Full Text

Available online at www.sciencedirect.com
Surface & Coatings Technology 202 (2008) 3947 – 3953
www.elsevier.com/locate/surfcoat
Surface alloying of an Mg alloy subjected to surface
mechanical attrition treatment
H.Q. Sun a , Y.N. Shi a,⁎, M.-X. Zhang b , K. Lu a
a
Shenyang National Laboratory for Materials Science, Institute of Metal Research, Chinese Academy of Sciences, Shenyang 110016, China
b
Division of Materials, School of Engineering, University of Queensland, St Lucia, Brisbane, Qld 4072, Australia
Received 19 November 2007; accepted in revised form 7 February 2008
Available online 4 March 2008
Abstract
A 100 ∼ 200 μm thick Al-enriched surface alloyed layer was formed on an AZ91D Mg alloy subjected to surface mechanical attrition treatment
and diffusion coating at temperature as low as 400 °C. Transmission electron microscopy observations indicated the formation of a large volume
fraction of pearlite-like lamellar microstructure within the surface alloyed layer, which was identified to be Mg17Al12 precipitates (γ phase) in Mg
solid solution matrix. The Al-enriched alloyed layer enhanced the wear resistance of the alloy in comparison with the un-treated AZ91D Mg alloy
substrate under the same dry sliding wear condition. Examination of the worn surface indicated that the enhanced wear resistance of the alloyed
layer was mainly attributed to the strengthening effect of γ phase.
© 2008 Elsevier B.V. All rights reserved.
Keywords: Magnesium alloy; Surface alloying; Surface mechanical attrition treatment (SMAT); Wear resistance.
1. Introduction
Nowadays, magnesium alloys are attracting great attention in
automobile and aerospace industries for their potential applications in the promotion of fuel efficiency. However, magnesium
alloys also suffer from low resistance to wear and corrosion,
which limit their applications. Surface modification has been
found to be an efficient way to improve the corrosion and wear
properties of metallic materials without changing the substrate
properties [1]. Surface alloying, which leads to the formation of
an integrated alloyed layer on the substrate, is an efficient and
reliable technique of surface modification. Some conventional
technologies, such as laser and electron beam surface alloying,
[2,3] have disadvantages of high cost and low efficiency.
Surface alloying by diffusion coating represents an inexpensive
and practical process, particularly, surface alloying of Mg alloys
with aluminium is a commonly-used technique in the present.
Previous experiment results [4,5] have shown that the formation
of large volume fraction of Mg17Al12 intermetallic compound
(γ phase) can not only increase the hardness, but can also
⁎ Corresponding author. Tel.: +86 24 23971948; fax: +86 24 23998660.
E-mail address: [email protected] (Y.N. Shi).
0257-8972/$ - see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.surfcoat.2008.02.010
improve the corrosion resistance of Mg alloys, for γ phase can
act as an anodic barrier to inhibit the overall corrosion of the
alloy [6,7]. However, the major challenge for diffusion coating
of Mg alloys is to lower the treatment temperature to avoid the
effect on the microstructure and properties of the substrate.
Nanocrystalline (NC) materials, structurally characterized by a
large volume fraction of grain boundaries, often exhibit an
enhanced atomic diffusion compared with their coarse-grained
(CG) counterparts [8,9] owing to the “fast” grain boundary
diffusion [10]. Surface mechanical attrition treatment (SMAT) has
been proven [11,12] to be able to generate NC grains in the top
surface layer of bulk materials. A low temperature nitriding
process for pure iron after SMAT was reported by Tong et al. [13].
The nitriding can be achieved at temperature as low as 300 °C,
which is 200 °C lower than that of the conventional nitriding
process. Similarly, the diffusion temperature of chromium into the
surface layer of iron can also be pronouncedly reduced by SMAT
[14]. It was also recognized that SMAT can be applied to coat
metals where an enhanced bonding between the substrate and the
coating due to mechanical activation has been reported [15,16].
Previous work [17] showed that a 100 μm thick NC layer with an
average grain size of 30 ± 5 nm was obtained in an AZ91D Mg
alloy after SMAT. Our recent work [18] on surface alloying
3948
H.Q. Sun et al. / Surface & Coatings Technology 202 (2008) 3947–3953
treatment on the NC layer generated by SMAT also found that the
surface alloying of the AZ91D Mg alloy had been accomplished
at 380 °C, which is 50 °C lower than the reported temperatures
without SMAT by using similar approach [19]. The combined
process of SMAT and the surface alloying led to the formation of
large volume fraction of Mg17Al12 phase within the top surface
layer. However, the detailed microstructure of the surface alloyed
layer remains unknown. The present work aims to characterize the
microstructure of the alloyed layer by using X-ray diffraction
(XRD) and transmission electron microscopy (TEM), to further
understand the formation process of the alloyed layer and the
effects of SMAT on the subsequent surface alloying process. The
wear behaviour of the alloyed layer will also be comprehensively
investigated followed by discussion on wear mechanism of the
alloy layer based on scanning electron microscopy (SEM)
examination of the worn surfaces.
2. Experimental
The composition of the AZ91D magnesium alloy used is
8.47%Al–0.69 %Zn–0.14%Mn (wt.%) with the balance Mg.
SMAT was performed at room temperature on SNC-II SMAT
equipment where spherical balls are placed in a chamber
vibrated by an ultrasonic vibration generator. Samples are fixed
to the opposite of the generator in the chamber, the detailed setup and procedure can be found in Refs. [11,12]. As optimised
previously [17], SMAT was executed for 20 min for all the
samples to achieve the finest NC grains and the thickest NC
layer. After SMAT, 20 × 20 × 10 mm small blocks were cut off
for surface alloying. The surface alloying process is described
as follows: The small blocks were buried in a powder mixture of
70 wt.% Al, 25 wt.% Al2O3 and 5 wt.% Zn in a crucible, which
occupied two-thirds of the crucible. Then the crucible was
topped up with a mixture of foundry sand and coke to reduce
oxidation of the underlying metallic powder and alloy. Al2O3
powders were used to prevent the metallic powders from
consolidating in the subsequent heating process. The surface
alloying process was conducted at 400°C for 24 h to get the
thickest alloyed layer [18]. To prepare TEM specimens, 0.3 mm
thick slices were cut off from the alloyed surface. After
mechanical grinding to 40–50 μm from the cutting side of the
slices only, 3 mm-diameter discs were punched from the thin
foils, which were then ion-milled in a Gatan PIPS with a small
incident angle till perforation at − 40 °C to minimize the effect
of temperature elevation. All TEM specimens were examined in
a JEOL 2010 microscope operated at 200 kV. X-ray diffraction
(XRD) analysis was carried out in a Rigaku DMAX/2400
diffractometer with an operation voltage of 40 kV using Cu-Kα
radiation.
SEM observations of the microstructure in the alloyed layer
and the morphologies of the worn surfaces were conducted on a
Quanta 600 MK2 SEM. The chemical composition scanning
from the top alloyed surface to the substrate was carried out on
energy dispersive X-ray spectra (EDS) attached to the Quanta
600 MK2 SEM.
Hardness tests were performed along depth of the alloyed
layer by using nanoindentation, which was conducted on a
Nano Indenter XP (MTS) system using a diamond Berkovich
indenter at a constant loading rate. The maximum load was
equal to 20 mN. At least four measurements were taken under
each depth.
Differential scanning calorimetry (DSC) experiments were
conducted on a Perkin-Elmer Pyris Ι instrument to detect the
thermal stability of the SMAT AZ91D Mg alloy samples. The
sensitivity of the energy measurement was about 0.01 mJ/g; the
temperature of the calorimeter was calibrated by using pure In
and Zn standard specimens, with an accuracy of ± 0.02 °C.
Samples cut from the top surface side of the SMAT AZ91D Mg
alloy with a thickness of less than 100 μm were heated in an
argon atmosphere at a constant heating rate of 5 °C/min. The
maximum temperature was 420 °C. The baseline for DSC
curves was determined by reheating the sample at the same
heating rate.
Optimol SRVIII oscillating friction and wear tester was
employed for wear performance examination. Experiments
were conducted in a ball-on-block contact configuration at room
temperature (25 °C) in air with a relative humidity of 40–50%.
The balls used in the experiments were made of WC-Co
composite with a hardness of Hv = 1750. The diameter of the
balls was 10 mm and the size of the blocks was 8 × 8 × 3 mm.
Wear tests were carried out under dry sliding condition with an
oscillating stroke of 2 mm and a frequency of 5 Hz under
normal loads of 5–15 N for 30 min. Wear volume loss was
determined by formula raised by Qu et al. [20].
3. Results and discussion
3.1. Microstructure of the alloyed layer
Our previous work [18] indicated that the thickness of the
alloyed layer is a function of surface alloying temperature and
the thickest alloyed layer can be obtained at 400 °C. Two
regions can be identified from the optical micrographs of the
surface alloyed layer (Fig. 1(a)), the top bright region (A) and
the lamellar structure region (B). XRD patterns of the top
surface layer (A), 50 μm beneath the top surface (B) and
250 μm underneath the top surface (C, the substrate) indicate
that C is a single-phased Mg solid solution while spectrums A
and B provide evidence of the co-existence of Mg17Al12 (γ)
phase and Mg solid solution in the surface alloyed layers (Fig.
1(b)). The intensity of γ phase in the top bright layer is higher
than that in the lamellar structure layer, which implies that the
volume fraction of γ phase in the top bright layer is possibly
higher than that in the lamellar structure layer. Furthermore,
irregular shaped small “island” type γ phase is observed in the
top bright layer, which is different from γ phase in the lamellar
layer. To further understand the atom diffusion process and the
microstructure formation mechanism in the alloyed layers,
EDS linear scan was carried out to investigate the chemical
composition variation from the top surface to the substrate. As
shown in Fig. 2, the concentration of Al in the bright layer is
much higher than that in the lamellar layer, but Mg content
increases dramatically across the boundary between the top
bright region and the lamellar region, and is kept at a constant
H.Q. Sun et al. / Surface & Coatings Technology 202 (2008) 3947–3953
3949
and the single γ phase region forms at the top surface layer.
Therefore, at the late stage of the diffusion process at 400 °C,
there would be two single-phase regions formed within the
alloyed layer, the γ phase region at the top surface layer and the
Mg–Al solid solution region between the substrate and the γ
phase region. During the subsequent cooling process, Mg solid
solution precipitates from the single γ phase region and forms
the bright layer as shown in Fig. 1(a). Similarly, lamella γ phase
precipitates in the Mg solid solution and results in the formation
of the lamella region. Although the lamella region looks like a
eutectic structure, in fact, it is the lamella γ precipitates in Mg
matrix. Hence, the average chemical composition of the lamellar
structure significantly deviates from the Mg–Al eutectic
composition. Based on this understanding, the EDS results in
our previous work [18], which indicated the composition of the
lamellar structure is 83 at.%Mg–15.5 at.%Al–1.5 at.%Zn, can
now be easily understood.
Fig. 1. (a) Cross-sectional microstructure of the SMAT AZ91D Mg alloy after
surface alloying treatment at 400 °C. A is the top surface alloyed layer, C is the
substrate, B is the alloyed layer between A and the substrate C. (b) XRD
spectrums taken from region A, B and C respectively. Formation of Mg17Al12
phase in A and B layers is obvious.
in the lamellar region. The variation of Zn content is not
detectable probably due to the limited resolution of EDS.
These chemical composition variation and microstructure
change within the surface alloyed layer can be explained
based on the understanding of reaction diffusion at 400 °C
during surface alloying treatment. At the early stage of the
diffusion process, Al atoms with small amount of Zn atoms
diffuse from the surrounding powders into the surface of
AZ91D Mg alloy blocks and consequently lead to an increase
in Al content in the surface and the formation of an Al
concentration gradient from the surface to the substrate. As the
diffusion process continues, the Al concentration within the
top surface layer keeps rising and the alloyed layer is
thickened. However, in order to fulfil the requirement of
reaction diffusion to maintain the thermodynamic equilibrium
between phases in the diffusion process, two phase regions can
not be generated. Thus, according to the Al–Mg phase diagram
[21], once the Al concentration in the top surface exceeds the
solid solubility of Al in Mg at 400 °C (close to 13 wt.% Al), the
concentration of Al dramatically increases to around 43 wt.%
Fig. 2. SEM micrograph of the AZ91D Mg alloy after surface alloying treatment
at 400 °C (a) and EDS linear scanning spectrum showing the variation of Mg, Al
and Zn content with the distance from the surface (b).
3950
H.Q. Sun et al. / Surface & Coatings Technology 202 (2008) 3947–3953
Fig. 3. Bright field images of lamella γ precipitates in the alloyed layer (Both (a)
and (b)). (c) is the corresponding SAED pattern taken from the precipitate (as
indicated by the white circle in Fig. 4(b)) indicating a BCC structure of γ phase.
TEM observation of the lamellar structure in the alloyed
layer found the γ lamella precipitates as shown in Fig. 3(a)
and (b). A selected area diffraction pattern (SAED) shown in
Fig. 3(c) taken from a precipitate in Fig. 3(b) confirms a BCC
structure of γ precipitate. These results are consistent with
previous works [22–24] on the precipitation of γ phase in
Mg–Al alloys.
Fig. 4. Cross-sectional optical micrograph of the surface alloyed layer of the CG
AZ91D treated at 400 °C for 24 h.
For comparison, a CG sample without SMAT was surface
alloyed at the same condition as used for SMAT samples. The
cross-sectional optical microstructure of the surface alloyed
layer (Fig. 4) illustrates that although the lamellar structure
could be observed along grain boundaries, the volume fraction
of γ precipitates in Fig. 4 is much less than that in the SMAT
sample shown in Fig. 1(a). In fact, it is very similar to the
microstructure of AZ91D Mg alloys after solution treatment and
ageing [22,23], where the γ phase precipitates along grain
boundaries. In addition, it is also indicated that the diffusion of
Al atoms in CG Mg alloy during surface alloying treatment
takes place along grain boundaries. But due to the large grain
size and lack of bulk/volume diffusion, the volume fraction of
the lamellar structure in the CG sample is much less than that in
SMAT samples and the microstructure is discontinued. On the
contrary, SMAT samples with a large volume fraction of grain
boundaries in the surface layer provide more diffusion channels
and the formed lamellar structure layer is continuous.
Another critical question in NC materials is grain growth, i.e.
thermal stability of NC grains in the surface layer. After SMAT,
severe plastic deformation in the surface layer results in a
metastable structure, which will recover in the subsequent
annealing process. In addition, grain growth will also take place
during annealing. DSC was introduced to determine the
temperature at which grain growth would take place. For
comparison, the original AZ91D CG sample was also detected.
At a heating rate of 5 °C/min, a heat release peak appears at about
130 °C, as shown in Fig. 5, corresponding to grain growth in the
NC surface layer of SMAT samples. Hence, it is impossible to
avoid grain growth when the surface alloying treatment was
performed at 400 °C. The results in the present work and
previous work [18] also show that the actual grain sizes in the
alloyed layer are in micro-meter scale rather than nano-meter
scale. But, they are still much smaller than those in the substrate.
Besides, it is reasonable to believe that some of the Al atoms
have already diffused into the surface layer before the grain
growth begins, this preceding diffusion might be quite crucial for
the formation of the thick surface alloyed layer even though the
grain boundary diffusion may be weakened as the grains growth
begins. Hence, any approach that can suppress grain growth
during surface alloying process might be able to further lower
the surface alloying temperature.
Fig. 5. DSC curves of the SMAT surface layer of AZ91D Mg alloy (Solid) and
the CG sample (dashed) at a heating rate of 5 °C/min. The arrow in the figure
indicates the exothermic heat release.
H.Q. Sun et al. / Surface & Coatings Technology 202 (2008) 3947–3953
Fig. 6. Variation of hardness with depth from the top surface alloyed layer. The
designated A. B. C regions correspond to those indicated in Fig. 1. The hardness
of CG AZ91D Mg alloy (designated as CG) is also presented for comparison.
3.2. Hardness of the alloyed layer
Hardness measurements on the cross-section of the alloyed
layer from the top surface to the substrate by using
nanoindentation are illustrated in Fig. 6. The average hardness
of the bright layer can be as high as 3.0 ± 0.4 GPa; at the
depth of 50 μm to about 200 μm, the hardness remains
almost unchanged with an average value of 1.5 GPa; at the
depth above 200 μm, the hardness gets down to below
1.0 GPa. These three values respectively correspond to A and
B regions of the alloyed layer and the substrate C as
designated in Fig. 6. The hardness variation along depth of
the alloyed layer to substrate is consistent with the
microstructure analysis based on XRD and EDS. The top
bright layer contains a larger fraction of γ phase and thus
exhibits a much higher hardness than that of the lamellar
structure alloyed layer with fewer γ phases. The thickness of
the two alloyed layer also corresponds quite well with the
optical micrograph in Fig. 1(a).
3951
Fig. 7. Variations of the wear volume loss with the applied loads for the CG, CG400 and SMAT-400 samples at an applied load of 7 N and the sliding duration of
30 min (CG-400 and SMAT-400 respectively denotes the CG and the SMAT
sample after surface alloying at 400 °C).
is relatively lower than the CG sample. With the increasing of
the applied load, the wear volume loss of these three samples
increases, and the difference of wear volume loss between the
SMAT-400 and the CG, the CG-400 gets smaller. As presented
earlier, the surface alloying treatment at 400°C subsequent to
3.3. Wear behaviour of the surface alloyed layer
3.3.1. Wear volume
Wear volume loss of CG samples (CG), CG samples after
surface alloying at 400 °C (CG-400) and SMAT samples after
surface alloying at 400 °C (SMAT-400) were determined at
various applied loads on ball-on-block sliding configuration.
Considering the thickness of the alloyed layer of the SMAT400 sample, the wear test were carried out at the applied load
ranges from 4 to 11 N, under which the wear depth of the
SMAT-400 sample is less than 150 μm. Fig. 7 illustrates the
variation of the wear volume loss with the applied load for the
three samples. It is evident that the SMAT-400 sample exhibits
the lowest wear volume and the highest wear resistance in
these three samples over the entire applied load range. When
the applied load is 5 N, there is almost no detectable material
loss for the SMAT-400 sample, the wear volume loss of the
CG-400 sample is much higher than that of the SMAT-400, but
Fig. 8. Morphologies of the typical worn surface of the CG-400 (a) and the
SMAT-400 (b) sample after sliding for 30 min under a load of 7 N.
3952
H.Q. Sun et al. / Surface & Coatings Technology 202 (2008) 3947–3953
As indicated previously [25], the formation of debris in
abrasion, depends on strength and hardness of materials, and in
most cases, the Archard's law [26] is obeyed. This means that
the wear rate (the total wear loss per unit sliding distance) is
inversely proportional to the hardness of the material. Higher
hardness of materials will lead to smaller wear volume loss.
The hardness of the lamellar structure in the surface alloyed
layer is approximately two times higher than that of the
substrate. Since the SMAT-400 sample contains much more γ
lamella precipitates within the alloyed layer than the CG-400
does, it exhibits much higher wear resistance than the CG-400
or the CG sample. In addition, examination on the crosssection of the CG-400 sample after sliding observed high
density deformation twins underneath the alloyed layer while
the twin density in the SMAT-400 sample was quite low, as
Fig. 9 illustrated. This suggested that during sliding the
substrate of the CG-400 sample underwent heavier plastic
deformation than that of the SMAT-400 sample. In other
words, the surface alloyed layer of the SMAT sample
efficiently prevented the substrate from such deformation.
Fig. 9. Cross-sectional optical micrographs of the CG-400 (a) and the SMAT400 (b) sample after the wear test presents the deformation twins in the substrate
underneath the alloyed layer.
SMAT resulted in the formation of a 100 μm to 200 μm-thick
alloyed layer on the surface of the AZ91D Mg alloy, under the
applied load of 5 N, sliding may undergo mainly on the top
bright layer with a hardness of 3.0 GPa, as the applied load
increases, the reciprocal movement will gradually get into the
lamellar structure layer. Due to the existence of lamella γ
precipitates within this layer, the wear resistance is pronouncedly improved in comparison with CG-400 and CG samples.
The wear volume loss of the CG-400 sample is very close to
that of the CG sample although the former exhibits slightly
lower wear volume loss due to the formation of lamella γ
phase along the coarse grain boundaries through grain
boundary diffusion. However, compared with the SMAT-400
sample, the CG-400 sample presents a much lower loadbearing ability. Hence it is reasonable to conclude that surface
alloying of AZ91D Mg alloy with Al as a subsequence of
SMAT is an efficient approach to improve the wear resistance
of the alloy.
3.3.2. Wear mechanism
SEM examination on the worn surfaces of both CG-400 and
SMAT-400 samples indicates a similar surface configuration
within the applied load ranging from 5 to 11 N. Fig. 8 is the
typical worn surface morphologies at the applied load of 7 N of
both the CG-400 (a) and the SMAT-400(b) sample. Ridges and
grooves that are parallel to the sliding direction imply the debris
abrasive wear mechanism. The formation of the grooves is
resulted from the hard asperities on the ball or/and hard particles
in-between the ball and the block. The hard asperities or the
hard particles plough or cut into the block and cause wear
through the removal of small fragments, as a result, scratches
formed on material surface.
4. Conclusions
1. Surface alloying of AZ91D Mg alloy subjected to SMAT
leads to the formation of a much thicker alloyed layer than
the CG sample under the same alloying condition. The
alloyed layer of SMAT AZ91D Mg alloy is composed of two
layers with different compositions. The top surface layer
consists of γ matrix and Mg solid solution precipitates while
the lamellar structure layer is composed of γ precipitates and
the Mg solid solution matrix. The formation of these two
layers is associated with the reaction diffusion that takes
place during surface alloying process.
2. The surface alloyed layer after SMAT can efficiently
improve the wear resistance of the AZ91D Mg alloys
compared with the conventional CG AZ91D Mg alloys.
Acknowledgements
The authors are indebted to Dr. W. Wang for nanoindentation
measurements. Financial support from NSFC of China
(No.50431010 and No.50621091), Ministry of Science &
Technology of China (No. 2005CB623604), Shenyang Science
and Technology Project(No.1071107-1-00) and Australia
Research Council (ARC) Discovery Project (DP0557213) is
greatly acknowledged by the authors.
References
[1] X. Liu, Surf. Coat. Technol. 131 (2000) 261.
[2] M.H. Staia, M. Cruz, N.B. Dahotre, Wear 251 (2001) 1459.
[3] V.P. Rotshtein, Y.F. Ivanov, A.B. Markiv, et al., Surf Coat. Technol. 200
(2006) 6378.
[4] Y.P. Ma, K.W. Xu, W.X. Wen, X.P. He, P.F. Liu, Surf. Coat. Technol. 190
(2005) 165.
[5] L. Bourgois, B.C. Muddle, J.F. Nie, Acta Mater. 49 (2001) 2701.
[6] N. Pebere, C. Riera, F. Dabosi, Electrochim. Acta 35 (1990) 555.
[7] G.L. Song, A. Atrens, Adv. Eng. Mater. 1 (1999) 11.
H.Q. Sun et al. / Surface & Coatings Technology 202 (2008) 3947–3953
[8] J. Horváth, R. Birringer, H. Gleiter, Solid State Commun. 62 (1987) 319.
[9] R. Würschum, P. Farber, R. Dittmar, P. Scharwaechter, W. Frank, H.-E.
Schaefer, Phys. Rev. Lett. 37 (1997) 1171.
[10] R.W. Balluffi, Grain boundary diffusion mechanisms in metals, in
Diffusion in Crystalline Solids, edited by G.E. Murch and A.S. Nowick,
Academic Press Inc. Orlando, 1984.
[11] K. Lu, J. Lu, J. Mater. Sci. Technol. 15 (1999) 193.
[12] K. Lu, J. Lu, Mater. Sci. Eng. A. 375–377 (2004) 38.
[13] W.P. Tong, N.R. Tao, Z.B. Wang, J. Lu, K. Lu. Science 299 (2003) 686.
[14] Z.B. Wang, N.R. Tao, W.P. Tong, J. Lu, K. Lu, Acta Mater. 51 (2003)
4319.
[15] A. Torosyan, L. Takacs, J. Mater. Sci. 39 (2004) 5491.
[16] Á. Révész, L. Takacs, J. All. Compd, 441 (2007) 111.
3953
[17] H.Q. Sun, Y.N. Shi, M.X. Zhang, K. Lu, Acta Mater. 55 (2007) 975.
[18] M.X. Zhang, Y.N. Shi, H.Q. Sun, P.M. Kelly, J. Nanosci. Nanotechnol 8
(2007) 1.
[19] M.X. Zhang, P.M. Kelly, J. Mater. Res. 17 (2002) 2477.
[20] J. Qu, J.J. Truhan, Wear 261 (2006) 848.
[21] T.B. Massalski, Second Edition, Binary alloy phase diagrams, vol. 1, ASM
International, December 1990, p. 169.
[22] D. Duly, M.C. Cheynet, Y. Brechet, Acta Metal. Mater. 42 (1994) 3843.
[23] S. Celotto, Acta Mater. 48 (2000) 1775.
[24] A.F. Crawley, K.S. Milliken, Acta Metall. 22 (1974) 557.
[25] H.Q. Sun, Y.N. Shi, M.X. Zhang, Surf. Coat. Technol. 202 (2008) 2859.
[26] J.F. Archard, J. Appl. Phys. 24 (1953) 981.