prechtl, J. Mater. Process. Technol. 157–158 (2004) 274.
[19] W. Jacob, Thin Solid Films 326 (1998) 1.
[20] M. Eckert, E. Neyts, A. Bogaerts, Chem. Vap. Deposition 14 (2008) 213.
S302 T. Schmidtová et al. / Surface & Coatings Technology 205 (2011) S299–S302
Non-monotonous evolution of hybrid PVD–PECVD process
characteristics onhydrocarbon supply
T. Schmidtová, P. Souček, V. Kudrle, P. Vašina ⁎
Department of Physical Electronics, Kotlářská 2, Masaryk University, 61137 Brno, Czech Republic
a b s t r a c ta r t i c l e i n f o
Article history:
Received 25 February 2013
Accepted in revised form 13 May 2013
Available online 21 May 2013
Keywords:
Magnetron sputtering
Diagnostics of deposition process
Nanocomposites
Metal–carbon coatings
Hybrid PVD–PECVD process of titanium sputtering in argon and acetylene atmosphere combines aspects of
both conventional techniques: sputtering of titanium target (PVD) and acetylene as a source of carbon
(PECVD). This process can be used for preparation of metal carbon nanocomposites (MeC/C(:H)) or DLC
layers doped with metal (DLC:Me). The aim of this paper is to describe and understand elementary processes
inﬂuencing the hybrid PVD–PECVD process. A non-monotonous dependence of cathode voltage and current,
total pressure and spectral line intensities on acetylene supply ﬂow is reported. Explanation of nonmonotonous
evolutions through the analysis of the target state correlating the process characteristics with
properties of coatings prepared by this process is proposed.
© 2013 Elsevier B.V. All rights reserved.
1. Introduction
Hybrid deposition systems combine more than one deposition
technique [1]. These techniques are mutually independent and they
are being used often in separate deposition chambers [1–4]. An overview
of various deposition systems can be found in [2]. The sputtering
of a metal target in a mixture of an inert gas and a hydrocarbon gas
can be classiﬁed as a hybrid process combining PVD and PECVD techniques
[5–8]. The source of carbon in the hybrid PVD–PECVD process
is dissociated hydrocarbon vapour.
Hybrid PVD–PECVD process is being used for synthesis of metal
hydrogenated carbon nanocomposite (MeC/C(:H)) [6,8–13] layers
or DLC layers doped with metal (DLC:Me) [14–17]. For example
nc-WC/a-C:H [18–21] or nc-TiC/a-C:H [9,22–24] are reported to
have good mechanical and tribological properties combining high
hardness, good toughness with low friction coefﬁcient and wear
which makes these materials industrially attractive for different
applications. For the speciﬁc case of the titanium target sputtering
in hydrocarbon containing atmosphere there exits variety of papers
that are focused on the inﬂuence of process parameters on properties
of deposited coating [8,25–29]. Process features are often omitted.
Hybrid PVD–PECVD process has been previously compared to the
conventional reactive magnetron sputtering [30]. Even though the
addition of the hydrocarbon is very similar to the adding of oxygen
or nitrogen as a reactive gas in reactive sputtering, it was found [30]
that hysteresis behaviour typical for reactive magnetron sputtering
described by Berg model [31] was completely suppressed during hybrid
PVD–PECVD process. Therefore, the hybrid PVD–PECVD depositions do
not require complex feedback control. The absence of the hysteresis
region was explained [30] by the capability of surfaces to be covered
by thick carbon rich layers. In these thick carbon rich layers the carbon
can be bound mutually in contrary to metal oxides/nitrides formation
on surfaces during classical reactive magnetron sputtering where
oxygen/nitrogen can be bound only to the metal. Nevertheless the
target covering by these carbon rich layers during the hybrid PVD–
PECVD process is still similar to the target poisoning by a reactive gas
because it prevents the sputtering of the metal. So for description of
both processes the phrases ‘target poisoning’ and ‘target cleaning’ can
be understood and used.
The aim of this paper is to describe and understand elementary
processes inﬂuencing the hybrid PVD–PECVD process governing the
properties of the deposited coatings. We report a non-monotonous
dependence of cathode voltage and current, total pressure and spectral
line intensities on acetylene supply ﬂow. We report on the dependence
of the chemical composition and mechanical properties of the
coatings on the acetylene ﬂow. We provide an explanation of nonmonotonous
dependences through the analysis of the target state.
The role of RF bias on the substrate and magnetron DC power on
the process characteristics is also discussed.
2. Experimental conﬁguration
Experimental measurements were done on semi-industrial magnetron
sputtering system Alcatel SCM 650. Experimental setup drawing
is shown in Fig. 1. A titanium target (purity 99.99%) 20 cm in
diameter was mounted on a well balanced magnetron head, sputteringup
geometry was used. The magnetron plasma was well localized near
the target, it does not extent to the substrate. The target-to-substrate distance
was 7 cm. The chamber was pumped by turbomolecular pump
Surface & Coatings Technology 232 (2013) 283–289
⁎ Corresponding author: Tel.: +420 549 496 479.
E-mail address: vasina@physics.muni.cz (P. Vašina).
0257-8972/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.surfcoat.2013.05.014
Contents lists available at ScienceDirect
Surface & Coatings Technology
journal homepage: www.elsevier.com/locate/surfcoat
(700 l s−1
) backed by Roots pump. The base pressure was typically
10−4
Pa. During the deposition the pump was throttled to the
pumping speed of 40 l s−1
. The argon (purity 99.999%) ﬂow was
kept constant at 20 sccm for all experiments. The typical working
pressure was around 1 Pa.
For the study of the hybrid PVD–PECVD process acetylene (purity
99.6%) was used as carbon source. The acetylene ﬂow ranged from 0 to
16 sccm. Magnetized plasma was ignited by DC power (1 kW or
2 kW) combined with or without RF substrate bias (225 W ~ –100 V).
To enable comparison of the deposition process and deposited coating
properties, the values of DC power, RF bias and supply ﬂows of Ar and
C2H2 were selected to correspond to values used in paper [23,24]
where properties of deposited n-TiC/a-C:H coatings were studied in
details. The DC generator was operated in constant power mode. Prior
to each experiment, the target was thoroughly cleaned both mechanically
and by Ar bombardment until the discharge voltage typical for
pure Ti target was reached.
Spectrometer Avantes SD2000 was used as a diagnostic tool.
Atomic spectral lines of titanium at 365 nm, argon at 420 nm and hydrogen
at 656 nm were chosen for the study. There were no lines attributed
to carbon or other fragments of acetylene in observed
spectral range of 250–670 nm. Discharge parameters used for electrical
diagnostics were the target voltage and the discharge current. The
total pressure in the vacuum chamber was also recorded using MKS
Baratron.
The depositions were carried out for 45 min at constant acetylene
ﬂow on high speed steel (HSS) and cermet tungsten carbide (WC)
substrates. The DC power to the magnetron was set at the desired
value and the substrate bias was −100 V. Chemical composition of
the coatings was determined by Rutherford backscattering (RBS)
and elastic recoil detection analysis (ERDA) methods using a Van de
Graaff generator and TANDETRON with linear electrostatic accelerators.
Fischerscope H100 depth sensing indenter equipped with a
Berkovich tip was used to study the indentation response of the coating
samples. The hardness and the indentation modulus were evaluated
using the standard procedure proposed by Oliver and Pharr
[32]. The EDX measurements were performed on JEOL JSM 6460
with Inca Energy EDX detector at 20 kV acceleration voltage.
For the study of the target state the following experimental settings
were used. The depositions for various C2H2 ﬂows were performed at
conditions described in [23,24]. Target photo-documentation was
done after each opening of the vacuum chamber. In order to quantify
the target state on the evolution on the acetylene supply the pixel
counting for target areas of interest were done for each taken photograph.
The pixel counting determined the relative amount of studied
parts. The pixel count was done at the racetrack part only, because the
racetrack, the target erosion zone, plays the key role for the magnetron
discharge. For the EDX (Energy-dispersive X-ray spectroscopy) study
of the target racetrack coverage by Ti–C phases, Ti square samples
(1 cm × 1 cm) were used and placed on the racetrack zone. One edge
of Ti sample was placed in the racetrack centre and the other at the
racetrack outer edge.
3. Results and discussion
3.1. Process characteristics
Typical overall hybrid PVD–PECVD process characteristics are
given in Fig. 2. There are shown the dependences of cathode voltage
and discharge current together with total pressure and spectral emission
line intensities on the acetylene supply ﬂow for 2 kW DC power
applied on the magnetron target with −100 V RF bias.
Before measurement of each point, the target was thoroughly
cleaned by Ar bombardment without the presence of acetylene to
ensure the same target conditions and to eliminate the inﬂuence of
previous measurements. The cleaning of the target covered by thick
carbon rich layers takes long time [30]. The steady-state conditions
of the hybrid PVD–PECVD process are reached in order of tens of
minutes after acetylene injection. In Fig. 2 the full black square points
denote the measured values with increasing acetylene ﬂows — on the
way from the clean target. When measuring with the decreasing
acetylene ﬂows, the cleaned target was exposed for 3 min to
30 sccm of C2H2 for full poisoning. The Fig. 2 the hollow red circle
points denote measured values with decreasing acetylene ﬂows —
on the way from the poisoned target.
Fig. 2 demonstrates that the hybrid PVD–PECVD process does not
exhibit hysteresis behaviour. The electrical characteristics in Fig. 2(a)
and (b) are non-monotonous; a certain local minimum in voltage
and local maximum in discharge current evolutions is observed. The
discharge voltage and current are increasing and decreasing respectively
while the DC generator was operated in constant power mode
meaning that the current and voltage dependencies are inverted.
There is also no rapid evolution of total pressure in the chamber
with acetylene ﬂow, see Fig. 2(c). The pressure is increasing with increasing
acetylene ﬂow almost linearly.
Dependence on acetylene supply ﬂow of hydrogen, titanium and
argon emission lines given by optical emission spectra are shown in
Fig. 2(d)–(f). In Fig. 2(d) the non-monotonous evolution of hydrogen
alpha line is plotted. It is increasing rapidly with increasing acetylene
ﬂow but for higher ﬂows it is sharply decreasing, a maximum of
hydrogen line intensity is observed for certain acetylene supply
ﬂow. The evolution of titanium emission line in Fig. 2(e) is different,
it is almost linear. Looking at the argon emission line, see Fig. 2(f),
its intensity is slightly decreasing, increasing and then decreasing as
acetylene ﬂow increases.
From the general overview of presented data, one can conclude
that there exist three distinctive zones in which studied parameters
evolve in a characteristic way. For 2 kW DC power applied on the
magnetron target there is zone I from 0 to approximately 6 sccm of
C2H2, zone II approximately from 6 to 12 sccm of C2H2 and zone III
for higher ﬂows than 12 sccm of C2H2.
The zone I is characterized by the increasing discharge voltage,
pressure and hydrogen emission intensity and by the increase in
discharge current, titanium emission intensity and slight decrease in
argon emission. The zone II is characterized by the increase in
discharge current, pressure, hydrogen emission intensity and slight
increase in argon emission together with decreasing discharge voltage
and titanium emission. In the zone III again the increase in the
target voltage together with the slight increase in the total pressure
can be observed, the decreasing values of the discharge current, titanium,
hydrogen and argon is characteristic. Summarizing overview of
each parameter evolutions within three deﬁned zones is given in
Table 1.
Fig. 1. Drawing of the experimental setup.
284 T. Schmidtová et al. / Surface & Coatings Technology 232 (2013) 283–289
3.2. Film properties
A set of coatings was prepared using this hybrid PVD–PECVD process
at acetylene ﬂows of 8–14 sccm (i.e. in zones II and III) and analysed
[24]. The chemical composition of the coatings behaves more monotonously
as can be seen in Fig. 3. Titanium content decreases while carbon
content increases with higher acetylene ﬂow. Hydrogen content remains
stable b10% and also the oxygen contamination remains low
for the whole deposition range. The stoichiometric composition where
a) Discharge voltage b) Discharge current
c) Total pressure d) Hydrogen emission
e) Titanium emission f) Argon emission
Fig. 2. Dependences of discharge voltage, discharge current, total pressure and spectral emission lines on acetylene ﬂow. Black full square points are values obtained with increasing
acetylene ﬂow – on the way from clean target and red hollow circular points are values obtained with decreasing acetylene ﬂow – on the way from poisoned target.
Table 1
Summarizing overview of each parameter evolutions within zones I–III.
Voltage
(current)
Spectral intensity Relative density
H Ti Ar H Ti Ar
Zone I ↗ (↘) ↗ ↘ ↘ ↗ ↗ →
Zone II ↘ (↗) ↗ ↗ ↗ ↗ ↘ →
Zone III ↗ (↘) ↘ ↘ ↘ → ↗ →
285T. Schmidtová et al. / Surface & Coatings Technology 232 (2013) 283–289
Ti/C = 1 was achieved for acetylene ﬂow of ~10.5 sccm, which belongs
to aforementioned zone II.
The mechanical properties show a distinct peak at acetylene ﬂow
of 12 sccm (see Fig. 4). At this point the hardness reaches 46 ±
2 GPa and Young's modulus 414 ± 7 GPa. The mechanical properties
decrease with both lower and higher acetylene ﬂows. The ﬂow of
12 sccm lies on the border between zones II and III.
More detailed study of the properties of the thin ﬁlms deposited
by the hybrid PVD–PECVD process including evolution of the deposition
rate, residual stress, microstructure, etc. with acetylene supply
was presented in the associated paper [24].
3.3. The state of the target
For reactive magnetron sputtering, it was reported that the
electrical characteristics are related to the state of the target; the
target poisoning by a compound of different coefﬁcient of secondary
electron emission from the original target material results in a change
in the target voltage [33]. In classical reactive magnetron sputtering
the cathode voltage decreases/increases monotonously with the
partial pressure of the reactive gas as there is a monotonous and continuous
coverage of the target surface by a compound. However the
electrical characteristics in Fig. 2(a) and (b) have distinctive evolution
on the acetylene ﬂow. A characteristic local minimum (maximum)
in voltage (current) evolution is observed. As the pressure is evolving
in Fig. 2(c) only a little its inﬂuence on the electrical characteristics
should be negligible and the presence of the local minimum (maximum)
should be related to the evolution of the state of the target in
a different way during hybrid PVD–PECVD process than during classical
reactive magnetron sputtering.
For hybrid PVD–PECVD process a target surface photography is
shown in Fig. 5. The target was covered by three differently ‘coloured’
regions. The following can be distinguished: a metallic region, a black
region, and a grey region. The metallic region is of the same colour as
the original titanium target, so it is denoted as titanium (Ti). The black
layer which was easily removable by hand or an abrasive resembles
the soots and so it is denoted as carbon (C). The grey region which
is matt, compact and very difﬁcult to remove by an abrasive is a transition
between the metallic and black regions. It is similarly coloured
to titanium carbide so this region is denoted as titanium carbide (TiC).
It was of interest to see the target surface changes after sputtering
in the same conditions for various acetylene ﬂows. In Fig. 6 are shown
photographs of three target halves for selected 3, 9 and 13 sccm of
acetylene ﬂows. For better visualization the target racetrack is
delimited by red highlights. For 3 sccm of C2H2 the target racetrack
was mostly Ti, for 9 sccm of C2H2 the target was mainly grey — mainly
TiC and for 13 sccm of C2H2 the racetrack was carbon rich with black
and grey regions. Clearly some evolution of the target state with the
acetylene ﬂow is visible.
Calculation of the relative target racetrack coverage by Ti, TiC and
C parts as was done by the pixel counts as described in the experimental
section. In Fig. 7 shown are racetrack coverage percentages
Fig. 3. Chemical composition of nc-TiC/a-C:H coatings depending on acetylene ﬂow
during deposition.
Fig. 4. Mechanical properties of nc-TiC/a-C:H coatings depending on acetylene ﬂow
during deposition.
Fig. 5. Photograph of the target after sputtering by 2 kW DC power with RF substrate
bias −100 V for 12 sccm of acetylene at steady-state conditions. Along the racetrack
region three differently ‘coloured’ regions are visible. The pure metallic region is
denoted as Ti, the grey as TiC and black as a-C.
Fig. 6. Photographs of three selected target halves for 3, 9 and 13 sccm of acetylene
after achievement of steady-state conditions during deposition. The target racetrack
region is highlighted.
286 T. Schmidtová et al. / Surface & Coatings Technology 232 (2013) 283–289
for the three studied parts and the results are also correlated with the
three zone behaviour of voltage characteristics.
In the zone I from 0 to approximately 6 sccm of C2H2 Ti regions are
diminishing on behalf of TiC regions, the racetrack is titanium rich. In
the zone II approximately from 6 to 12 sccm of C2H2 the racetrack is
composed mainly by TiC region. For higher ﬂows of C2H2 in the
zone III the TiC parts are overtaken by C, the racetrack becomes
carbon rich. Combining information from discharge voltage characteristics
with the racetrack coverage we can conclude that the voltage
increase in the zone I is connected with the creation of TiC on the
target racetrack. In the zone II, the voltage slightly decreases with
acetylene inﬂow (by about 5%) and the voltage local minimum is
observed at 11 sccm of acetylene supply. In the zone II, mainly the
presence of the grey colour TiC was detected in the racetrack region
(see Fig. 7) leading us to a conclusion that the voltage decrease should
be related with the evolution of TiC Invoking the phase diagram of
Ti–C [34,35], we suppose the voltage decrease to be caused by the
evolution of the TiC stoichiometry in the racetrack region from the
substoichiometric TiC (at 6 sccm of C2H2) to stoichiometric TiC
(around 11 sccm of C2H2). The rapid voltage increase with higher
acetylene ﬂows in the zone III is clearly related to the carbon formation
in the racetrack. Its composition is therefore carbon rich TiC/C.
Therefore increasing of the acetylene supply ﬂow results in continuous
phase change from Ti-rich Ti/TiC (zone I) to mainly single phase
TiC (zone II) and ﬁnally to carbon-rich TiC/C(zone III). The evolution
of the target state is similar to phase changes in the growing TiC/C
ﬁlms with increase carbon content as can be found for example in
[10,35].
For the better understanding of the racetrack coverage by Ti–C
phases the analysis was performed on the Ti square samples placed
on the cathode surface. The results for the three studied zones of acetylene
ﬂows are shown in Fig. 8. Note that the EDX method gathers the
information from approximately 1 C2H2 volume of the top layer of the
analysed material. As the thickness of the target layer which has been
modiﬁed by the process is unknown and probably varies with acetylene
supplies, only qualitative information of Ti and C compositions
can be derived. From Fig. 8 the racetrack homogeneity is clearly
observed. For low acetylene ﬂows in the zone I the Ti/C ratio is very
high and approaching the zone II the Ti and C abundances become
equal. In the zone III where the amorphous carbon enters the racetrack
region a clear evolution of composition is seen from racetrack
centre to its edge where Ti/C ratio is reversed.
In our case, the maximal hardness of 46 ± 2 GPa and indentation
modulus of 415 ± 7 GPa were measured for coatings with 55 at.% C
deposited at the conditions close to the frontiers between zones II
and III, where the target surface is covered by stoichiometric TiC.
According to the generic design concept [36,37], this high hardness
is based on the combination of the absence of dislocation activity in
the small TiC nanocrystals and blocking of a-C:H grain boundary
sliding by the formation of a strong interface between the two phases.
Thus, the coatings with maximal hardness should show certain excess
of carbon over titanium to form nanocomposite structure. In hybrid
PVD–PECVD process, both the target and the substrate are being
exposed to the ﬂux of carbonaceous species originating from acetylene
dissociations. Simultaneously to target poisoning, the target is
being sputter cleaned providing additional ﬂux of carbon atoms on
the substrate. Similarly to reactive magnetron sputtering [31], higher
relative amount of carbon could be expected to be found in the growing
ﬁlm than at the target surface which has been experimentally
proved for sputtering of Ti in mixture of Ar and CH4 [25]. In our experimental
conﬁguration, deposition process at conditions between
zones II and III caused the optimal conditions for the growth of the
nc-TiC/a-C:H with the maximal hardness. Depending on the experimental
conditions, such as the magnetron head design and position
of argon and acetylene gas inlet, the conditions to deposit coatings
with maximal hardness could be slightly shifted, however they should
be expected to be located always close to the frontiers between the
zones II and III, where the target is covered by TiC and coating composition
is slightly overstoichiometric with respect to carbon.
3.4. Relative evolution of concentrations of Ar, H and Ti in plasma
Assuming direct excitation of a ground state particle by an electron
impact, the intensity of a spectral line depends on the concentration
of electrons in the plasma bulk, concentration of ground state
particles and excitation rate which is a factor proportional to overlap
of excitation cross section and electron energy distribution function
(EEDF). The ﬂux of ions to the sheath edge is proportional to the ion
density (which is equal to the electron density in the plasma) times
the ion acoustic velocity, typically with a factor of 0.6 according to
Bohm presheath diffusion criterion [38]. The ions are then accelerated
by the sheath potential and bombard the cathode/target surface
causing secondary electron emission and sputtering. Thus, the evolution
of the concentration of electrons in the plasma is reﬂected by the
discharge current. Assuming that electron energy distribution function,
i.e. excitation rate and ion acoustic velocity does not depend
on C2H2 supply, ratio of the atomic line intensities and discharge
current provides the evolution of the ground state particle densities
in plasma.
Fig. 7. Racetrack coverage by Ti, TiC and C on acetylene supply. Voltage characteristics
with three zones behaviour is correlated with plotted racetrack coverage.
Fig. 8. EDX analysis of three Ti square samples (1 cm × 1 cm) placed in the target racetrack
during the deposition at different acetylene ﬂows. Position 0 marks the racetrack
centre, position 11 marks the racetrack outer edge.
287T. Schmidtová et al. / Surface & Coatings Technology 232 (2013) 283–289
The evolution of the relative concentrations of Ar, H and Ti in
plasma volume is plotted in Fig. 9 together with marked zones I, II
and III. The comparison of evolutions of the relative concentrations
with other process parameters can be found in Table 1. Despite that
the Ar line intensity evolves with C2H2 ﬂow the calculated Ar concentrations
are constant for all zones supporting the hypotheses on EEDF
and ion acoustic velocity to be independent on C2H2 supply. Decrease
of Ar line intensity in zone I, increase in zone II and decrease in
zone III that follow the evolution of the discharge current perfectly
(compare Fig. 2(b) and (f)) is caused by evolution of the plasma
density and not by the evolution of the buffer gas concentration
with the acetylene supply. The titanium concentration is decreasing
almost linearly though all zones showing that by increasing the
C2H2 ﬂow the amount of sputtering Ti is decreasing linearly without
any rapid transition.
The hydrogen concentration is increasing in zones I and II and
despite that the well pronounced maximum of hydrogen line intensity
was observed at the frontiers between the zones II and III, in the zone
III the hydrogen concentration is not increasing anymore but it stays
rather independent on C2H2 supply. The decrease of hydrogen emission
in the zone III (see Fig. 2(d)) is not caused by decrease of hydrogen
atom concentration (see Fig. 9) but should be related to the steep
decrease of discharge current (i.e. plasma density) linked with a steep
increase in the discharge voltage to maintain the discharge (see
Fig. 2(a)) while the generator was operating in the constant power
mode. For high amount of acetylene in the volume it is difﬁcult to
estimate the extent of the volume reactions of acetylene fragments.
The constant hydrogen atom density in the zone III could be caused
by two opposite effects that act simultaneously — an increase of
acetylene supply is followed by the reduced ability of the plasma to
dissociate acetylene molecules and produce hydrogen atoms due to
lower plasma density (lower discharge current, higher voltage to
sustain plasma). It is therefore possible that the production of atomic
hydrogen is constant or even decreasing with the acetylene supply
in the zone III. Other possible explanation for constant hydrogen
atom concentration with acetylene supply could be that the substrate
ﬁlm is growing with high content of carbon matrix where signiﬁcant
amount hydrogen could be stored. However the ﬁlm analysis in the
acetylene in the range between 9 and 14 sccm showed, that the
hydrogen content in the ﬁlms is constant around 5% [23,24].
3.5. Inﬂuence of magnetron power and substrate bias
The effect of DC power applied on the magnetron target on the
hybrid PVD–PECVD process evolution on acetylene supply was studied.
By lowering the applied DC power from 2 kW to 1 kW the general
trends (three zones behaviour) observed at electrical characteristics
Fig. 2(a), (b) and Fig. 7 are preserved, see Fig. 10. The three zones
are only shifted towards lower acetylene ﬂows, as applying half
power on the magnetron target shifts the boundaries of three zones
roughly to half the acetylene ﬂows.
The effect of applying RF bias to the substrate can also be seen in
Fig. 10. Substrate bias of −100 V was set by 225 W RF power which
was a quarter of the 1 kW DC power applied on the magnetron target.
Such relatively high RF power applied on the substrate could be
inﬂuencing the evolution of hybrid PVD–PECVD process on acetylene
supply. Fig. 10 demonstrates that the three zones of the target state
and entire process characteristics are not shifted with respect to
C2H2 ﬂow providing only negligible inﬂuence of RF plasma near the
substrate on the evolution of the trends of deposition process
characteristics.
4. Conclusions
The hybrid PVD–PECVD process of titanium sputtering in argon
and acetylene atmosphere does not show the hysteresis behaviour.
The absence of hysteresis region avoids process stability problems
met in conventional reactive magnetron sputtering making the
hybrid PVD–PECVD process relatively easy to control. Three distinctive
zones were identiﬁed in the process evolution with the acetylene
supply.
For the low range of acetylene ﬂows (zone I) the voltage is
increasing due to TiC formation on the surface of the target racetrack
at the expense of Ti and the relative concentration of Ti in plasma is
decreasing while the H concentration is increasing due to acetylene
fragmentation in plasma volume. For medium range of acetylene
ﬂows (zone II) the target voltage stops increasing or it is slightly
decreased. The racetrack is covered mainly by TiC. Ti concentration
in plasma phase is still decreasing and hydrogen concentration continues
to increase. Stoichiometric TiC coatings are being deposited
at this zone. For high acetylene ﬂows (zone III) the target voltage is
rapidly increasing, the carbon layers start to penetrate into the target
racetrack and the composition of the racetrack surface becomes carbon
rich. The Ti concentration in plasma phase is still decreasing but
H concentration is not evolving further, it remains more or less constant.
On the border of zones II and III the nc-TiC/a-C:H coatings
with highest hardness and Young's modulus were prepared. For the
studied range of acetylene supply the Ar concentration remains the
same.
Reduction of the power applied on the target keeps the three zone
behaviour of the process. It shifts zone boundaries towards lower
acetylene ﬂows. Halving the magnetron power shifts the boundaries
to half of the acetylene ﬂows. RF bias does not signiﬁcantly inﬂuence
Fig. 9. Dependence of the relative concentrations of Ti, Ar and H in plasma on acetylene
supply.
Fig. 10. Voltage characteristics on the acetylene supply ﬂow for 1 kW DC with and
without applied RF bias (225 W ~ –100 V).
288 T. Schmidtová et al. / Surface & Coatings Technology 232 (2013) 283–289
the hybrid PVD–PECVD process characteristics. By varying the RF bias
for constant magnetron power desired self-bias on the substrate can
be set without modifying the overall process behaviour.
Acknowledgement
This research has been supported by Czech Science Foundation
contract 104/09/H080 and 205/12/0407 and R&D centre for lowcost
plasma and nanotechnology surface modiﬁcations CZ.1.05/
2.1.00/03.0086 funded by European Regional Development Fund.
Pavel Souek acknowledges the Brno City Municipality as holder of
Brno PhD Talent Financial Aid. The authors would like to acknowledge
Dr. V. Peřina from the Nuclear Physics Institute of the ASCR for
RBS measurements and Dr. J. Buršík from the Institute of Physics of
Materials of the ASCR for EDX measurements.
References
[1] D.M. Mattox, The Foundations of Vaccum Coating Technology, William Andrew
Publishing, 2003. , (ISBN 1-800-932-7045).
[2] J.P. Celis, D. Drees, M.Z. Huq, P.Q. Wu, M. De Bonte, Surf. Coat. Technol. 113 (1999)
165.
[3] A. Matthews, A. Leyland, Surf. Coat. Technol. 71 (1995) 88.
[4] A. Daniel, T. Duguet, T. Belmonte, Appl. Surf. Sci. 253 (2007) 9323.
[5] J.T. Harnack, F. Thieme, C. Benndorf, Surf. Coat. Technol. 39 (40) (1989) 285.
[6] J.T. Harnack, C. Benndorf, Mater. Sci. Eng. A140 (1991) 764.
[7] C. Benndorf, T.J. Harnack, U. Fell, Surf. Coat. Technol. 48 (1993) 345.
[8] P.Y. Tessier, R. Issaoui, E. Luias, M. Boujtita, A. Granier, B. Angleraud, Solid State
Sci. 11 (2009) 1824.
[9] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, A. Cavaleiro, Surf. Coat. Technol. 198 (2005)
44.
[10] W. Gulbinski, S. Mathus, H. Shen, T. Suszko, A. Gilewicz, B. Warcholinski, Appl.
Surf. Sci. 239 (2005) 302.
[11] W. Kulisch, P. Colpo, F. Rossi, D.V. Shtansky, E.A. Levashov, Surf. Coat. Technol.
118–119 (2004) 714.
[12] T. Ghodselahi, M.A. Vesaghi, A. Shaﬁekhani, A. Baradaran, A. Karimi, Z. Mobini,
Surf. Sci. Technol. 202 (2008) 2731.
[13] J. Bruckner, T. Mantyla, Surf. Coat. Technol. 59 (1993) 166.
[14] C. Donnet, Surf. Coat. Technol. 100–101 (1998) 180.
[15] B. Oral, K.-H. Ernst, C.J. Schmutz, Diamond Relat. Mater. 5 (1996) 932.
[16] N. Yao, A.G. Evans, C.V. Cooper, Surf. Coat. Technol. 179 (2004) 306.
[17] G. Ma, S. Gong, G. Lin, L. Zhang, G. Sun, Appl. Surf. Sci. 258 (2012) 3045.
[18] S. Zhou, L. Wang, S.C. Xang, Q. Xue, Appl. Surf. Sci. 257 (2011) 6971.
[19] S.J. Park, K.-R. Lee, D.-H. Ko, K.Y. Eun, Diamond Relat. Mater. 11 (2002) 1747.
[20] A. Czyzniewski, Thin Solid Films 433 (2003) 180.
[21] C. Rebholtz, J.M. Schneider, H. Ziegele, B. Rahle, A. Leyland, A. Matthews, Vacuum
49 (1998) 265.
[22] Y. Wang, X. Zhang, X. Wu, H. Zhang, X. Zhang, Appl. Surf. Sci. 254 (2008) 5085.
[23] P. Vasina, P. Soucek, T. Schmidtova, M. Elias, V. Bursikova, M. Jilek, M. Jilek Jr., J.
Schafer, J. Bursik, Surf. Sci. Technol. 205 (2011) S53.
[24] P. Soucek, T. Schmidtova, L. Zabransky, V. Bursikova, P. Vasina, O. Caha, M. Jilek, A.
El Mel, P.Y. Tessier, J. Schafer, J. Bursik, V. Perina, R. Miksova, Surf. Coat. Technol.
211 (2012) 111.
[25] J.-E. Sundgren, B.-O. Johansson, S.-E. Karlsson, Thin Solid Films 105 (1983) 353.
[26] J.-E. Sundgren, B.-O. Johansson, S.-E. Karlsson, H.T.G. Hentzell, Thin Solid Films
105 (1983) 367.
[27] J.-E. Sundgren, B.-O. Johansson, H.T.G. Hentzell, S.-E. Karlsson, Thin Solid Films
105 (1983) 385.
[28] O.A. Fouad, A.K. Rumaiz, S.I. Shah, Thin solid Films 517 (2009) 5689.
[29] V.Yu. Kulikovsky, F. Fendrych, L. Jastrabik, D. Chvostova, Surf. Coat. Technol. 91
(1997) 122.
[30] T. Schmidtova, P. Soucek, P. Vasina, J. Schafer, Surf. Coat. Technol. 205 (2011)
S299.
[31] S. Berg, T. Nyberg, Thin Solid Films 476 (2005) 215.
[32] W.C. Oliver, G.M. Pharr, J. Mater. Res. 19 (2004) 3.
[33] D. Depla, G. Buyle, J. Haemer, R. DeGryse, Surf. Coat. Technol. 200 (2006) 4329.
[34] K. Frisk, Calphad 27 (2003) 367.
[35] D.A. Andersson, P.A. Korzhavyi, B. Johansson, Calphad 32 (2008) 543.
[36] S. Veprek, S. Reiprich, Thin Solid Films 268 (1995) 64.
[37] S. Veprek, J. Vac. Sci. Technol. A 17 (1999) 2401.
[38] D. Bohm, in: A. Guthrie, R.K. Wakerling (Eds.), McGraw-Hill, New York, 1949,
p. 77.
289T. Schmidtová et al. / Surface & Coatings Technology 232 (2013) 283–289
Depth proﬁle analyses of nc-TiC/a-C:H coating prepared by balanced
magnetron sputtering
Petr Vašina a
, Pavel Souček a,
⁎, Tereza Schmidtová a
, Marek Eliáš a
, Vilma Buršíková a
, Mojmír Jílek b
,
Mojmír Jílek Jr. b
, Jan Schäfer c
, Jiří Buršík d
a
Department of Physical Electronics, Faculty of Science, Masaryk University, Kotlářská 2, CZ-61137, Brno, Czech Republic
b
SHM Šumperk, Průmyslová 3020/3, CZ-787 01, Šumperk, Czech Republic
c
INP Greifswald e.V., Felix-Hausdorff-Str. 2, D-17489, Greifswald, Germany
d
Institute of Physics of Materials, Academy of Sciences of the Czech Republic, Žižkova 22, CZ-61662, Brno, Czech Republic
a b s t r a c ta r t i c l e i n f o
Available online 18 February 2011
Keywords:
Composite
Mechanical properties
Magnetron
Carbon
Titanium
Nanocrystalline titanium carbide embedded in an hydrogenated amorphous carbon matrix (nc-TiC/a-C:H)
shows high hardness and Young's modulus together with low wear and low friction coefﬁcient. In this paper,
we report on the preparation of well adherent nc-TiC/a-C:H coatings ~5 μm thick on stainless steel substrates
using a well balanced magnetic ﬁeld conﬁguration and only very low power RF bias on the substrate. Hardness
and Young's modulus of these coatings are 43 GPa and 380 GPa, respectively. The mechanical properties –
hardness and Young's modulus – measured from the coating's top reach the values obtained at optimized
experiments where the unbalanced magnetic ﬁeld conﬁguration was used. A simple method of depth
proﬁling suitable for evaluation of mechanical properties of several micrometers thick coatings is developed
and employed. The paper reports on the depth proﬁle analyses of the coating hardness, Young's modulus,
composition and morphology.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
Nanocomposite materials consisting of nanocrystalline titanium
carbide embedded in (hydrogenated) amorphous carbon matrix – ncTiC/a-C(:H)
show interesting mechanical properties such as high
hardness and fracture toughness combined with low friction and
wear. Typically n-TiC/a-C(:H) coatings show maximum hardness up
to 40 GPa, reduced Young's modulus up to 300 GPa [1], low friction
coefﬁcient [2,3] around 0.1 and wear rate below [4] 2∙10−7
mm3
/Nm.
A usually reported considerable deﬁciency of these coatings is the
insufﬁcient adhesion to the substrates, which is often improved by
using either Cr [5,6], Ti [7] or graded adhesion promoting layer [6,8].
These materials are usually produced either by the co-sputtering of
titanium and graphite targets [2,5,9–11] or by the sputtering of the
titanium target in a hydrocarbon containing environment [1,3,5,8,12].
While depositing nc-TiC/a-C(:H) ﬁlms, most of the authors use either
the unbalanced [1], the close ﬁeld unbalanced [5,8] magnetic ﬁeld
conﬁguration or the IPVD concept [3,12] (Ionized PVD – deposition
ﬂux contains signiﬁcant amount of ionized sputtered species [13])
which results in very high ion bombardment of the growing ﬁlm
during the deposition. The typical coating thickness ranges from
several hundreds of nm up to 3 μm.
In our experiment the sputtering of the titanium target in a
hydrocarbon (acetylene) containing environment was performed
using the well balanced magnetic ﬁeld conﬁgured magnetron and a
low power RF substrate bias (~0.03 W/cm2
) in order to prepare ncTiC/a-C:H
coatings at much lower ion bombardment. The goal was to
reduce internal stress in the coating and prepare well adherent and
hard thin ﬁlms with a thickness of at least 5 μm on Si and stainless
steel substrates. As the ion bombardment was reduced, the coating
was heated much more slowly during the deposition and the
mechanical properties, coating morphology or the composition
could have evolved relatively slowly with the coating thickness. In
this paper we report the depth proﬁle analyses of coatings ~5 μm thick
prepared at the conditions where the studied mechanical properties –
hardness and Young's modulus – measured from the coating top reach
the results obtained at experiments where the unbalanced or even the
close ﬁeld unbalanced magnetic ﬁeld conﬁguration was used.
2. Experimental
The deposition of nc-TiC/a-C:H was performed using an industrial
sputtering system Alcatel SCM 650. A well balanced sputtering source
was mounted by Ti (20 cm in diameter, purity 99.99%) target and was
powered by a DC generator. Si and polished steel substrates were
placed on a rotary biasable (RF 13.56 MHz) substrate holder located
6.5 cm from the Ti target. Prior to each experiment the chamber was
evacuated by a turbomolecular pump backed by a Roots pump to the
Surface & Coatings Technology 205 (2011) S53–S56
⁎ Corresponding author. Tel.: +420 549 49 3062.
E-mail address: soucek@physics.muni.cz (P. Souček).
0257-8972/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.surfcoat.2011.02.038
Contents lists available at ScienceDirect
Surface & Coatings Technology
journal homepage: www.elsevier.com/locate/surfcoat
base pressure in the order of 10−4
Pa. The position of the Ar and C2H2
gas inlet was located between the target and the substrate at 2 cm
distance from the Ti target.
Si (111) and polished steel samples were chemically cleaned before
being inserted into the deposition chamber. In order to always ensure
the same target condition prior to each deposition, the target was
cleaned (mechanically by an abrading agent followed by sputtering to
shutter) until the carbon layer from previous experiments was fully
removed. Several hours later, after cooling down of the chamber, both
the target and the substrates separated by the shutter were always
cleaned for 35 min to reach always the same temperature conditions
[14] at the deposition chamber prior to the deposition process.
The cleaning was performed at the following conditions: 4 Pa of Ar,
1 kW DC, 500 W RF. The deposition procedure then proceeded by the
setting of the following parameters: 20 sccm of Ar, pressure ~1.2 Pa,
2 kW DC, 200 W RF (bias −100 V). Then the shutter was opened and a
3 min deposition of pure titanium adhesion promoting layer ~700 nm
thick was performed. Then 10 sccm of acetylene gas was introduced
and the deposition of the nc-TiC/a-C:H coating started. After 45 min
the discharge and every gas supply was turned off and the samples
were cooled down under vacuum for several hours.
The thicknesses of both the nc-TiC/a-C:H and the adhesion
promoting titanium layer were determined by the means of a Calotest
equipment (Platit CT50). A Rockwell C-scale diamond-tipped indenter
from Meopta was used to evaluate the adhesion of the coating
according to German standard DIN CEN/TS 1071-8. Indentation tests
were performed using a Fischerscope H100 depth sensing indentation
tester equipped with a Berkovich indenter. The applied load varied in
the range from 10 to 1000 mN. The accuracy of the depth measurement
was about ±2 nm. The correct function of the measuring
equipment was veriﬁed on certiﬁcated (DKD Dortmund) BK7
standard. From the load–penetration curves it is possible to determine
the material resistance against plastic deformation HUpl = L max
At
, (so
called plastic hardness), where At is the area of the remained
indentation print created by irreversible deformation under maximum
load Lmax, and the reduced elastic modulus E⁎ =
E
1−ν2
, where ν is the
Poisson's ratio and E is the Young's modulus. The reduced elastic
modulus was evaluated using the standard procedure proposed by
Oliver and Pharr [15]. The hardness HUpl and Young's modulus E values
obtained for on fused silica using the above described method were
9.5 GPa and 72.6 GPa, respectively. The ﬁlms were tested by means of
both, the common method of testing provided for several different
indentation depths (i.e. several different applied loads in the range
from 10 to 1000 mN) and the method of testing the sample in the
abraded dimple from the calotest (calotte) along its diameter with
equidistant steps using applied loads of 10, 20 and 75 mN. The distance
between two indents was chosen so, that it was in each case at least
three times larger, than the estimated radius of plastically deformed
zone. The ﬁrst method enabled us to map the indentation response
from increasing volume of coating material from near surface almost
up to ﬁlm-substrate interface. The second method gave us information
about indentation response of different parts of coatings bulk carried
out at the same testing conditions. The vertical structure of samples
deposited on Si substrate was analyzed by a ﬁeld-emission Scanning
Electron Microscope JEOL JSM-7500Fa, which employs a ﬁeldemission
gun, a semi-in lens conical objective lens and a secondary
electron in lens detector, enabling to perform image at a maximum
speciﬁed resolution of 1.0 nm without the use of a conductive coating.
Operating voltage of the SEM device was 2 kV. The ﬁlm composition
was studied using a scanning electron microscope JEOL JSM 6460
equipped with an Inca Energy (Oxford Instruments) analyser. EDX
spectra collected in the SEM were evaluated using standard ZAF
corrections (referring to the three effects: atomic number Z,
absorption A and ﬂuorescence F) [16]. Raman measurements were
performed using the Jobin-Yvon T64000 system with laser operating at
532 nm. X-ray diffraction was performed using a setup for X-ray
reﬂectometry and diffraction with high collimation.
3. Results and discussions
According to the literature [1,2,5,6,8,11], mechanical properties of
the nanocomposite nc-TiC/a-C:H coatings depend strongly on the
carbon content in the deposited coating. Typically n-TiC/a-C:H
coatings show maximum of the hardness at Ti/C composition ratio
around 2/3 [1]. Recently Martínez-Martínez et al. [11] pointed out the
importance of relative amount of nc-TiC and a-C phases on
mechanical properties as it was shown that the highest hardness
(27 GPa in their particular case) was obtained at the relative amount
of a-C phase about 40%, which is slightly higher than the previous
mentioned Ti/C composition ratio of 2/3.
In order to optimize our deposition process to reach the maximal
hardness of the coating, a set of samples deposited at various acetylene
ﬂow rate was prepared and analysed using the nanoindentation test.
The hardness and Young's modulus of the coating prepared at 10 sccm
of C2H2 were determined from indentation tests as (43±2) GPa and
(380±10)GPa, respectively and were found satisfactory and these
samples were chosen for further research. The hardness and Young's
modulus of the samples were constant within the experimental errors
for indentation loads from 20 to 75 mN, which correspond to
maximum indentation depths from 150 to 410 nm. Nanocomposite
character of the coating was conﬁrmed using X-ray diffraction and
Raman measurement. The size of the TiC nanocrystallites was
estimated to be ~9 nm from X-ray diffraction using the Scherer
formula and the presence of D and G peaks at Raman signal revealed
the amorphous carbon phase.
The adhesion of the coating deposited on the stainless steel sample
was tested using the Rockwell C-scale diamond-tipped indenter and
the coating prepared on the Si sample was broken along the vertical
crystal plane of silicon in order to reveal the coating morphology by
SEM. The results from Rockwell-C tests show (Fig. 1) only ﬁne radial
cracks in the indentation area without delamination in the vicinity of
the crater – typical disruption pattern which can be related to the HF 1
adhesion strength quality according to German standard DIN CEN/TS
1071-8 proving very good adhesion of the coating system to the
polished steel substrates. The applied load was 1500 N. The cross
section SEM image (Fig. 2) demonstrates two structural phases of the
coating deposited on Si samples. The thickness of the coating is ~5 μm.
The Ti layer (700±100 nm) is characterized with a ﬁne vertical
texture, where the mean elementary length of the columns is
approximately 100 nm in the imaged fracture area. The n-TiC/a-C:H
Fig. 1. Residual Rockwell adhesion test crater with ﬁne cracks around its perimeter
signalizing good adhesion of the coating system to the polished steel substrate.
S54 P. Vašina et al. / Surface & Coatings Technology 205 (2011) S53–S56
structure shows more massive columns running through the whole
coating thickness very similar to those reported in the previously
mentioned research [3,5,6,8]. The bottom Ti adhesive layer gradually
merges with the n-TiC/a-C:H coating so no sharp interface exists
between the two layers. The total thickness, Ti adhesion layer and the
n-TiC/a-C:H layer thicknesses were estimated to be 5.6 μm, 0.7 μm and
4.9 μm, respectively.
The coating deposited on the stainless steel sample was studied by
means of the Calotest where a ball of a known diameter is rotated
against the coating surface and the addition of the abrasive slurry
causes a spherically-shaped crater that is abraded through the coating
and into the substrate. Subsequent inspection of the calotte with
optical microscopy revealed the projected surfaces of the abraded
coating and substrate sections as well as an interface between the Ti
adhesive layer and the n-TiC/a-C:H coating. The thickness of the
whole system was determined to be 6.2 μm: the thicknesses of the Ti
adhesion layer and the n-TiC/a-C:H coating were 0.7 μm and 5.5 μm,
respectively.
Moreover, the Calotest crater enabled the study of the thickness
dependence of the indentation response of the coating by indenting
the sample in the calotte along its diameter with equidistant steps.
The experiment was performed at three different loads – 75, 20 and
10 mN. The indent distance from the calotte centre gives the ﬁlm
thickness at the measured position. In Fig. 3 the sets of 75 mN
(horizontal and vertical direction) and 20 mN (visible at angles of
approximately 45° from previous directions) indents are shown. The
set of 10 mN indents is only scarcely visible. It is evident that the
indents become deepest at the positions where the inﬂuence of the
titanium layer dominates. The hardness and Young's modulus of Ti
adhesive layer are substantially lower than hardness and Young's
modulus of both the n-TiC/a-C:H coating and the stainless steel
substrate. Indentation tests at lower loads, for example at 10 mN, give
information about smaller volumes of the tested material, therefore
using such load enabled to study the structure response of the sample.
On the other hand, the fact that these tests sensed the structure
response of deposited material caused higher scattering in the
measured values (see Fig. 4). Moreover, the results obtained by
measurements on the top or near to the sample surface at 10 mN are
more inﬂuenced with the surface structure (e.g. roughness). The
elastic deformation zone is much larger than that of the plastic
deformation. Therefore the inﬂuence of the underlying titanium layer
appeared sooner on Young's modulus dependence than on the
dependence of the hardness. Fig. 4 also shows, that either the changes
in the coating hardness or more probably the substantial inﬂuence of
the titanium coated steel substrate on the measured hardness
appeared when the n-TiC/a-C:H coating thickness decreased to
approximately 0.75 μm. This value is in very good accordance with
the plastic zone radius (radius of irreversibly deformed coating
Fig. 3. The sets of 75 mN (horizontal and vertical direction) and 20 mN (visible at angles
of approximately 45° from previous directions) residual indentation marks in the
calotte. The more distinct residual indents correspond to areas with the lowest
hardness and vice versa.
Fig. 4. Hardness and reduced Young's modulus measured through the calotte at the load of 10 mN.
Fig. 2. SEM cross section image with visible columnar structure of the coating on Si
substrate.
S55P. Vašina et al. / Surface & Coatings Technology 205 (2011) S53–S56
volume) calculated according to Johnson's contact model [17]. It
means that the hardness of the coating reached the value of 40 GPa,
which corresponds well with the hardness measured from the top of
the coating, even earlier, i.e. very fast during the coating growth, at the
latest when the coating reaches the total thickness of ~1.5 μm.
In order to evaluate the sample composition, the EDX measurements
were performed along the calotte diameter. The results in Fig. 5
were obtained with accelerating voltage of 15 kV. The absolute
accuracy of EDX analyses across the calotte is however limited by the
fact, that each ‘point’ analysis is averaged over the interaction volume
(about 1 μm3
). On the other hand, the EDX analysis is a nondestructive
method contrary to techniques based on ion milling. Its
advantage lies in fact, that it enabled us to get the indentation
response from the same place as the EDX line scan was done. This
method of chemical composition analysis was applied mainly because
the interaction volume of the EDX measurement is comparable with
the plastically deformed zone around the indents in the nanoindentation
measurements (estimated according to Johnson's contact model
[17]) for applied loads of 10 and 20 mN. The ﬁlm consisted of titanium
and carbon; the substrate contained of iron, carbon, chromium and
vanadium. Although the carbon content evaluation using EDX is
rather qualitative, from the composition proﬁle in Fig. 5 it is possible
to deduce, that the titanium to carbon ratio slightly increases with
coating thickness. This evolution of composition however does not
signiﬁcantly inﬂuence the mechanical properties of the coating as the
measured hardness reaches steady state value as soon as the total
thickness (Ti+nc-TiC/a-C:H) reaches ~1.5 μm. We can furthermore
see that at the top of the coating a Ti/C ratio of about 45/55 was
reached. The decrease in hardness values obtained from the
measurements at 75 mN (Fig. 5) appears much earlier than in case
of 10 mN, already when the n-TiC/a-C:H coating thickness decreased
to approximately 2.2 μm, what again corresponded well with the
estimated radius of the plastically deformed zone according to
Johnson's contact model [17].
4. Conclusion
The nc-TiC/a-C:H coatings were prepared in the low ion bombardment
conditions by the means of Ti target sputtering in acetylene
containing atmosphere. In order to promote the adhesion, 700 nm Ti
layer was deposited on stainless steel samples prior the deposition of
the 5.5 μm nc-TiC/a-C:H layer. The adhesion of this layered system
was HF 1 according to German standard DIN CEN/TS 1071-8. The
hardness and Young's modulus measured from the top of the coating
were 43 GPa and 380 GPa, respectively. Depth analyses of mechanical
properties were performed by the nanoindentation measurements
throughout the calotte. Hardness reached the value of 40 GPa, which
corresponds well with the hardness measured from the top of the
coating at the latest as the total thickness (Ti+nc-TiC/a-C:H) reached
~1.5 μm. Chemical composition was slightly evolving with thickness
of the coating, however mechanical properties were not signiﬁcantly
affected by this evolution.
Acknowledgements
This research has been supported by the MSM contracts
0021622411, OP R&DI CZ.1.05/2.1.00/03.0086 and by GACR contracts
202/08/P038 and 104/09/H080. Pavel Souček is acknowledged as
Holder of the Brno PhD Talent Financial Aid.
References
[1] A. Czyżniewski, W. Prechtl, J. Mater. Process. Technol. 157–158 (2004) 274.
[2] J.C. Sánchez-López, D. Martínez-Martínez, C. Lópes-Cartes, A. Fernández, Surf.
Coat. Technol. 202 (2008) 4011.
[3] D.M. Cao, B. Feng, W.J. Meng, L.E. Rehn, P.M. Baldo, M.M. Khonsari, Appl. Phys. Lett.
79/3 (2001) 329.
[4] J. Musil, P. Novák, R. Čerstvý, Z. Soukup, J. Vac. Sci. Technol. A 28 (2010) 244.
[5] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, A. Cavaleiro, Surf. Coat. Technol. 198 (2005)
44.
[6] D. Galvan, Y.T. Pei, J.Th.M. De Hosson, Acta Mater. 53 (2005) 3925.
[7] Y. Hu, L. Li, X. Cai, Q. Chen, P.K. Chu, Diamond Relat. Mater. 16 (2007) 181.
[8] D. Galvan, Y.T. Pei, J.Th.M. De Hosson, J. Vac. Sci. Technol. A 24 (4) (2006) 1441.
[9] W. Gulbinski, S. Mathur, H. Shen, T. Suszko, A. Gilewicz, B. Warcholinski, Appl.
Surf. Sci. 239 (2005) 302.
[10] H. Wang, S. Zhang, Y. Li, D. Sun, Thin Solid Films 516 (2008) 5419.
[11] D. Martínez-Martínez, C. López-Cartes, A. Fernández, J.C. Sánchez-López, Thin
Solid Films 517 (2009) 1662.
[12] A.A. El Mel, B. Angleraud, E. Gautron, A. Granier, P.Y. Tessier, Surf. Coat. Technol.
204 (2010) 1880.
[13] J. Hopwood, Phys. Plasmas 5 (1998) 1624.
[14] P. Vašina, T. Hytková, M. Eliáš, Plasma Sources Sci. Technol. 18 (2009) 0250118
(8pp).
[15] W.C. Oliver, G.M. Pharr, J. Mater. Res. 19 (2004) 3.
[16] J. Goldstein, D.E. Newbury, D.C. Joy, C.E. Lyman, P. Echlin, E. Lifshin, L. Sawyer, J.R.
Michael, Scanning Electron Microscopy and X-Ray Analysis, third ed., Kluwer,
New York, 2003.
[17] K.L. Johnson, J. Mech. Phys. Solids 18 (1970) 115.
Fig. 5. Chemical composition measured by EDX and the hardness measured by
nanoindentation at two different loads (10 and 75 mN) as a function of the coating
thickness.
S56 P. Vašina et al. / Surface & Coatings Technology 205 (2011) S53–S56
Evaluation of composition, mechanical properties and structure of nc-TiC/a-C:H
coatings prepared by balanced magnetron sputtering
Pavel Souček a,
⁎, Tereza Schmidtová a
, Lukáš Zábranský a
, Vilma Buršíková a
, Petr Vašina a
, Ondřej Caha b
,
Mojmír Jílek c
, Abdelaziz El Mel d
, Pierre-Yves Tessier d
, Jan Schäfer e
, Jiří Buršík f
,
Vratislav Peřina g
, Romana Mikšová g
a
Department of Physical Electronics, Faculty of Science, Masaryk University, Kotlářská 2, CZ-61137, Brno, Czech Republic
b
Department of Condensed Matter Physics, Faculty of Science, Masaryk University, Kotlářská 2, CZ-61137, Brno, Czech Republic
c
SHM Šumperk, Průmyslová 3020/3, CZ-787 01, Šumperk, Czech Republic
d
Université de Nantes, CNRS, Institut des Matériaux Jean Rouxel, UMR 6502, 2 rue de la Houssinière B.P. 32229-44322 Nantes cedex 3, France
e
INP Greifswald e.V., Felix-Hausdorff-Str. 2, D-17489, Greifswald, Germany
f
Institute of Physics of Materials, Academy of Sciences of the Czech Republic, Žižkova 22, CZ-61662, Brno, Czech Republic
g
Nuclear Physics Institute, Academy of Sciences of the Czech Republic, v.v.i., Řež 130, CZ-25068, Řež, Czech Republic
a b s t r a c ta r t i c l e i n f o
Article history:
Received 6 May 2011
Accepted in revised form 8 September 2011
Available online 16 September 2011
Keywords:
Nanocomposite
Magnetron sputtering
Titanium
Carbon
Mechanical properties
Nanocomposites like nc-TiC/a-C:H show promising combinations of properties such as high hardness and
Young's modulus combined with low friction and wear. These properties can be furthermore tailored to
suit particular industrial needs. In this paper, we report on the preparation and analyses of several μm
thick and well adherent nc-TiC/a-C:H coatings with different stoichiometry. These coatings were deposited
in a relatively low ion bombardment environment using magnetron sputtering of titanium target operated in
well-balanced magnetic ﬁeld conﬁguration on high speed steel and hardmetal substrates. Gaseous hydrocarbon
(acetylene) was added to the deposition chamber as source of carbon. Hardness and Young's modulus of 46 GPa
and 415 GPa respectively were reached.
© 2011 Elsevier B.V. All rights reserved.
1. Introduction
nc-TiC/a-C(:H) nanocomposite is composed of TiC nanocrystallites
embedded in an amorphous (hydrogenated) carbon matrix. Depending
on the amount of titanium atoms added to a-C and on the synthesis
procedure the composition varies from pure titanium via TiC known
since 1960s [1] to almost pure DLC ﬁlm doped with Ti atoms. The
properties of nc-TiC/a-C(:H) ﬁlms can be tailored from very hard
ﬁlms with hardness reaching over 40 GPa and Young's modulus
higher than 300 GPa respectively [2,3] to low friction ﬁlms [4,5] with friction
coefﬁcient around 0.1 and wear rate lower than 2∙10−7
mm3
/Nm
[6]. However the adhesion of nc-TiC/a-C(:H) ﬁlms to industrially attractive
substrates is often poor. An adhesion promoting interlayer is
routinely introduced into the coating system to overcome this
problem. Usually chromium [7,8], titanium [9] or graded [8] interlayers
are employed. Usual synthesis procedures of these coatings are either
co-sputtering of graphite and titanium targets [4,7,10-12] or sputtering
of titanium target in an hydrocarbon (mostly methane or acetylene)
containing environment [2,5,7,13]. Other methods of nc-TiC/a-C(:H)
are also reported such as ﬁltered vacuum arc [3], sputtering combined
with pulsed laser deposition [14], carbon sputtering combined
with titanium evaporation [1] etc. The coatings are deposited at wide
range of ion bombardment conditions which are unfortunately not
always completely speciﬁed. Reports on process using balanced [5,7],
unbalanced magnetic ﬁeld conﬁguration sputtering [2,11], closed ﬁeld
unbalanced sputtering [7,8,16,17] or an IPVD concept [5,13,15] can be
found. Biased [2,3,7-9,11,15-20] or ﬂoating [4,7,9,12,13] substrates are
both used. The thicknesses of these coatings range typically from
several hundred nanometers up to about 3 μm.
Sputtering of a titanium target in acetylene (C2H2) containing
atmosphere was used in our experiments. In order to prepare
nc-TiC/a-C:H coatings with thickness around 5 μm or more, thus
coatings with low internal stress and good adhesion, a wellbalanced
magnetic ﬁeld conﬁguration, relatively low RF substrate
bias power and deposition of Ti adhesion layer was employed.
Acetylene ﬂow was varied to obtain ﬁlms with different chemical
compositions. The structure and mechanical properties of these
coatings were thoroughly analyzed to observe the dependencies
of composition, structure and mechanical properties on the acetylene
ﬂow [2,6,7,9-15]. The composition of these coatings was optimized in
order to obtain the highest possible hardness.
Surface & Coatings Technology 211 (2012) 111–116
⁎ Corresponding author. Tel.: +420 549 49 3062.
E-mail address: soucek@physics.muni.cz (P. Souček).
0257-8972/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.surfcoat.2011.09.012
Contents lists available at SciVerse ScienceDirect
Surface & Coatings Technology
journal homepage: www.elsevier.com/locate/surfcoat
2. Experimental
The depositions were carried out using industrial sputtering system
Alcatel SCM 650. A well balanced sputtering source equipped with
titanium target (purity 99.99%, 20 cm in diameter) was driven by
Huettinger TruPlasma Bipolar 4010 generator operated in DC mode.
High speed steel and hardmetal (cemented tungsten carbide) samples
were placed on a biasable (13.56 MHz RF) substrate holder located
6.5 cm from the target. The deposition chamber was evacuated by a
turbomolecular pump backed by a Roots pump to the base pressure
in the order of 10−4
Pa. Gas inlet was located between the target and
the substrate at the distance of 2 cm from the target. Argon (purity
99.999%) and acetylene (purity 99.6%) gasses were dosed via a mass
ﬂow controller.
Prior the deposition the high speed steel (HSS), hardmetal (WC)
and Si substrates were ultrasonically cleaned in a degreasing agent.
The titanium target was cleaned by an abrasive to mechanically
remove any carbon layer formed during previous depositions. After
placing the samples into the chamber and separating them from the
target by a shutter, both the target and the substrates were further
cleaned by argon ion bombardment. This cleaning phase was performed
until the discharge voltage stabilized indicating that the target
was completely cleaned of carbon contaminants. The next day, after
the target and the chamber cooled down to room temperature, the
deposition procedure commenced with already cleaned target by
a short pre-deposition cleaning. This was done for 35 min ensuring
removal of any adsorbed contaminants over the night and the substrates
were pre-heated at the same time by the ion bombardment.
The conditions were as follows: 65 sccm Ar (~3 Pa), 1 kW DC, 500 W
RF. Then the deposition took place. At ﬁrst the shutter between the
substrates and the target was opened and a pure titanium adhesion
layer was deposited for 3 min (~700 nm thick Ti layer) at 20 sccm
Ar, 2 kW DC, 225 W RF (~0.03 W/cm2
; ~−100 V bias). After that,
the desired acetylene ﬂow was set ranging from 8 to 14 sccm to
obtain coatings with speciﬁc chemical compositions (corresponding to
30 at.%C to 68 at.%C respectively in the nc-TiC/a-C:H coating), all
other parameters were kept constant, i.e. 20 sccm Ar, 2 kW DC, 225 W
RF (~−100 V bias). After 45 min of deposition time, the discharge and
gas supplies were turned off and the samples were let to cool down
for several hours under vacuum.
Atomic composition was determined by Rutherford backscattering
(RBS) and elastic recoil detection (ERDA) methods [21,22] using a
Van der Graaff generator and TANDETRON with linear electrostatic
accelerators. The contents of Ti, C and O were measured by RBS
using 1.74 MeV and 2.2 MeV protons and 3.04 MeV alpha particles
as projectiles. The ERDA with an incident beam of 2.75 MeV alpha
particles at 75° to the surface normal was used for the measurement of
the H content. Hydrogen atoms recoiled under 30° were detected with a
surface barrier detector covered with a 12 μm thick mylar stopping foil
which prevented detector from reﬂected alpha particles. The charge of
impinging alpha particles was measured using beam chopper with
Au layer and monitored by RBS with regard to precise live time.
The accessible depth for ERDA is less than 1 μm. The H contend is
compared with 11.9% H:Si standard. The RBS and ERDA measurements
were evaluated by computer codes GISA 3 [23] and SIMNRA [24], using
enhanced non-Rutherford cross-section values from a SigmaBase (also
part of SIMNRA Code). Thickness measured in at/cm3
compared with
the thickness evaluated using a calotest Platit CT 50 leads to density
(g/cm3
) estimation.
A Rockwell test was used to evaluate the adhesion of the ﬁlms
according to German DIN CEN/TS 1071-18 standards. This test was
performed with a Rockwell C-scale diamond tipped indenter from
Meopta. Fischerscope H100 depth sensing indenter equipped with a
Berkovich tip was used to study the indentation response of the
samples. The samples were polished before the nanoindentation
tests (approximately 5% of the coating's thickness was removed) in
order to eliminate the inﬂuence of the roughness. A large number of
indentation tests (at least 60 indentations per sample) were performed
at applied load of 75 mN on each sample. The hardness and the elastic
modulus were evaluated using the standard procedure proposed by
Oliver and Pharr [25]. In order to obtain reliable values of mechanical
properties such testing condition were used in nanoindentation
measurements under which the substrate did not inﬂuence the
measured values. The maximum indentation depth did not exceed
500 nm, therefore the very often used “rule-of-thumbs” like “indentation
to 10% of the ﬁlm thickness samples the ﬁlm alone” was fulﬁlled.
Moreover, we estimated the radius of the plastically deformed zone
(radius of irreversibly deformed volume of the coating) according to
Johnson's contact model [26]. The calculated radius was in each case
signiﬁcantly lower, than the coating thickness, which conﬁrms that
the measured data were not inﬂuenced with the substrate. More details
of the evaluation methods of the plastic hardness and Young's modulus
can be found elsewhere [27].
Residual stress was calculated from curvature of rectangular silicon
substrates before and after deposition. The curvature was determined
using proﬁlometer DEKTAK VEECO 8. The stress was calculated from
the Stoney formula [28]. The vertical structure of hardmetal samples
was analyzed on fracture surfaces by a ﬁeld-emission Scanning Electron
Microscope LYRA 3 XMU FEG/SEM×FIB by Tescan. Operating voltage of
the SEM was 15 kV. A thin sheet for TEM measurement was cut out from
hardmetal samples using a focused ion beam gun integrated in the same
scanning electron microscope. Transmission electron microscope
CM12 STEM Philips operating at accelerating voltage of 120 KV was
used to image the microstructure of the coating. X-ray diffraction was
measured using copper X-ray tube (wavelength 1.542 Å) in the parallel
beam setup with a parabolic multilayer mirror as a collimator and
monochromator of the primary beam and parallel plate collimator
with secondary graphite monochromator as an analyzer. Grazing
incidence detector scan was used for measurement of all samples.
Additionally also θ-2θ scan was used for selected samples.
3. Results and discussion
Mechanical properties of nc-TiC/a-C:H coatings depend primarily
on the carbon content [2,4,7,8] in the coating or more precisely on
the amount of the soft a-C:H matrix embedding the nc-TiC grains
[10]. In our research, the acetylene ﬂux was varied from 8 sccm to
14 sccm to obtain coatings with different carbon contents. Summary of
deposition conditions and resulting coating properties is provided in
Table 1. The coating's thicknesses varied from 6 μm for ﬁlms deposited
with the lowest acetylene ﬂux to about 4 μm for the samples deposited
at the highest acetylene ﬂux. Thus the deposition rate decreased
from ~150 nm/min to ~100 nm/min The deposition rate of the titanium
adhesion layer was ~230 nm/min. The deposition rate of nc-TiC/a-C:H is
comparable to reported deposition rates, i.e. around 100 nm/min [2,11].
The observed decrease of deposition rate with increasing acetylene
ﬂux may be caused due to formation of a-C layer on the target
(target poisoning) that prevents Ti sputtering. Within the range of the
studied experimental conditions the Ti ﬂux on the substrate is probably
not compensated enough by the ﬂux of carbon fragments.
Chemical composition of the nc-TiC/a-C:H coatings on hardmetal
substrates determined by RBS and ERDA measurements as a function of
acetylene supply is plotted in Fig. 1. The samples on HSS yielded similar
composition. Titanium content is steadily decreasing from ~62 at.% for
8 sccm of acetylene to ~28 at.% for 14 sccm, while carbon content steadily
increases from ~30 at.% to ~68 at.%. Oxygen contamination is less than
5 at.%. The main sources of oxygen are probably the post-deposition
contamination and contaminations in low purity acetylene used in the
deposition. Hydrogen content stays low around or less than 10 at.%. The
ratio of Ti/CN1 is obtained for coatings deposited at acetylene ﬂuxes
of 10 sccm and lower and Ti/Cb1 for coatings deposited at ﬂuxes
of 11 sccm or higher. The coating's density as determined from
112 P. Souček et al. / Surface & Coatings Technology 211 (2012) 111–116
RBS measurements using thickness evaluated from the calotest
showed a monotonous decrease from 4.7 g/cm3
for the coating
with the highest amount of titanium to 3.3 g/cm3
for the ﬁlm
with the least incorporated titanium. Density of stoichiometric
TiC is 4.9 g/cm3
[29] and the density of amorphous carbon is
2.0 g/cm3
[30].
Nanoindentation measurements were performed to evaluate the
plastic hardness and Young's modulus. The results for coatings deposited
on hardmetal and the applied load of 75 mN are summarized in Table 1
and plotted in Fig. 2. Hardness lower than 20 GPa is typical for ﬁlms with
either too high or too low carbon content [2,4,5,7–12]. In our experiment,
the maximal hardness and Young's modulus of 46±2 GPa and
415±7 GPa respectively are reached simultaneously at the acetylene
ﬂow of 12 sccm,1
signiﬁcantly exceeding the typical values in the literature
[1–12,14]. The residual compressive stress, measured for selected
samples only, stayed under 0.5 GPa for all coatings as can be seen in
Table 1.
High values of hardness were also reached for up to 2.5 μm thick
TiC/a-C:H coatings by Zehnder et al. [16,17] using closed ﬁeld unbalanced
magnetron sputtering conﬁguration. They reached hardness up to
~35 GPa for samples prepared at −240 V bias. About the same hardness
of ~35 GPa was also reached by Hu et al. [9] for coatings up to ~1 μm,
preparation of thicker coatings was limited by high internal stress.
Czyżniewski and Precht [2] reached hardness of ~42 GPa for nc-TiC/a-C:
H coatings prepared at −100 V bias. Wiklund et al. [1] reached hardness
over 40 GPa for up to 2.6 μm thick TiC/a-C coatings using combination
of electron beam evaporation of Ti target together with magnetron
sputtering of C target with rotating substrates. This way a thin titanium
ﬁlm and carbon ﬁlm are deposited in succession and are mixed during
the process. However this high hardness is coupled with high internal
compressive stress of 6–8 GPa.
Acetylene ﬂow of 12 sccm corresponds to 41 at.% titanium content
and 55 at.% carbon content thus to the Ti/C ratio approximately 3/4.
The highest values of hardness and Young's modulus in our research
are reached only for narrow range of acetylene. Both quantities are
signiﬁcantly lowered for acetylene ﬂow change of as little as 1 sccm.
Similar sharp maximum of hardness can be found in Ref. [16],
where the Ti concentration window for the hardest coatings was
only about 5 at.% wide.
It has been shown [9,10,16,17], that the relative amount of TiC and
a-C(:H) phases has more direct inﬂuence in determining the properties
of the coatings rather than the amount of titanium and carbon.
The content of TiC and a-C(:H) phases depends on the carbon content,
but also on other deposition conditions such as applied bias. Hu et al.
[9], Zehnder at al. [17] and Czyżniewski and Precht [2] observed highest
hardness for samples with only about 10–20% of a-C:H phase content,
on the other hand Martínez-Martínez et al. [10] and Zhang et al.
[18] observed highest hardness for samples containing about 35–45%
a-C phase content.
SEM and TEM imaging was performed for the sample with highest
hardness (HUpl=46.2 GPa), i.e. prepared at 12 sccm of acetylene. Both
SEM (Fig. 3) and TEM (Fig. 4) revealed presence of dense columnar
structure. The width of the elongated TiC grains is about 15 nm. Selected
area diffraction pattern (inset of Fig. 4) from circular area 0.5 μm in
diameter conﬁrms the face centered cubic structure of TiC with
lattice parameter close to that of TiC in database [31]. Only very
low a-C phase content compared to TiC phase is visible in the
TEM image of the sample with highest hardness (Fig. 6), which
is in accordance with literature [10,18].
Adhesion of the coatings on HSS substrates was measured using
Rockwell C-scale test at the load of 1500 N according to DIN CEN/TS
1071-18. All ﬁlms were viable according to the DIN CEN/TS 1071-18
industrial standards and showed very good adhesion in the HF0–HF1
Table 1
Summary of deposition conditions and resulting chemical composition and mechanical properties of nc-TiC/a-C:H coatings. All results except for the crystallite size are for hardmetal
substrates. Crystallite size is determined for high speed steel substrates. The roughness was measured over 8 mm long line a thus represents roughness at macroscale.
C2H2
[sccm]
Chemical
composition [at.%]
Thickness
[μm]
Dep.
rate
[nm/min]
Density
[g/cm3
]
Ra
(macro)
roughness
[μm]
HUpl
[GPa]
E [GPa] Mean
crystallite
size
[nm]
Ti C
8 62.2 29.8 6.1 151 4.7 0.12 13±1 222±5 13
9 60 27.5 5.8 144 4.3 0.15 22±3 283±20 15
10 55.4 35.6 6.0 149 4.2 0.17 24±2 289±12 42
11 41.4 54.1 4.3 111 3.8 0.07 35±1 315±6 15
12 40.6 55.2 5.0 127 4.2 0.07 46±2 415±7 19
13 30.1 59.1 3.4 91 3.7 0.1 37±2 316±10 5.3
14 27.5 62.5 4.5 116 3.3 0.05 23±1 225±5 4.5
Lattice parameter [Å] Adhesion Compressive stress [GPa]
4.326 HF 0
4.349 HF 0 0.34
4.357 HF 1
4.357 HF 0 0.37
4.354 HF 1 0.42
4.363 HF 1 0.45
4.356 HF 0
1
In our previous paper [20] the maximal hardness and Young's modulus are
obtained at lower acetylene ﬂow of 10 sccm. This shift is caused by replacing of an
old worn out DC generator by a new one providing now really 2 kW of DC power. Fig. 1. Chemical composition of nc-TiC/a-C:H coatings as measured by RBS and ERDA.
113P. Souček et al. / Surface & Coatings Technology 211 (2012) 111–116
range, as is shown in Table 1. Typical Rockwell indents are shown in
Fig. 5. Macro-roughness of samples on hardmetal substrates was evaluated
using a proﬁlometer over length of 8 mm and summarized in
Table 1. Roughness in the macroscale of the sample prepared at
8 sccm of acetylene is Ra =0.12 μm and it increases with increasing
acetylene supply (i.e. carbon content). A sharp decrease in the roughness
from Ra =0.17 to Ra =0.07 was observed when the acetylene ﬂux increases
from 10 sccm to 11 sccm. Further increase of the acetylene supply
results in roughness which doesn't signiﬁcantly alter being 0.05 μm
for samples with higher carbon content. The roughness of the uncoated
hardmetal substrates was measured to be 0.05 μm.
X-ray diffraction measurements were performed on HSS samples
to estimate the crystallite size and texture. The diffraction patterns
show strong peaks corresponding to TiC along with weak peaks
corresponding to substrate metal (iron) and adhesion interlayer
(titanium). Fig. 6 shows the diffraction patterns for samples prepared
at 8 sccm, 12 sccm and 14 sccm. The TiC crystallite size shown in
Table 1 was calculated from the Scherrer formula attributing the
total broadening of the (111) peak to the crystallite size only. Results
in Table 1 thus report the smallest possible value of the mean crystallite
size in the crystallographic direction [111]. Crystallite sizes in the
in-plane direction are smaller than the determined size in the
growth direction, as is shown by the TEM. More accurate values
were obtained for samples prepared for the acetylene ﬂux of 12
and 13 sccm using Williamson–Hall analysis of the diffraction maxima
111 and 222, which correspond to the same crystallographic plane.
The mean crystallite size in these samples turned out to be about
twice as big as the minimal values given in Table 1 and the mean
deformation was about 0.5% to 1.5%. The TEM image of a lamella
cut perpendicularly to the sample surface shows grains from
about 5 nm to around 30 nm often forming larger clusters (see Fig. 6).
The dependence of the crystallite size on the acetylene ﬂux behaves
similarly as the roughness. The crystallite size increases with increasing
Fig. 3. SEM image of a nc-TiC/a-C:H coating prepared at 12 sccm of acetylene on a
hardmetal substrate.
Fig. 4. TEM bright ﬁeld image and selected area diffraction pattern of the nc-TiC/a-C:H
coating prepared at 12 sccm of acetylene.
Fig. 5. Typical Rockwell indents showing a) adhesion level 0 for coating prepared at
10 sccm of acetylene b) adhesion level 1 for coating prepared at 11 sccm. Both indents
show only very ﬁne cracks around the edge of the crater.
Fig. 2. Dependence of plastic hardness and Young's modulus on the acetylene ﬂow. A
sharp maximum in both dependencies at 12 sccm is observed.
114 P. Souček et al. / Surface & Coatings Technology 211 (2012) 111–116
acetylene supply from 15 nm for the smallest acetylene ﬂow up to
42 nm for the acetylene ﬂow of 10 sccm. Then a sudden drop to
15 nm at acetylene ﬂow of 11 sccm and later well under 10 nm
is observed. Similar, however less sharp maximum in the crystallite
size at ~50 at.% titanium is reported in Ref. [11]. Also a signiﬁcant decrease
of grain size from more than 20 nm for samples with b4% a-C:
H phase content to 3–5 nm for samples with N4% a-C:H phase content
for samples prepared at −240 V bias is reported in Ref. [17]. The same
but less pronounced behavior observed for samples prepared at
−91 V bias in the same reference. Determined crystallite sizes in our research
are in accordance with values reported in the literature —
several nanometers for Ti contents up to 46 at.% [2,9].
Only very weak diffraction peaks corresponding to titanium were
found in the XRD patterns of those samples deposited at higher supply
of the acetylene gas. We attributed the presence of weak Ti peaks in
these patterns to the Ti adhesive layer as the coatings deposited at
higher acetylene supply are the thinnest in the series. The absence of
titanium peaks together with no peaks corresponding to Ti1.6 C phases
[4] in the XRD patterns (see Fig. 7) for samples where Ti/CN1 indicates
that a substoichiometric TiC was formed as can be deduced from a
Ti–C phase diagram [32]. This is also conﬁrmed by the lattice parameter
evolution — see Table 1. Samples prepared at acetylene ﬂows of
10 sccm or higher have lattice parameter around 4.36 Å (error for all
measurements is around 0.005 Å). Sample prepared at 9 sccm C2H2
(~59 at.% Ti) shows lattice parameter reduced by 0.2% and sample
prepared at 8 sccm (~63 at.% Ti) even by 0.8%. The lattice parameter
according to PDF card #32-1383 is 4.3274 Å [33]. Only the sample
prepared at 8 sccm of acetylene is close to this value, although lattice
parameters larger than the reference value are quite often observed
[19,20,34].
Also the texture of the crystallites was evaluated by the XRD measurements.
The material is observed to be highly textured for the least carbon
containing ﬁlms and becomes more isotropic with increasing carbon
content until 10 sccm of acetylene, where again preferential texture is
observed. This sudden change in the texture correlates with decrease of
crystallite size and the coating roughness. Increase of the acetylene
supply over 12 sccm results in nearly isotropic coating prepared at
14 sccm of acetylene containing 62 at.% carbon.
4. Conclusions
Several μm thick nc-TiC/a-C:H coatings were prepared at relatively
low ion bombardment environment using well-balanced magnetron
sputtering of a titanium target in an acetylene containing atmosphere.
High speed steel and hardmetal substrates were used. Acetylene ﬂow
was varied to obtain ﬁlms with different chemical compositions ranging
from 30 at.%C to 68 at.%C. Approximately 700 nm thick titanium
layer was deposited prior the nc-TiC/a-C:H growth to promote
the adhesion of the coating. Very good adhesion corresponding
to HF0-HF1 was reached according to the German DIN CEN/TS
1071-8 standard. The highest hardness and Young's modulus of
46 GPa and 415 GPa respectively were reached for the nanoindentation
measurements at 75 mN load at the Ti/C ratio approximately 3/4. SEM
and TEM images of this sample show a dense columnar structure with
elongated grains ~15 nm wide. XRD proves the presence of TiC grains
of characteristic size ranging from several tens of nm to several nm
depending of the carbon content in the coating. Shift in TiC peak position
and the fact that no Ti peaks were detected in XRD patterns of
the coatings, where Ti/C N1 indicates that a substoichiometric TiC was
formed. The orientation of the TiC grains changes non-monotonously
from highly preferential orientation for the coatings with the lowest
carbon content to nearly uniform orientation distribution of TiC
nanocrystallites in the coatings with highest carbon contents.
Acknowledgment
This research has been supported by the MSM contract
0021622411, by GACR contract 104/09/H080 and R&D center project
for low-costplasma and nanotechnology surface modiﬁcations
CZ.1.05/2.1.00/03.0086 funded by European Regional Development
Fund. Pavel Souček acknowledges the Brno City Municipality as Holder
of Brno PhD Talent Financial Aid.
References
[1] U. Wiklund, M. Nordin, O. Wänstrand, M. Larsson, Surf. Coat. Technol. 124 (2000)
154.
[2] A. Czyżniewski, W. Precht, J. Mater. Process. Technol. 157–158 (2004) 274.
[3] Y. Wang, X. Zhang, X. Wu, H. Zhang, X. Zhang, Appl. Surf. Sci. 255 (2008) 1801.
[4] J.C. Sánchez-López, D. Martínez-Martínez, C. Lópes-Cartes, A. Fernández, Surf.
Coat. Technol. 202 (2008) 4011.
[5] D.M. Cao, B. Feng, W.J. Meng, L.E. Rehn, P.M. Baldo, M.M. Khonsari, Appl. Phys.
Lett. 79/3 (2001) 329.
[6] J. Musil, P. Novák, R. Čerstvý, Z. Soukup, J. Vac. Sci. Technol. A 28 (2010) 244.
[7] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, A. Cavaleiro, Surf. Coat. Technol. 198 (2005)
44.
[8] D. Galvan, Y.T. Pei, J.Th.M. De Hosson, Acta Mater. 53 (2005) 3925.
[9] Y. Hu, L. Li, X. Cai, Q. Chen, P.K. Chu, Diam. Relat. Mater. 16 (2007) 181.
[10] D. Martínez-Martínez, C. López-Cartes, A. Fernández, J.C. Sánchez-López, Thin
Solid Films 517 (2009) 1662.
Fig. 6. HR-TEM image of a lamella cut perpendicularly to the sample surface shows
grains from about 5 nm to around 30 nm often forming larger clusters.
Fig. 7. XRD patterns for samples prepared at acetylene ﬂow of 8 sccm, 12 sccm and
14 sccm. The curves for 8 sccm and 12 sccm are shifted to higher values for clarity.
The positions of the TiC peaks are indicated by their Laue indices at the top of the
graph.
115P. Souček et al. / Surface & Coatings Technology 211 (2012) 111–116
[11] W. Gulbinski, S. Mathur, H. Shen, T. Suszko, A. Gilewicz, B. Warcholinski, Appl.
Surf. Sci. 239 (2005) 302.
[12] K. Polychronopoulou, C. Rebholz, M.A. Baker, L. Theodorou, N.G. Demas, S.J. Hinder,
A.A. Polycarpou, C.C. Doumanidis, K. Böbel, Diamond Relat. Mater. 17 (2008) 2054.
[13] A.A. El Mel, B. Angleraud, E. Gautron, A. Granier, P.Y. Tessier, Surf. Coat. Technol.
204 (2010) 1880.
[14] A.A. Voevodin, J.S. Zabinski, J. Mater. Sci. 33 (1998) 319.
[15] W.J. Meng, R.C. Tittsworth, L.E. Rehn, Thin Solid Films 377–378 (2000) 222.
[16] T. Zehnder, J. Patscheider, Surf. Coat. Technol. 133–134 (2000) 138.
[17] T. Zehnder, P. Schwaller, F. Munnik, S. Mikhailov, J. Patscheider, J. Appl. Phys. 95
(8) (2004) 4327.
[18] S. Zhang, X.L. Bui, J. Jiang, X. Lim, Surf. Coat. Technol. 198 (2005) 206.
[19] G. Li, L.F. Xia, Thin Solid Films 396 (2001) 16.
[20] E. Lewin, E.O.Å. Persson, M. Lattemann, M. Stüber, M. Gorgoi, A. Sandell, C. Ziebert,
F. Schäfers, W. Braun, J. Halbritter, S. Ulrich, W. Eberhardt, L. Hultman, H. Siegbahn,
S. Svensson, U. Jansson, Surf. Coat. Technol. 202 (2008) 3563.
[21] W.K. Chu, J.W. Mayer, M.A. Nicolet, Back scattering Spectrometry, Acad. Press,
New York, 1978.
[22] J.R. Tesmer, M. Nastasi, J.Ch. Barbour, in: C.J. Maggiore (Ed.), Handbook of Modern
Ion Beam Materials Analysis, Material Research Society, Pitsburg, Pensylvania, 1995.
[23] J. Saarilahti, E. Rauhala, Nucl. Instrum. Methods Phys. Res. Sect. B 64 (1992) 734.
[24] M. Mayer, SIMNRA User's Guide, Technical Report IPP 9/113, Max-Planck-Institut
fuer Plasmaphysick, Garching, Germany, 1997.
[25] W.C. Oliver, G.M. Pharr, J. Mater. Res. 19 (2004) 3.
[26] K.L. Johnson, J. Mech. Phys. Solids 18 (1970) 115.
[27] P. Vašina, P. Souček, T. Schmidtová, M. Eliáš, V. Buršíková, M. Jílek, M. Jílek Jr., J.
Schäfer, J. Buršík, Surf. Coat. Technol. 205 (2011) S53.
[28] G.G. Stoney, Proc. R. Soc. Lond. Ser. A 82 (1909) 172.
[29] J.F. Shackelford, W. Alexander (Eds.), CRC Materials Science and Engineering
handbook, 3 rd edition, CRC Press, Boca Raton, 2001.
[30] National Institute of Standards and Technology, www.nist.gov.
[31] P. Villars, K. Cenzual, Pearson's Crystal Data, Release 2009/10, ASM International,
Ohio, USA, 2009.
[32] D.A. Andersson, P.A. Korzhavyi, N. Johansson, Calphad 32 (2008) 543.
[33] M.C. Morris, H.F. McMurdie, E.H. Evans, B. Paretzkin, H.S. Parker, N.C. Panagiotopoulos,
C.R. Hubbard, NBS Monograph 25, Sec. 18, 1981.
[34] E. Lewin, M. Råsander, M. Klintenberg, A. Bergman, O. Eriksson, U. Jansson, Chem.
Phys. Lett. 496 (2010) 95.
116 P. Souček et al. / Surface & Coatings Technology 211 (2012) 111–116
On the control of deposition process for enhanced mechanical properties
of nc-TiC/a-C:H coatings with DC magnetron sputtering at low or high
ion ﬂux
Pavel Souček a,
⁎, Tereza Schmidtová a
, Lukáš Zábranský a
, Vilma Buršíková a
, Petr Vašina a
, Ondřej Caha b
,
Jiří Buršík c
, Vratislav Peřina d
, Romana Mikšová d
, Y.T. Pei a,e
, J.Th.M. De Hosson e
a
Department of Physical Electronics, Faculty of Science, Masaryk University, Kotlářská 2, CZ-61137 Brno, Czech Republic
b
Department of Condensed Matter Physics, Faculty of Science, Masaryk University, Kotlářská 2, CZ-61137 Brno, Czech Republic
c
Institute of Physics of Materials, Academy of Sciences of the Czech Republic, Žižkova 22, CZ-61662 Brno, Czech Republic
d
Nuclear Physics Institute, Academy of Sciences of the Czech Republic, v.v.i., Řež 130, CZ-25068, Řež, Czech Republic
e
Department of Applied Physics, Materials Innovation Institute M2i, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands
a b s t r a c ta r t i c l e i n f o
Available online 9 November 2013
Keywords:
Nanocomposite
Magnetron sputtering
Ion ﬂux on the substrate
Titanium carbide
Enhanced mechanical properties
Nanocomposite coatings consisting of nanocrystallites embedded in an amorphous matrix can be tailored to
exhibit unusual combination of properties such as high hardness and modulus combined with low friction and
wear. The properties of the coatings are governed by parameters of the deposition plasma, which are a nontrivial
function of the deposition process parameters. Energy ﬂux onto the growing ﬁlms provided by ionized species is
one of the key parameters inﬂuencing the properties of coatings. This study is focused on the inﬂuence of ion ﬂux
on the properties of the growing nc-TiC/a-C:H thin ﬁlm in order to achieve enhanced mechanical properties. Two
different magnetic ﬁeld conﬁgurations were used. The saturated ion current on the substrate was found to be
seven times higher in the strongly unbalanced magnetic conﬁguration as compared to well-balanced conﬁguration.
The deposition temperature was constant at ~320 °C, which is higher than the usual temperature for similar
depositions of nc-TiC/a-C(:H) coatings. The structure and mechanical properties of the coatings prepared at low or
high ion ﬂux were studied and compared. It is concluded that the level of the impinging ion ﬂux in DC magnetron
sputtering is not a signiﬁcant factor inﬂuencing the mechanical properties of the coatings in the presented setup.
© 2013 Elsevier B.V. All rights reserved.
1. Introduction
Nanocrystalline titanium carbide embedded in amorphous (hydrogenated)
carbon matrix – nc-TiC/a-C(:H) – is a versatile material combining
the properties of hard titanium carbide nanocrystallites and relatively
soft hydrogenized carbon matrix. Depending on the content of the phases
the properties of the nanocomposite can be tailored from hard ﬁlms with
hardness reaching over 40 GPa and Young's modulus higher than
300 GPa [1–3] to low friction ﬁlms [4,5] with friction coefﬁcient around
0.1 and wear rate lower than 2 × 10−7
mm3
/Nm [6,7]. In order to improve
the adhesion of nc-TiC/a-C(:H) ﬁlms an adhesion-promoting interlayer
is introduced into the coating system. Chromium [8,9], titanium
[10] or graded [9] interlayers are most often employed.
The synthesis of hydrogenated nc-TiC/a-C:H composite is mostly
performed by sputtering of titanium target in hydrocarbon containing
plasma by a hybrid PVD–PECVD process [11]. Methane [12] and acetylene
[3,5,7,8,13] are most often used. Hydrogen free nc-TiC/a-C
nanocomposites are usually deposited by co-sputtering of graphite
and titanium targets [4,8,14–16].
The properties of nc-TiC/a-C(:H) nanocomposite coatings are largely
determined by the volumetric fraction and size distribution of TiC
nanocrystallites [10,14,21] or by the concentration of titanium and
carbon [1,3,7–10,15,17–21]. It was reported that hardness of the nanocomposites
with similar chemical composition can vary signiﬁcantly
among different authors showing that in addition to coating composition,
deposition temperature or the ion impingement during the deposition
should have impact. Typically, the deposition temperatures are
below or around 200 °C [1,3,4,7–10,14,16,18–21], i.e. low compared to
the melting temperature of TiC (3067 °C) [1] and also below the temperature
where the amorphous carbon matrix starts to deteriorate
(b350 °C) [22]. The variance of coating hardness from some of the
authors can be seen in Fig. 1. Comparing the hardness of nc-TiC/a-C(:H)
coatings of Gulbinski et al. [15] (substrate temperature of 450 °C and
−10 V substrate bias voltage) with those of Martínez-Martínez et al.
[14] (150–250 °C, ﬂoating substrate) both deposited at low or ﬂoating
substrate bias voltage at distinctively different temperatures, similar
hardness for similar composition is found. Therefore, the deposition temperature
by itself does not explain differences in hardness. Flux of impinging
ions and the ion energy distribution function (or average
Surface & Coatings Technology 255 (2014) 8–14
⁎ Corresponding author. Tel.: + 420 549 49 2587.
E-mail address: soucek@physics.muni.cz (P. Souček).
0257-8972/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.surfcoat.2013.11.001
Contents lists available at ScienceDirect
Surface & Coatings Technology
journal homepage: www.elsevier.com/locate/surfcoat
energy per ion) have both inﬂuence on the coating properties. The ion
energy can be tuned by applying various substrate bias voltage levels
or by using different waveforms of voltage applied on the magnetron
cathode. The effect of applying various bias levels on coating hardness
was shown by Kulikovsky et al. [12], who observed a maximum hardness
at −100 V bias and lower hardness at other bias voltages, although in his
work higher bias voltage was linked with higher deposition temperature.
Zehnder and Patscheider [21] showed a distinct increase in hardness
when substrate bias voltage increased from −91 V to −240 V. Also
the structure can be altered by applying substrate bias [18]. Pei et al.
[19] showed the inﬂuence of pulsed DC (p-DC) frequency on the microstructure
of nc-TiC/a-C(:H) nanocomposite coatings. The p-DC frequency
inﬂuenced both the energy distribution and the ﬂux of argon ions in the
plasma. The microstructure changed from columnar at 100 kHz to
column-free microstructure at 300 and 350 kHz. This microstructure
evolution was attributed to the effect of ion impingement of higher energy
input as proven by the measurement of ion energy distribution function
on substrate. p-DC substrate bias also leads to higher ion current on
the substrate [23].
Nevertheless, it is difﬁcult to conclude from the published work the
inﬂuence of impinging ion ﬂux on the properties of resulting coatings
because not all the experimental details were always fully presented
(i.e. substrate bias voltage, magnetic ﬁeld conﬁguration, target size
and target to substrate distance). Depending on the target–substrate
distance, unbalanced magnetron does not necessarily lead to a considerable
plasma immersion of the substrates. For instance, a distance of
100–150 mm between substrates and 2 in. magnetrons (not uncommon
in research systems) gives very low plasma densities at the substrates
(no visible plasma) even for highly unbalanced magnetrons.
The aim of this work is to single out the effect of the impinging ion
ﬂux on the microstructure and properties of nc-TiC/a-C:H nanocomposite
coatings with changing magnetron strength in order to pinpoint
optimal deposition conditions for production of hard and adherent coatings.
Two different magnetic ﬁeld conﬁgurations, well balanced and
strongly unbalanced, were employed.
2. Experimental procedures
Depositions were carried out using industrial sputtering system
Alcatel SCM 650. DC magnetron sputtering was employed to sputter a
titanium target Ø 200 mm in argon/acetylene atmosphere. High speed
steel (Ø 20 mm × 4.7 mm) and tungsten carbide cermet substrates
(20 mm equilateral triangle base × 4.7 mm) were used. Further details
about the equipment can be found in [2].
In order to achieve different ion ﬂuxes on the growing ﬁlm, two
different sets of magnets were used to conﬁgure well-balanced and
strongly unbalanced magnetic ﬁeld, respectively. The magnetic ﬁeld
structure was measured by a Hall probe and is depicted in Fig. 2. In
this measurement, the Hall probe was ﬁxed and the magnets were
placed on a movable tray. The radial- and the z-component of the magnetic
ﬁeld were measured in an array 125 mm long (from the center of
the magnet to the edge of the magnet and further 25 mm beyond) and
110 mm high with 5 mm steps in both axes. The magnetic ﬁeld structure
was then calculated numerically from this data. The substrates
were placed at the position directly facing the racetrack on the target,
as indicated by the red lines in Fig. 2. The balanced magnetic ﬁeld conﬁguration
resulted in a saturated ion current of 0.3 mA/cm2
measured
at the substrate position, while the unbalanced magnetic ﬁeld resulted
in an ion current of 2.1 mA/cm2
. The probe was biased during the measurement
with −60 V DC, which is enough to shield the bulk electrons
in the plasma from the probe and achieve ion current saturation [24].
A titanium adhesion layer ~700 nm thick was deposited directly before
the deposition of the coating. After that, the desired acetylene ﬂow
was set to obtain coatings with speciﬁc chemical compositions, all other
parameters were kept constant, i.e. 20 sccm Ar ﬂow, −100 V substrate
bias voltage. The DC power supplied to the cathode was chosen to sustain
stable deposition temperature ~320 °C for both magnetic ﬁeld conﬁgurations
used in this work. The power necessary to achieve the
desired substrate temperature was 2–2.5 kW. Relatively high magnetron
power corresponding to the deposition temperature of ~320 °C
Fig. 2. Magnetic ﬁeld structure of the magnetrons used in the experiments: (a) well
balanced magnetron and (b) strongly unbalanced magnetron. The red line indicates the
approximate position of the substrates. The black line indicates the target surface.
Fig. 1. Hardness comparison of nc-TiC/C(:H) coatings from selected authors.
9P. Souček et al. / Surface & Coatings Technology 255 (2014) 8–14
was chosen to achieve high deposition rates of the coatings. The deposition
was carried out for 45 min. Further details about the deposition
procedure can be found in [2]. The two series of coatings deposited are
thus named according to the magnetron conﬁgurations as BAL-xxC and
UNB-xxC, where BAL and UNB mean well-balanced and strongly unbalanced,
respectively, and xx stands for the carbon concentration of a coating
in atomic percentage. Depositions with unbalanced magnetic ﬁeld
conﬁgurations were carried out to higher acetylene ﬂuxes (i.e. higher
carbon concentration of the coating), acetylene ﬂux during the balanced
series was limited due to excessive arcing.
Chemical composition of the coatings was determined by Rutherford
backscattering (RBS) and elastic recoil detection analysis (ERDA) using
a Van der Graaff generator and TANDETRON linear electrostatic accelerators.
The contents of Ti, C and O were measured by RBS using 1.74 MeV
and 2.2 MeV protons and 3.04 MeV alpha particles as projectiles. The
ERDA with an incident beam of 2.75 MeV alpha particles at 75° to the
surface normal was used for the measurement of the H content. Hydrogen
atoms recoiled under 30° were detected with a surface barrier
detector covered with a 12 μm thick mylar stopping foil which blocked
the detector from reﬂected alpha particles. The charge of impinging
alpha particles was measured using beam chopper with golden layer
and monitored by RBS with regard to precise live time. The accessible
depth for ERDA is less than 1 μm. The H content is compared with
11.9% H:Si standard. The RBS and ERDA measurements were evaluated
by computer codes GISA 3 [25] and SIMNRA [26], using enhanced nonRutherford
cross-section values from SigmaBase (also part of SIMNRA
Code).
Fischerscope H100 depth sensing indenter equipped with a
Berkovich tip was used to study the indentation response of the coating
samples. A large number of indentation tests (at least 60 indentations
per sample) were performed at the applied load on each sample. The
hardness and the elastic modulus were evaluated using the standard
procedure proposed by Oliver and Pharr [27]. In order to obtain reliable
values of mechanical properties such testing conditions were used in
nanoindentation measurements under which the substrate did not
inﬂuence the measured values. The maximum indentation depth did
not exceed 500 nm, therefore, the very often used “rule-of-thumb” —
“indentation to 10% of the ﬁlm thickness” was fulﬁlled. Moreover, the
radius of the plastically deformed zone (radius of irreversibly deformed
volume of the coating) was estimated according to Johnson's contact
model [28]. The calculated radius was in each case signiﬁcantly lower
than the coating thickness, which conﬁrmed that the measured data
were not inﬂuenced by the substrate.
The microstructure of nc-TiC/a-C:H coatings was characterized on
fracture cross section with a ﬁeld-emission scanning electron microscope
(SEM) (LYRA 3 XMU FIB-FESEM, Tescan, CZ). The operating voltage
of the SEM was 15 kV. Transmission electron microscope (TEM)
CM12 STEM Philips operating at accelerating voltage of 120 kV was
used for TEM imaging. Grazing incidence X-ray diffraction using copper
X-ray tube (wavelength 1.542 Å) was employed to measure the size
and lattice parameter of TiC nanocrystallites.
3. Results and discussion
In order to keep the desired substrate temperature at the same bias
voltage for both balanced and strongly unbalanced magnetron conﬁgurations,
different magnetron power was applied. According to our
expectation that higher power applied on the magnetron cathode will
result in higher sputtering rate of titanium, it was necessary to inject
more hydrocarbon gas to deposit nanocomposite coatings of similar
composition. Therefore, the acetylene supply was normalized to the
magnetron power in order to compensate for the sputtering rate difference.
Dependence of the chemical composition on the normalized acetylene
ﬂow is plotted in Fig. 3. Titanium composition steadily decreased
from ~65 at.% to ~10 at.% and carbon composition steadily increased
from ~30 at.% to ~70 at.% with higher acetylene ﬂow to magnetron
power ratio. Saturation of carbon concentration in UNB-xxC coating
series was observed at high acetylene supply to magnetron power
ratio. Hydrogen concentration stayed around 10 at.% for the depositions
with acetylene ﬂow to magnetron power ratio b6.5 sccm·kW−1
and
then somewhat increased with further increasing the acetylene ﬂow
at the expense of carbon concentration so that the aforementioned carbon
saturation is observed. Oxygen contamination stayed b4 at.% at the
whole deposition range. Matrix-assisted laser desorption/ionization
analyses of our coatings proved that hydrogen in the nc-TiC/a-C:H coating
was bonded not only in the a-C:H matrix but also in TiH compound
[29], which can explain presence of hydrogen in coatings where no matrix
should be formed. The Ti/C concentration ratio of unity was reached
at the acetylene ﬂow to magnetron power ratio of 4.8 sccm·kW−1
for
the UNB-xxC series deposited with unbalanced magnetron at high ﬂux
of impinging Ar ions and at 5.2 sccm·kW−1
for the BAL-xxC series
deposited with the balanced magnetron, respectively, as indicated in
Fig. 3. In unbalanced magnetron, the magnetized plasma is less conﬁned
close to the magnetron target which leads to less intensive sputtering of
titanium from the target at the same applied power than in the case of
the balanced magnetron. Thus, lower acetylene supply is required in unbalanced
magnetron sputtering to achieve the same composition of the
nanocomposite coatings as in balanced magnetron reactive sputtering
and the stoichiometric point is reached at lower normalized acetylene
ﬂow. The rate of the carbon content increase for coatings of around
50 at.% C was similar for both series unlike in the work of Samuelsson
et al. [17], who showed much slower evolution of C/Ti ratio in the case
of High-Power Impulse Magnetron Sputtering (HiPIMS) with very
high ion ﬂuxes compared with DC magnetron sputtering.
It has been shown [10,14,21,30], that the relative amount of TiC and
a-C(:H) phases can be used to determine the properties of the coatings
as well as the amount of titanium and carbon. Zhang et al. [30]
employing co-sputtering of titanium and graphite targets at −150 V
substrate bias voltage reported on 19% of a-C matrix at 58 at.% C and
an increase to 98% of a-C matrix with total carbon content of 92 at.%.
Hu et al. [10] employing Ti target sputtering in acetylene containing atmosphere
reported on a linear increase of a-C matrix content form ~10%
at 54 at.% C to ~90% at 96 at.% C. The linear increase was the same for
unbiased substrates and substrates biased with −100 V. Gulbiński
et al. [15] used a pulsed DC (100 kHz) unbalanced Ti sputtering
(1–3 mA/cm2
ion current) in Ar/C2H2 atmosphere and reported on an
increase of a-C matrix content in even sub-stoichiometric coatings
(~20% a-C:H matrix at 40 at.% C content) with a fast increase to ~70%
of a-C:H matrix at 60 at.% C with further though slower increase of
a-C:H matrix content with total carbon content to ~97% of a-C:H at
85 at.% C. The content of TiC and a-C(:H) phases depends on the carbon
content, but also on other deposition conditions such as applied substrate
bias. Thus, data in Fig. 3 can not be simply transformed to phase
composition.
Fig. 3. Dependence of the chemical composition on the normalized acetylene ﬂow.
10 P. Souček et al. / Surface & Coatings Technology 255 (2014) 8–14
Thick coatings were prepared. The thickness ranges from approximately
3 μm to 6.5 μm depending on the carbon content. Fig. 4 shows
the plot of the deposition rate normalized to magnetron power versus
the carbon content of the coatings. With balanced sputtering, the normalized
deposition rate shows a steady decrease with increasing carbon
content of the coating. This is attributed to evolution of target state as is
discussed in Schmidtová et al. [31]. Similar trend of normalized deposition
rate decrease is also observed with unbalanced conﬁguration up to
carbon content of ~65 at.%. Up to ~68 at.% C lower thickness and deposition
rate is observed for the unbalanced series of similar carbon
content. This can be explained by lower titanium sputtering rate at unbalanced
magnetron compared to the balanced one. Furthermore, the
plasma sustained at higher discharge voltage but lower current in unbalanced
magnetic ﬁeld conﬁguration is less dense in the volume and
lower dissociation rate of acetylene can be expected than in the plasma
driven at comparable power in balanced magnetic ﬁeld conﬁguration at
lower voltage and higher current, i.e. higher plasma density. In the unbalanced
conﬁguration the plasma is spread more to the substrate so
that the acetylene can be dissociated closer to the substrate and more
carbon can be incorporated into the growing ﬁlm as well. A sharp increase
in the normalized deposition rate from ~33 nm·min−1
·kW−1
at 65 at.% C to ~50 nm·min−1
·kW−1
at 69.9 at.% C is then observed.
Similar increase of the deposition rate of nc-TiC/a-C:H with carbon carrying
gas can be seen in the literature [3,16,17].
Fig. 5 shows the cross-section of BAL-55C and UNB-57C. Both exhibit
columnar microstructure with columns ~200 nm thick. The columns
are slightly tilted from vertical growth by ~15° because the cross sections
were fractured near the sample edge that was slanted from the
racetrack of the target. The presence of the columnar structure can be
also indicated by a cauliﬂower surface pattern of the coatings. Within
the studied composition range of both series, coatings with similar
chemical composition show similar structure for both the series, i.e. a
clear columnar structure with column thickness in the order of few hundred
nanometers. Fig. 6 shows the microstructure of UNB-68C and UNB-
70C with high carbon content deposited with the strongly unbalanced
magnetron on cermet substrates. UNB-68C (Ti/C = 0.451
) shows a
ﬁne columnar structure. In contrast, UNB-70C (Ti/C = 0.10) is of
column-free microstructure. The high difference in Ti/C ratios with similar
carbon content is due to high hydrogen content in the most carboncontaining
coating. These results are in accordance with the trend observed
previously. Pei et al. [7,8] show pronounced columnar structure
in coatings with 66.6 at.% C (Ti/C = 0.48), almost invisible columns in
the coating of 81 at.% C (Ti/C = 0.22) and glassy structure in the coating
containing 87.2 at.% C (Ti/C = 0.14), all the coating samples were
prepared with −100 V bias as in our experiment. Lin et al. showed
similar result, namely pronounced columns in coatings of low carbon
content and a shift to glassy structure at 80.5 at.% C concentration
(Ti/C = 0.24 assuming only Ti and C concentrations measured). A
Transmission electron microscope (TEM) imaging of sample BAL-55C
(lying in the composition interval, where hard coatings can be prepared)
from the balanced series was performed in order to see the
structure more clearly (see Fig. 7). This imaging clearly revealed the
presence of the dense columnar structure with TiC grains ranging
from about 5 nm to around 30 nm in size and often forming larger clusters,
relatively large distance between the clusters separated by the a-C:
H matrix can be seen.
The grain size and lattice parameter of TiC nanocrystallites were investigated
using grazing angle X-ray diffraction (XRD). The grain size
was calculated using Scherrer equation attributing the total broadening
of the (111) peak to the crystallite size only and is plotted in Fig. 8a.
Scherrer formula thus reports the smallest possible value of the mean
crystallite size in the crystallographic direction [111]. Crystallite sizes
from the Scherrer formula in the in-plane direction given by XRD
were smaller than the determined size in the growth direction, as was
shown by the TEM. More accurate values of the crystallite size were
obtained for selected samples using Williamson–Hall analysis of the
diffraction maxima 111 and 222, which correspond to the same crystallographic
plane. The mean crystallite size in these samples turned out to
be about twice as big as the minimal values given by the Scherrer equation.
The crystallite size remains stable around ~15 nm for carbon content
of b60 at.%, including the hypostoichiometric TiCx (x b 1) region (C
content b 50 at.%). Then a drop of the grain size to ~5 nm is observed in
the coatings with carbon content N 60 at.%. This is attributed to the
competition between grain nucleation and grain growth, i.e. the
nanocrystallites grow rather in number than in size at high carbon content
[19]. The evolution of the crystallite size with carbon content is
identical for both series regardless of the ﬂux of ions bombarding the
Fig. 4. Evolution of the normalized deposition rate of nc-TiC/a-C:H nanocomposite
coatings.
Fig. 5. SEM images revealing the microstructure of nc-TiC/a-C:H nanocomposite coatings
of similar composition on cermet substrates: (a) BAL-55C and (b) UNB-57C coating.
1
Ti/C ratio in the text is presented where coatings with similar carbon content show
very different titanium content or vice versa.
11P. Souček et al. / Surface & Coatings Technology 255 (2014) 8–14
growing coating. Similar trend in the grain size can be also found in
literature. Martínez-Martínez et al. [14] showed an increase in TiC crystallite
size from 6 nm to 30 nm when titanium concentration reached
49 at.%. Zehnder and Patscheider [21] demonstrated a sharp increase
from ~5 nm to over 20 nm with increasing titanium concentration
from 40 at.% to 55 at.%. Gulbinsky et al. [15], on the other hand, showed
a maximal TiC crystallite size of 75 nm at nearly stoichiometric TiC composition
and a decrease in TiC crystallite size for both higher and lower
concentration of titanium. The evolution of lattice parameter of TiC
nanocrystallites in both series is plotted in Fig. 8b. A clear increase in
the lattice parameter from ~4.259 Å at highly substoichiometric TiC
composition (28 at.% C) to 4.375 Å at nearly stoichiometric TiC composition
(48 at.% C). This increase of the lattice parameter can be explained
by reduced vacancies in the substoichiometric TiC lattice with increasing
C content. The lattice parameter remained around 4.375 Å with further
increasing the C content up to 70 at.% C. The lattice parameter of TiC
phase formed under equilibrium condition is 4.3274 Å [32], although
the lattice parameter larger than the reference value is quite often
observed in sputtered coatings [33,34]. Lewin et al. reported that expansion
of the lattice parameter was connected to charge transfer at the
interface from the carbide phase to the more electronegative carbon
matrix phase [35,36]. It is concluded that the evolution of the grain
size and of the lattice parameter is identical for both series of coatings
regardless of the ﬂux of bombarding ions.
The mechanical properties were evaluated using nanoindentation.
The maximal hardness of 46 ± 2 GPa and indentation modulus of
415 ± 7 GPa were achieved at 55 at.% C in the case of BAL-55C deposited
with the well balanced magnetron conﬁguration (see Fig. 9). In
the case of unbalanced magnetron sputtered coatings, the maximal
hardness and modulus were 45 ± 2 GPa and 420 ± 20 GPa, respectively,
measured at the carbon content of 53.5 at.% (UNB-54C). The
maximal hardness and modulus were reached at nearly the same composition
and are almost identical in value. No broadening of the optimal
composition range to reach high hardness and modulus was observed
even with strongly unbalanced magnetron sputtering, different from
the result reported by Samuelsson et al. [17] in HiPIMS deposition, in
which ionized species are delivered onto substrate and thus, mayFig. 6. SEM micrographs showing the fracture cross section of coatings deposited on cermet
substrates with a strongly unbalanced magnetron: (a) columnar structure of UNB-
68C coating (Ti/C = 0.45) and (b) column-free microstructure of UNB-70C coating (Ti/
C = 0.1).
Fig. 8. Dependence of (a) crystallite size and (b) lattice parameter of TiC nanocrystallites
on the carbon content of the coatings as measured by GI-XRD.Fig. 7. TEM image showing TiC grains in a-C:H matrix of BAL-55C.
12 P. Souček et al. / Surface & Coatings Technology 255 (2014) 8–14
change the microstructure and property of the coatings. A report on the
indentation response of the coatings can be found in [37].
The hardest samples from both series exhibited larger grains
~18 nm, which is contrary to the Hall–Petch theory in poly-crystallites
[38–40] which shows that the hardness is inversely proportional to
the crystal size. TEM image of the hardest coating from the balanced
series revealed the presence of non-negligible amount (much more
than a few atomic layers between grains) of a-C:H matrix and grain sliding
may occur at higher C/Ti ratios (see Fig. 7). Further optimization of
deposition parameters may lead to grain reﬁning at similar composition
thus mechanical properties may be enhanced as was proposed in
[36,41]. A similar trend, that is hardest coating with large TiC crystallite
size, was reported by Gulbiński et al. [15].
Also other parameters like toughness and resilience of the coating
play an important role in industrial applications of protective coatings.
Toughness can be deﬁned as the ability of a material to absorb energy
during deformation up to rupture. It has been proposed that the toughness
depends on the hardness to Young's modulus ratio H/E [8,14,42].
The prediction of the wear of the coatings can be based on the resilience
of the coating, which is deﬁned as the resistance of a material against
plastic deformation and relates to the ratio of H3
/E2
. Fig. 10 shows the
plot of H/E and H3
/E2
on the total carbon content. The H/E parameter
varies from 0.06 to 0.12 for the balanced series and from 0.08 to 0.12
for the unbalanced series (see Fig. 8). Both series show similar trends,
i.e. a low H/E (b0.1) in the coatings with a high Ti content and a high
H/E (~0.11–0.12) achieved in coatings with high carbon content. The
maximum H/E ratio of 0.12 is reached in coatings BAL-59C and UNB-
59C of the same carbon content. Both maxima of H/E lie in the region
with a higher C concentration than corresponds to the maximum hardness.
The values of H/E are in good accordance with Pei et al. [8], where
H/E ranges from 0.87 to 0.123 and the maximum achieved in the coating
of 87 at.% C and with Martínez-Martínez et al. [14] where H/E ranges
from 0.05 for pure titanium coating to 0.11 for nc-TiC/a-C coating with
79% of a-C matrix. For comparison the highest H/E values achievable
with heat-treated tool steel is ~0.04, for ceramics like Al2O3, ZrO2,
Si3N4 or SiC ~0.06 and nitride based coatings like TiN and CrN ~0.08
[43]. The composition of the coatings for the maximum H3
/E2
corresponds
with that for the maximum hardness for both series. BAL-55C
with a hardness of 46 GPa shows H3
/E2
of 0.57 GPa and UNB-54C
with hardness of 45 GPa shows H3
/E2
of 0.52 GPa. Martínez-Martínez
et al. [14] reported a maximum H3
/E2
ratio of 0.31 for the coating with
47 mol% of a-C matrix (66% C and 34% Ti).
Good mechanical performance of the nc-TiC/a-C:H coatings is illustrated
in Fig. 11 that shows residual indents made with a 1 N load for
the unbalanced series. No cracks were propagating from the tips of the
Fig. 10. Dependence of (a) H/E and (b) H3
/E2
on carbon content.Fig. 9. (a) Hardness and (b) modulus of nc-TiC/a-C:H nanocomposite coatings showing a
pronounced maximum at ~55 at.% C for both series of coatings.
Fig. 11. SEM image of an indent at 1 N on coating UNB-55C (Ti/C = 0.74).
13P. Souček et al. / Surface & Coatings Technology 255 (2014) 8–14
indent, only some minor cracks parallel to the indenter edges within the
indent itself were visible (marked with white arrows) for coatings with
high hardness. No cracks were visible for other coatings in the series.
The trend was the same in the balanced series. This was an indication
of high toughness. Pei et al. [44] observed distinct cracks propagating
in the coatings of columnar microstructure from the corners of indents
made with a cube-corner indenter at lower load of 190 mN.
4. Conclusions
Thick adherent nc-TiC/a-C:H nanocomposite coatings were prepared
with magnetron sputtering at high deposition rates with −100 V substrate
bias at low and high ﬂux of impinging Ar ions using well balanced
and strongly unbalanced magnetic ﬁeld conﬁguration at elevated deposition
temperature of ~320 °C. Super-hard coatings were prepared at
both magnetic conﬁgurations with maximal hardness and modulus of
45 GPa and 420 GPa. The coating structure, grain size, lattice parameter
and mechanical properties were governed primarily by the composition
of the coatings. The level of the ion ﬂux did not play a signiﬁcant role in
the deposition process in the presented setup, though it cannot be
ruled out that the effect of the ion ﬂux would be more noticeable
for coatings deposited at ambient temperatures as higher deposition
temperature associated with high deposition rate can partially negate
these effects. It can be concluded that at elevated temperatures superhard
coatings were prepared by optimizing the chemical composition
as the ion ﬂux played only a minor role in the determining of the mechanical
properties.
Acknowledgment
This research has been supported by GACR contracts 104/09/H080
and 205/12/0407 and R&D center project for low-cost plasma
and nanotechnology surface modiﬁcations CZ.1.05/2.1.00/03.0086
funded by European Regional Development Fund. This work was also
supported by RVO61389005. Pavel Souček acknowledges the Brno
City Municipality as recipient of Brno PhD Talent Financial Aid.
References
[1] Y. Wang, X. Zhang, X. Wu, H. Zhang, X. Zhang, Appl. Surf. Sci. 255 (2008) 1801–1805.
[2] P. Souček, T. Schmidtová, L. Zábranský, V. Buršíková, P. Vašina, O. Caha, M. Jílek, A.A.
El Mel, P.Y. Tessier, J. Schäfer, J. Buršík, V. Peřina, R. Mikšová, Surf. Coat. Technol. 211
(2012) 111–115.
[3] A. Czyżniewski, W. Prechtl, J. Mater. Process. Technol. 157-158 (2004) 274–283.
[4] J.C. Sánchez-López, D. Martínez-Martínez, C. Lópes-Cartes, A. Fernández, Surf. Coat.
Technol. 202 (2008) 4011–4018.
[5] D.M. Cao, B. Feng, W.J. Meng, L.E. Rehn, P.M. Baldo, M.M. Khonsari, Appl. Phys. Lett.
79 (3) (2001) 329–331.
[6] J. Musil, P. Novák, R. Čerstvý, Z. Soukup, J. Vac. Sci. Technol. A 28 (2010) 244–249.
[7] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, C. Strondl, J. Eur. Ceram. Soc. 26 (2006)
565–570.
[8] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, A. Cavaleiro, Surf. Coat. Technol. 198 (2005)
44–50.
[9] D. Galvan, Y.T. Pei, J.Th.M. De Hosson, Acta Mater. 53 (2005) 3925–3934.
[10] Y. Hu, L. Li, X. Cai, Q. Chen, P.K. Chu, Diamond Relat. Mater. 16 (2007)
181–186.
[11] T. Schmidtová, P. Souček, P. Vašina, Surf. Coat. Technol. 205 (2011) S299–S303.
[12] V. Kulikovsky, A. Tarasenko, F. Fendrych, L. Jastrabik, D. Chvostova, F. Franc, L.
Soukup, Diamond Relat. Mater. 7 (1998) 774–778.
[13] A.A. El Mel, B. Angleraud, E. Gautron, A. Granier, P.Y. Tessier, Surf. Coat. Technol. 204
(2010) 1880–1883.
[14] D. Martínez-Martínez, C. López-Cartes, A. Fernández, J.C. Sánchez-López, Thin Solid
Films 517 (2009) 1662–1671.
[15] W. Gulbinski, S. Mathur, H. Shen, T. Suszko, A. Gilewicz, B. Warcholinski, Appl. Surf.
Sci. 239 (2005) 302–310.
[16] K. Polychronopoulou, C. Rebholz, M.A. Baker, L. Theodorou, N.G. Demas, S.J. Hinder,
A.A. Polycarpou, C.C. Doumanidis, K. Böbel, Diamond Relat. Mater. 17 (2008)
2054–2061.
[17] M. Samuelsson, K. Sarakinos, H. Högberg, E. Lewin, U. Jansson, B. Wälivaara, H.
Ljungcrantz, U. Helmersson, Surf. Coat. Technol. 206 (2012) 2396–2402.
[18] D. Galvan, Y.T. Pei, J.Th.M. De Hosson, Surf. Coat. Technol. 201 (2006) 590–598.
[19] Y.T. Pei, C.Q. Chen, K.P. Shaha, J.Th.M. De Hosson, J.W. Bradley, S.A. Voronin, M. Čada,
Acta Mater. 56 (2008) 696–709.
[20] J. Lin, J.J. Moore, B. Mishra, M. Pinkas, W.D. Sproul, Thin Solid Films 517 (2008)
1131–1135.
[21] T. Zehnder, J. Patscheider, Surf. Coat. Technol. 133–134 (2000) 138–144.
[22] A. Grill, Diamond Relat. Mater. 8 (1999) 428–434.
[23] K.E. Cooke, J. Hampshire, W. Southwall, D.G. Teer, Surf. Coat. Technol. 177–178
(2004) 789–794.
[24] M.A. Lieberman, A.J. Lichtenberg, Principles of Plasma Discharges and Materials
Processing, second ed. John Wiley & sons, Hoboken, 2005.
[25] J. Saarilahti, E. Rauhala, Nucl. Instrum. Meth. Phys. Res. Sect. B 64 (1992)
734–741.
[26] M. Mayer, SIMNRA User's Guide, Technical Report IPP 9/113, Max-Planck-Institut
fuer Plasmaphysick, Garching, Germany, 1997.
[27] W.C. Oliver, G.M. Pharr, J. Mater. Res. 19 (2004) 3–20.
[28] K.L. Johnson, J. Mech. Phys. Solid 18 (1970) 115–126.
[29] F. Amato, N.R. Panyala, P. Vašina, P. Souček, J. Havel, Rapid Commun. Mass Spectrom.
27 (2013) 1196–1202.
[30] S. Zhang, X.L. Bui, J. Jiang, X. Li, Surf. Coat. Technol. 198 (2005) 206–211.
[31] T. Schmidtová, P. Souček, V. Kudrle, P. Vašina, Surf. Coat. Technol. (2013), http:
//dx.doi.org/10.1016/j.surfcoat.2013.05.014.
[32] M.C. Morris, H.F. McMurdie, E.H. Evans, B. Paretzkin, H.S. Parker, N.C.
Panagiotopoulos, C.R. Hubbard, Standard X-ray Diffraction Powder Patterns:
Section 18 — Data for 58 Substances, NBS, Washington, 1981.
[33] G. Li, L.F. Xia, Thin Solid Films 396 (2001) 16–22.
[34] E. Lewin, E.O.Å. Persson, M. Lattemann, M. Stüber, M. Gorgoi, A. Sandell, C. Ziebert, F.
Schäfers, W. Braun, J. Halbritter, S. Ulrich, W. Eberhardt, L. Hultman, H. Siegbahn, S.
Svensson, U. Jansson, Surf. Coat. Technol. 202 (2008) 3563–3570.
[35] E. Lewin, M. Rasender, M. Klintenberg, A. Bergman, O. Eriksson, U. Jansson, Chem.
Phys. Lett. 496 (1–3) (2010) 95–99.
[36] U. Jansson, E. Lewin, Thin Solid Films 536 (2013) 1–24.
[37] L. Zábranský, V. Buršíková, J. Buršík, P. Vašina, P. Souček, O. Caha, M. Jílek, V. Peřina,
Chem. List. 106 (2012) s1508–s1511.
[38] E.O. Hall, Proc. Phys. Soc. Lond. Sect. B 64 (1951) 747–753.
[39] N.J. Petch, J. Iron Steel Inst. 174 (1953) 25–28.
[40] P.H. Mayrhofer, C. Mitterer, L. Hultman, H. Clemens, Prog. Mater. Sci. 51 (2006)
1032–1114.
[41] U. Jansson, E. Lewin, M. Rasander, O. Eriksson, B. André, U. Wiklund, Surf. Coat.
Technol. 206 (2011) 583–590.
[42] A. Leyland, A. Matthews, Wear 246 (2000) 1–11.
[43] In: C. Donnet, A. Erdemir (Eds.), Tribology of Diamond-like Carbon Films, Springer,
New York, 2008.
[44] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, Acta Mater. 53 (2005) 4505–4521.
14 P. Souček et al. / Surface & Coatings Technology 255 (2014) 8–14
On the signiﬁcance of running-in of hard nc-TiC/a-C:H coating for
short-term repeating machining
Radek Žemlička a
, Pavel Souček a,
⁎, Petr Vogl b
, Mojmír Jílek a,b
, Vilma Buršíková a
, Petr Vašina a
, Y.T. Pei a,c
a
Department of Physical Electronics, Faculty of Science, Masaryk University, Kotlářská 2, CZ–61137, Brno, Czech Republic
b
Platit a.s., Průmyslová 3, CZ-78701 Šu3mperk, Czech Republic
c
Advanced Production Engineering, Engineering and Technology Institute Groningen, University of Groningen, Nijenborgh 4, 9747 AG Groningen, The Netherlands
a b s t r a c ta r t i c l e i n f o
Article history:
Received 26 September 2016
Revised 14 January 2017
Accepted in revised form 7 February 2017
Available online 09 February 2017
The importance of the often omitted running-in phase of tribological testing was studied on hard nc-TiC/a-C:H
coating system selected as a representative of protective tribological coatings. The tribological measurements
were periodically interrupted and the counterpart ball was replaced by a new one after every interrupted test
to simulate machining applications, where the coatings are repeatedly in contact with a new material. The changes
in the coefﬁcient of friction and the wear rate of the counterpart ball were found to be dependent mainly on the
roughness of the coating at any stage. It was found out that after substantial smoothening of the wear track of the
measured coating, the coefﬁcient of friction stabilized at the steady-state value attained in an uninterrupted long
tribotest. This steady state value was reached at the very beginning of the test and initial instabilities in the coefﬁcient
of friction were no longer observed. It was also concluded that only after the nc-TiC/a-C:H coating gets
smoothened enough within the wear track, the steady-state coefﬁcient of friction can describe even applications
where the counterpart is being periodically replaced by a new one.
© 2017 Elsevier B.V. All rights reserved.
Keywords:
TiC
Tribology
Run-in period
Machining
1. Introduction
Tribological studies are usually focused on the long term evolution of
the wear behavior and friction of a sliding couple [1], and the so-called
running-in stage that is the initial phase of sliding contact between
fresh, unworn solid surfaces is rarely studied and is often omitted [2].
The running-in period is usually a short transition phase and signiﬁcantly
differs from the steady-state of sliding contact that is generally studied
and quantiﬁed [1]. The running-in stage usually takes much less
time than the rest of the measurement and the value of the coefﬁcient
of friction (CoF) is often unstable in this initial measurement phase.
The omitting of the initial running-in period is easily justiﬁable if the
tested coating is designed for the applications where the same parts of
the machine are in constant contact with each other during their
whole life time, typically in car engines. On the other hand, there are applications
such that the coating gets in contact with a new counterpart
over and over again, such as drilling, milling, forming etc. For example,
a drill makes a hole usually just for several seconds and then it starts
to drill a new hole in a different place or a different workpiece. If there
is a transfer ﬁlm created between the contact surfaces of the drill and
one hole, this transfer ﬁlm does not play a role in drilling another
hole. What coefﬁcient of friction is more relevant for such applications?
Is it the one measured in the running-in phase or the steady-state coefﬁcient
of friction? To answer this question, the running-in stage was
studied in detail in both continuous and interrupted tribotests.
Hard nc-TiC/a-C:H nanocomposite coatings consisting of TiC
nanocrystallites embedded in amorphous carbon matrix have lately
been the focus of many studies for an unusual combination of high hardness
[3–5] and low friction [6–9]. The tribological properties of these
coatings have been widely studied and are well understood [10–14].
The CoF was reported to depend on several coating parameters with
the most important being the content of carbon and the coating structure
[7,8,15]. In general, initially high CoF usually decreases to a lower
stable value at a long sliding distance. The achievement of a low and stable
CoF is typically explained by creation of a carbon rich transfer ﬁlm in
between the counterpart ball and the testing coating [16].
In this study we investigated the signiﬁcance of the running-in stage
and if the value of the coefﬁcient of friction measured after the initial
stabilization is even relevant for short-term repeated applications such
as drilling. Furthermore, we studied the inﬂuence of the coating surface
roughness on the evolution of the CoF to investigate the origin of the
running-in aspect.
2. Experimental
Hybrid PVD-PECVD technique was used for the deposition of nc-TiC/aC:H
coating with carbon content of approximately 90 at.%. The deposition
Surface & Coatings Technology 315 (2017) 17–23
⁎ Corresponding author.
E-mail address: soucek@physics.muni.cz (P. Souček).
http://dx.doi.org/10.1016/j.surfcoat.2017.02.020
0257-8972/© 2017 Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Surface & Coatings Technology
journal homepage: www.elsevier.com/locate/surfcoat
was carried out in an industrial coating machine Pi 411 from Platit. A
single central rotating cylindrical titanium cathode (Ø110 × 510 mm,
20 mm thick, impurities max. 0.17 mass %) was sputtered in argon
(purity N 99.999%) and acetylene (purity 99.8%) atmosphere. Hardened
M2 steel substrates (Ø30 × 6 mm) were pre-cleaned in a chemical bath
and then ultrasonicated in a degreasing agent. The pre-cleaned samples
were put into the chamber and placed in substrate holders around the
central cathode.
The chamber was ﬁrst evacuated to the base pressure of 1 × 10−3
Pa
and subsequently heated to 450 °C. Then the samples were cleaned by
argon ion bombardment. The deposition was then carried out with the
following deposition parameters: 20 kW of magnetron power,
−100 V bias, 25 sccm of Ar (~0.42 Pa) and a substrate temperature of
450 °C. The samples performed a 3 axes planetary rotation to ensure
good coating homogeneity. Firstly, a titanium nitride layer ~1 μm thick
was deposited using sputtering of titanium target in argon/nitrogen
atmosphere. The TiN initial layer was followed by a hard interlayer of
nc-TiC/a-C:H coating 0.5 μm thick with a hardness of 38 GPa and carbon
content of 50 at.%. This layer was created using sputtering of titanium
target in argon/acetylene atmosphere as described in [17]. A nc-TiC/aC:H
top coat with ~90 at.% carbon and thickness of 1 μm was ﬁnally
deposited by increasing the ﬂow of acetylene. Thus, the total thickness
of the deposited coating system was 2.5 μm with the tribological topcoat
thickness of 1 μm.
Tribotests were done using a CSM high temperature tribometer in a
ball-on-disc conﬁguration and in circular mode at sliding speeds of 0.1
and 0.3 m/s and normal loads of 5 N and 10 N, respectively, against uncoated
Ø6 mm 100Cr6 steel ball (hardness of 7 GPa). The tribotests were
always carried out for a speciﬁc number of laps (10,000 laps–sliding distance
of approximately 180 m). The tests were carried out at humidity
of 50% and at room temperature of 20 ± 1 °C. The roughness and
wear were quantiﬁed using a laser confocal optical microscope OLYMPUS
OLS 4000. The morphology of nc-TiC/a-C:H coatings was further
characterized using DektakXT 3D proﬁlometer and Philips XL30 ESEM
scanning electron microscope operating at acceleration voltage of 10 kV.
3. Results
Fig. 1 shows the evolution of the coefﬁcient of friction in dry sliding,
where the test was repeatedly interrupted after every 2000 laps. The experiment
was carried out at sliding speed of 0.3 m/s and normal load of
5 N. The counterpart ball was changed for each new run and the wear
track was ultrasonically cleaned by each interruption to remove any debris
and also possible carbon rich transfer ﬁlm formed in the wear track.
It was observed that during the ﬁrst run the CoF remained ~0.25 in the
whole test of 2000 laps (7 min). At the start of the second run the CoF
ﬁrst increased to 0.3 and then slowly decreased to 0.2. The third run
started again with a high value of CoF of 0.3 and this value gradually decreased
to 0.17. The CoF was 0.25 at the start of the fourth run and stabilized
at 0.15 after ~500 laps. The ﬁfth run was similar to the fourth run
with a less pronounced initial CoF increase. The runs 6 and 7 exhibited a
relatively stable value of CoF of ~0.18 with no pronounced initial CoF
evolution during the ﬁrst laps of each run. Small peaks in the CoF
curve followed by gradual decrease are typical manifestation of the tearing
off of the transfer ﬁlm and its gradual building up thereafter. For the
few ﬁrst interrupted runs, the initial CoF measured at the beginning of
the successive run was higher than the steady-state value of the CoF
measured at the end of the previous run. It was concluded that until
the ﬁfth run, the replacement of a new counterpart ball by the interruption
played a role in the evolution of the CoF during the running-in period.
These measurements were repeated several times and the same
behavior was always observed such that the initial increase of the coefﬁcient
of friction typically disappeared in the fourth or ﬁfth run.
In order to investigate the role of possible carbon rich transfer ﬁlm a
similar experiment was performed with the wear track not being
cleaned by the individual interruptions but only the counterpart ball
Fig. 2. Evolution of the coefﬁcient of friction in an interrupted tribo-test carried out at the sliding speed of 0.3 m/s and normal load of 5 N. The test was repeatedly interrupted after every
2000 laps. The wear track was not cleaned by each interruption.
Fig. 1. Evolution of the coefﬁcient of friction in an interrupted tribo-test carried out at the sliding speed of 0.3 m/s and normal load of 5 N. The test was repeatedly interrupted after every
2000 laps. The wear track was ultrasonically cleaned by each interruption.
18 R. Žemlička et al. / Surface & Coatings Technology 315 (2017) 17–23
was replaced with a new one every interruption. The measured evolution
of the CoF is plotted in Fig. 2. There was no appreciable difference
from the previous group of experiments where the wear track was
cleaned after every interruption. Again the CoF was high and stable
(~0.25) in the ﬁrst run. The CoF was initially high and gradually decreased
in the second to fourth run, and the value of the CoF was relatively
stable (~0.15–0.2) during all the subsequent runs. One can thus
conclude that the cleaning of the wear track did not play a signiﬁcant
role in the measured evolution of the CoF during the running-in phase
of the measurement. The debris that remained on the surface of the tested
sample did not affect the friction and possible formation of the
transfer ﬁlm. This is highly advantageous for the transfer of laboratory
results to industrial applications, where the cutting tools are not usually
cleaned during work. The runs six and seven as seen in Fig. 2 were further
prolonged from 2000 to 10,000 laps in order to test a long term stability
of the CoF. The slight gradual increase of the CoF during the run
sixth and seventh can be attributed to a gradual wear and roughening
of the steel ball with the transfer ﬁlm being thinner or even missing at
the edges of the contact area of the ball and the coating [18].
Fig. 3 depicts the wear scar of counterpart balls taken after each run
corresponding to the CoF plotted in Fig. 1. The wear volume of the steel
ball decreased at every consecutive run with most pronounced changes
Fig. 4. Optical image of the wear scar of counterpart balls from the interrupted tribo-test corresponding to the CoF curves plotted in Fig. 2. Five runs of equal sliding length of 2000 laps and
two runs of 10,000 laps were performed with the wear track not being cleaned after each run.
Fig. 3. Optical image of the wear scar of counterpart balls from the interrupted tribo-test corresponding to the CoF curves plotted in Fig. 1. Seven runs of equal sliding length of 2000 laps
were performed with the wear track being cleaned after each run.
19R. Žemlička et al. / Surface & Coatings Technology 315 (2017) 17–23
between the ﬁrst several runs. Typically, the runs where higher CoF was
measured resulted in several times larger wear volume of the steel ball.
It was also found out that the wear scars on the counterpart ball observed
after testing both with and without cleaning of the racetrack
yielded were of similar size after the same number of laps, as is illustrated
in Fig. 4, where the wear scars of the balls used in the experiment in
Fig. 2 are shown. Furthermore, even if the sixth and seventh runs were
prolonged from 2000 laps (experiment in Fig. 1) to 10,000 laps (experiment
in Fig. 2) the wear scars were still comparable in size (see images
of wear scars from the seventh run in Figs. 3 and 4) This indicates that
the wear of the counterpart ball is most pronounced at the start of the
measurement and is minor during the later stages of the test, as
prolonging the sliding distance ﬁve times in the last two runs led to a little
increase of counterpart ball wear.
The length and the character of the running-in period were further
studied for different normal loads and sliding speeds. The normal load
in the previous experiments with repeated interruptions was always
set to 5 N. Fig. 5 shows the CoF evolution with a normal load of 10 N.
The wear track was not cleaned in this test. The comparison of the experimental
results between 5 N load and 10 N load showed that increasing
the applied load led to shortening of the initial high CoF running-in
phase. A low and stable CoF was achieved by the ﬁfth run with 5 N load,
while the same was achieved by third run when the load was doubled to
10 N.
Furthermore, a repeatedly interrupted tribo-test for comparison was
carried out where the wear track was cleaned by each interruption but
the ball was kept unchanged for all the runs. The CoF evolution (not
shown) exhibited similar behavior as in the case where the ball was
changed after every interruption. The CoF exhibited low and stable
value from the fourth run onwards, although a very short running-in
period can be now detected in all runs. The apparent size of the wear
scar on the ball was almost identical after the ﬁrst run of 2000 laps till
the end of the experiment at 30000 laps.
The evolution of the CoF plotted in Fig. 6 compares two experiments
with different sliding speeds without any interruptions. Fig. 6a shows
the CoF measured at load of 5 N and sliding speed of 30 cm/s as used
in previous experiments, while Fig. 6b shows the CoF measured at
lower sliding speed of 10 cm/s and 5 N load. The number of laps needed
to achieve a stable CoF was very close in both cases. As the sliding speed
was signiﬁcantly faster in Fig. 6a, the running-in period took markedly
shorter time – approximately 15 min at sliding speed of 30 cm/s as compared
to approximately 48 min with sliding speed of 10 cm/s.
The evolution of wear morphology of the coating was studied on the
same area of a wear track by each run. Fig. 7 shows the CoF evolution in
the interrupted test performed under the same sliding conditions as
shown in Fig. 2. The sample was not cleaned between the interruptions
and the same area on the wear track was examined by scanning electron
microscopy before the test and after selected runs in order to correlate
the evolution of the wear morphology and the CoF. The O symbol denotes
the original status of as-deposited coating before the tribotest,
letter A indicates the second group of micrographs taken after run 2.
The B group and C group, of micrographs were taken after the fourth
run and the ﬁnal seventh run, respectively. Fig. 8 depicts the micrographs
of the wear track taken at different moments as described previously.
The surface of the as-deposited coating was covered by typical
cauliﬂower patterns that result in a high roughness Ra of about
0.18 μm. As the tribotest progressed, the asperities were being polished
and the CoF was decreasing at the same time. When the asperities
Fig. 6. CoF curve measured at 5 N normal load and different sliding speeds: (a) 0.3 m/s and
(b) 0.1 m/s. The tests were not interrupted.
Fig. 5. Evolution of the coefﬁcient of friction measured at the applied load of 10 N and sliding speed of 0.3 m/s. The test was repeatedly interrupted. The wear track was not cleaned by each
interruption.
20 R. Žemlička et al. / Surface & Coatings Technology 315 (2017) 17–23
Fig. 8. SEM images showing the evolution of wear morphology of the same area in the wear track taken during consecutive runs indicated in Fig. 7.
Fig. 7. Evolution of the CoF in an interrupted tribo-test performed at 0.3 m/s sliding speed and 5 N load. The indexes O, A, B and C indicate SEM observations of the wear morphology of the
coating as shown in Fig. 8.
21R. Žemlička et al. / Surface & Coatings Technology 315 (2017) 17–23
appeared to be almost completely polished after four runs - resulting in
much smoother contact with the counterpart ball as seen in the C group
of micrographs - the CoF decreased to a low steady-state value.
The decrease of the surface roughness is closely correlated with the
decrease of the CoF and of the wear rate of the ball counterpart, as
shown in Fig. 9. The gradual polishing of the wear track by the ball during
sliding leads to a signiﬁcant decrease in the surface roughness of the
wear track, which in turn leads to considerable decrease of the coefﬁcient
of friction. When the coating was polished enough, the CoF stayed
stable at the same level even after the interruption so that the change of
the counterpart did not play a signiﬁcant role in the value of the CoF
during the studied running-in phase.
4. Discussion
The inﬂuence of the initial roughness on the tribological properties
of titanium carbon coatings was investigated by Shaha et al. [16]. The
time needed for the initial decrease and stabilization of the CoF was correlated
with the initial surface roughness. The initial decrease of the CoF
was explained by the formation of carbon-rich transfer ﬁlm between
the wear track and the counterpart. It was found out that a higher initial
coating roughness correlated with longer time needed to achieve a stable
CoF value. It was concluded that high initial roughness of the coating
impedes the creation of the transfer ﬁlm and the drop and the stabilization
of the CoF takes longer time in the case of rough coatings [16].
Similar explanation can be also used to describe the behavior of the
CoF in this work. The high initial roughness of the as-deposited coatings
(Ra ~ 0.18 μm) caused an initial roughening of the ball counterpart and
high wear of the ball. In this case the carbon-rich transfer ﬁlm did not
sufﬁciently cover the contact area and therefore the CoF value was
high. As the counterpart was changed to a new one for the next run,
the CoF increased at the beginning of the second run because of the
roughening of the surface of the new counterpart. During the second
run, the CoF started to decrease as the carbon rich transfer ﬁlm started
to form when the roughening of the contact area was less signiﬁcant
due to lower roughness of the coating wear track as compared to the
very ﬁrst run. As the coating got smoother after the very ﬁrst runs, the
time needed for the formation of a new transfer ﬁlm got shorter. Finally,
when the surface of the coating wear track was smoothed enough the
change of the counterpart ceased to inﬂuence the value of the CoF and
the CoF value was low and stable from the beginning of the run. The formation
of the carbon rich transfer layer in our experiments can be deduced
from the behavior of the CoF observed in the later runs of all
tests. A sudden increase in the CoF followed by a slower gradual decrease
to the steady-state value is a typical for a sudden breakdown
and gradual rebuilding of a carbon rich transfer layer between the sliding
bodies [18,19]. Also the steady state value of the CoF is consistent to
that of nc-TiC/a-C(:H) coatings with high carbon content that are
characterized by creation of the carbon transfer ﬁlm [7]. Moreover, residual
carbon debris was detected in and around the wear track after
the tribotests.
No signiﬁcant change in the CoF evolution was observed in the case
where the ball was kept unchanged for all the interrupted tests but the
wear track was cleaned after each interruption. This indicates that the
roughness of the much softer counterpart ball plays only a little role in
affecting the CoF behavior. Also as the low value of the CoF was achieved
at the very beginning of the later runs regardless of the experiment type,
no effect of chemical aging of the wear track and repeated creation of
disadvantageous contaminated and oxidized layer on the coating as
proposed by Rani et al. [20] was observed. Thus the CoF behavior can
be attributed only to the coating roughness.
The counterpart ball in tests with the signiﬁcantly smoothened coating
exhibited such a small wear rate that the size of its wear scar was
nearly unchanged with the length of sliding. The apparent wear volume
of the ball achieved in the last two runs of 10,000 laps each depicted in
Fig. 4 was very close to those depicted in Fig. 3 where the sliding distance
was only 2000 laps. Apart from the prolonged runs, moreover, the wear
track was not cleaned after every interruption in this case. As the CoF
was low and stable even at the start of the last two runs for both cleaned
and uncleaned wear track, the transfer ﬁlm was assumed to form fast at
the start of the measurement even when the wear track was cleaned.
The ever rapidly reached low and stable CoF in the test with doubled
load 10 N can be explained by a faster polishing of the coating due to
higher applied load. The same explanation holds for the experiments
verifying the inﬂuence of sliding speed. Although the time needed to
smoothen the coating and thus reach the low and stable CoF was different
at sliding speed of 0.3 m/s versus 0.1 m/s, the total number of laps
needed to reach such a regime was nearly the same. It implies that sliding
at lower speed leads to corresponding increase of running-in period
necessary to achieve low enough roughness for fast formation of transfer
ﬁlm that helps to reduce friction and wear.
5. Conclusions
Nanocomposite nc-TiC/a-C:H coating with high carbon content of
~90 at.% was deposited by a hybrid PVD-PECVD process in industrial conditions.
The running-in phase of ball-on-disc tribotests was analyzed in
detail in order to establish the validity of omitting the running-in phase
for short term repeated applications such as drilling, milling, forming, etc.
The roughness of the nc-TiC/a-C:H coating was found to be the primary
cause of the instabilities of the initial part of the tribo-testing.
The initial unstable CoF was identiﬁed to be caused by the initial decrease
of the roughness of the as-deposited coating. High initial roughness
of the coating led to slow or no formation of a carbon rich transfer
ﬁlm needed for achieving of low coefﬁcient of friction. After the roughness
of the hard coating decreased to a certain value, transfer ﬁlm could
form within a short time and the resulting CoF was stable. Roughness of
the softer steel ball and possible oxidation of the wear track were found
out to be of little inﬂuence.
When the CoF had stabilized, the interruption of sliding test and the
change of counterpart had no effect on further CoF evolution. Furthermore,
the steady-state CoF measured in the standard long uninterrupted
tests was identical to the one measured in interrupted short tests
after the coating became smoothened. Therefore, the standard CoF analysis
procedure involving quantiﬁcation of the value of CoF measured in
the steady state region after the running-in period is relevant even for
short term repeating applications such as drilling, milling or forming.
Acknowledgment
This research has been supported by GACR contract 205/12/0407, by
project CZ.1.05/2.1.00/03.0086 funded by European Regional Development
Fund, and project LO1411 (NPU I) funded by Ministry of Education,
Youth and Sports of Czech Republic.
Fig. 9. Evolution of CoF against the roughness of the wear track formed on the coating and
the wear volume of the ball counterpart after each run of 2000 laps.
22 R. Žemlička et al. / Surface & Coatings Technology 315 (2017) 17–23
References
[1] B. Bhushan, Principles and Applications of Tribology, John Wiley & Sons, 2013
321–401.
[2] P.J. Blau, On the nature of running-in, Tribol. Int. 38 (2005) 1007.
[3] Y. Wang, X. Zhang, X. Wu, H. Zhang, X. Zhang, Compositional, structural and mechanical
characteristics of nc-TiC/aC: H nanocomposite ﬁlms, Appl. Surf. Sci. 255
(5) (2008) 1801–1805.
[4] P. Souček, T. Schmidtová, L. Zábranský, V. Buršíková, P. Vašina, O. Caha, M. Jílek, A.A.
El Mel, P.Y. Tessier, J. Schäfer, J. Buršík, V. Peřina, R. Mikšová, Evaluation of composition,
mechanical properties and structure of nc-TiC/aC: H coatings prepared by balanced
magnetron sputtering, Surf. Coat. Technol. 211 (2012) 111–116.
[5] A. Czyżniewski, W. Precht, Deposition and some properties of nanocrystalline, nanocomposite
and amorphous carbon-based coatings for tribological applications, J.
Mater. Process. Technol. 157 (2004) 274–283.
[6] F.S. James, F. Shackelford, W. Alexander, Materials Science and Engineering Handbook,
CRC Press, LLC, 2001 471.
[7] J.C. Sánchez-López, D. Martínez-Martínez, C. López-Cartes, A. Fernández, Tribological
behaviour of titanium carbide/amorphous carbon nanocomposite coatings: from
macro to the micro-scale, Surf. Coat. Technol. 202 (16) (2008) 4011–4018.
[8] D.M. Cao, B. Feng, W.J. Meng, L.E. Rehn, P.M. Baldo, M.M. Khonsari, Friction and wear
characteristics of ceramic nanocomposite coatings: titanium carbide/amorphous
hydrocarbon, Appl. Phys. Lett. 79 (3) (2001) 329–331.
[9] J. Musil, P. Novák, R. Čerstvý, Z. Soukup, Tribological and mechanical properties of
nanocrystalline-TiC/aC nanocomposite thin ﬁlms, J. Vac. Sci. Technol. A 28 (2)
(2010) 244.
[10] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, A. Cavaleiro, Nanostructured TiC/aC coatings
for low friction and wear resistant applications, Surf. Coat. Technol. 198 (1)
(2005) 44–50.
[11] K.P. Shaha, Y.T. Pei, D. Martínez-Martínez, J.C. Sánchez-López, J.Th.M. De Hosson, Effect
of process parameters on mechanical and tribological performance of pulsedDC
sputtered TiC/a-C:H nanocomposite ﬁlms, Surf. Coat. Technol. 205 (7) (2010)
2633–2642.
[12] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, C. Strondl, Advanced TiC/aC: H nanocomposite
coatings deposited by magnetron sputtering, J. Eur. Ceram. Soc. 26 (4) (2006)
565–570.
[13] K. Polychronopoulou, C. Rebholz, M.A. Baker, L. Theodorou, N.G. Demas, S.J. Hinder,
A.A. Polycarpou, C.C. Doumanidis, K. Böbel, Nanostructure, mechanical and tribological
properties of reactive magnetron sputtered TiCx coatings, Diam. Relat. Mater. 17
(12) (2008) 2054–2061.
[14] T. Zehnder, P. Schwaller, F. Munnik, S. Mikhailov, J. Patscheider, Nanostructural and
mechanical properties of nanocomposite nc-TiC/aC: H ﬁlms deposited by reactive
unbalanced magnetron sputtering, J. Appl. Phys. 95 (8) (2004) 4327–4334.
[15] D. Martínez-Martínez, C. López-Cartes, A. Fernández, J.C. Sánchez-López, Inﬂuence
of the microstructure on the mechanical and tribological behavior of TiC/aC nanocomposite
coatings, Thin Solid Films 517 (5) (2009) 1662–1671.
[16] K.P. Shaha, Y.T. Pei, D. Martínez-Martínez, J.Th.M. De Hosson, Inﬂuence of surface
roughness on the transfer ﬁlm formation and frictional behavior of TiC/aC nanocomposite
coatings, Tribol. Lett. 41 (1) (2011) 97–101.
[17] R. Žemlička, M. Jílek, P. Vogl, P. Souček, V. Buršíková, J. Buršík, P. Vašina, Understanding
of hybrid PVD-PECVD process with the aim of growing hard nc-TiC/a-C: H coatings
using industrial devices with a rotating cylindrical magnetron, Surf. Coat.
Technol. 255 (2014) 118–123.
[18] Y.T. Pei, D. Galvan, J.Th.M. De Hosson, Nanostructure and properties of TiC/aC: H
composite coatings, Acta Mater. 53 (17) (2005) 4505–4521.
[19] K.P. Shaha, Y.T. Pei, D. Martinez-Martinez, J.Th.M. De Hosson, Inﬂuence of hardness
and roughness on the tribological performance of TiC/a-C nanocomposite coatings,
Surf. Coat. Technol. 205 (2010) 2624–2632.
[20] R. Rani, N. Kumar, A.T. Kozakov, K.A. Googlev, K.J. Sankaran, P.Kr. Das, S. Dash, A.K.
Tyagi, I.- Nan Lin, Superlubrication properties of ultrananocrystalline diamond ﬁlm
sliding against a zirconia ball, RSC Adv. 5 (2015) 100663–100673.
23R. Žemlička et al. / Surface & Coatings Technology 315 (2017) 17–23
Superhard nanocomposite nc-TiC/a-C:H coatings: The effect of HiPIMS on
coating microstructure and mechanical properties
Pavel Souček ⁎, Josef Daniel, Jaroslav Hnilica, Katarína Bernátová, Lukáš Zábranský, Vilma Buršíková,
Monika Stupavská, Petr Vašina
Department of Physical Electronics, Faculty of Science, Masaryk University, Kotlářská 2, CZ-61137 Brno, Czech Republic
a b s t r a c ta r t i c l e i n f o
Article history:
Received 1 July 2016
Revised 5 January 2017
Accepted in revised form 7 January 2017
Available online 09 January 2017
Nanocomposite coatings consisting of titanium carbide crystallites embedded in amorphous carbon matrix were
deposited utilizing direct current magnetron sputtering (DCMS) and high power impulse magnetron sputtering
(HiPIMS) of titanium target in argon/acetylene atmosphere. DC driven depositions of coatings with high amount
of carbon were accompanied by signiﬁcant arcing during the deposition. HiPIMS utilization made reliable arc-free
deposition of coatings with high carbon content possible. Analyses of the coating structure revealed that utilization
of HiPIMS favours carbon incorporation into the TiCx grains rather than to the matrix. HiPIMS deposited coatings
exhibited stoichiometry of the TiCx grains closer to unity with grains exhibiting lower lattice parameter and a
lower fraction of the a-C matrix. This led to a smaller mean TiCx grain separation by the a-C:H matrix. Also the
surface structure of the HiPIMS deposited coatings was much ﬁner. The structure reﬁnement resulted in overall
higher hardness of HiPIMS deposited coatings with the hardest coatings reaching the superhard region of
N40 GPa and in 40% increase of H/E ratio and nearly threefold increase of H3
/E2
ratio.
© 2017 Elsevier B.V. All rights reserved.
Keywords:
Nanocomposite
TiC
HiPIMS
Structure
Mechanical properties
1. Introduction
Coatings consisting of nanocrystalline titanium carbide grains embedded
in amorphous (hydrogenated) carbon matrix – nc-TiC/a-C(:H)
can be tailored from super-hard coatings [1,2,3] with hardness reaching
up to 46 GPa [4] to low friction coatings [5,6,7] by varying the phase
composition. Such a wide range of functional properties can be achieved
owing to nanocomposite nature of such coatings.
Hydrogen free nc-TiC/a-C coatings are usually deposited by
employing co-sputtering of carbon and titanium targets [8,9,10,11,12,
13,14], although other PVD methods such as electron beam titanium
evaporation with carbon magnetron sputtering [15], titanium magnetron
sputtering with carbon pulsed laser deposition [16,17], compound
TiC target sputtering with possible carbon target co-sputtering [5,18,19,
20], ICP assisted PVD/CVD [21] or combination of IPVD/PECVD [22]. Hydrogenated
nc-TiC/a-C:H coatings are most often deposited by
sputtering of titanium target in hydrocarbon containing plasma by a hybrid
PVD–PECVD process with methane [23,24] or acetylene [1,3,4,6,25,
26,27,28,29,30,31] as carbon precursors. If hydrogenated nc-TiC/a-C:H
coatings are deposited by direct-current (DC) magnetron sputtering
with argon/acetylene atmosphere, TiH phase can also be present in
the coating [32]. Typically, DC [3,4,5,6,7,21,22,23,24,25,26,28,30] or
mid-frequency pulsed DC (p-DC) [1,10,11,12,33] magnetron sputtering
is employed and the resulting deposited coatings have been thoroughly
studied in recent years. However, signiﬁcantly fewer reports dedicated
to nc-TiC/a-C:H coatings deposited by High Power Impulse Magnetron
Sputtering (HiPIMS) can be found [34,35]. Thus the HiPIMS deposition
process is much less described and the employment of HiPIMS on the
structure and mechanical properties of the deposited nc-TiC/a-C:H coatings
is not fully understood yet.
HiPIMS employs short intense power pulses on the sputtered target
with a typical duty cycle of a few percent [36,37,38,39]. This power time
distribution leads to plasma several orders of magnitude denser during
the pulse than it is for a DC magnetron sputter deposition process. Denser
plasma generally leads to high fraction of ionized sputtered species
[39,40,41]. Correspondingly, such deposition process usually leads to
total ion ﬂux on the substrate several orders of magnitude higher during
the pulse than it is in DC magnetron sputtering [36,42]. Furthermore,
the much denser plasma can enhance the plasma chemistry [43]
which can alter the dissociation of acetylene in the hybrid PVD-PECVD
process of nc-TiC/a-C:H deposition using acetylene gas as carbon source
for the coating. Such a high level of ion bombardment of the growing
coating and altered plasma chemistry can affect the structure and overall
properties of the resulting coating [44].
The present paper explores possibilities of nc-TiC/a-C:H deposition
by a hybrid PVD-PECVD process of Ti target sputtering in argon/acetylene
atmosphere. Coatings were deposited with DC and HiPIMS at the
same mean power. The conditions for HiPIMS were chosen in a way to
achieve high and stable current in the pulse for at least half of the
Surface & Coatings Technology 311 (2017) 257–267
⁎ Corresponding author.
E-mail address: soucek@physics.muni.cz (P. Souček).
http://dx.doi.org/10.1016/j.surfcoat.2017.01.021
0257-8972/© 2017 Elsevier B.V. All rights reserved.
Contents lists available at ScienceDirect
Surface & Coatings Technology
journal homepage: www.elsevier.com/locate/surfcoat
pulse duration. In addition, biasing conditions were chosen to attain
similar grain microstrain in DCMS and HiPIMS. This allows us to focus
only on the differences in the structure of the deposited coatings and
its effects on the mechanical properties without introducing further
stress induced hardening effects.
The evolution of the discharge state with acetylene ﬂow is described.
The possibilities of carbon rich coating deposition by DCMS and HiPIMS
will be discussed. Coatings deposited with DC, HiPIMS are mutually
compared, and the differences of the coatings properties are discussed.
Structural parameters leading to mechanical properties enhancement
for HiPIMS depositions are identiﬁed and the advantages and disadvantages
of HiPIMS depositions process to synthesize nc-TiC/a-C:H coatings
are shown.
2. Experimental setup
All depositions were carried out using industrial sputtering system
Alcatel SCM 650 with a cylindrical vacuum chamber ∅ 65 × 35 cm.
The deposition chamber was evacuated by a turbomolecular pump
backed by a Roots pump to the base pressure in the order of 10−4
Pa.
A well balanced magnetron source (see ref. [4] for speciﬁc magnetic
ﬁeld structure measurements) was equipped with a 99.95% purity ∅
20 cm Ti target. For direct current (DC) deposition the target was driven
by a Hüttinger TruPlasma Bipolar 4010 generator operated in DC mode.
Melec SIPP 2000 USB generator was used for High Power Impulse
Magnetron Sputtering (HiPIMS) experiments. Voltage and current
waveforms for HiPIMS discharge were measured using integrated
Melec measuring system containing voltage and current transducers
using Tektronix TDS 2014C digital oscilloscope.
Cemented tungsten carbide (hardmetal) substrates were placed on a
biasable (13.56 MHz RF) substrate holder located 6.5 cm from the target
above the target racetrack. Argon (purity 99.999%) was used as the
working gas and acetylene (purity 99.6%) was used as a carbon source.
Gasses were dosed via mass ﬂow controllers. The gas inlet was located
between the target and the substrate at a distance of 2 cm from the
target.
First, the hardmetal substrates were ultrasonically cleaned in a
degreasing agent. The titanium target was mechanically cleaned. Then
the vacuum chamber was evacuated. The target and the substrates
were further simultaneously cleaned by argon ion bombardment after
being separated by a shutter. The cleaning was performed until the discharge
voltage stabilized indicating that the target was completely
cleaned. This usually took 1–2 h. After the target and the chamber
cooled down to room temperature during the night, the deposition procedure
commenced by a short pre-deposition cleaning. This was done
for 35 min ensuring removal of any contaminants adsorbed during the
night. The substrates were pre-heated to ~150 °C at the same time by
the ion bombardment. The cleaning phase conditions were as follows:
65 sccm Ar (~3 Pa), 1 kW DC on the Ti target, 500 W RF on the substrate
holder.
Then the shutter between the substrate and the target was opened
and the argon ﬂow was set to 20 sccm (~1.1 Pa). For DCMS depositions
the power on the titanium target was set to 1.25 kW and the self-bias on
the RF driven substrate was set to −100 V. A pure titanium ~500 nm interlayer
was deposited before the nc-TiC/a-C:H coating in order to promote
adhesion. The acetylene ﬂow was then set to the desired value.
Coatings in DCMS conditions were deposited for 60 min in order to prepare
at least two micrometres thick ﬁlms. Although no intentional
heating was applied, the temperature on the substrate during the
deposition was measured to be ~300 °C due to close proximity of the
substrate and the target. For HiPIMS experiments, the sample holder
was disconnected after the cleaning phase and was left at ﬂoating potential.
The potential time averaged through the pulse was measured
to be ~−3 V during the deposition. HiPIMS depositions were performed
similarly to DCMS depositions at the same mean power of 1.25 kW with
several exceptions - thinner ~100 nm Ti interlayer was deposited due to
lower deposition rate of HiPIMS, 400 μs long pulse with repetition frequency
of 30 Hz (1.2% duty cycle) was used instead of DC, the deposition
time was doubled to 120 min to again achieve at least ~2 μm thin ﬁlm
and ﬂoating substrate was used instead of RF biasing. In DCMS the
power of 1.25 kW corresponded to the discharge voltage of ~−360 V
and discharge current of ~3.5 A for a clean Ti target. The temperature
of the substrates during the deposition was measured also to be
~300 °C.
The surface of nc-TiC/a-C:H coatings was imaged using a Tescan
MIRA 3 ﬁeld-emission scanning electron microscope (SEM) with electron
accelerating voltage of 10 kV. The thickness was determined by
abrading a spherical cap by a calotester and consecutive imaging of
the spherical cap by LEXT OLS4000 3D Laser Measuring Confocal Microscope.
The chemical composition was measured using Oxford Instruments
EDX attached to the SEM. The quantiﬁcation of the elemental
composition was made using internal standards library supplied by
Oxford Instruments within AZtec Software. The chemical composition
referred to throughout the text was always evaluated by EDX. Grazing
angle of incidence X-ray diffraction (GIXRD) utilizing RIGAKU SmartLab
diffractometer was performed to analyse the crystalline structure. The
angle of incidence of the copper Kα X-ray (λ = 1.54056 Å) source was
set to 0.5°. GIXRD was chosen to eliminate WC diffraction peaks from
the cemented tungsten carbide substrate as WC and TiC diffraction
peaks overlap. After correction for the instrumental broadening, the
grain size was calculated from the most intensive TiC diffraction peak
corresponding to (111)diffraction using the Scherrer formula attributing
the total broadening of the peak to the crystallite size only
(neglecting the strain broadening). The Scherrer formula thus reports
the smallest possible value of the mean crystallite size. The microstrain
of the crystallites was estimated by thy Williamson-Hall plot. XPS analysis
were performed to analyse the bonding state of the carbon atoms,
to calculate the amount of the a-C:H phase, the grain stoichiometry factor
and the mean grain separation. The XPS data were recorded using a
Thermo Scientiﬁc K-Alpha XPS system equipped with a micro-focused,
monochromatic Al Kα X-ray source (1486.6 eV). High resolution spectra
were recorded with pass energy of 50 eV. The samples were cleaned by
0.5 keV Ar ion bombardment for 30 min before every XPS measurement.
Low ion energy ensured that the coating itself was not affected by the
cleaning and a long time of cleaning ensured removal of all surface contaminants.
The recorded spectra were calibrated by setting the C1 s peak
to 285 eV. Horriba LabRAM HR Evolution microraman spectrometer
with 532 nm laser was used to analyse the bonding state of the a-C:H
matrix. The acquired Raman spectra were ﬁtted with four peaks – the
D peak located ~1350 cm−1
, the G peak located ~1570 cm−1
and the
D1 and G1 peaks located ~1220 and 1350 cm−1
, respectively - similarly
to [30,45,46,47]. The G and D peaks arise due to stretching or breathing
of the sp2
carbon sites [47]. D1 and G1 peaks (alternatively can be found
denoted as ν1 and ν2) can be attributed to trans-polyacetylene chains
[46,47]. Fischerscope H100 depth sensing indenter equipped with a
Berkovich tip was used to study the indentation response of the coated
samples. At least 12 indents were made at each load on every sample.
The hardness and the elastic modulus were evaluated using the standard
procedure proposed by Oliver and Pharr [48]. Such testing conditions
were used in nanoindentation measurements under which the
substrate did not inﬂuence the measured values of the hardness and
the effective elastic modulus (Eeff = E / (1 − ν2
), where E is the Young's
modulus and ν is the Poisson ratio).
3. Results
A typical waveform of the discharge current and cathode voltage acquired
for a clean Ti target during the HiPIMS pulse is plotted in Fig. 1.
The pulse voltage in this case was ~−525 V. The discharge current
was increasing for the ﬁrst 175 μs and saturated at a nearly steady
state value of ~260 A for the remaining 225 μs. This means that a high
discharge current density of 3 A/cm2
of the racetrack surface was
258 P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
achieved for ~50% of each pulse. The values of the discharge current and
cathode voltage were evolving slightly after the addition of acetylene
and were stabilized in approximately 15 min for both the DCMS and
the HiPIMS process signifying that the coatings were grown in stabilized
conditions for 75% of the deposition time for DCMS and 88% of the deposition
time for HiPIMS. In the DCMS the voltage shifted from ~−360 V
for no acetylene supply to −370–−470 V for acetylene ﬂow range
used. The voltage shift in HiPIMS case was from −525 V for no acetylene
to −480 V to −580 V for process with acetylene. These conditions resulted
in a nearly homogeneous depth proﬁle of coating composition
and mechanical properties. This was veriﬁed by a technique described
in [49], where EDX and nanoindentation measurements in an abraded
crater (calotte) throughout the coating are proposed for in-depth analysis
of chemical composition and mechanical properties. Achieving a
homogeneous depth proﬁle of the coating properties simpliﬁes evaluation
of the EDX and indentation measurements performed from the top
of the coating.
Fig. 2a depicts the plot of the acetylene supply needed to achieve a
speciﬁc carbon content in the deposited coatings and Fig. 2b depicts
the deposition rate of the corresponding coatings. The carbon content
on the x-axis for Fig. 2 as well as all other graphs is derived from EDX
analyses. The speciﬁc values of the carbon content as well as other coating
parameters are also summarized in Table 1. For both the DCMS and
HiPIMS processes, coatings with carbon content lower than ~55 at.% exhibited
approximately linear increase of the carbon content with increasing
acetylene supply and a nearly stable deposition rate.
However, approximately 2–3 times higher acetylene ﬂow was required
to deposit coatings with the same composition in DCMS as in HiPIMS.
The deposition rate of HiPIMS deposited samples was ~20 nm/min,
whereas it was ~40 nm/min for DCMS depositions. Signiﬁcantly lower
deposition rate of coatings with the same carbon content deposited in
HiPIMS conditions together with higher acetylene supply needed to
achieve the same coating composition in DCMS was explained by
lower titanium ﬂux onto the coating in HiPIMS depositions compared
to DCMS case. Because high discharge current density was reached
and maintained for a signiﬁcant part of the pulse, high ionization of
sputtered titanium atoms and their back-attraction to the target can
be expected [50]. This agrees well with the available literature, as the
deposition rate of metals in HiPIMS is often decidedly lower than in
DCMS [34,51,52]. A small reduction of the deposition rate of HiPIMS deposited
coatings around 55 at.% of C was observed, similarly as in [34].
Samuelsson et al. [34] also observed that for depositions of TiC/C:H coatings
the Ti/C ratio ~ 1 could have been reached at much broader interval
of acetylene supplies in HiPIMS than was possible in DCMS process,
however opposite behaviour was observed in this work. In our case a
much smaller change of the acetylene gas supply was needed to alter
the composition of the deposited coating in the HiPIMS process compared
to DCMS. Thus such a convenient broadening of the acetylene
supply interval as described by Samuelsson et al. [34] is not universal
for HiPIMS sputtering of Ti target in argon/acetylene mixture and that
such a trend was rather speciﬁc for the employed experimental setup
and conditions.
Coatings with carbon content higher than ~55 at.% deposited by
DCMS also showed a pronounced decrease of the deposition rate with
increasing carbon content. In case of reactive magnetron sputtering,
similar drop of the deposition rate is being explained by changes of
the target state [53,54] which are usually reﬂected in the evolution of
the discharge voltage [55]. Fig. 3 shows the corresponding dependence
of the discharge voltage on the carbon content in the coating deposited
using hybrid PVD-PECVD process with both the DC and HiPIMS plasma
excitation. Whereas the discharge voltage in DCMS is nearly constant
for coatings with carbon content lower than 50 at.%, a very shallow minimum
at carbon content of ~50 at.% was observed. The hybrid deposition
process behaviour and the existence of the local voltage
minimum in DC driven process was studied in detail in our previous
work and was attributed to the formation of the TiC on the target racetrack
[56]. The shallow local minimum was followed by a steep increase
of the discharge voltage that was attributed to the formation of the carbon
rich layer at the target racetrack [56]. The decrease of the deposition
rate at higher carbon contents in the DCMS deposited coatings can be
explained by lower ﬂux of the sputtered titanium from the target as
the target racetrack got covered by carbon residue which prevented
the titanium sputtering. The effect of lower sputtering rate was probably
not sufﬁciently compensated by the increase of carbon ﬂux on the
Fig. 1. Voltage and current waveform of the HiPIMS pulse for a clean Ti target.
Fig. 2. a) Acetylene ﬂux needed to deposit coatings with speciﬁc composition b)
Deposition rate of corresponding coatings. Lines are added as guides for the eye.
259P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
coating. Furthermore, with DCMS sputtering it was not possible to prepare
coatings with carbon content higher than ~72 at.% because the
racetrack got almost completely covered with black carbon residue
and signiﬁcant arcing occurred during the deposition at these conditions.
A photograph of the target surface after deposition of coating
with 72 at.% is depicted in Fig. 4a. It shows the racetrack to be almost entirely
covered by black carbon rich layer.
Fig. 4b shows the target racetrack after HiPIMS deposition process
leading to deposition of a coating with 94 at.% C. The whole target exhibited
signiﬁcantly lower amount of carbon contamination than in the
case of DCMS. We supposed that strong back attraction of the sputtered
and ionized metallic atoms to the target in HiPIMS conditions kept the
target surface more metallic and clean of carbon contaminates as compared
to DCMS case, where the number of titanium ions is low. This
led to postponement of the discharge voltage rise until depositions leading
to coatings with ~80 at.% C in HiPIMS were carried out as compared
to ~50 at.% for DCMS (see Fig. 3). Before the onset of the voltage rise, the
HiPIMS discharge voltage exhibited a decrease when the carbon content
was increased from ~55 at.% to 80 at.%. This can be again explained by
formation of the TiC on the target racetrack as in the DCMS case, although
shifted to higher carbon contents of the coatings.
Using HiPIMS it was possible to perform controlled arc-free depositions
of coatings with much higher carbon content than in the DCMS
case. Furthermore, the deposition rate in HiPIMS increased for coatings
with high carbon content. Similar though less pronounced increase of
the deposition rate in HiPIMS sputtering of Ti in argon/acetylene atmosphere
was shown by Samuelsson et al. [34]. The increase of the deposition
rate at high acetylene ﬂuxes with HiPIMS can be explained by the
enhanced chemistry in the plasma utilized by HiPIMS, as the deposition
rate of carbon coatings was shown to increase with the number of active
carbon species generated in the plasma [57,58].
Table 1
Overview of the deposited coatings.
C [at.%] Grain size [nm] Lattice parameter [Å] a-C phase [%] Stoichiometry H [GPa] Eeff [GPa]
DCMS
30 16 4.276 17.2 0.35 11 ± 1 238 ± 16
38 13 4.293 21 ± 2 261 ± 12
39 14 4.323 26.4 0.48 21 ± 2 279 ± 11
48 15 4.348 28 ± 3 386 ± 18
49 19 4.358 29 0.68 24 ± 2 339 ± 15
53 14 4.363 33 ± 4 326 ± 13
56 15 4.355 25.7 0.94 27 ± 1 254 ± 8
61 8 4.337 33 ± 3 268 ± 27
61 11 4.348 34.8 1.03 27 ± 3 227 ± 13
61 6 29 ± 1 236 ± 6
65 4 4.339 21 ± 3 183 ± 17
68 66.5 0.69
HiPIMS
25 20 4.251 12 ± 2 242 ± 20
40 17 4.295 5.8 0.63 32 ± 3 392 ± 12
48 25 4.319 39 ± 3 428 ± 17
49 20 4.318 14.1 0.83 38 ± 3 343 ± 18
53 17 4.322 16.6 0.94 40 ± 3 378 ± 30
57 9 4.313 25.2 0.99 45 ± 4 309 ± 12
68 4 55.4 0.95 32 ± 4 230 ± 12
79 2 4.311 85.3 0.55 20 ± 1 186 ± 6
85 1 4.313 91.9 0.46 20 ± 1 167 ± 3
94 20 ± 1 143 ± 3
Fig. 3. Steady state discharge voltage for HiPIMS and DC depositions. The onset of the
voltage increase is shifted to higher carbon contents for HiPIMS depositions. Lines are
added as guides for the eye.
Fig. 4. Target racetrack after deposition for processes leading to coatings with high carbon
content a) DC b) HiPIMS. Black carbon residue in the case of DC is present.
260 P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
SEM images of deposited coating surfaces are shown in Fig. 5. DCMS
deposited samples are shown in the left column and HiPIMS deposited
samples in the right column. Samples with similar carbon content are
always compared. Coatings with ~40 at.% C, coatings with ~53 at.% C
and coatings with ~65–70 at.% C were selected to demonstrate the evolution
of the coating surface morphology. All images were taken with
identical magniﬁcation of 50 kx and view ﬁeld of 5.54 μm. There was
no appreciable difference between DCMS and HiPIMS deposited coatings
with ~40 at.% of carbon. Both exhibited patterned structure with
surface features several hundred nanometres in size. DCMS deposited
coating with 53 at.% of carbon exhibited larger clusters with the size in
the order of hundreds of nanometres with secondary smaller structures
with dimensions in the order of tens of nanometres. HiPIMS deposited
coating with the same carbon content exhibited signiﬁcantly more reﬁned
structure indicating a denser coating structure. The biggest difference
between the DCMS and HiPIMS deposited coatings was observed
for coatings with carbon content of ~65 at.%. DCMS deposited coating
with 65 at.% carbon content exhibited patterned scaly surface with
scales several hundred nanometres in size. HiPIMS deposited coating
with 68 at.% C exhibited a very signiﬁcant reﬁnement of the surface
structure size in parts of the substrate surface, but the surface was not
homogenous. Larger plains and narrower ledges were present on the
sample surface. The size and shape of such plains corresponded well
with the size and shape of tungsten carbide grains in the hardmetal substrate
and the ledges corresponded with the WC grain boundary ﬁlled
by a cobalt based matrix. The thickness of the HiPIMS deposited coating
was measured to be 2.4 ± 0.2 μm. Thus, the coating surface was significantly
affected by the substrate inhomogeneity even for a very thick
coating. The inhomogeneous substrate surface proﬁle was probably
caused by the preceding mechanical polishing and/or preferential
sputtering of the cobalt matrix in the substrate during pre-deposition
plasma cleaning. Utilizing of HiPIMS thus led to surface structure reﬁnement
especially in samples with higher carbon content, however, the
coating surface was affected by the substrate surface inhomogeneity
even for N2 μm thick coating.
Comparison of the XRD diffraction patterns for coatings with similar
carbon content deposited by DC and HiPIMS is plotted in Fig. 6a and b,
respectively. The observed TiC diffractions are labelled with their Miller
indices. Reference positions of the TiC diffractions as per PDF card no.
00-032-1383 are plotted as dashed vertical lines. All the measured
data exhibited only peaks corresponding to TiC, no diffractions corresponding
to metallic hcp Ti were observed in any of the XRD patterns
even for coatings with low carbon content. The XRD results are plotted
in Fig. 7. The mean TiC crystallite size calculated from the (111) diffraction
peak is plotted in Fig. 7a and the lattice parameter in Fig. 7b. The
DCMS and HiPIMS deposited samples exhibited similar trends in the
mean crystallite size. Coatings with ≲55 at.% C exhibited relatively
large TiC grains (10–20 nm). The mean crystallite size was somewhat
higher for HiPIMS deposited coatings containing ≲55 at.% C (~18 nm)
compared to DC deposited samples (~14 nm). The mean TiC crystallite
size decreased with increasing carbon content for the HiPIMS as well
as the DCMS deposited coatings with carbon content ≳ 55 at.% C. A decrease
of crystallite size with increasing carbon content is typical for
nc-TiC/a-C:H coatings prepared by DC magnetron sputtering with carbon
content ≳ 50 at.% C [3,9,13,27,28,29], as high amounts of carbon
act as crystallite boundary sealant preventing further crystallite growth
and instead promote nucleation of higher number of smaller TiC crystallites
[3]. Samuelsson et al. [34] observed sharpening of TiC diffraction
peaks with increasing carbon content and suggested that in contrast
to the DCMS process, the increase of C content in the ﬁlms did not hinder,
but even promoted the TiC grain growth in HiPIMS deposited coatings.
Such an effect was not seen in our work as the trends for DCMS and
HiPIMS depositions both showed a marked decrease of the crystallite
Fig. 5. SEM images of coating surface of coatings with similar composition for DC (left
column) and HiPIMS (right column). Surface feature reﬁnement was observed for
coatings deposited using HiPIMS.
Fig. 6. Comparison of the XRD data for coatings with similar carbon content deposited by
a) DC and b) HiPIMS. Reference positions of the TiC diffractions as per PDF card no. 00-032-
1383 are plotted as dashed vertical lines.
261P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
size with increasing carbon content. The microstrain of the crystallites
was analysed by Williamson-Hall plot and it was found out that the
DCMS deposited as well as HiPIMS deposited nc-TiC/a-C:H coatings exhibited
similar strain of 0.4–0.8%. Thus the grain size calculated by the
Scherrer equation is not systematically shifted between the DCMS and
HiPIMS samples.
The lattice parameter of the TiC crystallites calculated from the X-ray
diffractograms is plotted in Fig. 7b. The value of the lattice parameter of
bulk TiC (4.33 Å) was taken from PDF card no.00-031-1400 [59] and is
plotted in the graph as a horizontal dashed line for comparison. The lattice
parameter was smaller than the bulk value for both the HiPIMS and
DCMS depositions for samples with carbon content ≲ 50 at.% probably
due to carbon vacancies in the lattice. The lattice parameter increased
with increasing carbon content for both DCMS and HiPIMS deposited
coatings as the vacancies in the TiC grains were ﬁlled with carbon
atoms until carbon content of ~55 at.%. When this composition was
reached, the lattice parameter stopped to increase for both DCMS and
HiPIMS deposited samples. DCMS deposited coatings moreover always
exhibited higher lattice parameter than HiPIMS deposited coatings
with similar composition. The lower lattice parameter of the TiC grains
in nc-TiC/a-C:H coatings deposited using HiPIMS can be correlated to a
densiﬁcation of the crystal structure [60] suggesting a stronger ion bombardment
in HiPIMS conditions.
XPS studies were performed to analyse the phase composition of the
coatings. An example of a ﬁtted spectra (DCMS deposited coating with
65 at.% C) is plotted in Fig. 8 and shows the C1s peak ﬁtted by its
C\\C, C\\Ti and C\\Ti* components [22,34]. C\\C bonds corresponded
to peak centred at 285 eV, Ti\\C bonds corresponded to the peak located
~282 eV. The middle peak located at ~283 eV corresponded to the so
called C\\Ti* bonds located on the edge of the TiC grains and the matrix
[61]. Fig. 9a and b shows the comparison of the XPS spectra of coatings
deposited by DCMS and HiPIMS, respectively. DCMS as well as HiPIMS
deposited coatings exhibit increase of the intensity of the C\\C peak as
Fig. 7. XRD results: a) mean crystallite size, b) lattice parameter of HiPIMS and DC
deposited coatings. Lines are added as guides for the eye.
Fig. 8. XPS C1s peak of DC deposited sample with 65 at.% of carbon ﬁtted by its C\\C, C\\Ti
and C\\Ti* components.
Fig. 9. Comparison of XPS C1s peaks of coatings prepared by a) DC b) HiPIMS.
262 P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
the carbon content increases, indicating forming of increasing amounts
of the a-C:H matrix. Furthermore, the HiPIMS deposited coatings exhibit
comparatively less prominent C\\C peak corresponding to DCMS deposited
coatings with similar carbon content, particularly at lower carbon
contents. Fig. 10 shows the plot of the a-C phase content on the
total carbon content DC deposited coatings always contained a signiﬁcant
amount of the a-C:H matrix, even for coatings with ≲50 at.% C.
The amount of the a-C:H phase slowly increased from ~15 to ~30%
with carbon content increase from 29 at.% to 56 at.%. The amount of
the a-C matrix was signiﬁcantly lower for HiPIMS coatings with carbon
content ≲ 60 at.%. The amount of the a-C phase was even nearing zero at
the lowest carbon contents. For coatings with carbon content ≳ 60 at.%,
the amount of a-C matrix signiﬁcantly increased with increasing carbon
content for both the DCMS and HiPIMS deposited coatings. Lower
amount of a-C matrix, although in contrast in the whole studied composition
range up to ~70 at.% of carbon, was also shown by Samuelsson
[62]. The amount of the a-C phase in HiPIMS deposited coatings in the
present study is also close to the amount predicted by thermodynamic
equations for deposition under thermodynamic equilibrium [62].
The thickness of the matrix separating the individual TiC crystallites
(i.e. the mean grain separation) was estimated from the phase ratios
and the grain size using a model assuming cube shaped grains [14,30,
63] and the density of the a-C:H phase of 2.2 g·cm−3
[30]. The resulting
mean grain separation as a function of the total carbon content is plotted
in Fig.11. Even though the grain size in DCMS deposited coatings
was slightly smaller, the DCMS deposited coatings generally exhibited
higher mean grain separation, especially for samples with lower carbon
content. This was due to much higher fraction of the a-C phase in DCMS
deposited coatings. The mean grain separation was increasing only
slightly with increasing carbon content for DCMS deposited coatings
and was in the interval of 0.75–1.1 nm. The mean grain separation
evolved much more for HiPIMS deposited coatings. The mean grain separation
was lower than 0.25 nm for sample with 40 at.% C and rose up to
1 nm for sample with 85 at.% C with a much steeper slope than was observed
for DCMS deposited samples.
The stoichiometry factor x in TiCx was estimated from the total carbon
and titanium content evaluated by EDX and the TiC and the a-C:H
phase ratio acquired by XPS measurements according to equation.
x ¼ C ðat:%Þ
Ti ðat:%Þ Á AC−Ti
AC−C þAC−Ti
, where C (at.%), Ti (at.%) denote the total concentration
of carbon and titanium, respectively, AC − Ti denotes the
area of the C\\Ti and C\\Ti* peaks ﬁtted in the C1s XPS spectrum and
AC − C denotes the area of the C\\C peak in the C1s XPS spectrum. The
stoichiometry factor as a function of the total carbon content is plotted
in Fig. 12. The stoichiometry factor increased with increasing total
carbon content for both the DCMS and HiPIMS cases for coatings with
≲60 at.% of carbon. When the carbon content was lower than the titanium
content there was not enough carbon atoms present in the coatings
for formation of stoichiometric TiC crystals. Moreover, the amorphous
carbon matrix was formed even when the coating contained low
amount of carbon (see Fig. 10). Thus, highly sub-stoichiometric TiCx
crystals are formed in coatings with low carbon content. The stoichiometry
improved with increasing carbon content as more carbon is available
to ﬁll the vacancies in the TiC lattice. Lower amount of
amorphous carbon matrix was formed when HiPIMS was utilized for
coatings with ≲60 at.% carbon. This resulted in HiPIMS deposited TiCx
crystallites stoichiometry generally exhibiting closer to unity. The TiCx
crystallites reached the highest stoichiometry factor at around 55–
60 at.% C for both DCMS and HiPIMS cases. If the total carbon content
further increased, the stoichiometry factor of the TiCx grains decreased.
This indicates that for coatings with high amount of carbon, the carbon
rather formed the a-C:H matrix phase than the TiCx phase.
The Raman spectra of the deposited coatings are plotted in Fig. 13.
The sp3
/sp2
hybridization of carbon in the a-C:H matrix was qualitatively
estimated from the position of the G peak and intensity ratio of I(D)/I(G)
as proposed for a-C ﬁlms by Ferrari [64] and extended to a-C:H ﬁlms by
Casiraghi et al. [65]. The position of the G peak for all coatings regardless
of their carbon content or method of plasma excitation was in the
interval of 1555–1575 cm−1
and the I(D)/I(G) ratio was within the
Fig. 10. The a-C phase fraction in HiPIMS and DC deposited coatings. Lines are added as
guides for the eye.
Fig. 11. Mean grain separation in HiPIMS and DC deposited coatings. Lines are added as
guides for the eye.
Fig. 12. Stoichiometry of TiC crystallites in HiPIMS and DC deposited coatings. Lines are
added as guides for the eye.
263P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
interval of 0.6–0.8. Thus, the sp3
carbon hybridization was estimated
to always be lower than ~10%.
The hardness and the effective elastic modulus of the coatings are
plotted in Fig. 14a and b, respectively. The trends in hardness were similar
for coatings prepared by both DCMS and HiPIMS. The hardness was
low for samples with very low carbon content (b15 GPa). As the carbon
content increased, the hardness increased correspondingly. This continued
up to ~55 at.% carbon, where the hardness was the highest. The
highest hardness of 33 ± 4 GPa was achieved in the case of DCMS deposited
coatings, whereas the hardness of 45 ± 4 GPa was achieved in
the case of HiPIMS deposited coatings. If the amount of carbon was increased
even more above 55 at.%, the hardness again signiﬁcantly decreased.
The hardness of coatings with high amount of carbon was
15–20 GPa which is in good accordance with hardness of magnetron
sputtered DLC or Ti-DLC coatings [66,67,68]. Although the trend is the
same for DCMS and HiPIMS deposited coatings, the hardness of the
HiPIMS coatings was shifted to higher values compared to coatings prepared
by DCMS in all cases. The elastic modulus exhibited similar trends
as the hardness – that is low the relatively values at the lowest carbon
content followed by an increase up to ~425 GPa with increasing carbon
content continued by a decrease with further carbon increase in carbon
content. The highest elastic modulus was obtained for coatings with
slightly sub-stoichiometric chemical composition (~48 at.% C). The elastic
modulus was generally higher for coatings prepared by HiPIMS, albeit
the difference was less pronounced than in the hardness. It has been
proposed that the toughness of the material depends on the hardness
to elastic modulus ratio H/E, and that the prediction of the wear of the
coatings can be based on the resilience of the coating which is deﬁned
as the resistance of the material against plastic deformation and relates
to the ratio of H3
/E2
[11,18,69]. Because of signiﬁcantly higher hardness
and only slightly higher elastic modulus, the HiPIMS deposited coatings
exhibited also higher values of H/E and H3
/E2
ratios. The hardest HiPIMS
coating with 57 at.% C exhibited H/E ratio of 0.14 and H3
/E2
ratio of
0.94 GPa, while the hardest DCMS deposited coating with 53 at.% C
exhibited lower values of 0.10 and 0.35 GPa, respectively. The
highest values of the H/E ratio for nc-TiC/a-C(:H) reported in the literature
are usually in the interval of 0.1–0.12 [1,10,26,70]. HiPIMS
deposited coatings presented in this work thus exhibited high
hardness N 40 GPa, H/E ratio was 30% higher than the highest literature
value for coatings with similar composition [71]. The H3
/E2
ratio
was even trebled compared to the highest reported value [10]. Thus
superior toughness and resistance to plastic deformation of the
HiPIMS deposited nc-TiC/a-C:H coatings are strongly implied.
4. Discussion
The hardness of a nanocomposite coating is given by the properties
of the phases forming the nanocomposite and by the properties of the
phase interface. In the case of nc-TiC/a-C:H coatings it is the coating
structure, the crystallite stoichiometry and size, the mean grain separation,
matrix properties and anchoring of the grains in the matrix. The
mechanical properties of a nanocomposite can be enhanced due the
so-called grain boundary strengthening, i.e. effect similar to the socalled
Hall-Petch effect described in polycrystalline materials [72,73].
Fig. 13. Comparison of Raman spectra of coatings with similar carbon content prepared by
a) DC and b) HiPIMS.
Fig. 14. a) Hardness and b) elastic modulus of HiPIMS and DC deposited coatings. Lines are
added as guides for the eye.
264 P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
For this effect to take place, the grain size should be in the size range of
5–10 nm. Furthermore, the grains should be separated by only a single
or at most a few monolayers of the matrix. It was shown that a single
monolayer of the matrix was optimal for reaching the highest hardness
of nanocomposite nc-TiN/a-SiNx system [74]. The sp3
/sp2
bonding in the
matrix might also play a role as carbon based coatings with higher sp3
content generally exhibit higher hardness [75,76]. The macrostructure
of the coatings inﬂuences the mechanical properties of the coatings as
well, as it has been shown that coatings with glassy structure exhibit
higher hardness and elastic modulus than coatings with columnar
structure [6,71].
The properties of the hardest coatings deposited by DCMS and
HiPIMS are compared in Table 2. Two samples in the DCMS experiments
are listed as both exhibited the same hardness within the margin of
error. One of the DCMS prepared samples contained 53 at.% C and the
other 61 at.% C. The surface topography of the coating with carbon content
of 53 at.% and hardness of 33.3 GPa was coarse, indicating columnar
macrostructure with column width of 600–700 nm. The lattice parameter
was 0.7% larger than the equilibrium value, indicating less dense
structure of the grains. The TiCx grains were ~14 nm large, the stoichiometric
factor was 0.9 and the grains were separated by ~1 nm of the matrix
phase. Considering that the average C\\C bond length in
amorphous carbon is in the range of 0.143–0.152 nm [75], the mean
grain separation corresponded to ~5–6 bond lengths, approximately
corresponding to the same number monolayers of the matrix. The
grain strain was similar in DCMS and HiPIMS deposited coatings. The
content of the sp3
C\\C bonding in the a-C:H matrix was approximately
5–10%. The mean grain separation, the macrostructure and the sp3
/sp2
bond ratio in the matrix were similar in both hardest samples. However,
the grain size in the sample containing 61 at.% C decreased to 8 nm and
the lattice parameter was only 0.2% larger than the equilibrium value.
Also the stoichiometry factor was increased to 0.95. Despite this, the
hardness was not signiﬁcantly affected. Thus the mean grain separation,
the macrostructure and low sp3
bond content in the matrix seem to be
the limiting factors for achieving higher hardness in DC driven process.
Decreasing the amount of carbon in the DCMS deposited coatings
below the optimum interval, i.e. below 53 at.% C, led to no signiﬁcant
change of the surface topography and the grain size also stayed approximately
the same. The sp3
/sp2
ratio of the C\\C bonds in the matrix was
not signiﬁcantly changed. The mean grain separation decreased to ~4–5
monolayers. However, the stoichiometry factor of the TiCx grains decidedly
decreased. The hardness decreased as the effect of smaller mean
grain separation was overshadowed by the loss of TiCx grain stoichiometry.
On the other hand, carbon content increase above 61 at.% led to a
fast drop of the TiCx grain size below the value optimal for grain boundary
strengthening. The stoichiometry factor of the grains decreased and
the mean grain separation notably increased. Consequently, the hardness
of the coatings decreased due to adverse effects of bad stoichiometry,
possible inverse Hall-Petch effect [77] and grain sliding [78,79].
Thus in order to reach higher hardness, the macrostructure should
be reﬁned, the crystals should be made denser and should be separated
by lower number of a-C:H matrix layers. The stoichiometry of the TiCx
grains should be maintained or increased. Using DC driven plasma deposition
process it is not possible to optimize the structure of the ncTiC/a-C:H
coatings for achieving hardness higher than 33 GPa.
The hardest coating prepared by HiPIMS contained 57 at.% of C
and thus exhibited similar carbon content as the hardest samples
prepared by DCMS. The TiCx grain size is comparable to that of the
samples prepared by DCMS and the sp3
bond ratio also did not significantly
change. However, the stoichiometry factor was slightly increased
to 0.99, the lattice parameter was 0.7% smaller than the
equilibrium bulk value, indicating denser structure of the crystals,
the mean grain separation decreased to 3–4 monolayers of the aC:H
matrix and the surface topography was decidedly reﬁned indicating
a denser macrostructure. This structure reﬁnement led to
the hardness of 44.7 GPa which is an increase of 33%. As the grain
size was not reﬁned compared to DCMS case and the grain stoichiometry
was improved only very slightly, the crystallite density, the
mean grain separation and the macrostructure were the most important
parameters responsible for the hardness enhancement.
Decreasing the carbon content below 53 at.% as well as increasing it
above 57 at.% had similar effects on the coating structure as in DCMS
case. However, the HiPIMS deposited coatings always exhibited grain
stoichiometry closer to unity, lower mean grain separation, ﬁner macrostructure
and lower lattice parameter than their DCMS deposited counterparts.
This resulted in overall higher hardness of the HiPIMS
deposited nc-TiC/a-C:H coatings.
5. Conclusion
Nanocomposite nc-TiC/a-C:H coatings were deposited with the
same mean power utilizing DC magnetron sputtering and HiPIMS in
argon/acetylene atmosphere. The deposition rate was lower in HiPIMS,
however it increased at higher acetylene ﬂows where DCMS process
was unstable. Signiﬁcantly lower acetylene ﬂow was needed to produce
coatings with similar total carbon content in HiPIMS compared to
DCMS. It was also shown that HiPIMS employment was highly advantageous
for controlled and repeatable deposition of nc-TiC/a-C:H coatings
with high carbon content. Even though HiPIMS discharge with a relatively
long pulse length is generally prone to arc occurrence due to impurities
or debris on the cathode, no such behaviour was observed.
Whereas the total carbon content in DCMS deposited samples was limited
to ≲70 at.% due to poisoning processes on the titanium target, coatings
with carbon contents of ≳90 at.% were possible to be deposited by
HiPIMS.
HiPIMS deposited coatings exhibited similar sized grains as DCMS
but with exhibited lower lattice parameter and a ﬁner surface structure.
Utilization of HiPIMS also favoured carbon incorporation into
the TiCx grains as HiPIMS deposited coatings exhibited stoichiometry
factor of the TiCx grains closer to unity and lower fraction of the a-C
matrix. This corresponded to a smaller mean grain separation of
the TiCx grains by the a-C matrix in HiPIMS deposited coatings. This
was attributed to higher ion bombardment and thus higher energy
inﬂux to the growing ﬁlm inherent to the HiPIMS process. TiCx
grain stoichiometry factor closer to density and higher grain density
combined with lower mean grain separation and macrostructure reﬁnement
led to overall higher hardness of HiPIMS deposited coatings.
The mean grain separation, TiCx grain density and the
macrostructure were found to play the key role in hardness enhancement
of the HiPIMS deposited nc-TiC/a-C:H coatings. H/E and H3
/E2
ratios were substantially higher for HiPIMS deposited coatings
indicating enhanced toughness and tribological properties. Thus
HiPIMS can be successfully used to inﬂuence the growth conditions
of nc-TiC/a-C:H coatings in order to alter the microstructure in a
way favourable for achieving enhanced mechanical properties.
Table 2
Comparison of the hardest coatings deposited by DC and HiPIMS.
H [GPa] C [at.%] Crystallite size [nm] Lattice parameter [Å] Stoichiometry Mean grain separation [nm] Macrostructure
DC 33 ± 4 53 14 ± 2 4.363 ± 0.001 0.90 1 Coarse
32.9 ± 3 61 8 ± 1 4.337 ± 0.001 0.95 1 Coarse
HiPIMS 45 ± 4 57 9 ± 1 4.313 ± 0.001 0.99 0.5 Fine
265P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
Acknowledgment
This research has been supported by project LO1411 (NPU I) funded
by Ministry of Education, Youth and Sports of Czech Republic.
References
[1] A. Czyzniewski, W. Precht, Deposition and some properties of nanocrystalline, nanocomposite
and amorphous carbon-based coatings for tribological applications, J.
Mater. Process. Technol. 157–158 (2004) 274–283.
[2] Y. Wang, X. Zhang, X. Wu, H. Zhang, X. Zhang, Compositional, structural and mechanical
characteristics of nc-TiC/a-C:H nanocomposite ﬁlms, Appl. Surf. Sci. 255
(2008) 1801–1805.
[3] T. Zehnder, J. Patscheider, Nanocomposite TiC/a-C:H hard coatings deposited by reactive
PVD, Surf. Coat. Technol. 133–134 (2000) 138–144.
[4] P. Souček, T. Schmidtová, L. Zábranský, V. Buršíková, P. Vašina, O. Caha, J. Buršík, V.
Peřina, R. Mikšová, Y.T. Pei, J.T.M. De Hosson, On the control of deposition process
for enhanced mechanical properties of nc-TiC/a-C:H coatings with DC magnetron
sputtering at low or high ion ﬂux, Surf. Coat. Technol. 255 (2014) 8–14.
[5] J. Musil, P. Novák, R. Čerstvý, Z. Soukup, Tribological and mechanical properties of
nanocrystalline-TiC/a-C nanocomposite thin ﬁlms, J. Vac. Sci. Technol. A 28 (2010)
244–249.
[6] Y.T. Pei, D. Galvan, J.T.M. De Hosson, C. Strondl, Advanced TiC/a-C:H nanocomposite
coatings deposited by magnetron sputtering, J. Eur. Ceram. Soc. 26 (2006) 565–570.
[7] P. Souček, T. Schmidtová, V. Buršíková, P. Vašina, Y.T. Pei, J.T.M. De Hosson, O. Caha,
V. Peřina, R. Mikšová, P. Malinský, Tribological properties of nc-TiC/a-C:H coatings
prepared by magnetron sputtering at low and high ion bombardment of the growing
ﬁlm, Surf. Coat. Technol. 241 (2014) 64–73.
[8] J.C. Sánchez-López, D. Martínez-Martínez, C. López-Cartes, A. Fernández, Tribological
behaviour of titanium carbide/amorphous carbon nanocomposite coatings: from
macro to the micro-scale, Surf. Coat. Technol. 202 (2008) 4011–4018.
[9] D. Martínez-Martínez, C. López-Cartes, A. Fernández, J.C. Sánchez-López, Inﬂuence
of the microstructure on the mechanical and tribological behavior of TiC/a-C nanocomposite
coatings, Thin Solid Films 517 (2009) 1662–1671.
[10] C.Q. Chen, Y.T. Pei, K.P. Shaha, J.T.M. De Hosson, Nanoscale deformation mechanism
of TiC/a-C nanocomposite thin ﬁlms, J. Appl. Phys. 105 (2009) 114314.
[11] J. Lin, J.J. Moore, B. Mishra, M. Pinkas, W.D. Sproul, Syntheses and characterization of
TiC/a:C composite coatings using pulsed closed ﬁeld unbalanced magnetron
sputtering (P-CFUBMS), Thin Solid Films 517 (2008) 1131–1135.
[12] Y.T. Pei, C.Q. Chen, K.P. Shaha, J.T.M. De Hosson, J.W. Bradley, S.A. Voronin, M. Čada,
Microstructural control of TiC/a-C nanocomposite coatings with pulsed magnetron
sputtering, Acta Mater. 52 (2008) 696–709.
[13] S. Zhang, X.L. Bui, J. Jiang, X. Li, Microstructure and tribological properties of magnetron
sputtered nc-TiC/a-C nanocomposite, Surf. Coat. Technol. 198 (2005) 206–211.
[14] E. Lewin, O. Wilhelmsson, U. Jansson, Nanocomposite nc-TiC∕aC thin ﬁlms for electrical
contact applications, J. Appl. Phys. 100 (2006), 054303.
[15] U. Wiklund, M. Nordin, O. Wänstrand, M. Larsson, Evaluation of a ﬂexible physical
vapor deposited TiC–C coating system, Surf. Coat. Technol. 124 (2000) 154–161.
[16] A.A. Voevodin, J.S. Zabinski, Load-adaptive crystalline–amorphous nanocomposites,
J. Mater. Sci. 33 (1998) 319–327.
[17] A.A. Voevodin, S.V. Prasad, J.S. Zabinski, Nanocrystalline carbide/amorphous carbon
composites, J. Appl. Phys. 82 (1997) 855–858.
[18] Y.T. Pei, D. Galvan, J.T.M. De Hosson, A. Cavaleiro, Nanostructured TiC/a-C coatings
for low friction and wear resistant applications, Surf. Coat. Technol. 198 (2005)
44–50.
[19] M. Stüber, H. Leiste, S. Ulrich, H. Holleck, D. Schild, Microstructure and properties of
low friction TiC–C nanocomposite coatings deposited by magnetron sputtering, Surf.
Coat. Technol. 150 (2002) 218–226.
[20] A. Mani, P. Aubert, F. Mercier, H. Khodja, C. Berthier, P. Houdy, Effects of residual
stress on the mechanical and structural properties of TiC thin ﬁlms grown by RF
sputtering, Surf. Coat. Technol. 194 (2005) 190–195.
[21] W.J. Meng, R.C. Tittsworth, L.E. Rehn, Mechanical properties and microstructure of
TiC/amorphous hydrocarbon nanocomposite coatings, Thin Solid Films 377–378
(2000) 222–232.
[22] A.A. El Mel, B. Angleraud, E. Gautron, A. Granier, P.Y. Tessier, Microstructure and
composition of TiC/a-C:H nanocomposite thin ﬁlms deposited by a hybrid IPVD/
PECVD process, Surf. Coat. Technol. 204 (2010) 1880–1883.
[23] V. Kulikovsky, A. Tarasenko, F. Fendrych, L. Jastrabik, D. Chvostova, F. Franc, L.
Soukup, The mechanical, tribological and optical properties of Ti-C:H coatings, prepared
by DC magnetron sputtering, Diam. Relat. Mater. 7 (1998) 774–778.
[24] V. Kulikovsky, P. Bohac, F. Franc, D. Chvostova, A. Deineka, V. Vorlicek, L. Jastrabik,
Degradation and stress evolution in a-C, a-C:H and Ti-C:H ﬁlms, Surf. Coat. Technol.
142-144 (2001) 702–706.
[25] D. Galvan, Y.T. Pei, J.T.M. De Hosson, Reactive magnetron sputtering deposition and
columnar growth of nc-TiC/a-C:H nanocomposite coatings, J. Vac. Sci. Technol. A 24
(2006) 1441–1447.
[26] Y.T. Pei, D. Galvan, J.T.M. De Hosson, Nanostructure and properties of TiC/a-C:H composite
coatings, Acta Mater. 53 (2005) 4505–4521.
[27] W. Gulbiński, S. Mathurb, H. Shen, T. Suszko, A. Gilewicz, B. Warcholiński, Evaluation
of phase, composition, microstructure and properties in TiC/a-C:H thin ﬁlms deposited
by magnetron sputtering, Appl. Surf. Sci. 239 (2005) 302–310.
[28] D. Galvan, Y.T. Pei, J.T.M. De Hosson, Inﬂuence of deposition parameters on the
structure and mechanical properties of nanocomposite coatings, Surf. Coat. Technol.
201 (2006) 590–598.
[29] Y. Hu, L. Li, X. Cai, Q. Chen, P.K. Chu, Mechanical and tribological properties of
TiC/amorphous hydrogenated carbon composite coatings fabricated by DC
magnetron sputtering with and without sample bias, Diam. Relat. Mater. 16
(2007) 181–186.
[30] T. Zehnder, P. Schwaller, F. Munnik, S. Mikhailov, J. Patscheider, Nanostructural
and mechanical properties of nanocomposite nc-TiC/a-C:H ﬁlms deposited by
reactive unbalanced magnetron sputtering, J. Appl. Phys. 95 (8) (2004)
4327–4334.
[31] P. Souček, T. Schmidtová, L. Zábranský, V. Buršíková, P. Vašina, O. Caha, M. Jílek, A.A.
El Mel, P.Y. Tessier, J. Schäfer, J. Buršík, V. Peřina, R. Mikšová, Evaluation of composition,
mechanical properties and structure of nc-TiC/a-C:H coatings prepared by balanced
magnetron sputtering, Surf. Coat. Technol. 211 (2012) 111–116.
[32] F. Amato, N.R. Panyala, P. Vašina, P. Souček, J. Havel, Laser desorption ionisation
quadrupole ion trap time-of-ﬂight mass spectrometry of titanium-carbon thin
ﬁlms, Rapid Commun. Mass Spectrom. 27 (2013) 1196–1202.
[33] Y.T. Pei, K.P. Shaha, C.Q. Chen, R. van der Hulst, A.A. Turkin, D.I. Vainsthein, J.T.M. De
Hosson, Growth of nanocomposite ﬁlms: from dynamic roughening to dynamic
smoothening, Acta Mater. 57 (2009) 5156–5164.
[34] M. Samuelsson, K. Sarakinos, H. Högberg, E. Lewin, U. Jansson, B. Wälivaara, H.
Ljungcrantz, U. Helmersson, Growth of Ti-C nanocomposite ﬁlms by reactive high
power impulse magnetron sputtering under industrial conditions, Surf. Coat.
Technol. 206 (2012) 2396–2402.
[35] M. Balzer, M. Fenker, Three-dimensional thickness and property distribution of TiC
ﬁlms deposited by DC magnetron sputtering and HIPIMS, Surf. Coat. Technol. 250
(2014) 37–43.
[36] V. Kouznetsov, K. Macak, J.M. Schneider, U. Helmersson, I. Petrov, A novel pulsed
magnetron sputter technique utilizing very high target power densities, Surf. Coat.
Technol. 122 (1999) 290–293.
[37] U. Helmersson, M. Lattemann, J. Bohlmark, A.P. Ehiasarian, J.T. Gudmundsson, Ionized
physical vapor deposition (IPVD): a review of technology and applications,
Thin Solid Films 513 (2006) 1–24.
[38] K. Sarakinos, J. Alami, S. Konstantinidis, High power pulsed magnetron sputtering: a
review on scientiﬁc and engineering state of the art, Surf. Coat. Technol. 204 (2010)
1661–1684.
[39] A. Anders, Discharge physics of high power impulse magnetron sputtering, Surf.
Coat. Technol. 205 (2011) S1–S9.
[40] D.V. Mozgrin, I.K. Fetisov, G.V. Khodachenko, High-current low-pressure quasi-stationary
discharge in a magnetic ﬁeld: experimental research, Plasma Phys. Rep. 21
(5) (1995) 400–409.
[41] K. Macák, V. Kouznetsov, J. Schneider, U. Helmersson, I. Petrov, Ionized sputter deposition
using an extremely high plasma density pulsed magnetron discharge, J.
Vac. Sci. Technol. A 18 (2000) 1533–1537.
[42] J. Bohlmark, M. Lattemann, J. Gudmundsson, A. Ehiasarian, Y. Arandagonzalvo, N.
Brenning, U. Helmersson, The ion energy distributions and ion ﬂux composition
from a high power impulse magnetron sputtering discharge, Thin Solid Films 515
(2006) 1522–1526.
[43] J. Benedikt, R.V. Woen, S.L.M. van Mensfoort, V. Perina, J. Hong, M.C.M. van de
Sanden, Plasma chemistry during the deposition of a-C:H ﬁlms and its inﬂuence
on ﬁlm properties, Diam. Relat. Mater. 12 (2003) 90–97.
[44] J. Musil, S. Kadlec, J. Vyskočil, V. Poulek, Reactive deposition of hard coatings, Surf.
Coat. Technol. 39–40 (1989) 301–314.
[45] J. Patscheider, T. Zehnder, M. Diserens, Structure-performance relations in nanocomposite
coatings, Surf. Coat. Technol. 146–147 (2001) 201–208.
[46] F. Piazza, A. Golanski, S. Schulze, G. Relihan, Transpolyacetylene chains in hydrogenated
amorphous carbon ﬁlms free of nanocrystalline diamond, Appl. Phys. Lett. 3
(2003) 358–360.
[47] A.C. Ferrari, J. Robertson, Resonant Raman spectroscopy of disordered, amorphous,
and diamondlike carbon, Phys. Rev. B 64 (2001), 075414.
[48] W.C. Oliver, G.M. Pharr, Measurement of hardness and elastic modulus by instrumented
indentation: advances in understanding and reﬁnements to methodology,
J. Mater. Res. 19 (2004) 3–20.
[49] P. Vašina, P. Souček, T. Schmidtová, M. Eliáš, V. Buršíková, M. Jílek, M. Jílek Jr., J.
Schäfer, J. Buršík, Depth proﬁle analyses of nc-TiC/a-C:H coating prepared by balanced
magnetron sputtering, Surf. Coat. Technol. 205 (2011) S53–S56.
[50] D.J. Christie, Target material pathways model for high power pulsed magnetron
sputtering, J. Vac. Sci. Technol. A 23 (2005) 330–335.
[51] A. Anders, Deposition rates of high power impulse magnetron sputtering: physics
and economics, J. Vac. Sci. Technol. A 28 (2010) 783–790.
[52] J. Čapek, M. Hála, O. Zabeida, J.E. Klemberg-Sapieha, L. Martinů, Deposition rate enhancement
in HiPIMS without compromising the ionized fraction of the deposition
ﬂux, J. Phys. D. Appl. Phys. 46 (2013), 205205. (10 pp.).
[53] S. Berg, H.-O. Blom, T. Larsson, C. Nender, Modeling of reactive sputtering of compound
materials, J. Vac. Sci. Technol. A 5 (1987) 202–207.
[54] S. Berg, T. Nyberg, Fundamental understanding and modeling of reactive sputtering
processes, Thin Solid Films 476 (2005) 215–230.
[55] D. Depla, G. Buyle, J. Haemer, R. DeGryse, Discharge voltage measurements during
magnetron sputtering, Surf. Coat. Technol. 200 (2006) 4329–4338.
[56] T. Schmidtová, P. Souček, V. Kudrle, P. Vašina, Non-monotonous evolution of hybrid
PVD–PECVD process characteristics on hydrocarbon supply, Surf. Coat. Technol. 232
(2013) 283–289.
[57] G. Fedosenko, A. Schwabedissen, J. Engemann, E. Braca, L. Valentini, J.M. Kenny,
Pulsed PECVD deposition of diamond-like carbon ﬁlms, Diam. Relat. Mater. 11
(2002) 1047–1052.
[58] N. Ravi, V.L. Bukhovets, I.G. Varshavskaya, G. Sundararajan, Deposition of diamondlike
carbon ﬁlms on aluminium substrates by RF-PECVD technique: inﬂuence of process
parameters, Diam. Relat. Mater. 16 (2007) 90–97.
266 P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
[59] M.C. Morris, H.F. McMurdie, E.H. Evans, B. Paretzkin, H.S. Parker, N.C.
Panagiotopoulos, C.R. Hubbard, Standard X-ray Diffraction Powder Patterns: Section
18 — Data for 58 Substances, NBS, Washington, 1981.
[60] D. Binder, W.J. Sturm, Equivalence of X-ray lattice parameter and density changes in
neutron-irradiated LiF, Phys. Rev. 96 (1954) 1519–1522.
[61] E. Lewin, P.O.Å. Persson, M. Lattemann, M. Stüber, M. Gorgoi, A. Sandell, C. Ziebert, F.
Schäfers, W. Braun, J. Hallbirter, S. Ulrich, W. Eberhardt, L. Hultman, H. Siegbahn, S.
Svensson, U. Jansson, On the origin of a third spectral component of C1s XPS-spectra
for nc-TiC/a-C nanocomposite thin ﬁlms, Surf. Coat. Technol. 202 (2008)
3563–3570.
[62] M. Samuelsson, Fundamental Aspects of HiPIMS Under Industrial
Conditions(Dissertation No. 1461) Linköping University, Sweden, 2012.
[63] U. Jansson, E. Lewin, Sputter deposition of transition-metal carbide ﬁlms — a critical
review from a chemical perspective, Thin Solid Films 536 (2013) 1–24.
[64] A.C. Ferrari, Determination of bonding in diamond-like carbon by Raman spectroscopy,
Diam. Relat. Mater. 11 (2002) 1053–1061.
[65] C. Casiraghi, A.C. Ferrari, J. Robertson, Raman spectroscopy of hydrogenated amorphous
carbons, Phys. Rev. B 72 (2005), 085401.
[66] X.L. Bui, Y.T. Pei, J.T.M. De Hosson, Magnetron reactively sputtered Ti-DLC coatings
on HNBR rubber: the inﬂuence of substrate bias, Surf. Coat. Technol. 202 (2008)
4939–4944.
[67] K. Yamamoto, K. Matsukado, Effect of hydrogenated DLC coating hardness on the
tribological properties under water lubrication, Tribol. Int. 39 (2006) 1609–1614.
[68] S. Zhang, X.L. Bui, Y. Fu, Magnetron sputtered hard a-C coatings of very high toughness,
Surf. Coat. Technol. 167 (2003) 137–142.
[69] A. Leyland, A. Matthews, On the signiﬁcance of the H/E ratio in wear control: a nanocomposite
coating approach to optimised tribological behaviour, Wear 246 (2000)
1–11.
[70] D.M. Cao, B. Feng, W.J. Meng, L.E. Rehn, P.M. Baldo, M.M. Khonsari, Friction and wear
characteristics of ceramic nanocomposite coatings: titanium carbide/amorphous
hydrocarbon, Appl. Phys. Lett. 3 (2001) 329–331.
[71] K.P. Shaha, Y.T. Pei, D. Martínez-MArtínez, J.C. Sánchez-López, J.T.M. De Hosson, Effect
of process parameters on mechanical and tribological performance of pulsedDC
sputtered TiC/a-C:H nanocomposite ﬁlms, Surf. Coat. Technol. 205 (2010)
2633–2642.
[72] E.O. Hall, The deformation and aging of mild steel: III discussion of results, Proc.
Phys. Soc. London B 64 (1951) 747–753.
[73] N.J. Petch, The cleavage strength of polycrystals, J. Iron Steel Inst 174 (1953) 25–28.
[74] S. Veprek, R.F. Zhang, M.G.J. Veprek-Heijman, S.H. Sheng, A.S. Argon, Superhard
nanocomposites: origin of hardness enhancement, properties and applications,
Surf. Coat. Technol. 204 (2010) 1898–1906.
[75] J. Robertson, Diamond-like amorphous carbon, Mater. Sci. Eng. R 37 (2002) (2002)
129–137.
[76] P. Písařík, M. Jelínek, T. Kocourek, J. Remsa, J. Zemek, J. Lukeš, J. Šepitka, Inﬂuence of
diamond and graphite bonds on mechanical properties of DLC thin ﬁlms, J. Phys.
Conf. Ser. 594 (2015), 012008. (6 pp.).
[77] C.E. Carlton, P.J. Ferreira, What is behind the inverse Hall–Petch effect in nanocrystalline
materials? Acta Mater. 55 (2007) 3749–3756.
[78] J. Schiotz, F.D. Di Tolla, K.W. Jacobsen, Softening of nanocrystalline metals at very
small grain sizes, Nature 391 (1998) 561–563.
[79] Z.W. Shan, E.A. Stach, J.M.K. Wiezorek, J.A. Knapp, D.M. Follstaedt, S.X. Mao, Grain
boundary-mediated plasticity in nanocrystalline nickel, Science 305 (2004)
654–657.
267P. Souček et al. / Surface & Coatings Technology 311 (2017) 257–267
Research Article
Investigation of the Influence of Ni Doping on the Structure and
Hardness of Ti-Ni-C Coatings
J. Daniel, P. SouIek, K. Bernátová, L. Zábranský, M. Stupavská, V. Buršíková, and P. Vašina
Department of Physical Electronics, Faculty of Science, Masaryk University, Kotlarska 2, 611 37 Brno, Czech Republic
Correspondence should be addressed to P. Vaˇsina; vasina@physics.muni.cz
Received 31 January 2017; Revised 16 April 2017; Accepted 30 May 2017; Published 6 July 2017
Academic Editor: Albano Cavaleiro
Copyright © 2017 J. Daniel et al. This is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Nanocomposite nc-TiC/a-C:H thin films exhibit unique combination of mechanical properties, high hardness, low friction, and
wear. Selective doping by weak-carbide forming element can be used in order to specifically design the physical and chemical
properties of nc-TiC/a-C:H coatings. In this paper we report on an effect of nickel addition on structure and hardness of the ncTiC/a-C:H
coatings. The effect of Ni alloying on the coating structure under conditions of DCMS and HiPIMS depositions was
studied. The coating structure was correlated with the coating hardness. The grain size, the grain carbon vacancy concentration,
and the mean grain separation were found to be the key parameters determining the coating hardness. Ni doping proved to have a
significant effect on the coating microstructure which resulted in changes of the hardness of the deposited coatings.
1. Introduction
Nanocomposite transition metal carbide based coating systems
have been systematically studied because of advantageous
combination of their physical, chemical, and mechanical
properties such as high hardness, high wear resistance,
low friction coefficient, and good thermal stability. This
nanocomposite structure consists of hard nanocrystalline
metal carbide grains surrounded by amorphous carbon
phase. One of the most widely studied transition metal
carbide nanocomposite systems, denoted as nc-TiC/a-C:H,
consists of nanocrystalline TiC grains embedded in an amorphous
hydrogenated carbon matrix [1]. The nc-TiC/a-C:H
coatings with high hardness [2–15] or with low coefficient of
friction and wear [16–19] were described in the literature. The
synthesis of hydrogenated nc-TiC/a-C:H nanocomposites is
generally performed utilizing a hybrid PVD-PECVD process
of sputtering of titanium target in hydrocarbon containing
plasma. This process was described in detail in [20].
To further enhance the mechanical properties of titanium
and carbon based coatings, a strategy utilizing doping of the
coatings by a non-carbine forming element such as nickel was
proposed [21]. If nickel is introduced in specific quantities
under specific deposition conditions, nickel can be incorporated
into the Ti-C grains and therefore metastable Ti-NiC
grains embedded in carbon matrix can be formed [22].
It was calculated that Ti-Ni-C/a-C system with ≳25% of Ni
can transit to a more thermodynamically favorable state:
the metastable grains will release carbon to the amorphous
phase. This effect can result in changes of the coatings
structure and thus in their mechanical properties [22].
Jansson et al. prepared nanocomposite Ti-Ni-C coatings with
constant [Ni]/[Ti] ratio and different carbon contents by
DC cosputtering of Ti, C, and Ni targets and compared
them with nickel-free nc-TiC/a-C coatings prepared using
the same deposition techniques [22]. Addition of ∼25 at. %
of Ni led to the grain size decrease and increase of the
amount of the amorphous carbon phase. Also, a moderate
hardening of the coatings was observed. This effect was
attributed to the nanocomposite hardening effect described
by Zehnder et al. [3, 12]. Furthermore, if the Ti-Ni-C system
contains <25 at. % of Ni, application of external forces, for
example, via indentation of friction, can lead to overcoming
the thermodynamical barrier and releasing of carbon from
the grain to the matrix. Therefore, nc-Ti-Ni-C/a-C:H system
could be used as a smart lubricating coating.
Hindawi
Journal of Nanomaterials
Volume 2017,Article ID 6368927, 13 pages
https://doi.org/10.1155/2017/6368927
2 Journal of Nanomaterials
In this work hybrid PVD-PECVD process consisting
of sputtering of a metallic target in reactive hydrocarbon
atmosphere [4, 23–26] is used to synthesize nanocomposite
hydrogenated nc-Ti-Ni-C/a-C:H coatings. The hybrid process
instead of cosputtering was used to further expand
the concept proposed for hydrogen free coatings [22]. Also,
the hybrid process has a better future applicability as it is
already industrially used [23] exhibiting higher deposition
rates. Significantly lower amounts of Ni (less than ∼10 at. %)
than in the case of Jansson et al. [22] were used to purposely
synthesize coatings with metastable Ti-Ni-C grains. To study
the influence of the plasma excitation on the possibilities of
synthesis of the metastable grains, Direct Current Magnetron
Sputtering (DCMS) and High Power Impulse Magnetron
Sputtering (HiPIMS) processes were used. HiPIMS process
is characterized by enhanced adatom diffusion, enhanced
ion and energy fluxes, and higher thin film forming ion
concentration compared to DCMS [26–28]. The influence of
Ni doping on the chemical composition, phase composition,
microstructure, and mechanical properties of nc-Ti-Ni-C/aC:H
coatings prepared by both DCMS and HiPIMS will be
discussed.
2. Experimental Details
Samples were deposited in a semi-industrial Alcatel SCM 650
sputtering system. Sputtering target (⌀ 20 cm) was driven
by a H¨uttinger TruPlasma Bipolar 4010 generator operated
in DC mode in the DCMS case or by Melec SIPP 2000
generator in the HiPIMS case. A well balanced magnetic
field configuration was used; for more details of the magnetic
field configuration see [4]. The substrates were mounted onto
a biasable (13.56 MHz RF) substrate holder 6.5 cm above
the racetrack of the target. Titanium target with purity of
99.7% was used for preparation of the Ni-free samples. A
compound target with 87.5% of Ti and 12.5% of Ni and purity
of 99.95% was used for the Ni-doped coating preparation.
The deposition chamber was evacuated by a turbomolecular
pump backed by a Roots pump to the base pressure in the
order of 10−4
Pa. Gas inlet was located between the substrate
holder and the target at the distance of 2 cm from the target.
Acetylene with purity of 99.6% was used as a source of carbon
and hydrogen; argon with purity of 99.999% was used as a
buffer gas. Gases were dosed via mass-flow controllers.
Cemented tungsten-carbide substrates (𝑅 𝑎 = 3 nm) were
precleaned in a chemical bath and ultrasonicated in a
degreasing agent. Both the target and the substrates were further
cleaned by Ar ion bombardment. During this cleaning
phase, the substrate was preheated prior to the deposition
to a temperature of ∼150∘
C. After the cleaning phase, the
argon flow rate was set to 20 sccm (∼1.1 Pa) and the deposition
of ∼750 nm thick adhesion interlayer took place. Then the
acetylene was introduced into the chamber. The coating’s
depositions were performed with deposition parameters
summarized in Table 1. Longer deposition time in HiPIMS
was used due to generally lower deposition rates [29]. For
DCMS depositions the bias of −100 V was used; for HiPIMS
deposition a substrate was left on floating potential. Substrate
temperature of ∼300∘
C was reached in both the DCMS and
Table 1: Parameters of the coatings deposition.
DCMS HiPIMS
Mean power (kW) 1.25 1.25
Pulse length (𝜇s) — 400
Pulse frequency (Hz) — 30
Maximal current (A) ∼2.5 ∼400
Substrate bias (V) −100 —
Argon flow rate (sccm) 20 20
Deposition duration (min) 60 120
the HiPIMS cases after approximately 15 minutes from the
start of the deposition. After the deposition the samples were
left to cool down for two hours under vacuum.
Spherical abrasions prepared by a calotester and measured
by laser confocal microscope LEXT OLS4000 3D
were used for determination of coating’s thickness. Energydispersive
X-ray spectroscopy (EDX) analyses using Tescan
MIRA FEG SEM equipped with an EDX detector from
Oxford Instruments were performed to determine the chemical
composition of the coatings. Internal library of standards
included in the Aztec software by Oxford Instruments was
used for quantitative evaluation of the composition. Xray
diffraction (XRD) utilizing Rigaku SmartLab Type F
diffractometer was used to determine the mean grain size
and the lattice parameter. The diffractometer operated with
CuK𝛼 radiation (𝜆 = 0.1541 nm). Grazing angle of incidence
(GIXRD) configuration with the angle of 0.5∘
was used for
the measurement. X-ray photoelectron spectroscopy (XPS)
was used to obtain information about the phase composition:
content of the amorphous phase and the nanocrystalline
phase. The XPS signals were recorded using a Thermo
Fischer Scientific K-Alpha XPS. This system equipped with a
microfocused monochromatic AlK𝛼 X-ray source (1486.6 eV)
was used for determination of phase composition of coatings.
An X-ray beam with 200 W power (650 microns spot size)
was used for determination of the lattice parameter as well
as the mean grain size. The mean size of the TiC grains
was calculated from the most intensive diffractogram peak
corresponding to the (111) crystallographic direction using
the Scherrer equation neglecting the effect of the grain
strain [30]. The lattice parameter was always calculated
from the three first intensive TiC diffraction peaks: (111),
(200), and (311). Survey and high resolution spectra were
collected with pass energy of 50 and 20 eV, respectively. The
charge shift of the spectra was then corrected by setting
the C1s peak to 284.8 eV. Fischerscope H100 depth sensing
indenter (DSI) equipped with a Berkovich tip was used to
measure the coating’s hardness. The load applied on the tip
and the corresponding indentation depths were recorded
simultaneously for both the loading and the unloading parts.
Hardness was determined using standard Oliver and Pharr
method [31]. The indentation load of 40 mN was chosen in
order to keep the radius of the deformed volume (estimated
according to Johnson’s contact model [32]) sufficiently low
to avoid influence of the substrate. The tip area function
was calibrated using fused silica and hard metal standards.
Journal of Nanomaterials 3
Typically, 32 indents were performed in different places on
the sample for hardness analysis.
Stoichiometry of TiC grains was calculated in a simplified
qualitative model, where all titanium atoms are assumed
to be forming Ti-C bonds in the titanium carbide grains.
Coefficient of stoichiometry of TiC grains in the Ni-free
coatings was calculated as
𝑥 =
[C] ⋅ 𝐴C-Ti
[Ti] ⋅ 𝐴C1s
, (1)
where [C] and [Ti] denote the atomic composition of the
sample taken from EDX, 𝐴C-Ti denotes the area under
(C-Ti + C-Ti∗
) peak, and 𝐴C1s denotes the area of the whole
XPS C1s peak [26]. Two extreme cases were calculated. The
first case assumed that there were no Ni-C bonds in the
grain. In this case the coefficient of stoichiometry could be
calculated directly from (1). The second case assumed that
all the titanium as well as the nickel atoms created bonds
with C atoms. In this case the coefficient of stoichiometry was
calculated using a modification of (1):
𝑥 =
[C] ⋅ (𝐴C-Ti + 𝐴C-Ni)
([Ti] + [Ni]) ⋅ 𝐴C1s
, (2)
where 𝐴C-Ni denotes area under (C-Ni) peak in XPS C1s
spectrum. The actual coefficient of stoichiometry should
therefore lie between those of the extreme cases.
The mean grain separation (MGS) was calculated using
information about the chemical composition, the mean grain
size, and the phase composition. For calculation of the MGS
a model assuming cubic grains uniformly distributed in the
matrix was utilized [12, 21]. The equation used for calculation
of the nc-TiC/a-C mean grain separation was adopted from
[21] and adapted for nc-(Ti,Ni)C/a-C:H coatings assuming
that Ni is included in TiC grains not forming regular carbide:
MGS = 𝑐 ( 3
√1 + 𝑟vol − 1) , (3)
𝑟vol =
𝐴C-C
𝐴C-Ti + 𝐴C-Ni
𝑀a-C
𝑥𝑀TiC + 𝑦𝑀Ni
𝑥𝜌TiC + 𝑦𝜌Ni
𝜌a-C
, (4)
where 𝑐 denotes the mean grain size, 𝐴C-C, 𝐴C-Ti, and 𝐴C-Ni
denote areas under (C-C), (C-Ti + C-Ti∗
), and (C-Ni) peaks
in XPS C1s spectrum, 𝑀a-C, 𝑀TiC, and 𝑀Ni denote atomic
weights of amorphous carbon, titanium carbide, and nickel,
and 𝜌a-C, 𝜌TiC, and 𝜌Ni denote mass densities of amorphous
carbon, titanium carbide, and nickel. The atomic weights
were taken as 𝑀a-C = 12.0 amu [12], 𝑀TiC = 59.5 amu [12],
and 𝑀Ni = 58.7 amu [33]. Mass densities 𝜌a-C = 2.20 g⋅cm−3
[12], 𝜌TiC = 4.91 g⋅cm−3
[12], and 𝜌Ni = 8.91 g⋅cm−3
[34] were
used. Symbols 𝑥 and 𝑦 in (4) denote the weighted average of
titanium and nickel content:
𝑥 =
[Ti]
[Ti] + [Ni]
, (5)
𝑦 =
[Ni]
[Ti] + [Ni]
, (6)
where [Ti] and [Ni] denote content of titanium and nickel in
the coating.
3. Results
Series of Ni-free nc-TiC/a-C:H as well as Ni-doped nc-(Ti,
Ni)C/a-C:H coatings were prepared by HiPIMS and DCMS
processes. The thickness of the DCMS prepared coatings was
in the range of 3.8–4.2 𝜇m for coatings with ∼30%–∼70%
of carbon. The thickness of the HiPIMS prepared coatings
deposited for twice as long as the DCMS deposited coatings
was in the range of 2.5–3.0 𝜇m for coatings with ∼20%–∼75%
of carbon. Deposition rate of the DCMS prepared coatings
was thus approximately three times higher than the deposition
rate of HiPIMS prepared coatings. This agrees well with
finding that the deposition rate is generally significantly lower
in HiPIMS compared to DCMS, mostly due to the titanium
ion back-attraction to the target [27, 29, 35]. No influence of
nickel addition on coating’s thickness and deposition rate was
observed.
As the total carbon content in the Ni-free as well as the Nidoped
coating increased, a total content of metallic elements,
that is, Ti in the Ni-free coatings and Ti and Ni in the Nidoped
coatings, linearly decreased. Figure 1 shows linear
decrease of the total content of titanium and nickel in the case
of the Ni-doped coatings prepared by DCMS (Figure 1(a))
and HiPIMS (Figure 1(b)).
An example of diffraction patterns of the Ni-free coating
with [C]/[Ti] ratio of ∼0.33 (black line) and the coating
containing ∼10% Ni with the same [C]/[Ti] ratio (red line) is
depicted in Figure 2. Only peaks corresponding to the TiC
phase were observed in the diffraction pattern; no diffraction
peaks corresponding to nickel or titanium phase were
observed. The same, that is, only peaks corresponding to the
TiC phase, were observed in case of all studied coatings. The
lattice parameter calculated from the most intense peak of Xray
diffractograms is plotted in Figure 3. The bulk TiC lattice
parameter of 4.33 ˚A [36] is plotted as a horizontal line for
comparison. Figure 3(a) shows the comparison of the lattice
parameter of the Ni-free and the Ni-doped coatings prepared
by the DCMS; Figure 3(b) shows the same comparison in the
HiPIMS case. The lattice parameter increased with increasing
carbon content for all the coatings with less than ∼50% of
C. This was followed by a decrease of the lattice parameter
for the DCMS and the Ni-free HiPIMS coatings. The
Ni-doped HiPIMS deposited coatings with carbon content
≳50% exhibited lattice parameter independent of the total
carbon content. The HiPIMS prepared coatings always exhibited
lattice parameter lower than the DCMS prepared coatings
and also lower than lattice parameter of bulk TiC. The
lattice parameter of the Ni-doped coatings with ≤50% C was
always higher than the lattice parameter of the corresponding
Ni-free coatings.
The comparison of the mean grain size of the Ni-free
and the Ni-doped coatings prepared by DCMS is shown in
Figure 4(a). The mean grain size of the Ni-free coatings did
not significantly change with increasing total carbon content
up to ∼50% and was ∼25 nm. Ni-doped coatings exhibited
pronounced increase of the grain size from ∼5 nm to ∼24 nm
as the carbon content increased from ∼30% to ∼50%. As the
carbon content increased further, the grain size decreased
both for the Ni-free and for the Ni-doped coatings and
4 Journal of Nanomaterials
Ti
Ni
0
10
20
30
40
50
60
70
80
Ti(at.%)
30 40 50 60 70 8020
C (at. %)
0
2
4
6
8
10
Ni(at.%)
(a)
Ti
Ni
0
10
20
30
40
50
60
70
80
Ti(at.%)
20 30 40
C (at. %)
50 60 70 80
0
2
4
6
8
10
Ni(at.%)
(b)
Figure 1: Dependence of titanium and nickel content on the total content of carbon. Comparison of (a) the DCMS and (b) the HiPIMS
prepared coatings.
Ni-free
9.7 at. % of Ni
Intensity(a.u.)
(111)
(200)
(220)
(311)
(222)
30 40 50 60 70 8020
2휃 (deg)
Figure 2: Diffraction patterns of the HiPIMS prepared Ni-free
(black line) coating with [C]/[Ti] ratio ∼0.33 and the HiPIMS
prepared Ni-doped coating with 9.7% of Ni (red line) with [C]/[Ti]
ratio ∼0.34. Crystallographic direction of the TiC peaks is denoted
above certain peaks.
the coating with the highest amount of carbon exhibited
grain size of ∼4 nm. The mean grain size of the HiPIMS
prepared coatings is plotted in Figure 4(b). The mean grain
size in the Ni-free coatings decreased with the increasing
total carbon content from ∼37 nm for coating with ∼25% of
carbon down to ∼5 nm for coating with ∼68% of carbon. The
grain size in the Ni-doped coatings increased from ∼4 nm
for coating with the lowest amount of carbon up to ∼15 nm
for coating with carbon amount of ∼48%. As the carbon
content increased further, the grain size decreased down to
∼2 nm and remained constant for the rest of the carbon rich
coatings. Ni-doped coatings showed generally smaller grains
compared to Ni-free coatings which was more pronounced
in the HiPIMS series. The grain size reduction effect was also
more pronounced in coatings with low amount of carbon.
High resolution XPS measurements were carried out on
the C1s, Ti2p, and Ni2p peak regions. Comparison of Ti2p
peaks of coatings with similar carbon content without and
with Ni prepared at both the DCMS and HiPIMS is plotted
in Figure 5. Samples with the lowest amount of carbon and
thus the highest amount of nickel are shown to illustrate the
highest possible effect of Ni doping on the measures Ti2p
peaks. The Ti2p peaks exhibit minimal differences between
the samples. The binding energy of the maximum intensity
of the Ti2p3/2 of 449.6–449.9 eV as well as that of the Ti2p1/2
peak of 460.9–461.3 eV agrees with the position of the TiC
bond [3, 10, 37] indicating majority of Ti being bonded
to carbon. Ti-O bonding is also present in the coating as a
minor component. No Ti-Ti bonding was detected. This was
true for all of the analyzed samples independently of their
carbon content, nickel presence, or the method of deposition.
The Ni2p peaks of the DCMS and HiPIMS deposited coatings
with the highest amount of nickel together with a reference
spectrum Ni (99.999% purity, Goodfellow) are presented in
Figure 6. The positions of the Ni2p3/2 as well as the Ni2p1/2
peaks in the deposited coatings are shifted by 0.8 eV to higher
energies compared to the Ni reference. The positions of the
Ni2p peaks are in good agreement with the literature value
for Ni-C bonds [37]. Further, no Ni-Ni bonds were detected,
which was in good agreement with findings from XRD
showing no presence of Ni crystallites. The amount of the
a-C:H phase was quantified using the curve deconvolution
of the C1s peak [15, 22, 37]. An example of deconvoluted
high resolution C1s peak of Ni-free (∼61% of carbon) and
Ni-doped (∼58% of carbon) coatings prepared by DCMS
is shown in Figure 7. An amount of the a-C:H phase was
estimated by comparison of C-C bond area to sum of C-Ti,
C-Ti∗
and C-Ni bond areas.
Journal of Nanomaterials 5
Without Ni
With Ni
Bulk
4.24
4.26
4.28
4.30
4.32
4.34
4.36
4.38
Latticeparameter(Å)
30 40 50 60 70 8020
C (at. %)
(a)
Without Ni
With Ni
Bulk
4.24
4.26
4.28
4.30
4.32
4.34
4.36
4.38
Latticeparameter(Å)
30 40 50 60 70 8020
C (at. %)
(b)
Figure 3: The lattice parameter of the Ni-free (black marks) and the Ni-doped (red marks) coatings. Comparison of (a) the DCMS prepared
coatings with (b) the HiPIMS prepared coatings. Line at 4.33 ˚A denoted the bulk TiC lattice parameter [36]. Lines are added as guides for the
eye.
Without Ni
With Ni
0
5
10
15
20
25
30
35
40
Meangrainsize(nm)
30 40 50 60 70 8020
C (at. %)
(a)
Without Ni
With Ni
0
5
10
15
20
25
30
35
40
Meangrainsize(nm)
30 40 50 60 70 8020
C (at. %)
(b)
Figure 4: The mean grain size of the Ni-free (black marks) coatings compared with the Ni-doped (red marks) coatings. Comparison of (a)
the DCMS prepared coatings and (b) the HiPIMS prepared coatings. Lines are added as guides for the eye.
Figures 8(a) and 8(b) show the comparison of the amount
of the a-C:H phase of the Ni-free and the Ni-doped coatings
prepared by DCMS and HiPIMS, respectively. Rapid increase
of amount of the a-C:H phase was observed in the DCMS
deposited coatings. However, the Ni-free coatings with ≲50%
exhibited significantly higher amount of the a-C:H phase
compared to corresponding Ni-doped coatings. This effect
is the most pronounced in the coatings with the lowest
amount of carbon where the Ni-free coatings exhibited
significantly higher amount of the a-C:H phase than the Nidoped
coatings. No effect of Ni addition on the total a-C:H
amount was observed in the HiPIMS deposited coatings.
Both the Ni-free and the Ni-doped coatings exhibited slight
increase of the amount of the a-C:H phase in the coatings with
total carbon content ≲50% followed by rapid increase up to
∼70% for coatings with the high amount of carbon.
Coefficient of stoichiometry in the DCMS prepared
coatings is plotted in Figure 9(a). The Ni-free coatings with
6 Journal of Nanomaterials
Intensity(a.u.)
22.9% C, 67.4% Ti, 9.7% Ni
25.0% C, 75.0% Ti
28.8% C, 64.1% Ti, 7.1% Ni
Carbidic
Metallic
Oxidic
29.5% C, 70.5% Ti
Metallic
CarbidicOxidic
DCMS
DCMS
HiPIMS
HiPIMS
470 465 460 455 450475
Eb (eV)
Figure 5: Comparison of high resolution XPS spectra of the Ti2p
peaks.
carbon content ≲50% exhibited increase of coefficient of
stoichiometry with increasing carbon content. When ≲50%
of carbon was reached, coefficient of stoichiometry of ∼1
was attained. As the carbon content increased further, the
coefficient of stoichiometry decreased. In the case of the Nidoped
coatings the stoichiometry coefficient was calculated
for two extreme cases. In both cases the Ni-doped coatings
exhibited a monotonous increase of the coefficient of
stoichiometry with increasing carbon content; stoichiometry
coefficient ∼1 was reached for coatings with ≳60% of carbon.
The coefficient of stoichiometry was higher for the Ni-doped
coatings with ≲45% of carbon. The Ni-doped coatings with
carbon content of ∼45–65% exhibited lower coefficient of
stoichiometry, as the increase of the coefficient of stoichiometry
was slower than in the case of the Ni-free coatings. The
coefficient of stoichiometry of the Ni-doped coatings was
higher for coatings with carbon content ≳60% as the Nifree
coatings already exhibited a decrease of the coefficient
of stoichiometry. Coefficient of stoichiometry in the Nifree
coatings prepared by HiPIMS (see Figure 9(b)) with
≲50% of C also increased with increasing carbon content.
Similarly to the DCMS case, the HiPIMS coatings with ∼50%
of C exhibited coefficient of stoichiometry ∼1. The Ni-doped
coatings with ≲50% of carbon exhibited increasing of the
stoichiometry coefficient with increasing carbon content.
The coefficient of stoichiometry in the Ni-doped coatings
reached maximum value ∼1 for coatings with 50–60% of
carbon and it decreased with further carbon content increase.
The coefficient of stoichiometry was lower for the HiPIMS
deposited Ni-doped coatings with ≲50% of carbon.
The mean grain separation is plotted in Figure 10. The
number of carbon monolayers separating the grains corresponding
to the calculated MGS is plotted on the right axis in
case of both graphs [38]. Increase of the MGS with increasing
carbon content was observed in all series of coatings. This
increase is, however, for coating with carbon content ≳50%
less pronounced than the increase of the amount of a-C:H
phase because of simultaneous decrease of the grain size. The
DCMS prepared Ni-free coatings exhibited higher MGS compared
to corresponding Ni-doped coatings (see Figure 10(a)).
The most pronounced difference of the MGS was observed
in coatings with the lowest amount of carbon. While MGS
Intensity(a.u.)
Ni-ref.
22.9% C, 67.4% Ti, 9.7% Ni
28.8% C, 64.1% Ti, 7.1% Ni
MetallicCarbidic
HiPIMS
DCMS
875 870 865 860 855 850 845880
Eb (eV)
Figure 6: Comparison of high resolution XPS spectra of the Ni2p
peaks.
of the Ni-free coating was ∼0.8 nm, which corresponded to
∼6 carbon monolayers, grains in the Ni-doped coating were
separated only by 1 carbon monolayer which corresponded
to MGS ≲ 0.1 nm. As the carbon content increased the
difference in MGS of the Ni-doped and the Ni-free coatings
decreased. In the case of the HiPIMS prepared coatings
(Figure 10(b)), the Ni-doped coatings exhibited lower MGS
than the corresponding Ni-free coatings. Employing the
HiPIMS, the MGS was reduced compared to DCMS which
was the most pronounced particularly for Ni-free coatings at
low carbon contents.
Figure 11 shows the hardness of the DCMS (Figure 11(a))
and the HiPIMS (Figure 11(b)) prepared coatings. In the
DCMS case the hardness of the Ni-free coatings increased
from ∼11 GPa up to ∼30 GPa with increasing content of
carbon up to ∼55%. As the carbon content increased further,
the hardness of the Ni-free coatings decreased. Hardness
of the Ni-doped coatings increased from ∼20 GPa up to
∼29 GPa with increasing total carbon content up to ∼45%.
Then the hardness was not significantly influenced by the
total carbon content and remained ∼25 GPa for coatings with
≳45% of carbon. The Ni-free coatings prepared by HiPIMS
exhibited increase of hardness from ∼15 GPa up to ∼45 GPa
with increasing content of carbon up to ∼57%. The hardness
decreased down to 30 GPa with further carbon increase up
to 70%. Hardness of the Ni-doped coatings prepared by
HiPIMS behaved similarly to the case of the Ni-free up
to ∼48% of carbon where the Ni-doped coatings exhibited
maximum hardness of ∼38 GPa. The hardness then decreased
with increasing carbon content down to 20 GPa for coatings
with 75% of carbon. HiPIMS prepared coatings exhibited
higher hardness than the DCMS prepared coatings for both
Ni-free and Ni-doped coatings. The maxima of the hardness
for HiPIMS prepared coatings were shifted to lower amount
of carbon for Ni-doped coatings with respect to the Ni-free
case.
All the results presented in previous chapter are summarized
in Table 2 (DCMS prepared coatings) and Table 3
(HiPIMS prepared coatings). Some parameters could not be
determined due to problems, for example, coating delamination.
Visibly erroneous data was discarded.
Journal of Nanomaterials 7
Table 2: Overview of the coatings prepared by DCMS.
C (at. %) Lattice parameter (˚A) Grain size (nm) a-C:H phase (%) MGS (nm) Hardness (GPa)
Ni-free
29.5 4.2761 26.6 17.2 0.799 11.4 ± 0.8
38.4 20.6
39.3 4.3233 17.8 26.4 0.904 20.7 ± 1.6
49.0 4.3578 27.5 29.0 24.4 ± 1.8
53.4 4.3634 17.8 13.7 33.3 ± 4.3
55.9 4.3547 20.7 25.7 1.016 26.8 ± 1.2
61.3 4.3483 14.4 34.8 1.065 26.8 ± 2.7
65.1 4.3405 4.2 66.5 0.991 21.2 ± 2.5
Ni-doped
28.8 4.3010 5.5 2.6 0.025 20.0 ± 1.1
36.9 4.3391 8.8 14.2 0.248 19.0 ± 1.9
39.4 4.3521 12.9 3.8 0.088 23.4 ± 2.3
41.9 4.3601 21.8 13.3 0.562 29.1 ± 3.1
46.3 4.3678 19.2 11.3 0.404 25.8 ± 2.3
50.5 4.3411 23.4 32.9 24.2 ± 2.2
52.4 4.3472
52.9 20.1 14.3 0.556 23.9 ± 2.6
56.5 4.3419 11.2 41.4 30.6 ± 1.8
57.5 4.3450 9.2 39.7 29.9 ± 2.0
57.8 4.3489 7.8 44.8 22.6 ± 3.5
64.4 4.3432 6.1 40.2 0.609 21.4 ± 1.9
66.4 4.3229 4.1 54.8 0.688 27.5 ± 2.3
Table 3: Overview of the coatings prepared by HiPIMS.
C (at. %) Lattice parameter (˚A) Grain size (nm) a-C:H phase (%) MGS (nm) Hardness (GPa)
Ni-free
25.0 4.2502 36.5 15.1 ± 1.3
40.0 4.2950 24.1 5.8 0.221 34.5 ± 2.5
49.0 4.3180 27.9 14.1 0.671 38.3 ± 2.7
53.0 4.3216 22.4 16.6 0.650 40.0 ± 3.0
57.0 4.3130 10.0 25.2 0.483 44.7 ± 3.8
68.0 4.2890 5.5 55.4 0.873 31.5 ± 3.6
Ni-doped
22.9 4.2600 4.3 5.7 0.043 17.3 ± 0.8
33.8 4.3033 5.5 8.6 0.084 26.8 ± 2.1
36 4.3118 7.5 33.8 ± 1.5
47.7 4.3185 15.5 12.1 0.341 38.2 ± 4.1
50.7 4.3120 1.6 13.9 0.041 33.6 ± 4.4
63 4.2982 3.0 37.0 0.264 24.6 ± 4.1
77.2 4.2930 2.0 70.2 0.576 20.4 ± 1.7
4. Discussion
The observed dependencies of the crystalline phase from
XRD analyses (Figures 2–4) can be summarized and
explained as follows. Only diffractions corresponding to fcc
TiC lattice were observed (see Figure 2) for both Ni-free
and Ni-doped coatings. Thus, Ni did not form independent
crystallites. It was incorporated in TiC grains or it was
dispersed in the carbon matrix. As the lattice parameter of
the Ni-doped coatings was influenced by Ni being always
larger than that of the Ni-free coatings (Figure 3) the nickel
incorporation in the grains can be presumed. The increase
8 Journal of Nanomaterials
Raw data
Fitted data
Background
C-Ti
C-C
Intensity(a.u.)
287 286 285 284 283 282 281 280288
Eb (eV)
C-Ti∗
(a)
C-NiRaw data
Fitted data
Background
C-Ti
C-C
Intensity(a.u.)
287 286 285 284 283 282 281 280288
Eb (eV)
C-Ti∗
(b)
Figure 7: Fitted XPS spectra of C1s peak of (a) Ni-free coating and (b) Ni/doped coating with similar carbon content of ∼60%.
30 40 50 60 70 8020
C (at. %)
0
15
30
45
60
75
a-C:Hphase(%)
Without Ni
With Ni
(a)
Without Ni
With Ni
0
15
30
45
60
75
a-C:Hphase(%)
30 40 50 60 70 8020
C (at. %)
(b)
Figure 8: The relative amount of the amorphous hydrogenated carbon (a-C:H) phase. Comparison of the Ni-free (black marks) and the
Ni-doped (red marks) coatings prepared by (a) the DCMS and (b) the HiPIMS. Lines are added as guides for the eye.
of the lattice parameter of the coatings with increasing total
carbon content in coatings with ≲50% can be explained by
vacancy filling [34]. The number of vacancies in the TiC
crystal lattice decreased with increasing carbon content. The
carbon atoms occupied the vacancies and the lattice parameter
increased accordingly. The DCMS prepared coatings with
≳50% of carbon exhibited higher lattice parameter than bulk
value [36]. Similar effect, that is, higher value of the lattice
parameter of DCMS deposited TiC/C coatings than the bulk
lattice parameter of TiC, was observed by several authors
[3, 26, 39–41]. The lattice parameter of the HiPIMS prepared
coatings (Figure 3(b)) was always lower than the lattice
parameter of the DCMS (Figure 3(a)) prepared coatings. This
was due to the generally higher kinetic energy of the particles
impinging the surface of the growing coatings in HiPIMS
[42]. Higher energy influx promoted the surface diffusion
of the adatoms and enabled them to reach energetically
more favorable position. Higher energy influx and promoted
surface diffusion resulted in denser crystallites as indicated
by the lower lattice parameter in the HiPIMS case [29].
A decrease of the grain size with increasing carbon
content was observed in DCMS as well as in HiPIMS
prepared coatings with carbon content ≳ 50% regardless of
Ni doping. This is due to a rapid increase of the a-C:H
Journal of Nanomaterials 9
Without Ni
With Ni (C-Ti) + (C-Ni)
With Ni (C-Ti)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Coefficientofstoichiometry
30 40 50 60 70 8020
C (at. %)
(a)
Without Ni
With Ni (C-Ti) + (C-Ni)
With Ni (C-Ti)
30 40 50 60 70 8020
C (at. %)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Coefficientofstoichiometry
(b)
Figure 9: Stoichiometry in TiC crystallites of the Ni-free (black marks) and of the Ni-doped (red and green marks) coating. Comparison of
(a) the DCMS and (b) the HiPIMS prepared sets of coatings. Lines are added as guides for the eye.
30 40 50 60 70 8020
C (at. %)
0.0
0.3
0.6
0.9
1.2
1.5
Meangrainseparation(nm)
Without Ni
With Ni
0
2
4
6
8
10
Carbonmonolayers
(a)
Without Ni
With Ni
0.0
0.3
0.6
0.9
1.2
1.5Meangrainseparation(nm)
30 40 50 60 70 8020
C (at. %)
0
2
4
6
8
10
Carbonmonolayers
(b)
Figure 10: The mean grain separation of the Ni-free (black marks) and of the Ni-doped (red marks) coating. Comparison of (a) the DCMS
and (b) the HiPIMS prepared sets of coatings. Lines are added as guides for the eye.
phase content in coatings with carbon content ≳ 50% (see
Figure 5). Thus the a-C:H phase covered the growing grains
and hindered their growth [29]. This led to formation of
higher number of smaller grains. Ni-doped coatings with
carbon content ≲ 50% exhibited smaller grains than the Nifree
coatings with similar carbon content (see Figure 4). This
effect was more pronounced in the coatings with low amount
of carbon and high amount of nickel. A similar phenomenon
of grain size decrease initiated by third-element doping was
observed in other carbide systems [21, 37, 43, 44]. R˚asander
et. al. calculated that nickel presence in the TiC lattice
leads to lowering of the formation energy and the migration
energy barriers of the carbon vacancies in comparison with
the Ni-free TiC lattice [45]. High concentration of crystal
defects, for example, vacancies, leads to the formation of
planar defects acting as grain boundaries which resulted in
smaller grains compared to a system with lower number of
vacancies [34]. Thus, we formulate a hypothesis that nickel
incorporation into the nc-TiC/a-C:H coatings modified the
vacancy distribution in the TiC 𝑥 grains which resulted in
arrangement of planar defects leading to formation of grain
boundaries and thus to the reduction of the grain size.
Jansson et al. [22] proposed that Ni addition should lead
to releasing of the carbon from the TiC grains into the aC
phase. Thus, Ni-doped coatings with the same amount of
carbon as Ni-free coatings should exhibit a higher amount
of a-C phase. However, in our study, no influence of nickel
addition on the amount of amorphous carbon phase was
observed for the HiPIMS prepared coatings (see Figure 6(b)).
In the case of the DCMS prepared coatings, the Ni-doped
coatings with ≲50% of carbon exhibited even lower amount
of the a-C(:H) phase than the Ni-free coatings (Figure 8(a)).
It has been observed that the a-C(:H) phase was formed
even for coatings with ≲50% of carbon [10, 16]. So the
10 Journal of Nanomaterials
Without Ni
With Ni
30 40 50 60 70 8020
C (at. %)
10
15
20
25
30
35
40
45
50Hardness(GPa)
(a)
Without Ni
With Ni
10
15
20
25
30
35
40
45
50
Hardness(GPa)
30 40 50 60 70 8020
C (at. %)
(b)
Figure 11: The hardness of the Ni-free (black marks) and the Ni-doped (red marks) coatings. Comparison of (a) the DCMS and (b) the
HiPIMS prepared coatings. Lines are added as guides for the eye.
coatings are composed of TiC grains with strong deficit of
carbon surrounded by small amount of amorphous carbon.
High temperature or strong ion bombardment may force
the carbon to pass from the a-C:H phase near the grain
surface to the grain itself. The Ni-doped coatings with ≲50%
C prepared by DCMS exhibited significantly smaller grains
(see Figure 4(a)) and lower amount of the a-C:H phase
(see Figure 8(a)) than the corresponding Ni-free coatings.
The lower amount of the a-C:H phase in case of the Nidoped
coatings can be explained by presence of smaller grains
with higher surface/volume ratio, facilitating the effect of the
carbon transfer from the matrix to the grains. In the case of
HiPIMS the ion bombardment was so intense that enough
energy is supplied where carbon was incorporated into the
grains independently of their size. The nickel induced grain
size reduction thus did not significantly affect the amount of
the a-C:H phase in HiPIMS deposited coatings. Furthermore,
in strongly substoichiometric TiC grains the nickel addition
may not lead to complete releasing of carbon atoms from
a grain, but rather to carbon atom relocation within a TiC
grain. The carbon atom could be trapped in a nearby vacancy
instead of being removed from the grain to the amorphous
phase.
The evolution of the mean grain size and the mean grain
separation of the Ni-free and the Ni-doped coatings prepared
by DCMS is schematically depicted in Figure 12. Analogous
scheme for the HiPIMS prepared coatings is depicted in
Figure 13. For simplicity, we can divide all the range of
carbon content into three zones. This division into three
zones illustrates the changes of the microstructure of the
deposited coatings affecting the mechanical properties.
In the case of the DCMS deposited coatings an increase
of the hardness was observed up to carbon content of ∼50%
(Zone 1 in Figures 12(a) and 12(b)). This increase was
Zone 1
(a)
(b)
Zone 2 Zone 3
Content of carbon
Figure 12: Scheme of the mean grain size and the mean grain
separation of (a) the Ni-free and (b) the Ni-doped coatings prepared
by DCMS. Zone 1 denoted coatings with low amount of carbon,
Zone 2 denoted medium amount of carbon, and Zone 3 denoted
high amount of carbon.
attributed to lowering of the grain vacancy concentration
similarly to [26], because only low vacancy concentration is
favorable for achieving a high hardness [46, 47]. Increase of
the initially low grain size in the case of the Ni-doped coatings
was also an important factor in the hardness enhancement
because of effects analogical to the Hall-Petch effect [48–
53]. The initial hardness increase with increasing carbon
content was observed despite a slight increase of the MGS
as high MGS facilitates the grain sliding [54]. The DCMS
prepared Ni-doped coatings with carbon content ≲ 50% also
Journal of Nanomaterials 11
Zone 1
(a)
(b)
Zone 2 Zone 3
Content of carbon
Figure 13: Scheme of the mean grain size and the mean grain
separation of (a) the Ni-free and (b) the Ni-doped coatings prepared
by HiPIMS. Zone 1 denoted coatings with low amount of carbon,
Zone 2 denoted medium amount of carbon, and Zone 3 denoted
high amount of carbon.
exhibited higher hardness than the Ni-free coatings. This
hardening was attributed to lower grain size as well as the
lower amount of a-C:H phase corresponding to lower MGS.
Also the Ni-doped coatings exhibited lower amount of the
a-C:H phase for the same total carbon content and thus lower
carbon vacancy concentration in the grains can be expected.
The hardness of the DCMS deposited coatings with carbon
content ≳ 50% was similar for the Ni-doped and the Ni-free
coatings due to comparable grain size as well as comparable
amount of the a-C:H phase (see Zone 3 in Figures 11(a) and
11(b)). Some slight differences in the grain size and in the aC:H
phase content led to differences in the MGS; however,
generally high MGS (≫1 monolayer) had no appreciable
effect on the measured hardness as only 1 monolayer is
optimal for the effect of nanocomposite hardening [51, 52, 55].
The coating microstructure governed also the hardness
of the coatings prepared by HiPIMS. The initial hardness
increase up to 50% of carbon was mainly caused by lowering
of the grain vacancy concentration and the grain size evolution
(Zone 1 in Figure 13). The initial TiC grain size of ∼5 nm
observed in the Ni-doped coatings was low and the effect
analogical to the inverse Hall-Petch effect in polycrystalline
films could occur [51–53, 55]. As the TiC grain size increased,
the hardness increased accordingly. The maximum grain size
of ∼15 nm was reached at 45% of carbon (Zone 2 in Figure 13).
The grains still contained lattice vacancies because of carbon
deficiency and preferential formation of the a-C:H matrix
phase. The individual grains were on average separated by
∼2 monolayers of the matrix. This combination of parameters
resulted in hardness of ∼37 GPa. On the other hand, the Nifree
coatings initially exhibited large grains with the size of
∼35 nm that is too high for achieving high hardness due the
effect analogous to the Hall-Petch effect in polycrystallites
[51–53]. The grain size steadily decreased while the hardness
increased with the carbon content increase (see Zone 3 in
Figures 13(a) and 13(b)). The combination of the grain size,
low vacancy concentration, and low mean grain separation
which was optimal for the highest hardness was reached in
coating with ∼55% of C. This coating exhibited ∼10 nm grains
separated by ∼3 monolayers of the matrix. Low vacancy
concentration can be expected as the carbon content is >50%
unlike in the case of the hardest Ni-doped coating. This
combination of parameters resulted in maximal hardness of
∼45 GPa.
The hardness of the coatings deposited by HiPIMS was
generally higher than that of the corresponding DCMS
deposited coatings. This can be partially attributed to grain
size decrease, but also to the lower amount of the a-C:H
phase formed under HiPIMS conditions. This led to a lower
mean grain separation as well as reduction of grain vacancy
concentration in the grains due to HiPIMS induced carbon
transfer from the matrix to the grains.
5. Conclusion
Ni-free and Ni-doped nc-TiC/a-C:H coatings were prepared
by a hybrid PVD-PECVD process of titanium and titaniumnickel
compound target sputtering in argon/acetylene atmosphere.
DCMS as well as the HiPIMS was used for the
depositions in order to study the influence of the plasma
excitation on the structure and hardness of the deposited
coatings.
Nickel was found to be incorporated into the grains.
Nickel addition also led to decrease of the grain size. This
effect was more pronounced for the coatings with carbon
content ≲ 50%. The grain size reduction was attributed to
grain boundary formation due to formation of lattice defects
induced by nickel incorporation into the grains. Smaller
grains in the Ni-doped coatings prepared by DCMS exhibited
higher surface to volume ratio than that exhibited by the
larger grains in the Ni-free coatings. This led to decrease of the
amount of the a-C:H phase in the Ni-free coatings as higher
surface area of the grains led to enhanced carbon transfer
from the matrix to the grains. The HiPIMS prepared coatings
exhibited no influence of nickel addition on the a-C:H
amount, due the high ion bombardment energy promoting
the carbon incorporation even into the bulk of the larger
grains. The resulting hardness of the deposited coatings was
shown to be governed primarily by the grain size, vacancy
concentration in the grains, and their mean grain separation.
Ni doping had a significant effect on the hardness of the coatings.
For DCMS deposited coatings with ≲50% of carbon the
hardness increased with Ni doping due to favorable structure
changes. The structure was not significantly changed and the
hardness was not changed. The effect of the Ni doping was
less pronounced for HiPIMS deposited coatings with ≲50%
of carbon. However, the hardness of the Ni-doped coatings
was strongly reduced for coatings with higher carbon content
primarily due to unfavorable grain size lowering induced by
Ni doping.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
12 Journal of Nanomaterials
Acknowledgments
This research has been supported by Project LO1411 (NPU
I) funded by Ministry of Education, Youth and Sports of
Czech Republic and the Czech Science Foundation (Project
15-17875S).
References
[1] J.-E. Sundgren, B.-O. Johansson, and S.-E. Karlsson, “Mechanisms
of reactive sputtering of titanium nitride and titanium
carbide I: influence of process parameters on film composition,”
Thin Solid Films, vol. 105, no. 4, pp. 353–366, 1983.
[2] D. Mart´ınez-Mart´ınez, C. L´opez-Cartes, A. Fern´andez, and
J. C. S´anchez-L´opez, “Influence of the microstructure on the
mechanical and tribological behavior of TiC/a-C nanocomposite
coatings,” Thin Solid Films, vol. 517, no. 5, pp. 1662–1671, 2009.
[3] T. Zehnder and J. Patscheider, “Nanocomposite TiC/a-C:H hard
coatings deposited by reactive PVD,” Surface and Coatings
Technology, vol. 133, no. 134, pp. 138–144, 2000.
[4] Pavel Souˇcek, T. Schmidtov´a, L. Z´abransk´y et al., “On the control
of deposition process for enhanced mechanical properties
of nc-TiC/a-C:H coatings with DC magnetron sputtering at low
or high ion flux,” Surface and Coatings Technology, vol. 255, pp.
8–14, 2014.
[5] Y. T. Pei, D. Galvan, J. T. M. De Hosson, and A. Cavaleiro,
“Nanostructured TiC/a-C coatings for low friction and wear
resistant applications,” Surface and Coatings Technology, vol.
198, no. 1–3, pp. 44–50, 2005.
[6] D. Galvan, Y. T. Pei, and J. T. M. De Hosson, “TEM characterization
of a Cr/Ti/TiC graded interlayer for magnetron-sputtered
TiC/a-C:H nanocomposite coatings,” Acta Materialia, vol. 53,
no. 14, pp. 3925–3934, 2005.
[7] Y. Wang, X. Zhang, X. Wu, H. Zhang, and X. Zhang, “Compositional,
structural and mechanical characteristics of nc-TiC/aC:H
nanocomposite films,” Applied Surface Science, vol. 255, no.
5, pp. 1801–1805, 2008.
[8] W. K. Chu, J. W. Mayer, and M. A. Nicolet, Back Scattering
Spectrometry, Academic Press, New York, NY, USA, 1978.
[9] U. Wiklund, M. Nordin, O. W¨anstrand, and M. Larsson,
“Evaluation of a flexible physical vapor deposited TiC-C coating
system,” Surface and Coatings Technology, vol. 124, no. 2-3, pp.
154–161, 2000.
[10] Y. Hu, L. Li, X. Cai, Q. Chen, and P. K. Chu, “Mechanical and
tribological properties of TiC/amorphous hydrogenated carbon
composite coatings fabricated by DC magnetron sputtering
with and without sample bias,” Diamond and Related Materials,
vol. 16, no. 1, pp. 181–186, 2007.
[11] W. J. Meng, R. C. Tittsworth, and L. E. Rehn, “Mechanical
properties and microstructure of TiC/amorphous hydrocarbon
nanocomposite coatings,” Thin Solid Films, vol. 377-378, pp.
222–232, 2000.
[12] T. Zehnder, P. Schwaller, F. Munnik, S. Mikhailov, and J.
Patscheider, “Nanostructural and mechanical properties of
nanocomposite nc-TiC/a-C:H films deposited by reactive
unbalanced magnetron sputtering,” Journal of Applied Physics,
vol. 95, no. 8, pp. 4327–4334, 2004.
[13] S. Zhang, X. L. Bui, J. Jiang, and X. Li, “Microstructure and
tribological properties of magnetron sputtered nc-TiC/ a-C
nanocomposite,” Surface and Coatings Technology, vol. 198, no.
1-3, pp. 206–211, 2005.
[14] G. Li and L. F. Xia, “Structural characterization of TiC 𝑥 films
prepared by plasma based ion implantation,” Thin Solid Films,
vol. 396, no. 1-2, pp. 16–22, 2001.
[15] E. Lewin, P. O. ˚A. Persson, M. Lattemann et al., “On the origin
of a third spectral component of C1s XPS-spectra for nc-TiC/aC
nanocomposite thin films,” Surface and Coatings Technology,
vol. 202, no. 15, pp. 3563–3570, 2008.
[16] D. M. Cao, B. Feng, W. J. Meng, L. E. Rehn, P. M. Baldo, and
M. M. Khonsari, “Friction and wear characteristics of ceramic
nanocomposite coatings: Titanium carbide/amorphous hydrocarbon,”
Applied Physics Letters, vol. 79, no. 3, pp. 329–331, 2001.
[17] J. C. S´anchez-L´opez, D. Mart´ınez-Mart´ınez, C. L´opez-Cartes, C.
Fern´andez-Ramos, and A. Fern´andez, “A nanoscale approach
for the characterization of amorphous carbon-based lubricant
coatings,” Surface and Coatings Technology, vol. 200, no. 1-4, pp.
40–45, 2005.
[18] J. Musil, P. Nov´ak, R. ˇCerstv´y, and Z. Soukup, “Tribological and
mechanical properties of nanocrystalline-TiC/a-C nanocomposite
thin films,” Journal of Vacuum Science & Technology A,
vol. 28, no. 2, pp. 244–249, 2010.
[19] Y. T. Pei, D. Galvan, J. T. M. De Hosson, and C. Strondl,
“Advanced TiC/a-C:H nanocomposite coatings deposited by
magnetron sputtering,” Journal of the European Ceramic Society,
vol. 26, no. 4-5, pp. 565–570, 2006.
[20] T. Schmidtov´a, P. Souˇcek, V. Kudrle, and P. Vaˇsina, “Nonmonotonous
evolution of hybrid PVD-PECVD process characteristics
on hydrocarbon supply,” Surface and Coatings Technology,
vol. 232, pp. 283–289, 2013.
[21] U. Jansson and E. Lewin, “Sputter deposition of transition-metal
carbide films—a critical review from a chemical perspective,”
Thin Solid Films, vol. 536, pp. 1–24, 2013.
[22] U. Jansson, E. Lewin, M. R˚asander, O. Eriksson, B. Andr´e, and U.
Wiklund, “Design of carbide-based nanocomposite thin films
by selective alloying,” Surface and Coatings Technology, vol. 206,
no. 4, pp. 583–590, 2011.
[23] R. ˇZemliˇcka, M. J´ılek, P. Vogl et al., “Understanding of hybrid
PVD-PECVD process with the aim of growing hard nc-TiC/aC:
H COATINGS using industrial devices with a rotating
cylindrical magnetron,” Surface and Coatings Technology, vol.
255, pp. 118–123, 2014.
[24] P. Souˇcek, T. Schmidtov´a, L. Z´abransk´y et al., “Evaluation of
composition, mechanical properties and structure of nc-TiC/aC:H
coatings prepared by balanced magnetron sputtering,”
Surface and Coatings Technology, vol. 211, pp. 111–116, 2012.
[25] P. Souˇcek, T. Schmidtov´a, V. Burˇs´ıkov´a et al., “Tribological
properties of nc-TiC/a-C: H coatings prepared by magnetron
sputtering at low and high ion bombardment of the growing
film,” Surface and Coatings Technology, vol. 241, pp. 64–73, 2014.
[26] P. Souˇcek, J. Daniel, J. Hnilica et al., “Superhard nanocomposite
nc-TiC/a-C:H coatings: the effect of HiPIMS on coating
microstructure and mechanical properties,” Surface and Coatings
Technology, vol. 311, pp. 257–267, 2017.
[27] U. Helmersson, M. Lattemann, J. Bohlmark, A. P. Ehiasarian,
and J. T. Gudmundsson, “Ionized physical vapor deposition
(IPVD): a review of technology and applications,” Thin Solid
Films, vol. 513, no. 1-2, pp. 1–24, 2006.
[28] K. Sarakinos, J. Alami, and S. Konstantinidis, “High power
pulsed magnetron sputtering: a review on scientific and engineering
state of the art,” Surface and Coatings Technology, vol.
204, no. 11, pp. 1661–1684, 2010.
[29] M. Samuelsson, D. Lundin, J. Jensen, M. A. Raadu, J. T. Gudmundsson,
and U. Helmersson, “On the film density using high
Journal of Nanomaterials 13
power impulse magnetron sputtering,” Surface and Coatings
Technology, vol. 205, no. 2, pp. 591–596, 2010.
[30] P. Scherrer, “Estimation of the size and internal structure of
colloidal particles by means of r¨ontgen,” Nachrichten von der
Gesellschaft der Wissenschaften zu G¨ottingen, vol. 2, pp. 96–100,
1918.
[31] W. C. Oliver and G. M. Pharr, “Measurement of hardness
and elastic modulus by instrumented indentation: advances
in understanding and refinements to methodology,” Journal of
Materials Research, vol. 19, no. 1, pp. 3–20, 2004.
[32] K. L. Johnson, “The correlation of indentation experiments,”
Journal of the Mechanics and Physics of Solids, vol. 18, no. 2, pp.
115–126, 1970.
[33] N. N. Greenwood and A. Earnshaw, Chemistry of the Elements,
Elsevier, 2012.
[34] A. I. Gusev, A. A. Rempel, A. Magerl, and J., Disorder and
Order in Strongly Nonstoichiometric Compounds: Transition
Metal Carbides, Nitrides and Oxides, vol. 47, Springer Science
& Business Media, 2013.
[35] J. Vlˇcek, P. Kudl´aˇcek, K. Burcalov´a, and J. Musil, “High-power
pulsed sputtering using a magnetron with enhanced plasma
confinement,” Journal of Vacuum Science and Technology A:
Vacuum, Surfaces and Films, vol. 25, no. 1, pp. 42–47, 2007.
[36] L. Toth, Ed., Transition Metal Carbides and Nitrides, Elsevier,
2014.
[37] E. Lewin, B. Andr´e, S. Urbonaite, U. Wiklund, and U. Jansson,
“Synthesis, structure and properties of Ni-alloyed TiC 𝑥-based
thin films,” Journal of Materials Chemistry, vol. 20, no. 28, pp.
5950–5960, 2010.
[38] J. Robertson, “Diamond-like amorphous carbon,” Materials
Science and Engineering: R: Reports, vol. 37, pp. 129–282, 2002.
[39] E. Lewin, M. Rsander, M. Klintenberg, A. Bergman, O. Eriksson,
and U. Jansson, “Design of the lattice parameter of
embedded nanoparticles,” Chemical Physics Letters, vol. 496, no.
1–3, pp. 95–99, 2010.
[40] E. Lewin, O. Wilhelmsson, and U. Jansson, “Nanocomposite ncTiC/a-C
thin films for electrical contact applications,” Journal of
Applied Physics, vol. 100, no. 5, Article ID 054303, 2006.
[41] E. Lewin, M. Gorgoi, F. Sch¨afers, S. Svensson, and U. Jansson,
“Influence of sputter damage on the XPS analysis of metastable
nanocomposite coatings,” Surface and Coatings Technology, vol.
204, no. 4, pp. 455–462, 2009.
[42] D. Binder and W. J. Sturm, “Equivalence of X-ray lattice parameter
and density changes in neutron-irradiated LiF,” Physical
Review, vol. 96, no. 6, pp. 1519–1522, 1954.
[43] O. Wilhelmsson, S. Bijelovic, M. Lindquist et al., “Deposition
and characterization of magnetic Ti-Fe-C nanocomposite thin
films,” Thin Solid Films, vol. 518, no. 10, pp. 2607–2616, 2010.
[44] E. Lewin, K. Buchholt, J. Lu, L. Hultman, A. L. Spetz, and U.
Jansson, “Carbide and nanocomposite thin films in the Ti-Pt-C
system,” Thin Solid Films, vol. 518, no. 18, pp. 5104–5109, 2010.
[45] M. R˚asander, B. Sanyal, U. Jansson, and O. Eriksson, “A first
principles study of the stability and mobility of defects in
titanium carbide,” 2013, https://arxiv.org/abs/1303.2848.
[46] H. W. Hugosson, P. Korzhavyi, U. Jansson, B. Johansson, and
O. Eriksson, “Phase stabilities and structural relaxations in
substoichiometric TiC1−𝑥,” Physical Review B, vol. 63, no. 16,
2001.
[47] X.-X. Yu, G. B. Thompson, and C. R. Weinberger, “Influence
of carbon vacancy formation on the elastic constants and
hardening mechanisms in transition metal carbides,” Journal of
the European Ceramic Society, vol. 35, no. 1, pp. 95–103, 2015.
[48] J. Musil, “Hard nanocomposite coatings: thermal stability,
oxidation resistance and toughness,” Surface and Coatings
Technology, vol. 207, pp. 50–65, 2012.
[49] E. O. Hall, “The deformation and ageing of mild steel: III
discussion of results,” Proceedings of the Physical Society Section
B, vol. 64, no. 9, pp. 747–752, 1951.
[50] N. J. Petch, “The cleavage strength of polycrystals,” Journal of the
Iron and Steel Institute, vol. 174, pp. 25–28, 1953.
[51] S. Zhang, H. L. Wang, S.-E. Ong, D. Sun, and X. L. Bui, “Hard
yet tough nanocomposite coatings—present status and future
trends,” Plasma Processes and Polymers, vol. 4, no. 3, pp. 219–
228, 2007.
[52] S. Vepˇrek and S. Reiprich, “A concept for the design of novel
superhard coatings,” Journal of Thin Solid Films, vol. 268, no. 1-
2, pp. 64–71, 1995.
[53] J. Musil and J. Vlˇcek, “Magnetron sputtering of hard nanocomposite
coatings and their properties,” Surface and Coatings
Technology, vol. 142-144, pp. 557–566, 2001.
[54] J. Schiøtz, F. D. Di Tolla, and K. W. Jacobsen, “Softening of
nanocrystalline metals at very small grain sizes,” Nature, vol.
391, no. 6667, pp. 561–563, 1998.
[55] C. E. Carlton and P. J. Ferreira, “What is behind the inverse HallPetch
effect in nanocrystalline materials?” Acta Materialia, vol.
55, no. 11, pp. 3749–3756, 2007.
Submit your manuscripts at
https://www.hindawi.com
ScientiﬁcaHindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Corrosion
International Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Polymer Science
International Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Ceramics
Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Composites
Journal of
Nanoparticles
Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
International Journal of
Biomaterials
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Nanoscience
Journal of
TextilesHindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Journal of
NanotechnologyHindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Journal of
Crystallography
Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
The Scientific
World JournalHindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Coatings
Journal of
Advances in
Materials Science and Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Smart Materials
Research
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Metallurgy
Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
BioMed
Research International
Materials
Journal of
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014
Eur. Phys. J. Appl. Phys. (2016) 75: 24716 DOI: 10.1051/epjap/2016150591
On the study of the mechanical properties of Mo-B-C coatings
Luka´sˇ Za´bransky´, Vilma Bursˇı´kova´, Pavel Soucˇek, Petr Vasˇina, and Jirˇı´ Bursˇı´k
Eur. Phys. J. Appl. Phys. (2016) 75: 24716
DOI: 10.1051/epjap/2016150591
THE EUROPEAN
PHYSICAL JOURNAL
APPLIED PHYSICS
Regular Article
On the study of the mechanical properties of Mo-B-C coatings
Luk´aˇs Z´abransk´y1,a
, Vilma Burˇs´ıkov´a1
, Pavel Souˇcek1
, Petr Vaˇsina1
, and Jiˇr´ı Burˇs´ık2
1
Department of Physical Electronics, Faculty of Science, Masaryk University, Kotl´aˇrsk´a 2, 61137 Brno, Czech Republic
2
Institute of Physics of Materials, Academy of Sciences of the Czech Republic, ˇZiˇzkova 22, 61662 Brno, Czech Republic
Received: 1 December 2015 / Received in ﬁnal form: 14 April 2016 / Accepted: 21 April 2016
c EDP Sciences 2016
Abstract. Mo2BC thin ﬁlms show a favourable combination of high stiﬀness, hardness and elastic modulus
together with moderate ductility. In this study we focused on the comparison of mechanical properties of
Mo-B-C thin ﬁlms with diﬀerent structures (nanocrystalline or amorphous). The thin ﬁlms were deposited
on steel, hard metal and silicon substrates using DC magnetron sputtering. The mechanical properties of
Mo-B-C ﬁlms were studied using indentation techniques under both quasistatic and dynamic conditions
using a wide range of loads from 50 µN up to 1 N. The results showed that even amorphous Mo-B-C
thin ﬁlms had high hardness of 19.5 ± 0.5 GPa and elastic modulus of 276 ± 5 GPa. Their hardness is
comparable with the common amorphous diamond-like carbon coatings. Moreover, their fracture toughness
is signiﬁcantly higher. The results of mechanical tests were correlated with microstructure observations
carried out using scanning and transmission electron microscopy. The images of the deformed area under
the residual indentation imprints showed no cracking even after high loads or after indentation with sharp
cube corner indenter.
1 Introduction
Hard and wear resistant coatings employed for protection
of cutting tools exhibit high hardness and stiﬀness which
are often accompanied with brittle deformation behaviour.
Cracking can consequently lead to coating failure and
thus reduces the lifetime of the coated tool. It is therefore
favourable to combine high stiﬀness, hardness and elastic
modulus with moderate ductility to reduce crack initiation
and propagation. There are two criteria which are
generally used for determination if the material is brittle
or ductile. Pugh [1] showed that if the ratio of bulk
modulus to shear modulus B/G is greater than 1.75, the
material exhibits ductile metal-like behaviour. Otherwise,
the material is considered brittle. Pettifor [2] showed that
the value of the Cauchy pressure can also give information
about ductility. Cauchy pressure is determined as the
diﬀerence between elastic constants C12–C44. If it is negative,
material is brittle. Positive Cauchy pressure, on the
other hand, implies ductile behaviour.
Based on previous ab initio calculations [3,4], the
material Mo2BC has a desired combination of high Young
modulus of 470 GPa and moderate ductility with B/G
= 1.74 and Cauchy pressure of +43 GPa [3]. The Mo2BC
material belongs to nanolaminates exhibiting similar
a
e-mail: zerafel@mail.muni.cz
Contribution to the topical issue “6th Central European
Symposium on Plasma Chemistry (CESPC-6)”, edited by
Nicolas Gherardi, Ester Marotta and Cristina Paradisi
structure as the well known MAX phases [5,6] (M = transition
metal, A is a group A element and X is either boron
or carbon). Emmerlich et al. [3] prepared thin ﬁlms of
this material on Al2O3 substrate (temperature of 900 ◦
C)
using DC magnetron sputtering. The surface roughness
measured by scanning probe microscope (SPM) was in
the order of nm. The chemical composition was determined
by Rutheford backscattering spectroscopy (RBS)
and elastic recoil detection analysis (ERDA) as 49 at.%
Mo, 27 at.% B and 24 at.% C. The Young modulus of
460 ± 21 GPa and hardness of 29 ± 2 GPa were measured
by nanoindentation using the load of 5 mN. 80% of the
deformation energy was dissipated during plastic deformation.
The maximum depth was 290 nm and the pile-up
around residual imprints was 20–50 nm high.
Bolvardi et al. [4] simulated properties of M2BC phases
for M = Ti, V, Zr, Nb, Mo, Hf, Ta, W. The highest density
of valence electrons and the highest Young modulus
were obtained for M = Mo, W or Ta. In another paper
Bolvardi et al. [7] managed to prepare crystalline Mo2BC
thin ﬁlms using high power pulsed magnetron sputtering
(HPPMS) with synthesis temperature of 380 ◦
C while the
synthesis temperature in case of standard DC magnetron
sputtering was found to be 550 ◦
C. Because the crystallization
temperature of as deposited amorphous Mo2BC
powder was measured by diﬀerential scanning calorimetry
as 820 ◦
C, the ion bombardment induced surface diﬀusion
was considered as the main reason for the lower synthesis
temperature.
24716-p1
The European Physical Journal Applied Physics
Table 1. Summarized deposition conditions, “(p)” means that the power was pulsed with frequency 350 kHz and duty cycle
65%. Otherwise, the power was DC.
Sample name Power on a target [W] Bias voltage [V] Deposition time [min] Temperature [◦
C]
B4C C Mo Mo2BC
A1 – – – 250 (p) – 180 No heating
A2 – – – 250 (p) −200 420 No heating
C1 128 278 (p) 91 – −200 480 350
C2 128 250 (p) 110 – – 480 500
C3 104 281 84 – −200 300 No heating
In this work the mechanical properties of nanocrystalline
(partially crystallized) and amorphous Mo-B-C
thin ﬁlms were compared. The ﬁlms were deposited on
steel, hard metal and silicon substrates using DC magnetron
sputtering. Even the amorphous ﬁlms showed relatively
high hardness and elastic modulus which could be
compared to common diamond-like carbon (DLC) material
discussed see for example in [8]. The study of residual
indentation imprints for the loads up to 1 N revealed that
no cracking occurred even after the indentation depth
exceeded the ﬁlm thickness. This implies very good
fracture toughness of the studied material.
2 Experimental
Several samples were prepared using magnetron sputtering
of (i) a single Mo2BC target (Mo-B-C coatings marked
with A) or (ii) three targets: B4C, C and Mo (Mo-B-C
coatings marked with C) in order to better control the
stoichiometry. The hard-metal (cemented tungsten carbide,
WC-Co), high speed steel (HSS) and silicon substrates
were ultrasonically cleaned in a degreasing agent
and then placed in the chamber using load-lock system.
Prior to the deposition process all substrates were cleaned
in argon plasma for 20 min. The deposition pressure was
0.3 Pa. The deposition conditions are summarized in
Table 1. In case of HSS substrate an 80 nm thick molybdenum
interlayer was deposited prior the deposition of
Mo-B-C ﬁlm. The temperature on the substrate holder
for sample C1 and C2 was 350 ◦
C and 500 ◦
C, respectively.
The RF bias −200 V was used in some cases (see
Tab. 1). The thickness of all coatings was in the range
from 1 to 2 µm.
The hardness and elastic modulus were measured and
evaluated by depth sensing nanoindentation technique
performed on a Hysitron TI950 Triboindenter equipped
with a Berkovich tip. The tip diameter was less than
50 nm. The nanoscale measuring head with resolution of
1 nN and load noise ﬂoor less than 30 nN was used for
this study. Several testing modes were used in the range
of indentation loads from 0.1 to 10 mN, namely quasistatic
nanoindentation test, quasistatic nanoindentation with
several unloading segments and nanodynamic mechanical
analysis (nanoDMA) in the range from 0.1 to 300 Hz. The
quasistatic indentation tests were carried out in load controlled
regime using constant loading rate of 0.2 mN/s.
Both the partial unload regime and the nanodynamical
analysis were carried out in constant strain rate regime.
The pre-load of 2 µN was set to ensure good contact
between the tip and the surface. The indenter tip calibration
was carried out for low loads (indentation depths
<100 nm). The depth proﬁles of the hardness and elastic
modulus were carried out in the load range from 0.25 to
10 mN. The standard procedure proposed by Oliver and
Pharr [9] was used for the evaluation of the hardness and
elastic modulus.
Further, the cube-corner indenter with tip radius of
about 30 nm and with centerline-to-face angle of 34.3◦
was used to study the prepared ﬁlms. This very sharp
indenter is commonly used to produce well-deﬁned cracks
around indentation prints in order to estimate the fracture
toughness of thin ﬁlms at very small scales.
Moreover, several quasistatic indentation test were
performed using a Fisherscope H100 instrumented indentation
tester at maximum load of 1 N in order to study
the diﬀerential hardness [10,11] and the fracture resistance
of the coating/substrate system. Berkovich indenter with
tip radius of 200 nm was used for these tests. The load
increase was controlled according to function L = Ct2
,
where L is the load, t is the time and C = 2.5 mNs−2
.
The diﬀerential hardness was calculated as the derivative
of the indentation load according to the contact area.
The diﬀerential hardness dependence on the indentation
depth is used to visualize the defect creation in the tested
materials.
X-ray diﬀraction was used in order to determine
whether the material is crystalline, nanocrystalline or
amorphous. Measurements of X-ray diﬀraction were performed
on Rigaku Smartlab X-ray diﬀractometer with a
ﬁxed angle of incidence. The grazing incidence conﬁguration
was used. The details of microstructure in the vicinity
of indentation prints made with indentation load of
1 N were studied by means of electron microscopy. Thin
lamellas were prepared using a focused ion beam (FIB)
technique in a scanning electron microscope (SEM) LYRA
3 XMU FEG/SEM×FIB by Tescan. The lamellas were
further studied using transmission electron microscopes
(TEM) Philips CM12 STEM (thermoemission source,
accelerating voltage 120 kV) and JEOL JEM-2100F
(FEG, accelerating voltage 200 kV). The surface morphology
was analyzed by an atomic force microscope Ntegra
Prima NT-MDT in a semi-contact mode. The area of
10 × 10 µm was studied using a 20 µm/s scan rate.
24716-p2
L. Z´abransk´y et al.: On the study of the mechanical properties of Mo-B-C coatings
20 30 40 50 60 70 80
0
1
2
3
C3
A1
A2
C1
C2
Ref [7]
Intensity[arb.u.]
2theta [°]
Fig. 1. The diﬀraction patterns of ﬁve ﬁlms compared with
Ref. [7].
3 Results and discussion
The microstructure of Mo-B-C ﬁlms deposited on WC
substrates was studied by both, X-ray diﬀraction and
TEM. The diﬀractograms presented in Figure 1 are
compared with results obtained by Bolvardi et al. [7] on
Mo2BC ﬁlm prepared using high power pulsed magnetron
sputtering (HPPMS) at substrate temperature of 480 ◦
C.
According to [7], the appearance of two distinctive peaks
at around 36◦
and 42◦
indicates the crystalline character
of the ﬁlms. The peak around 42◦
becomes stronger for
ﬁlms with increasing crystalline content in Mo-B-C ﬁlms
and it originates mainly from Mo2BC (1 1 1) signal, however,
it may be overlapped by Mo2BC (0 8 0). The wide
band at around 36◦
is an overlap of several peaks instead
of signal arisen from amorphous Mo-B-C matrix.
The diﬀractogram presented at the top in Figure 1 shows
only a wide band at around 36◦
suggesting the formation
of an amorphous structure for coating marked as A1.
The diﬀractogram of sample A2 contains one wide band
situated between 36◦
and 42◦
which may represent some
transition between amorphous and nanocrystalline state.
The combined Rutheford backscattering spectroscopy
(RBS) and the elastic recoil detection analysis (ERDA)
showed that all coatings prepared using sputtering of one
Mo2BC target were far from the stoichiometric composition
(the atomic percentage of Mo diﬀered from the stoichiometric
one by more than 10%). The sample group C
was prepared using sputtering of three targets (see Tab. 1).
What dominates in the diﬀractograms of this group is a
more or less asymmetric broad peak between 36◦
and 42◦
,
which arises from an overlap of several peaks including
an amorphous band. This peak suggests partially crystalline
(nanocomposite) structure for samples C1, C2 and
C3, which was also proven by TEM study. Moreover, the
diﬀractogram of sample C3 exhibits a signiﬁcant increase
in Mo2BC (1 1 1) signal indicating substantial increase in
volume fraction of Mo2BC crystallites. The sputtering of
three targets (see Tab. 1) enabled a better control of the
ﬁlm stoichiometry, therefore the C samples prepared
using this method exhibit a partial crystallinity even
without any sample heating as it is proven in case of
sample C3.
The microstructure of amorphous coating A1 deposited
on hard metal substrate is depicted on TEM micrographs
in Figure 2. A wavy coating/substrate interface
and heavily deformed grains of hard metal, but no cracks
or marks of adhesive failure are observed (Fig. 2a). Both
the high resolution image and the selected area diﬀraction
(SAD) pattern (Fig. 2b) conﬁrm the amorphous character
of the deposited layer.
The microstructure of nanocrystalline coating C3
deposited on an HSS substrate is shown in Figure 3. The
thin lamella prepared by FIB in SEM (Fig. 3a) represents
the central vertical section of an indentation print. Global
TEM view (Fig. 3b) shows the deformed structure free of
any cracks and interfacial failures. A detailed view (Fig. 3c
and a high resolution image in Fig. 3d) shows nanosized
columnar grains with amorphous material in between.
(a) (b)
Fig. 2. The microstructure of amorphous coating A1 deposited on hard metal substrate. A global TEM view of the volume
under indentation print (a) and a high resolution TEM micrograph of WC grain at the bottom part and an amorphous layer
above with SAD pattern (b).
24716-p3
The European Physical Journal Applied Physics
(a) (b)
(c) (d)
Fig. 3. The microstructure of the coating C3 deposited on an HSS substrate. The ﬁnal stage of lamella preparation using a
FIB in SEM, signal of backscattered electrons (a), a global TEM view of the volume under indentation print (b), a detail of the
columnar microstructure with inserted SAD pattern (c) and a high resolution TEM micrograph (d).
0 20 40 60 80 100 120 140 160
250
275
300
325
350
375
400
425
indentation depth [nm]
elasticmodulus[GPa]
A1
A2
C1
C2
0 20 40 60 80 100 120 140 160
20
25
30
35
40
45
hardness[GPa]
indentation depth [nm]
A1
A2
C1
C2
(a) (b)
Fig. 4. The evolution of (a) hardness and (b) elastic modulus as functions of maximum indentation depth. The behaviour of these
parameters diﬀered with depth for amorphous and nanocrystalline coatings due to diﬀerent intrinsic stress and microstructure.
The SAD pattern in Figure 3c conﬁrms the Mo2BC
orthorhombic lattice (ICSD entry no. 043318, [12]). The
diﬀraction patterns are assigned to (0 4 1), (1 1 1), (2 0 0),
(0 0 2) and (2 4 1) planes of the Mo2BC structure. The
structure of SAD pattern indicates the presence of preferential
orientation for Mo2BC nanocrystallites.
The evolution of hardness and elastic modulus for all
four samples deposited on hardmetal substrates is
presented in Figure 4 as a function of maximum indentation
depth. The error of these measurements was
always <10%. Because the thickness of the coatings was
larger than 1.0 µm, the values of hardness and elastic modulus
taken from depths < 100 nm should not be
inﬂuenced by the substrate. However, in case of partially
crystalline coatings C1 and C2 there is a signiﬁcant
increase in the hardness as well as elastic modulus at
24716-p4
L. Z´abransk´y et al.: On the study of the mechanical properties of Mo-B-C coatings
Table 2. Calculated hardness H, elastic modulus E and their
ratios H/E and H3
/E2
for all ﬁve ﬁlms.
Sample H [GPa] E [GPa] H/E H3
/E2
[GPa]
name
A1 19.4 ± 0.5 272 ± 5 0.071 0.099
A2 19.6 ± 0.5 280 ± 5 0.070 0.096
C1 27.2 ± 0.9 313 ± 11 0.087 0.205
C2 31.6 ± 0.8 345 ± 9 0.091 0.259
C3 20.6 ± 2.0 275 ± 16 0.075 0.116
the sample surface. The tip was carefully calibrated in
the region of low indentation depths (between 20 nm and
100 nm) on certiﬁcated fused silica standard, therefore
this increase does not arise from artefacts connected with
imperfect indenter tip. The increased surface hardness
compared to the bulk part of samples C1 and C2 is probably
caused by internal stress evolution during the growth
and due to nonhomogeneous microstucture through depth
proﬁle of the nanocomposite ﬁlms. For comparison of the
material parameters in Table 2, the values from greater
depths were taken. The optimum depth for bulk hardness
and elastic modulus evaluation was considered for the
maximum indentation depth of 100 nm. This indentation
depth is less than 10% of the ﬁlm thickness.
The measured and calculated hardness H, elastic modulus
E and their ratios H/E (measure of toughness) and
H3
/E2
(resilience) are presented in Table 2. The highest
hardness of 31.6 ± 0.8 GPa was found for nanocrystalline
sample C1, the highest elastic modulus was 345 ± 9 GPa
for the same sample. The average hardness for amorphous
samples was 19.5 ± 0.5 GPa and elastic modulus 276 ±
5 GPa.
The cube corner indenter was used and several
imprints with maximum load of 10 mN were performed in
order to study the crack resistance. The load-displacement
curves revealed no sudden pop-in events in any of the
samples. There was no cracking induced by the indentation
up to 10 mN which indicated very good crack resistance
[3]. In Figure 5a, there is a typical load-displacement
0 500 1000 1500 2000
4
6
8
10
12 sample C3
Load: 1 N
differentialhardness[GPa]
indentation depth [nm]
Fig. 6. The diﬀerential hardness dependence on the indentation
depth for sample C3. The sudden diﬀerential hardness
decrease from indentation depth of about 0.25 µm is
caused by the expansion of the deformation zone into the substrate.
The indentation response of the studied sample is fully
controlled by plastic deformation of the HSS substrate from
indentation depths of about 0.5 µm. Although the maximum
indentation depth exceeded the coating thickness, indentation
induced cracking was not observed.
curve, here for sample C1. Several measurements of differential
hardness for sample C1 are plotted in Figure 5b.
Diﬀerential hardness is calculated numerically from the
loading curves for the maximum load of 1 N and it
describes the instantaneous resistance of the material to
deformation. It is important to state that the initial part
(about ﬁrst 200 nm) of the diﬀerential hardness curves
strongly depends on the local structure, and is therefore
very sensitive to any inhomogeneities such as grain boundaries.
No large sudden drops in evolution of diﬀerential
hardness also implied no cracking events, even for the load
of 1 N. Here the maximum indentation depth exceeded
the coating thickness and indentation induced plastic
0 100 200 300 400
0
2
4
6
8
10
load[mN]
indentation depth [nm]
sample C1
0 500 1000 1500
8
9
10
11
12
13
14
15
sample C1
Load: 1 N
differentialhardness[GPa]
indentation depth [nm]
(a) (b)
Fig. 5. (a) The example of load/displacement curve for maximum load of 10 mN performed using a cube corner indenter, and
(b) three curves of diﬀerential hardness dependence on the indentation depth. No pop-in events in load-displacement curve or
sudden drops in diﬀerential hardness indicated no cracking inside. The ﬁlm C1 was deposited on hard metal substrate (WC).
24716-p5
The European Physical Journal Applied Physics
(a) (b)
Fig. 7. The AFM images of (a) surface of the sample C2 and (b) uncoated hardmetal substrate.
deformation was controlled by the plastic behaviour of
the substrate.
The hardness of amorphous Mo-B-C coatings is comparable
with the hardness of most often used amorphous
diamond-like carbon coatings [8,13–16]. However, the elastic
modulus is signiﬁcantly higher. Moreover, the crack
resistance of the coatings is also substantially enhanced.
Even though the maximum indentation depth exceeded
the thickness of the coating, only plastic deformation of
the coating and the substrate occurred. There were no
pop-in events on the loading curve suggesting that no
cracks occurred during the indentation (see Fig. 5a). This
was also conﬁrmed by the TEM results (see Figs. 2a, 3a
and 3b) as well as by depth proﬁles of diﬀerential hardness
(see Figs. 5b and 6) where sudden drops indicating
cracking were not detected. The sudden diﬀerential
hardness decrease from indentation depth around 0.25 µm
was caused by the expansion of the deformation zone into
the softer HSS substrate. The indentation response of the
studied sample was fully controlled by plastic deformation
of the HSS substrate from indentation depths around
0.5 µm. The hardness and elastic modulus of nanocrystalline
Mo-B-C coatings are comparable with common
hard coatings used for protective applications [17,18], for
example nc-TiC/a-C:H [11,19–24]. However, fracture
toughness is generally better for Mo-B-C coatings, especially
when compared to nc-TiC/a-C:H that contains more
carbon. The nc-TiC/a-C:H coatings with higher titanium
content have excellent fracture toughness but, on the other
hand, their hardness is generally lower [11,21]. The excellent
resistance against brittle failure during tensile testing
of Mo2BC ﬁlm on copper substrate was also reported [25].
The surface was studied by atomic force microscopy
and the results for the nanocrystalline sample C2 are
depicted in Figure 7a. The surface of uncoated polished
and plasma cleaned hardmetal is in Figure 7b. From
Figure 7b it is clear that the uncoated hardmetal contains
deeper holes. The peak-to-valley height of hardmetal
was 81 nm and the average roughness Ra = 5.1 nm.
On the other hand, the calculated peak-to-valley height
of the sample C2 was smaller and equaled approximately
36 nm. The average surface roughness of sample C2 was
Ra = 2.1 nm. The coating obviously ﬁlled the holes during
its growth and generally smoothed the surface.
4 Conclusion
Magnetron sputtering process was used for the preparation
of ﬁve Mo-B-C coatings with diﬀerent microstructures.
The correct methodology for hardness and elastic
modulus evaluation was developed. Diﬀerent evolutions
of hardness and elastic modulus depending on maximum
indentation depth for amorphous and nanocrystalline
samples were carried out. This was caused by diﬀerent
inner stress evolution during growth and by diﬀerent
inner structures. The Mo-B-C coating showed very favorable
combination of high hardness and elastic modulus
together with high fracture toughness. The average hardness
of Mo-B-C coatings with amorphous structure was
found as 19.5 GPa and their elastic modulus was around
276 GPa. The nanocrystalline Mo-B-C coatings have
approximately the same hardness and elastic modulus as
the typical nc-TiC/a-C:H coatings. However, the fracture
toughness is generally higher for Mo-B-C ﬁlms. The differential
hardness curves and TEM images from the area
under the indentation imprint proved that there was no
crack formation in the vicinity of the imprints. Also the
adhesion after indentation up to 1 N remained suﬃcient
and no delamination occurred. The surface of the
samples was studied using AFM. From the presented
images and calculated surface roughness it was proven
that the Mo-B-C coating smoothed the surface, surface
roughness decreased from 5.1 nm for uncoated hardmetal
to 2.1 nm.
This research has been supported by the project CZ.1.05/
2.1.00/03.0086 funded by European Regional Development
24716-p6
L. Z´abransk´y et al.: On the study of the mechanical properties of Mo-B-C coatings
Fund and project LO1411 (NPU I) funded by Ministry of
Education Youth and Sports of Czech Republic and The Czech
Science Foundation (Project 15-17875S).
References
1. S.F. Pugh, Phil. Mag. 45, 823 (1954)
2. D.G. Pettifor, Mater. Sci. Technol. 8, 345 (1992)
3. J. Emmerlich, D. Music, M. Braun, P. Fayek, F. Munnik,
J.M. Schneider, J. Phys. D: Appl. Phys. 42, 185406
(2009)
4. H. Bolvardi, J. Emmerlich, M. to Baben, D. Music, J. von
Appen, R. Dronskowski, J.M. Schneider, J. Phys.: Condens.
Matter 25, 045501 (2013)
5. D. Music, J.M. Schneider, JOM 59, 60 (2007)
6. M.W. Barsoum, T. El-Raghy, Am. Sci. 89, 334 (2001)
7. H. Bolvardi, J. Emmerlich, S. Mr´az, M. Arndt,
H. Rudigier, J.M. Schneider, Thin Solid Films 542, 5
(2013)
8. J. Robertson, Mat. Sci. Eng. R 37, 129 (2002)
9. W.C. Oliver, G.M. Pharr, J. Mater. Res. 19, 3 (2004)
10. B. Wolf, Phil. Mag. Phil. Mag. Lett. 86, 5251 (2006)
11. L. Z´abransk´y, V. Burˇs´ıkov´a, J. Daniel, P. Souˇcek,
P. Vaˇsina, J. Dug´aˇcek, P. St’ahel, O. Caha, J. Burˇs´ık,
V. Peˇrina, Surf. Coat. Technol. 267, 32 (2015)
12. G. Bergerhoﬀ, I.D. Brown, in “Crystallographic Databases”,
edited by F.H. Allen et al. (International Union
of Crystallography, Chester, England, 1987)
13. A. Grill, Diam. Relat. Mater. 8, 428 (1999)
14. D. Franta, V. Burˇs´ıkov´a, I. Ohl´ıdal, P. St’ahel, M. Ohl´ıdal,
D. Neˇcas, Diam. Relat. Mater. 16, 1331 (2007)
15. S. Neuville, A. Matthews, Thin Solid Films 515, 6619
(2007)
16. S. Neuville, Surf. Coat. Technol. 206, 703 (2011)
17. A.A. Voevodin et al., in Nanostructured Thin Films
and Nanodispersion Strengthened Coatings, edited by
A.A.Voevodin, D.V. Shtansky, E.A. Levashov, J.J. Moore
(NATO Science Series II. Vol. 155, Kluwer Academic
Sciences, Dordrecht, 2004)
18. J. Musil, Surf. Coat. Technol. 207, 50 (2012)
19. A. Czy ˙zniewski, W. Prechtl, J. Mater. Proc. Technol.
157–158, 274 (2004)
20. P. Souˇcek, T. Schmidtov´a, L. Z´abransk´y, V. Burˇs´ıkov´a,
P. Vaˇsina, O. Caha, M. J´ılek, A.E. Mel, P.Y. Tessier,
J. Sch¨afer, J. Burˇs´ık, V. Peˇrina, R. Mikˇsov´a, Surf. Coat.
Technol. 211, 111 (2012)
21. L. Z´abransk´y, V. Burˇs´ıkov´a, P. Souˇcek, P. Vaˇsina,
T. Gardelka, P. St’ahel, O. Caha, V. Peˇrina, J. Burˇs´ık,
Surf. Coat. Technol. 242, 62 (2014)
22. Y.T. Pei, D. Galvan, J.Th.M. De Hosson, C. Strondl,
J. Eur. Ceram. Soc. 26, 565 (2006)
23. Y.T. Pei, D. Galvan, J.Th.M. De Hosson, A. Cavaleiro,
Surf. Coat. Technol. 198, 44 (2005)
24. P. Vaˇsina, P. Souˇcek, T. Schmidtov´a, M. Eli´aˇs,
V. Burˇs´ıkov´a, M. J´ılek, M. J´ılek Jr., J. Sch¨afer, J. Burˇs´ık,
Surf. Coat. Technol. 205, s53 (2011)
25. S. Djaziri, S. Gleich, H. Bolvardi, C. Kirchlechner,
M. Hans, C. Scheu, J.M. Schneider, G. Dehm, Surf. Coat.
Technol. 289, 213 (2016)
24716-p7
Thermal stability of hard nanocomposite Mo-B-C coatings
L. Zabranský a, *
, V. Bursíkova a
, P. Soucek a
, P. Vasina a
, J. Dugacek a
, P. S
0
tahel a
, J. Bursík b
,
M. Svoboda b
, V. Perina c
a
Department of Physical Electronics, Faculty of Science, Masaryk University, Kotlarska 2, Brno, CZe61137, Czech Republic
b
Institute of Physics of Materials, Academy of Sciences of the Czech Republic, Zizkova 22, Brno, CZe61662, Czech Republic
c
Nuclear Physics Institute, Academy of Sciences of the Czech Republic, v.v.i., Rez 130, CZe25068, Czech Republic
a r t i c l e i n f o
Article history:
Received 15 July 2016
Received in revised form
1 December 2016
Accepted 13 December 2016
Available online 14 December 2016
Keywords:
Thermal stability
Mo2BC coatings
Hardness
Fracture resistance
a b s t r a c t
In the present work, nanocomposite Mo-B-C coatings were deposited on high speed steel and hard metal
substrates by magnetron sputtering of three targets. These coatings were subjected to annealing to ﬁnal
temperatures in the range from 500 
C to 1000 
C. It was found that the as deposited Mo-B-C coatings
exhibited hardness of ~20 GPa, nanocomposite microstructure with very ﬁne grains (~2 nm) and low
degree of crystallinity. The X-ray diffraction and transmission electron microscopy together with selective
area electron diffraction were used to study the temperature induced changes of the microstructure
of the coating and its crystallinity. The annealing process signiﬁcantly improved the hardness
(from ~20 GPa to ~30 GPa) and effective elastic modulus (from initial 330 GPae500 GPa) of coatings
while their resistance to fracture was kept sufﬁciently high.
© 2016 Elsevier Ltd. All rights reserved.
1. Introduction
Recently, there has been an increased interest in boron and
carbon based nanolaminate coatings such as Mo2BC with similar
structure to the so-called MAX phases [1,2] (M ¼ transition metal, A
is a group A element and X is either boron or carbon). The Mo2BC
material exhibits orthorhombic structure with the Cmcm space
group (unit cell with large ratio of b/a ¼ 5.6) and consists of facesharing
Mo6B trigonal prismatic and Mo6C octahedral (cubic rocksalt
coordination) building blocks. Previous ab initio calculations
[3,4] and subsequent experimental work [3,5,6] proved that Mo2BC
coatings exhibit a highly demanded combination of high hardness
and Young's modulus which is accompanied by moderate ductility.
These properties enable to reduce the crack initiation and propagation
in the coating. Emmerlich et al. [3] synthesized crystalline
Mo2BC thin ﬁlms on Al2O3 substrate using DC (direct current)
magnetron sputtering with the deposition temperature of 900 C.
The Young's modulus of 460 ± 21 GPa and hardness of 29 ± 2 GPa
were measured by nanoindentation using a load of 5 mN. Bolvardi
et al. [7] managed to prepare crystalline Mo2BC thin ﬁlms using
high power pulsed magnetron sputtering (HPPMS) with synthesis
temperature of 380 C while the synthesis temperature in case of
standard DC magnetron sputtering was found to be at least 550 C.
In our previous papers [5,8], we reported on a low temperature
deposition of Mo-B-C coatings using HPPMS and DC magnetron
sputtering of either one Mo2BC target or three targets: B4C, C and
Mo with or without additional substrate heating. In case of Mo2BC
target it was difﬁcult to control the stoichiometry of the deposited
samples, most of the coatings prepared by employing only the
Mo2BC target showed amorphous microstructure. Using sputtering
of three targets we were able to control better the stoichiometry
and the related microstructure of growing coatings. In general, the
synthesized coatings exhibited different mechanical properties:
their hardness ranged from ~19 GPa for amorphous coatings to
~32 GPa for nanocomposite coatings. According to our previous
TEM study [5], the nanocomposite coatings consisted of nanosized
Mo2BC grains (2e10 nm) embedded in amorphous matrix. The as
deposited coatings showed very good resistance against fracture
due to their moderate ductility, predicted and calculated in previous
studies [3,4]. It was concluded that these coatings have potential
application as protective coatings for cutting tools.
However, for cutting tool applications the hardness and toughness
are not the only important characteristics. The tools protected
by hard coatings often operate at elevated temperatures and thus
their thermal stability is also crucial. In general, the thermal activation
can lead to the enhancement of diffusion, e.g. outward
diffusion of hydrogen in case of hydrogenated DLC coatings [9e12]
or inter-diffusion at the interface [13], changes in size, shape or
* Corresponding author.
E-mail address: zerafel@mail.muni.cz (L. Zabranský).
Contents lists available at ScienceDirect
Vacuum
journal homepage: www.elsevier.com/locate/vacuum
http://dx.doi.org/10.1016/j.vacuum.2016.12.016
0042-207X/© 2016 Elsevier Ltd. All rights reserved.
Vacuum 138 (2017) 199e204
composition of grains such as recrystallization, segregation or grain
growth processes [14,15], phase transitions [16] or stress relaxations
[17]. These undesired effects often act as limiting factors
in applications of protective coatings and, moreover, the mismatched
coefﬁcients of thermal expansion (CTE) of the substrate
and the coating can often lead to adhesive failure. On the other
hand, the thermal treatment can also improve some of the mechanical
properties, e.g. age hardening of TiAlN due to phase
decomposition [16].
Although Mo2BC material in its bulk form is known from 1960s
[18], only few research papers have been published so far in the
area of thermal stability of Mo-B-C coatings. Bolvardi et al. [7]
discussed the thermal annealing of amorphous Mo2BC powder
which was studied by differential scanning calorimetry and X-ray
diffraction. It was shown that from 820 C the bulk diffusion
induced crystallization occurred which resulted in several distinguished
Mo2BC and Mo2C diffraction peaks. However, the crystallization
temperature was signiﬁcantly lower for HPPMS or DC
sputtering where the surface diffusion induced by ion bombardment
was reported to be the main reason for this decrease of the
crystallization temperature.
As it was shown by Bolvardi et al. [7], the annealing of the
Mo2BC powder resulted in the increase of the degree of crystallization.
One of the goals of the present work was to determine
whether the crystallization mechanism is performed in the same
way in the case of thin Mo-B-C coatings. In order to study the inﬂuence
of the annealing temperature on the degree of crystallization
and the subsequent changes in the mechanical properties,
several Mo-B-C coatings were prepared using HPPMS and DC
magnetron sputtering. The deposition conditions were chosen in
order to prepare coatings with similar initial properties and microstructures
as in our previous study [5], i.e. coatings with nanocomposite
microstructure, lower degree of crystallinity and
hardness of ~20 GPa. In order to study the thermally induced
changes the samples were measured using nanoindentation,
scanning electron microscopy (SEM) to study the potential fracture
around the residual imprints made by nanoindentation and GI-XRD
to study the crystallinity changes after the annealing process.
2. Experimental
All discussed Mo-B-C thin ﬁlms were prepared using magnetron
co-sputtering of three targets: B4C, Mo and C using sputtering
system with a cylindrical chamber ø 50 cm Â 50 cm. The hard metal
(cemented tungsten carbide WC-Co, with [Co] ¼ 10 at.%) and high
speed steel (HSS) were ultrasonically cleaned in a degreasing agent
and then placed in the chamber using a load-lock system. Prior to
the deposition process all substrates were cleaned in argon plasma
(pressure 0.3 Pa) for 20 min, here the radiofrequency (RF,
13.56 MHz) power of 100 W was applied on the substrate holder.
After the cleaning of the substrate, the deposition commenced. The
biasable substrate holder was located 28 cm below the targets and
its rotation was set to 5 rpm during the whole deposition process in
order to ensure good homogeneity of the growing ﬁlms. The
deposition conditions were: 128 W DC power on the B4C target,
110 W DC on the Mo target and pulsed power of 250 W with
repetition frequency of 350 kHz and off-time of 1 ms (duty cycle of
45%) applied on the C target. The deposition pressure was 0.1 Pa and
the duration 480 min. Neither additional external heating on the
substrate nor bias voltage were used during the deposition of the
studied coatings as well as no interlayer enhancing the adhesion to
above mentioned substrates was introduced.
The thermal stability of the ﬁlms was investigated as a function
of thermal annealing under UHV conditions. The furnace chamber
was evacuated to the base pressure of about 10À5
Pa. The ﬁlms were
annealed in the resistively heated laboratory furnace Classic Clare
4.0 to several different ﬁnal temperatures depending on the substrate
material. The studied samples were subjected to annealing
with constant heating rate of 10 C/min. The coatings on hard metal
substrate were annealed up to 1000 C and in case of the HSS
substrate, because of its lower thermal stability [15], the coatings
underwent annealing up to 500 C, 600 C and 700 C. After
achieving the desired temperature, it was kept constant for 30 min
in case of all experiments. Then the samples cooled down in vacuum
for approximately 12 h. Afterwards, the nanoindentation
measurements, X-ray diffraction, SEM and TEM studies were performed
in order to determine changes in microstructure and mechanical
properties.
The hardness and reduced elastic modulus were measured and
evaluated using a depth sensing indentation method performed on
a Hysitron TI950 Triboindenter equipped with a Berkovich tip. The
mounted nanoscale measuring head with resolution of 1 nN and
load noise ﬂoor lower than 30 nN allow to measure in the load
range from 50 mN to 11 mN. Several testing methods were used, e.g.
classic quasistatic test or quasistatic test with several partially
unloading segments. The quasistatic indentation tests were carried
out in load controlled regime using a constant loading rate of
0.2 mN/s. The indenter tip calibration was carried out for low loads
(indentation depths < 100 nm). For the evaluation of the measured
data, the standard procedure proposed by Oliver and Pharr [19] was
used. The differential hardness Hdif [20,21], calculated as Hdif ¼ kvL/
v(h2
), where L is the applied load, k is a constant depending on the
indenter and h is the indentation depth, was measured using
Fischerscope H100 microindentation tester equipped with a Berkovich
tip. The quasistatic test with the load of 1 N was used in
order to determine the loads or depths where the substrate starts to
inﬂuence the measurement of the hardness or to detect cracking
possibly emerging during deformation. The residual imprints after
microindentation tests with the Fischerscope H100 and the surface
of coatings as deposited and after annealing were analyzed on
scanning electron microscope MIRA 3 FEG SEM with the accelerating
voltage of 15 kV.
The coating microstructure was studied by means of electron
microscopy using a Tescan LYRA 3XMU SEM Â FIB scanning electron
microscope (SEM), a Philips CM12 STEM transmission electron
microscope (TEM) operating at 120 kV and a JEOL 2100F high resolution
TEM. Thin lamellar cross sections for TEM observations
were prepared using a focussed ion beam (FIB) in SEM.
Scratch tests were carried out with a Revetest Scratch Tester
(Anton Paar) equipped with a 200 mm diamond Rockwell indenter.
A progressive scratch of load with loading rate of 15 N/min from 1 N
to 16 N on a 8 mm path was applied to all samples. X-ray diffraction
was used in order to determine the inner structure and level of
crystallinity of coatings. The measurements were performed on
Rigaku Smartlab X-ray diffractometer with a grazing angle of incidence
conﬁguration with an incidence angle of 0.5. The Rutheford
backscattering spectroscopy (RBS) using ion beam provided by the
tandem accelerator Tandetron MC 4130 was applied to determine
the atomic composition of the studied samples. For this purpose,
the proton projectiles with energies of 1740 (higher sensitivity to C
content) and 2700 keV (higher sensitivity to B content) impinging
vertically on the samples were utilized. The backscattered protons
were analyzed at the scattering angle of 170.
3. Results and discussion
The atomic composition of Mo-B-C coatings on carbon substrates
was determined using RBS. The deposition conditions
resulting in coatings with composition near to stoichiomeric
Mo2BC ([Mo] ¼ 49.9 at.%, [B] ¼ 23.2 at.%, [C] ¼ 23.2 at.%,
L. Zabranský et al. / Vacuum 138 (2017) 199e204200
[Ar] ¼ 3.3 at.%, [O] ¼ 0.4 at.%) were chosen for preparation of the set
of coatings for the thermal stability studies. The RBS data obtained
on Mo-B-C coating deposited on steel substrate showed of
approximately 10% higher amount of Mo and of 10% lower amount
of carbon as in the case of coating on carbon substrate, however, in
case of steel substrate the sensitivity of B and C estimation is limited
by overlapping their signals in spectra by signal of the steel
substrate.
The inﬂuence of the annealing temperature on the degree of
crystallization of the selected Mo-B-C coating was studied using Xray
diffraction technique. If we compare the X-ray diffractograms
for Mo2BC coatings as deposited and annealed to 500 C, 600 C,
700 C, 900 C and 1000 C, one can see that there is no indication
for any microstructural change (see Fig. 1) up to 900 C. All
diffractograms up to 900 C are almost identical which means that
the thermally induced microstructural changes are very ﬁne and
subtle. However, the diffraction region between 35 and 45 consists
of many possible diffraction peaks from Mo2BC and Mo2C phases.
Taking into account the fact that the coating is composed of very
ﬁne grains of nanometer size, the particular diffraction peaks could
be wide and, therefore, their signals can be overlapped by each
other forming such broad band. This broad band can hinder the
observation of any small scale microstructural changes. This effect
was also observed in our previous work [5] where the studied
coatings also exhibited the ﬁne-grained microstructure. Signiﬁcant
change in the diffractogram occurred after annealing to 1000 C
where several sharp peaks emerged. The main cause of their origin
is the presence of the b-Mo2C phase with additional contribution of
the Mo2BC phase. The b-Mo2C peaks are sharp suggesting that a
pronounced grain growth occurred.
A Philips CM12 STEM transmission electron microscope (TEM)
was used to study the microstructure of as deposited coating and
the coating annealed to 1000 C. Although the TEM bright ﬁeld
image showed no contrast, the dark ﬁeld image (Fig. 2a) reveals
small crystalline objects ~2 nm in size. The respective selected area
electron diffraction (SAED) pattern indicates a nanocrystalline
material. Diffuse diffraction rings may point to Mo2C structure in an
amorphous MoBC matrix. The nanocrystalline character of the
annealed coating is apparent in a TEM bright ﬁeld image (Fig. 2b).
The layer consists of nanocrystals up to 10 nm in size. SAED pattern
with several diffuse rings and a week texture indicates Mo2BC
structure.
The as deposited Mo2BC coatings on HSS and hard metal substrates
exhibited a relatively high hardness of 20.5 ± 0.3 GPa,
comparable with the hardness of commonly used amorphous
diamond-like carbon protective coatings [9e12]. Their effective
elastic modulus Eeff, expressed as Eeff ¼ E/(1-n2
) where E is Young's
modulus of the coating and n is Poisson's ratio of the coating, was
330 ± 5 GPa. The ﬁlm thickness was around 1.5 mm.
The coatings on HSS substrates could withstand annealing up to
600 C without any signiﬁcant change in their adhesion. However,
Fig. 1. The comparison of diffractograms for Mo2BC coatings as deposited and
annealed to 500 C, 600 C, 700 C and 1000 C.
Fig. 2. TEM micrographs and respective SAED patterns: a) as deposited sample, dark ﬁeld image and SAED pattern with rings of Mo2C lattice; b) sample annealed to 1000 C, bright
ﬁeld image and SAED pattern with rings of Mo2BC lattice.
L. Zabranský et al. / Vacuum 138 (2017) 199e204 201
during annealing to 700 C the coating delaminated. On the other
hand, the adhesion of all coatings on hard metal substrates
remained sufﬁcient even after annealing to 1000 C. The reason for
this dissimilarity lies in different coefﬁcients of thermal expansion
for HSS (CTE of ~13 mm/mK) and hard metal (CTE of ~6 mm/mK)
substrates and the coating (estimated as ~ 8 mm/mK, based on
known CTE of Mo2C phase). If we assume that the intrinsic stress
from the deposition process is relaxed during the annealing process,
the estimated thermal stresses induced after cooling down
from 700 C are À1.4 GPa (compressive) in case of the coating on
HSS substrate and 0.6 GPa (tensile) in case of the coating on hard
metal.
The depth proﬁles of hardness and effective elastic modulus
obtained for as-deposited and annealed coatings on HSS and hard
metal substrates are presented in Fig. 3 and Fig. 4, respectively. Each
point in the graphs represents an averaged value from 16 measurements
performed by the quasistatic indentation test with 20
unloading segments in a load range from 0.5 to 11 mN. The error of
these measurements is <8%. In order to eliminate the size effect and
the inﬂuence of the indenter tip calibration for very low loads
(<20 nm), the hardness and elastic modulus data were taken into
account from indentation depths > 60 nm. As it can be seen, from
60 nm up to 160 nm both parameters are constant in the frame of
the experimental errors indicating both, that the coatings are homogeneous
and that the inﬂuence of the substrate on the measured
data is negligible up to indentation depths of 160 nm. There is an
increase of the hardness and effective elastic modulus measured
after annealing to 500 C, and in case of samples on HSS the
hardness increased from the former 20.5 ± 0.3 GPa to 23.5 ± 0.5 GPa
(see Fig. 3a) and the effective elastic modulus increased from
330 ± 5 GPa to 390 ± 7 GPa (see Fig. 4a). Annealing to 600 C did not
improve the hardness or, the effective elastic modulus signiﬁcantly
and all changes are within the experimental error from Figs. 3b and
4b we can see that similar behaviour is achieved also for annealed
coatings on hard metal, however, here the increase of the hardness
and effective elastic modulus is slightly smaller. The reason for this
discrepancy is the different nature of thermal stress that arose from
different CTEs of both substrates. The hardness improved to
22.0 ± 0.4 GPa and remained approximately constant after further
annealing up to 800 C. Although the crystallization temperature
[7] had not been achieved, the increase of both mechanical properties
upon annealing up to 800 C is evident and appropriate
discussion concerning several possible factors inﬂuencing the
mechanical properties is needed and provided in paragraphs below.
Further annealing of the Mo2BC coating on hard metal substrate to
ﬁnal temperature of 1000 C revealed considerable increase of
hardness to 29.4 ± 0.6 GPa and the effective elastic modulus to
500 GPa. Here the temperature was high enough to induce the
formation and growth of Mo2BC crystallites. Moreover, the acquired
values of the hardness and the effective elastic modulus are typical
for crystalline Mo2BC coatings [3].
There is a question that arose from nanoindentation and XRD
data presented in Figs. 1, 3 and 4 e what is the main reason for the
hardness increase at annealing temperatures starting from 500 C
where, according to XRD diffractograms, no signiﬁcant crystallization
occurred? The different thermal stress evolution may explain
20 40 60 80 100 120 140 160 180
320
340
360
380
400
420
440
460
480
500
520
as deposited
500 °C
600 °C
HSS substrate
effectivemodulus[GPa]
indentation depth [nm]
a)
20 40 60 80 100 120 140 160 180
320
340
360
380
400
420
440
460
480
500
520
hard metal substrate
effectivemodulus[GPa]
indentation depth [nm]
b)
as deposited
after 500 °C
after 700 °C
after 800 °C
after 1000 °C
Fig. 4. The dependency of the effective elastic modulus of coatings on a) HSS and b) hard metal substrates on indentation depth and its evolution upon annealing.
20 40 60 80 100 120 140 160 180
20
22
24
26
28
30
32
as deposited
after 500 °C
after 700 °C
after 800 °C
after 1000 °C
hard metal substrate
hardness[GPa]
indentation depth [nm]
b)
20 40 60 80 100 120 140 160 180
20
22
24
26
28
30
32
as deposited
500 °C
600 °C
HSS substrate
hardness[GPa]
indentation depth [nm]
a)
Fig. 3. The dependency of the hardness of coatings on a) HSS and b) hard metal substrates on indentation depth and its evolution induced by annealing.
L. Zabranský et al. / Vacuum 138 (2017) 199e204202
the slightly different hardness and elastic modulus observed for
HSS and hard metal substrates at 500 and 600 C, what we see from
Figs. 3 and 4, however, it does not provide explanation of step increase
of hardness for these temperatures. The load/displacement
curves obtained on as deposited and annealed coatings (see Fig. 5)
showed that the hardness increase after annealing to 500 C was
also accompanied by structural changes. The as deposited coating
exhibited smooth displacement response to loading and unloading
without any pop-ins or pop-outs. However, after annealing several
pop-ins, characteristic for crystalline materials, occurred for coatings
on both substrates. These pop-in events are most probably
caused by a grain boundary sliding mechanism [22,23] which
means that during annealing to 500 C additional crystallization of
amorphous matrix occurred. However, the newly formed grains are
too small to be distinguished in diffractograms presented in Fig. 3. It
means that the nanocomposite strengthening effect is most probably
the main reason for the hardness and effective elastic modulus
increase of coatings after annealing to 500 C.
The fracture resistance changes were studied using SEM imaging
of residual imprints caused by the microindentation test. The as
deposited samples showed no cracking around or inside the
imprint even if the load of 1 N was used. With this load, the plastic
zone already over-passed the coating-substrate interface which
resulted in the plastic deformation of the substrate. This deformation
of the substrate can induce cracking in the coating which is
usually visible on SEM as cracks inside the imprint [14,15]. Using
the differential hardness study it is possible to detect when this
substrate deformation occurred as well as the presence of the
subsequent cracking represented as distinct sudden drops in differential
hardness curves. Fig. 6a shows the SEM image of the
0 10 20 30 40 50 60 70 80
0,0
0,5
1,0
1,5
2,0
2,5
load[mN]
depth [nm]
as deposited
after 700 °C
pop-ins
coatings on
hard metal
substrate
Fig. 5. The comparison of load-displacement curves for as deposited and annealed
coatings on hard metal substrates with marked pop-ins.
Fig. 6. a) The SEM image of the typical residual 1 N imprint made in as deposited sample on HSS substrate and b) the corresponding measurements of the differential hardness.
Fig. 7. SEM images of a) the sample on HSS substrate after annealing to 600 C showing no cracking inside or around the imprint, b) the sample on hard metal after 1000 C where
cracks spreading from corners of the imprint are visible.
L. Zabranský et al. / Vacuum 138 (2017) 199e204 203
typical residual 1 N imprint made in as deposited sample on HSS
substrate, and Fig. 6b presents the corresponding measurements of
the differential hardness. No cracks are observed by SEM imaging,
as well as no sudden drops are present in differential hardness
curves meaning that, most probably, no indentation induced
cracking occurred. The decrease of differential hardness from
indentation depth of 0.5 mm in Fig. 6b was caused by inﬂuence of
the HSS substrate. The coatings kept their desired ductile properties
up to 600 C e see Fig. 7a. However, the indentation tests after
annealing to 1000 C induced cracks spreading from corners of the
imprints e see Fig. 7b. The decrease of ductility of the coatingsubstrate
system (see Fig. 7b) is caused mainly by a growth of
crystalline phase to the detriment of the amorphous phase which
was indicated by pop-ins emerging after annealing and which may
play an important role in the overall ductile properties of the
coating.
4. Conclusion
Nanocomposite Mo-B-C coatings, exhibiting ﬁne-grained
structure, were prepared employing magnetron sputtering of
three targets: B4C, C and Mo. Two different substrates were used e
high speed steel and hard metal (cemented tungsten carbide). The
samples were subjected to annealing in temperature range from
500 to 1000 C which in all cases signiﬁcantly improved their
hardness and elastic modulus. The hardness increased from original
~20 GPa to almost 30 GPa while annealed to 1000 C, the
effective elastic modulus increased from 330 GPa to about 500 GPa.
Even though the crystallization temperature of the amorphous
Mo2BC material was found to be 820 C, the ﬁne grains existing in
as deposited state could play the role of crystallization centers and
may induce nanometer scale grain growth even below 500 C.
Exceeding the temperature above 900 C additional growth of bMo2C
grains occurred according to X-ray diffractograms. According
to the selective area electron diffraction patterns and X-ray
diffraction technique the ﬁlms annealed at 1000 C consist of
mixture of nanoscale Mo2BC and b-Mo2C grains. The coatings kept
their desired ductile properties up to 600 C. However, the indentation
tests after annealing to 1000 C induced cracks spreading
from corners of the imprints.
Acknowledgement
This research has been supported by the project CZ.1.05/2.1.00/
03.0086 funded by European Regional Development Fund and
project LO1411 (NPU I) funded by Ministry of Education Youth and
Sports of Czech Republic and the Czech Science Foundation (Project
GACR 15-17875S).
References
[1] M.W. Barsoum, T. El-Raghy, Am. Sci. 89 (2001) 334e343.
[2] D. Music, J.M. Schneider, JOM 59 (2007) 60e64.
[3] J. Emmerlich, D. Music, M. Braun, P. Fayek, F. Munnik, J.M. Schneider, J. Phys. D.
Appl. Phys. 42 (2009) 185406.
[4] H. Bolvardi, J. Emmerlich, M. to Baben, D. Music, J. von Appen, R. Dronskowski,
J.M. Schneider, J. Phys. e Condens. Mat. 25 (2013) 045501.
[5] L. Zabranský, V. Bursíkova, P. Soucek, P. Vasina, J. Bursík, Eur. Phys. J. Appl.
Phys. 75 (2016) 24716.
[6] S. Djaziri, S. Gleich, H. Bolvardi, C. Kirchlechner, M. Hans, C. Scheu,
J.M. Schneider, G. Dehm, Surf. Coat. Tech. 289 (2016) 213e218.
[7] H. Bolvardi, J. Emmerlich, S. Mraz, M. Arndt, H. Rudigier, J.M. Schneider, Thin
Solid Films 542 (2013) 5.
[8] J. Bursík, V. Bursíkova, P. Soucek, L. Zabranský, P. Vasina, Rom. Rep. Phys. 68
(2016) 1069e1075.
[9] J. Robertson, Mat. Sci. Eng. R37 (2002) 129, 281.
[10] D.R. Tallant, J.E. Parmeter, M.P. Siegal, R.L. Simpson, Diam. Relat. Mater 4
(1995) 191e199.
[11] M. Benlahsen, B. Racine, K. Zellama, G. Turban, J. Non-Cryst. Solids 283 (2001)
47e55.
[12] R.K.Y. Fu, Y.F. Mei, M.Y. Fu, X.Y. Liu, P.K. Chu, Diam. Relat. Mater 14 (2005)
1489e1493.
[13] H.C. Barshilia, A. Jain, K.S. Rajam, Vacuum 72 (2004) 241e248.
[14] L. Zabranský, V. Bursíkova, P. Soucek, P. Vasina, T. Gardelka, P. S
0
t ahel, O. Caha,
V. Perina, J. Bursík, Surf. Coat. Tech. 242 (2014) 62e67.
[15] L. Zabranský, V. Bursíkova, J. Daniel, P. Soucek, P. Vasina, J. Dugacek, P. S
0
t ahel,
O. Caha, J. Bursík, V. Perina, Surf. Coat. Tech. 267 (2015) 32e39.
[16] I.A. Abrikosov, A. Knutsson, B. Alling, F. Tasnadi, H. Lind, L. Hultman, M. Oden,
Materials 4 (2011) 1599e1618.
[17] C. Fitz, A. Kolitsch, W. Fukarek, Thin Solid Films 389 (2001) 173e179.
[18] W. Jeitschko, H. Nowotny, F. Benesovsky, Monatsh. Chem. 94 (1963) 565e568.
[19] W.C. Oliver, G.M. Pharr, J. Mat. Res. 19 (2004) 3e20.
[20] B. Wolf, A. Richter, New J. Phys. 5 (2003) 15, 1-15.17.
[21] B. Wolf, Phil. Mag. Phil. Mag. Lett. 86 (2006) 5251e5264.
[22] E.N. Borodin, A.E. Mayer, Mat. Sci. Eng. A 532 (2012) 245e248.
[23] J. Shi, M.A. Zikry, Mat. Sci. Eng. A 520 (2009) 121e133.
L. Zabranský et al. / Vacuum 138 (2017) 199e204204