Wear 263 (2007) 1253–1258
Short communication
Friction and wear property of a-CNx coatings sliding
against Si3N4 balls in water
Fei Zhou a,∗ , Xiaolei Wang b , Koji Kato c , Zhendong Dai a
a
Institute of Bio-Inspired Structure and Surface Engineering (IBSS), Academy of Frontier Science, Nanjing University
of Aeronautics and Astronautics, Nanjing 210016, PR China
b School of Mechanical Engineering, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, PR China
c Laboratory of Tribology, School of Mechanical Engineering, Tohoku University, Sendai 980-8579, Japan
Received 27 August 2006; received in revised form 15 November 2006; accepted 15 November 2006
Available online 23 May 2007
Abstract
Amorphous carbon nitride coatings (a-CNx ) were deposited on Si(1 0 0) wafers and Si3 N4 disks using ion beam assisted deposition (IBAD),
and their composition and chemical bonding were determined by X-ray photoelectron spectroscopy (XPS). The a-CNx coatings’ hardness was
measured by nano-indentation and the friction and wear behavior of a-CNx coating sliding against a Si3 N4 ball in water was investigated. The results
indicated that the a-CNx coatings contained 12 at.% nitrogen and the major chemical bonding was sp2 C N and sp3 C–N. The nano-hardness of
the a-CNx coatings was 29 GPa. At a sliding velocity of 0.16 m/s and after running-in, the mean steady-state friction coefficient varied around 0.02
when the normal load was lower than 3.5 N, and then decreased abruptly from 0.018 to 0.007 at 5 N. For self-mated Si3 N4 , the specific wear rate of
a Si3 N4 ball was a little higher than that of a Si3 N4 disk, while for a-CNx /Si3 N4 , the specific wear rate of a Si3 N4 ball was slightly smaller than that
of a-CNx coating. Furthermore, the specific wear rate of Si3 N4 ball sliding against a-CNx coating was reduced by a factor up to 35 in comparison to
that against Si3 N4 in water. This indicated that the wear mechanism of a-CNx coating/Si3 N4 ball was the formation of a carbonaceous transfer film
on the a-CNx coatings via a tribochemical reaction between a-CNx coatings and water induced by friction, while that of self-mated Si3 N4 ceramics
was the formation of silica gel on the contact zone via the reaction of silicon nitride and water.
© 2007 Elsevier B.V. All rights reserved.
Keywords: Carbon nitride coatings; Si3 N4 ceramics; Friction; Wear; Water lubrication
1. Introduction
Since Liu and Cohen [1] in 1989 predicted theoretically that
the existence of a metastable covalent carbon nitride compound
(␤-C3 N4 ) with an analogous structure to ␤-Si3 N4 , and this carbon nitride compound, with a high bulk modulus, might have
higher hardness than diamond, many attempts at developing
new processing methods to obtain carbon nitride films have
been performed. But until now, nearly all CNx films grown at
room temperature are amorphous mixtures of carbon and carbon
nitride with x ranging from 0.1 to 0.5 [2,3]. Nitrogen incorporation in the carbon coatings decreases the fraction of sp3 carbon
bonds by the formation of C–N, C N and C N bonds. Recently,
∗ Corresponding author. Tel.: +86 25 84892581 803;
fax: +86 25 84892581 803.
E-mail addresses: [email protected],
[email protected] (F. Zhou).
0043-1648/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.wear.2006.11.048
Zhou et al. reported that the a-CNx coatings could enhance the
wear resistance of a SiC ball and shorten the running-in period as
sliding against a SiC ball in water at sliding velocities in the range
of 0.019–0.16 m/s [4,5] and the wear-mechanism map of the aCNx /SiC tribo-pair in water was developed [6]. Furthermore,
the friction and wear properties of the a-CNx /SiC tribo-couple
in water have already been found to be better than those of the aC/SiC tribo-pair under the same experimental conditions [7,8].
At the normal load of 5 N and the sliding velocity of 0.16 m/s
in water, the friction coefficients of 0.01–0.02 were obtained
as the a-CNx coatings slid against SiC and Si3 N4 balls, while
larger friction coefficients of 0.07–0.10 were acquired as the
a-CNx coatings slid against Al2 O3 , SUS440C and SUJ2 balls
[9]. It was indicated that the amorphous carbon nitride coatings sliding against Si-based non-oxide ceramics such as SiC
and Si3 N4 exhibited the lowest friction coefficient and lower
wear rate. Currently, the friction, lubrication and wear mechanisms of the a-CNx /SiC tribo-pair have already been reviewed
1254
F. Zhou et al. / Wear 263 (2007) 1253–1258
in detail [10], however, the friction and wear property of the aCNx /Si3 N4 tribo-pair in water has not yet been studied in detail.
The aim of this paper is to investigate the friction and wear
properties of the a-CNx /Si3 N4 tribo-pair in water and compare
their wear characteristics between the a-CNx /Si3 N4 tribo-pair
and Si3 N4 /Si3 N4 tribo-pair by using ball-on-disk tribo-meters
at room temperature. The influences of testing conditions on the
tribological behaviors of a-CNx /Si3 N4 tribo-pair in water were
analyzed.
2. Experimental procedures
The IBAD machine (Hitachi Ltd., Japan) consists of a cryogenically pumped chamber, a sputter deposition source, an
electron beam evaporator, two ion guns for sputtering and mixing, respectively, and a substrate holder (Ref. [5]). The diameter
of the ion beam irradiation area is about 80 mm. The substrate
holder consists of a water-cooled copper plate and can be rotated
at a speed of 4 rpm during deposition. Prior to the IBAD process, Si3 N4 disks (Ø 30 mm × t 8 mm) and Si(1 0 0) wafers
were ultrasonically cleaned in acetone for 30 min. Then, a high
purity carbon target was put into an electron beam evaporator and a substrate jig containing a Si3 N4 disk was fixed on
the substrate holder. After that, the deposition chamber doors
were closed and the vacuum chamber was subsequently evacuated to lower than 2.0 × 10−4 Pa. For further cleaning, the disk
surface was bombarded using nitrogen ions for 5 min. After
that, the a-CNx coatings were synthesized by mixing carbon
vapor and energetic N ions. Energetic N ions were produced
under the acceleration voltage of 1.5 kV with the acceleration current density of 90 ␮A/cm2 . Carbon vapor was formed
through heating a graphite target with an electron beam evap˚ which was controlled
orator. The deposition rate was 20 A/s,
by adjusting the emission current of carbon vapor. The coating
thickness was 0.5 ␮m. The deposition parameters are listed in
Ref. [5].
The composition and chemical bonding of the a-CNx coatings
were determined by a scanning ESCA microprobe (Quantum
2000, Physical Electronics Inc., USA). The coatings’ surface
roughness was measured by a Surfcom-1500DX profilometer,
and their hardness and Young’s modulus were evaluated using
a Nano Indenter ELIONIX ENT-1100A.
The diameter of all Si3 N4 balls was 8 mm and the balls’
roughness was determined by a Surfcom-1500DX profilometer
and its mechanical properties were obtained from the ball manufacturer. The data are listed in Table 1. Prior to each wear test,
all samples were ultrasonically cleaned in acetone and ethanol
for 30 min. The experiments were performed on the ball-on-disk
apparatus consisting of a rotating disk sliding on a stationary ball
at 0.16 m/s and 1.5–5 N. The rubbing surfaces were submerged
in purified water. The contact point was designed at an eccentricity of 7.5 mm from the center of the rotary motion, which
created a wear track of 15 mm in diameter on the a-CNx coated
Si3 N4 disks’ surface. In order to know the influence of the aCNx coatings on the wear behavior of silicon nitride ceramic,
the wear behavior of self-mated Si3 N4 ceramic tribo-pairs was
also studied under the same conditions. The total friction cycles
were 49,200 cycles (equal to a sliding distance of 2304 m).
The friction forces were detected by a LMA-A-10 N load cell
(Kyowa Co. Ltd., Japan). The load cell voltage was measured
by a DPM-700B strain amplifier (Kyowa Co. Ltd., Japan) and
recorded by NR-110/150 data collection system (Keyence Co.
Ltd., Japan) with a compatible personal computer. The diameter
of the wear scar on the SiC ball under each condition was measured using a Keyence VH-8000 optical microscope (Keyence
Co. Ltd., Japan). The cross-section area of the wear track on
disk, A, was determined using a Tencor P-10 surface profilometer
(Kurashiki Kako Co. Ltd., Japan). Thus, the specific wear rates
for balls and disks were determined using the same equations in
Refs. [5–9]. To know the wear mechanism of the a-CNx /Si3 N4
and self-mated Si3 N4 tribo-pairs in water, the wear scars on
the balls and the wear tracks on disks were observed by optical
microscopy.
3. Result
3.1. Surface roughness and mechanical properties of
a-CNx coatings
As seen in Table 1, the arithmetic mean roughness Ra of the aCNx coating was a little smaller than that of the Si3 N4 substrate.
This indicated that the energetic particle bombardment enhanced
the mobility of carbon atoms on the growing surface and induced
the smooth surface. Fig. 1 displays nano-indentation load versus indentation depth curves for a-CNx coatings. Based on the
standard Oliver and Pharr approach [11], the mean values of the
Table 1
Surface roughness and mechanical properties of Si3 N4 ball, a-CNx coatings and
Si3 N4 disk
Name
Si3 N4 ball
a-CNx
Si3 N4 disk
a
Ra (␮m)
H (GPa)
E (GPa)
0.0552
0.0251
0.0280
15.3a
308a
330 ± 20
290a
29 ± 2
16a
The data are from the sample manufacturer.
Fig. 1. Nano-indentation load vs. indentation displacement curves for a-CNx
coatings.
F. Zhou et al. / Wear 263 (2007) 1253–1258
1255
elastic modulus (E) and the hardness (H) for the a-CNx film were
calculated from the nano-indentation load–displacement curves
(Fig. 1) and are listed in Table 1. The results in Table 1 show
that the a-CNx coatings offered a combination of reasonably
high hardness and reduced stiffness with a remarkable elastic
recovery. It indicated that the nitrogen incorporation in carbon
increased the sp2 carbon bonds’ fraction so that the tribological
property of the films was improved (low friction coefficient and
better durability).
3.2. Composition and chemical bonding of a-CNx coatings
According to the XPS analysis, the a-CNx coatings contained
12% nitrogen atoms. To know the possible chemical bonding
configurations of nitrogen doped into the carbon network, the
individual C 1s and N 1s lines were deconvoluted into Gaussian line shapes (Fig. 2). The C 1s line was also deconvoluted
into three peaks at binding energies of 285.1, 286.5 and 288 eV,
and the N 1s line was deconvoluted into four peaks at binding
energies of 398.5, 400.3, 401.5 and 404 eV. Scharf et al. [12]
reported that, for the a-CN0.14 coatings, the peaks at binding
energies of 284.5, 285.2, 286.5 and 288.6 eV for the deconvoluted C 1s spectra were attributed to C–C, C N, C–N or
C N, and C–O bonds, respectively, while the peaks at 398.6,
400.1 and 402.3 eV for the N 1s line were assigned to C–N or
C N, C N and N–O bonds, respectively. Comparing the data of
Ref. [11], the peaks at 285.1, 286.5 and 288 eV in Fig. 2a were
assigned to C N, C–N or C N, and C–O bonds, respectively,
while the peaks at 398.5, 400.3, 401.5 and 404 eV in Fig. 2b
were marked as C–N or C N, C N and N–O bonds, respectively. The appearance of C–O and N–O bonds showed that the
coatings surface was contaminated by oxygen from air. Furthermore, since the C 1s and N 1s binding energies of urotropine
(N bonded to sp3 -hybridized C, C–N) and polyacrylonitrile (N
bonded to sp-hybridized C, C N) molecules were nearly identical [13,14], it was difficult to unequivocally distinguish the
bonding configuration of a C–N bond from that of a C N bond.
However, Ref. [12] pointed out that, as the nitrogen concentration of the a-CNx coatings was smaller than 14 at.%, the C N
bond could not be detected in the a-CNx coatings. Due to the
present a-CNx coating containing 12 at.% nitrogen, the sp3 C–N
Fig. 3. Variation of friction coefficient with sliding cycles as a-CNx coatings
sliding against Si3 N4 ball at 160 mm/s and various normal loads in water.
and sp2 C N bonds were the major CN component in the a-CNx
coatings.
3.3. Friction behaviors of a-CNx coatings sliding against
Si3 N4 ball in water
The friction behaviors of a-CNx coatings sliding against
Si3 N4 balls at 0.16 m/s with various normal loads in water lubrication are illustrated in Fig. 3. In general, the friction coefficient
decreased during the early stage of the test and then approached
a steady-state value. At a lower normal load of 1.5 N, the initial friction coefficient of a-CNx /Si3 N4 tribo-pair was 0.25, but
when the normal load was higher than 2.5 N, the initial friction coefficient of an a-CNx /Si3 N4 tribo-pair varied in the range
0.10–0.13. As seen in Fig. 3, the running-in period of an aCNx /Si3 N4 tribo-pair varied with the normal load at a constant
sliding velocity of 0.16 m/s. The running-in period was 12,500
cycles at 1.5 N, 13,000 cycles at 2.5 N, 9020 cycles at 3.5 N and
12,500 cycles at 5 N, respectively. After running-in, the friction
coefficient fluctuated in the range of 0.002–0.02. Fig. 4 shows the
influence of normal load on the mean steady-state friction coefficients after running-in for a-CNx /Si3 N4 tribo-pairs in water.
Fig. 2. XPS spectra of the C 1s (a) and N 1s (b) photoelectron peaks for a-CNx coatings.
1256
F. Zhou et al. / Wear 263 (2007) 1253–1258
Fig. 4. Influence of normal load on mean steady-state friction coefficients after
running-in for a-CNx /Si3 N4 tribo-pairs in water.
It is clear that the mean steady-state friction coefficient varied
around 0.02 or so when the normal load was lower than 3.5 N.
At the highest normal load of 5 N, the mean steady-state friction
coefficient decreased abruptly from 0.018 to 0.007. The above
friction coefficients indicate that the a-CNx /Si3 N4 tribo-pair has
excellent friction characteristics in water lubrication.
Fig. 5. Variation in specific wear rate at various normal loads for two kinds of
tribo-pairs in water.
3.4. Wear behaviors of two kinds of tribo-pairs
In order to know the influence of a-CNx coating on the specific
wear rate of Si3 N4 in water, the wear behavior of a-CNx /Si3 N4
tribo-pairs was compared with that of self-mated Si3 N4 ceramics. The experimental results are illustrated in Fig. 5. It is clear
Fig. 6. Wear scar on Si3 N4 ball (a and c) and wear track surface on a-CNx coatings (b and d) after sliding in water: (a and b) 2.5 N; (c and d) 5 N at 0.16 m/s.
F. Zhou et al. / Wear 263 (2007) 1253–1258
1257
Fig. 7. Wear scar on Si3 N4 ball (a and c) and wear track surface on Si3 N4 disks (b and d) after sliding in water: (a and b) 2.5 N; (c and d) 5 N at 0.16 m/s.
that, with an increase in normal load, the specific wear rate of
all tribo-materials decreased gradually as the normal load was
in the range of 1.5–5 N. For the a-CNx /Si3 N4 tribo-pair, the
specific wear rate of the a-CNx coatings varied in the range
of 3.89 × 10−8 to 7.89 × 10−8 mm3 /Nm, a little larger than
that of Si3 N4 balls. Moreover, the specific wear rate of the aCNx coatings and the Si3 N4 balls all were at a lowest level of
10−8 mm3 /Nm. But for self-mated Si3 N4 ceramics, the specific
wear rate of Si3 N4 balls fluctuated in the range of 1.28 × 10−6
to 2.79 × 10−6 mm3 /Nm, and was approximately twice as large
as that of the Si3 N4 disk, whose specific wear rate varied in the
range of 6.18 × 10−7 to 9.64 × 10−7 mm3 /Nm. The results in
Fig. 5 also show that the specific wear rate of Si3 N4 ball sliding
against a-CNx coating was reduced by a factor up to 35 in comparison to that against Si3 N4 in water. This indicates that the
a-CNx coatings can enhance the wear behavior of silicon nitride
ceramics in water.
surface with some shallower grooves except for some original micro-voids (Fig. 6b and d). But for the self-mated Si3 N4
ceramic tribo-pair, the wear scar surface on the Si3 N4 ceramic
ball became much smoother and the wear track surface on the
Si3 N4 ceramic disks also became smoother and flatter (Fig. 7).
As compared with the unworn surface on the Si3 N4 ceramic
disk, it is clear that the original micro-pits disappeared during
sliding of a Si3 N4 ball against a Si3 N4 disk in water. If Fig. 7 is
compared with Fig. 6, it is evident that the wear scar diameter for
a Si3 N4 ball in the a-CNx /Si3 N4 tribo-pair was smaller than that
in the self-mated Si3 N4 ceramic tribo-pair, but the wear track
surface on a Si3 N4 disk was smoother than that on the a-CNx
coatings. This indicated that the friction and wear mechanisms
of a self-mated Si3 N4 ceramic tribo-pair were different from
those of an a-CNx /Si3 N4 tribo-pair in water.
3.5. Observation of friction surfaces
For the a-CNx /Si3 N4 tribo-couple, the Si3 N4 ball hardness is
15.3 GPa, and the a-CNx coatings’ hardness is 29 GPa. According to normal friction and wear theory, the wear rate of a Si3 N4
ball should be larger than that of the a-CNx coatings. But here,
the wear rate of the Si3 N4 ball was a little smaller than that of the
a-CNx coating. A similar phenomenon has already been reported
For the a-CNx /Si3 N4 tribo-pair, at a low normal load of 2.5 N,
the wear scar surface on a Si3 N4 ball displayed a smooth surface with some shallow scratch lines (Fig. 6a and c). The wear
track surface of a-CNx coatings exhibited a smoother and flatter
4. Discussion
1258
F. Zhou et al. / Wear 263 (2007) 1253–1258
by Jia et al. [15] when they studied the tribology of a DLC coating sliding against a Si3 N4 ball. The wear rate of a DLC film was
higher than that of a Si3 N4 ball, which was related to transfer
of a tribo-layer from the disk to the ball. Furthermore, the specific wear rate of a Si3 N4 ball sliding against an a-CNx coating
was reduced by a factor up to 35 in comparison to that against
Si3 N4 in water. This indicated that the a-CNx could enhance the
anti-wear ability of silicon nitride ceramics in water. For selfmated Si3 N4 ceramic tribo-pairs, the wear track surface became
smooth and flat, as seen in Fig. 7. When a Si3 N4 or SiC ceramic
slides against itself in water, tribochemical reactions occurred
as follows:
SiC + 4H2 O = Si(OH)4 +CH4
(1)
G298
= −598.91 kJ/mol
f
(2)
Si3 N4 + 12H2 O = 3Si(OH)4 + 4NH3
(3)
G298
= −1268.72 kJ/mol
f
(4)
G298
f
where
is the reaction Gibbs free energy of formation at
298 K. From Eqs. (2) and (3), we could conclude that silicon
nitride is more easily hydrated than silicon carbide. The tribooxidatively formed amorphous hydrate Si(OH)4 is then either
dissolved water, known as tribo-chemical wear, or removed from
the interface, and an ultra flat contact surface is obtained easily.
Furthermore, water lubrication of ceramics is thin film hydrodynamic lubrication, and the water during friction is an electrolyte
solution, so these two ceramics were in charged states during
friction, and the effect of an electric double layer formed by
electrostatic charges may play a role in the hydrodynamic lubrication of the two ceramics [16]. Thus, the self-mated Si3 N4
ceramic tribo-pair exhibited the higher specific wear rate and
lower friction coefficient.
It is well-known that the a-CNx coatings offered higher values of H/E and a combination of reasonably high hardness and
suitable stiffness, so the a-CNx coatings possess excellent tribological properties [8]. If the Si3 N4 disk was covered with
amorphous carbon nitride coatings, friction transforms the surface layer of the a-CNx coating and gives it lower shear strength,
which is responsible for low friction and the transfer of material.
As Tanaka et al. [17–20] studied the water lubrication of DLC
coatings, they indicated that the structure of transferred materials was very different from that of the original DLC film and
similar to that of polymer-like carbon, which is softer in comparison to DLC film. The amount of transferred material with the
polymer-like structure was larger in water than that in air. But
now, the surface chemistry of this easy-shear transfer film was
determined in previous publications on CNx films to be formed
by C sp2 -bonding-rich structures [8,9,21,22]. Moreover, the aCNx coatings are hydrophilic, and the physisorption of water
seems to cause the formation of hydrogen bonds between water
molecules and nitrogen atoms [23]. Nitrogen atoms are removed
easily from the CNx coating by reaction with water. Hellgren et
al. [24] have indicated that if operated in the presence of oxygen
or hydrogen, those elements would react with a-CNx film and
promote decomposition. After the nitrogen atoms are removed
from a-CNx coatings, a carbonaceous transfer film will form on
the wear scar surface and carbon bonds can be terminated with
OH− in water, which is responsible for low friction for the aCNx /Si3 N4 tribo-couple, and low wear rate of the Si3 N4 ball in
water.
5. Conclusions
(1) The a-CNx coatings contained 12 at.% nitrogen and the
major chemical bonding was sp2 C N and sp3 C–N. The
nano-hardness of the a-CNx coatings was 29 GPa.
(2) After running-in, the mean steady-state friction coefficient
varied around 0.02 when the normal load was lower than
3.5 N, and then decreased abruptly from 0.018 to 0.007 at
5 N.
(3) The specific wear rate of a Si3 N4 ball sliding against an aCNx coating was reduced by a factor up to 35 in comparison
to that against Si3 N4 in water.
Acknowledgements
This work was supported by the Japan Society for the Promotion of Science under Grant-in-Aid for Scientific Research
(JSPS Fellows P03219) and the National Nature Science Foundation of China (NNSFC) (No. 50675102). We would like to
acknowledge JSPS and NNSFC for financial support.
References
[1] A.Y. Liu, M.L. Cohen, Science 245 (1989) 841.
[2] Z.J. Zhang, J. Huang, S. Fan, C.M. Lieber, Mater. Sci. Eng. A 209 (1996)
5.
[3] S.F. Yoon, J. Rusli, J. Ahn, Q. Zhang, C.Y. Yang, H. Yang, F. Watt, Thin
Solid Films 340 (1999) 62.
[4] F. Zhou, K. Kato, K. Adachi, Mater. Sci. Forum 475/479 (2005) 2899.
[5] F. Zhou, K. Kato, K. Adachi, Tribol. Lett. 18 (2005) 153.
[6] F. Zhou, K. Adachi, K. Kato, Surf. Coat. Technol. 200 (2006) 4909.
[7] F. Zhou, K. Adachi, K. Kato, Surf. Coat. Technol. 200 (2006) 4471.
[8] F. Zhou, K. Adachi, K. Kato, Thin Solid Films 514 (2006) 231.
[9] F. Zhou, K. Adachi, K. Kato, Diamond Relat. Mater. 14 (2005) 1711.
[10] F. Zhou, Z. Dai, K. Kato, Lubr. Eng. 7 (185) (2006) (in Chinese).
[11] W.C. Oliver, G.M. Pharr, J. Mater. Res. 7 (1992) 1564.
[12] T.W. Scharf, R.D. Ott, D. Yang, J.A. Barnard, J. Appl. Phys. 85 (1999)
3142.
[13] S. Bhattacharya, J. Hong, G. Turban, J. Appl. Phys. 83 (1998) 3917.
[14] G. Beamson, D. Briggs, High Resolution XPS of Organic Polymers, Wiley,
New York, 1992.
[15] K. Jia, Y.Q. Li, T.E. Fischer, B. Gallois, J. Mater. Res. 10 (1995) 1403.
[16] M. Chen, K. Kato, K. Adachi, Tribol. Int. 35 (2002) 129.
[17] M. Suzuki, A. Tanaka, T. Ohana, W. Zhang, Diamond Relat. Mater. 13
(2004) 1464.
[18] M. Suzuki, T. Ohana, A. Tanaka, Diamond Relat. Mater. 13 (2004) 2216.
[19] A. Tanaka, M. Suzuki, T. Ohana, Tribol. Lett. 17 (2004) 917.
[20] J. Wei, P. Hing, Z.Q. Mo, Wear 225–229 (1999) 1141.
[21] J.C. Sanchez-Lopez, C. Donnet, T. Le Mogne, Vacuum 64 (2002) 191.
[22] J.C. Sanchez-Lopez, M. Belin, C. Donnet, C. Quiros, E. Elizalade, Surf.
Coat. Technol. 160 (2002) 138.
[23] W. G¨opel, T. Jones, M. Kleitz, I. Lundstr¨om, T. Seiyama, Sensors, Part II,
vol. 3, Wiley-VCH, Weinheim, 1998.
[24] N. Hellgren, N. Lin, E. Broitman, V. Serin, S. Grillo, et al., J. Mater. Res.
16 (11) (2001) 3188.