Multilayered WCCo/Cu coatings by warm spray deposition

Makoto Watanabe , Masayuki Komatsu, Seiji Kuroda

High Temperature Materials Unit, National Institute for Materials Science, 1-2-1 Sengen, Tsukuba, Ibaraki, 305-0047, Japan
a b s t r a c t a r t i c l e i n f o
Article history:
Received 8 March 2011
Accepted in revised form 31 May 2011
Available online 15 June 2011
Keywords:
Warm spray
WCCo
Multilayered coatings
Fracture toughness
Damage tolerance
WCCo/copper multilayer coatings consisting of 8 layers were fabricated by warm spray deposition in order
to investigate the effect of ductile layer inclusion onto their fracture behavior. Bending strength, work of
fracture, and surface hardness of freestanding coatings were examined by three point bending tests after
removal of the substrates. The multilayered samples showed non-linear stressstrain curves due to crackings
in the WCCo layers and plastic elongation of the copper layers. The multilayered samples with lower volume
fraction of copper showed even lower bending strength than the monolithic samples of WCCo and copper
and no benecial feature in mechanical performance was found. On the other hand, the samples containing
higher volume fraction of copper exhibited more than twice higher work of fracture and moderately better
bending strength than the monolithic WCCo coatings, while the surface hardness was almost identical
among all samples instead of the monolithic copper. The ductility of copper layers and the plastic constraint
by the intact WCCo layers attributed to enhance their mechanical properties. It has been concluded that
cermet/metal laminate coatings can be one alternative approach to further improvement of the mechanical
properties of thermal sprayed cermet coatings.
2011 Elsevier B.V. All rights reserved.
1. Introduction
Thermally sprayed WCCo coatings have been applied to various
industrial components in order to protect them from aggressive
environmental damages such as abrasive wear, erosion, and foreign
object damages [13]. High velocity oxy-fuel (HVOF) spraying is one
of the most popular deposition techniques of cermet coatings in
recent days. While coatings deposited by HVOF show similar or even
higher hardness of 10001300 Hv than sintered bulk WCCo
materials, their fracture toughness is much inferior to the bulk
materials. The origins of the poor fracture properties are formations of
brittle phases such as W
2
C and phase (M
12
C and M
6
C) due to
decarburization of WC and dissolution of W and C into Co binder
during spraying.
In order to improve the fracture toughness and overall mechanical
performance of WCCo coatings, mainly two approaches have been
taken. One is to deposit WCCo coatings at lower process temperature
and to suppress those detrimental reactions during ying of particles.
Jacobs et al. [4,5] deposited WCCo and WCCoCr coatings by high
velocity air-fuel (HVAF) spraying. The coatings did not show the
formation of brittle phases. While the hardness of HVAF WCCo was
decreased, the WCCoCr coating showed improved hardness and
wear resistance. Kim et al. [6] successfully applied cold spray (CS)
deposition to fabricate WCCo coatings without the detrimental
phases mentioned above. While the coatings exhibited extremely
high hardness of 18202050 Hv, the overall performances of these
coatings such as toughness and wear properties are still unknown.
Watanabe et al. [7] and Chivavibul et al. [8,9] investigated several
mechanical properties of WCCo coatings with various Co contents
fabricated by warm spray (WS) and high velocity oxy-fuel (HVOF)
depositions and found the improved toughness for WC-17 and 25% Co
by WS. The wear performances of WS coatings such as abrasive wear
resistance were also better than those of HVOF coatings for a given
hardness value. However, there still remained a substantial gap
between the toughness of WS WCCo and that of a corresponding
sintered body.
Another approach to improve the mechanical properties of cermet
coatings is addition of metallic phase. Osawa et al. [10] deposited a
mixture of WC20Cr
3
C
2
7Ni agglomerated-sintered powder and Ni or
NiCr powder by HVOF and investigated the abrasive wear and impact
resistance of the coatings. The deposited coatings contained the
uniformly distributed Ni or NiCr particles surrounded by WCCrCNi
cermet matrix. The coatings showed a signicant improvement in
wear and impact properties due to energy absorption by the ductile
metallic particles. Hadad et al. [11] investigated adhesion strength and
impact behavior of HVOF cermet coatings consisting of three layers, in
which a metallic intermediate layer (Ni, NiCr, CoCr) was sand-
wiched by the WCCrCo layers. The intermediate layers were
deposited by either HVOF or electroplating. The impact resistance of
the coatings was tested by an experimental shooting device and the
cermet coating with the Ni-electroplated layer exhibited better
performance than the monolithic coating. The results implied the
Surface & Coatings Technology 205 (2011) 53585368
Corresponding author. Tel.: +81 29 859 2469; fax: +81 29 859 2401.
E-mail address: watanabe.makoto@nims.go.jp (M. Watanabe).
0257-8972/$ see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.surfcoat.2011.05.054
Contents lists available at ScienceDirect
Surface & Coatings Technology
j our nal homepage: www. el sevi er. com/ l ocat e/ sur f coat
possibility of further enhancement of fracture properties in cermet
coatings by introducing multilayer structures. Valarezo et al. [12]
investigated HVOF WCCo/stainless steel functionally graded coatings
(FGC) consisting of six different layers with a stepwise change in
composition from 100% stainless steel to 100% WCCo. The FGCs
exhibited high fracture resistance for vertical crackings due to energy
absorption of the distributed metal phases and due to relaxation of
residual stress in the coatings. In all the three cases, the essential point
is adding the ductile phases in the coatings.
Based on the previous studies, the design strategy to maximize the
fracture properties of WCCo coatings can be summarized as follows.
The rst is to suppress the detrimental reaction during spray and the
second is to add the metal reinforcements such as particles, bers, or
layers. Fiber reinforced composites [13] or laminates [14,15] are well
known to have greater fracture toughness than their monolithic
counterparts. Although it is of great interest to develop ber
reinforced coatings as studied in [16,17], multilayer coatings has
been chosen in the present study because of the difculty of
reinforcing coatings by continuous bers. However even for multi-
layer system, since it requires rather complex operations to deposit
many layers, only few works have been attempted in the past.
Ravichandran et al. [18] investigated thermal conductivity of
multilayer coatings of alumina and yttira-stabilized zirconia, which
were deposited by plasma spraying. The maximum number of layers
was 16. Although the effectiveness of multilayer structure on
reduction of thermal conductivity was small, it assured the potential
of thermal spray technology to make a multilayer structure. In the
present paper, WCCo/copper multilayer coatings containing 8 layers
have been developed by warm spray deposition aiming at damage
tolerance of cermet coatings with keeping high hardness on their
surfaces. WS copper coatings showed high strength with moderate
ductility in previous study [19,20] and hence copper was chosen as
metal reinforcements in the current work. In order to reveal the basic
and fundamental aspects of fracture characteristics of multilayer
coatings deposited by warm spraying, the strength and fracture
behavior of the freestanding coatings were investigated by three point
bending tests.
2. Experimental procedure
Tungsten carbidecobalt (WCCo) and copper (Cu) layers were
deposited one after the other onto a pure aluminum plate with a
dimension of 501002 mm
3
by warm spray (WS) deposition.
Powder of WC12wt.%Co (Fujimi Inc., Aichi, Japan) with a particle size
of 520 mand a carbide size of 0.2 m, and copper powder (Cu-ATW,
Fukuda metal foil and powder co. ltd., Japan) with a particle size under
45 m were used. The WCCo powder was manufactured by spray
drying and light sintering, and the copper powder was fabricated by
water atomization. The aluminum substrate was chosen because it is
easy to dissolve in alkali solution in order to obtain a freestanding
coating.
A WS system[21] has been developed by modifying a conventional
HVOF equipment (JP5000, Praxair Technology Inc., USA) by adding a
mixing chamber between a combustion chamber and powder feed
ports. It is capable to control temperature of combustion ame by
injecting nitrogen gas at the mixing chamber so that this process can
keep temperature of sprayed particles under their melting point with
moderately heated and thermally softened states. The detail of the
process can be found elsewhere [22]. The spraying conditions of WS
process are listed in Table 1. In these conditions, both powders of WC
Co and copper were sprayed as solid states and bonded by ultra-high
speed impact. Surfaces of the substrates were blasted by alumina grit
and degreased by ultrasonic cleaning in acetone before spraying. Five
combinations of multilayer coatings were fabricated as listed in
Table 2. Samples were labeled as A, B, C, D, E in terms of Cu volume
fraction, v
m
=0.0, 19.4, 31.8, 62.0, and 100.0%. The Cu volume fraction
was controlled by varying the powder feed rate of Cu powder
(Table 1) and thus changing their thickness ratio. Samples A and E are
monolithic WCCo and copper coatings respectively. In all cases
except for sample A, the rst layer on the substrate was Cu and the
outer most surface was WCCo. The total number of layers was xed
at 8 containing 4 WCCo and 4 copper layers in all laminate samples
BD. Since the number of layers was xed, the total thicknesses of
coatings were varied among samples as shown in Table 2. After
deposition of coatings, the samples were cut into rectangular bars
with a dimension of 505 mm
2
(lengthwidth). All the bars except
for sample E were immersed into NaOH aqueous solution (NaOH:
15 g, distilled water: 500 ml) at a temperature of 60 C, and the
aluminum substrates were dissolved in order to obtain freestanding
coatings. For sample E, the substrate was mechanically removed
because it was very easy to obtain Cu coatings with more than 2 mm
thick and thus the machining was handy for this sample.
Three point bending tests were conducted on those freestanding
coatings. The outer span L was 40 mm and the crosshead speed of
0.1 mm/min was applied. Three specimens were tested for each case.
For the cases of the multilayer coatings, the copper side surface, which
was the rst layer during deposition, was placed in the compression
side and thus the WCCo surface, which was the eighth layer, was
placed on the tension side. The load P and crosshead displacement
were recorded. Fracture behaviors were monitored during the
bending tests by a CCD camera. The strain
t
and the apparent
bending stress
t
on the outer most tension side for arbitrary were
calculated as [23]

t
=
6t
L
2
1

t
=
3
2
L
wt
2
P 2
where wand t is the width and thickness of the specimen. Note that
t
is only valid for elastic deformation and cannot represent correct
stress states after yielding, and thus this value should be considered as
the apparent parameter. In addition, apparent work of fracture was
dened and calculated as the area under the load-displacement curve
divided by the twice the cross section area of the specimen [24,25],
=

max
0
Pd
2wt
3
Table 1
Spray parameter.
Parameter Cu WCCo
Fuel (dm
3
/min) 0.35 0.38
Oxygen (dm
3
/min) 713 779
Nitrogen (dm
3
/min) 1000 500
Powder feed rate (g/min) 63, 32, 16 85
Spray distance (mm) 180
Barrel length (mm) 203.2
Table 2
Sample list.
Sample
ID
Number of layer Thickness of layer
(m)
Amount of Cu
(vol. %)
WCCo Cu WCCo Cu
A 1 0 600.0 0.0
B 4 4 98.9 23.7 19.4
C 4 4 93.7 43.6 31.8
D 4 4 74.0 121.0 62.0
E 0 1 1970.0 100.0
5359 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368
where
max
stands for the maximum displacement of the crosshead.
The microstructure and crack patterns were examined with a
scanning electron microscope (SEM, JSM-6500F, JEOL Ltd., Japan)
and the phases in the monolithic coatings (samples A and E) were
determined by X-ray diffraction analysis (XRD, RINT2500, Rigaku
Corp., Japan) on the surface of as-sprayed samples. Microhardness
tests were also carried out on the surface with a 300 g load and a dwell
time of 15 s. At least ten measurements were made for each sample.
Oxygen contents in the copper powder and coatings were analyzed by
the inert gas fusion method (LECO TC-600, Leco corp., US).
3. Results
3.1. Microstructural characterization of coatings
Fig. 1 shows the cross-sectional images of the monolithic coatings
of sample A (Fig. 1a,b) and sample E (Fig. 1c,d). The WS WCCo
coating has very dense microstructure with tightly packed blocky
carbides. In Fig. 1b, some carbide are larger than 1 m possibly due to
the crystal growth during light sintering after spray drying in the
powder manufacturing process. The images of the copper coatings
were taken after removal of the substrate. Although the microstruc-
ture of the Cu coating appears very clean and dense, some interfaces
among the deposited particles contain voids (Fig. 1d). This implies
lower mechanical properties of the WS copper coating than bulk
copper metal. XRD analysis results are plotted in Fig. 2. Sample A
consists of only WC and Co phases indicating lowprocess temperature
in WS and suppression of the formation of detrimental phases such as
W
2
C as already reported [79]. On the other hand, sample E showed
not only the peaks of copper but also of a copper oxide CuO. It
indicates that the particle temperatures were high enough to cause
oxidation during ight. The oxygen content in sample E was 0.24 mass
% while the value in the feedstock powder was 0.096 mass%.
The microstructures of samples B, C, and D are indicated in Fig. 3a
b, cd, and ef respectively. They consist of four WCCo (lighter gray)
and four copper (darker gray) layers. Since the feed rate of copper
powder has been varied in order to control the volume fraction of
copper, the average copper thicknesses are different among the three
samples while the thickness of WCCo layers was almost constant
(Table 1). The vertical cracks were observed in some WCCo layers in
all three samples (Fig. 2a,c,e). The cracks seemto have occurred at the
second WCCo layer from the interface between the coating and
substrate, and propagated toward the coating surface. Such cracking
was not observed in the monolithic WCCo sample. There are two
possible mechanisms behind, causing this phenomenon. One is the
larger thermal expansion of the Cu layers than WCCo. In order to
spray multilayer structures, it was necessary to change the powder
feeding lines for each layer and correspondingly the spray conditions,
which it took some time so in the present spraying system. As a result,
the coating-substrate couple was cooled down and reheated again
before the deposition of a next layer. Since the thermal expansion of
Cu [26] is much larger than WCCo [27,28], these thermal cycling
might cause such cracking. The other possible explanation is the
Cu
WC-Co
a
100 m 1 m
10 m
500 m
b
c d
Fig. 1. Cross-sectional images of monolithic coatings of (a)(b) WCCo and (c)(d) copper.
a
b
Fig. 2. Cross-sectional image of laminated coatings deposited by warm spraying: (a)(b)
sample B, (c)(d) sample C, (e)(f) sample D.
5360 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368
impact of solid Cu particles. Although sprayed particles are heated and
softened in WS process, the surface impacted by the high velocity
particles is subjected to substantial impact energy and it can be
cracked especially for a brittle material. Since such pre-existing cracks
will reduce the mechanical properties of coatings, it is necessary to
understand the mechanism and to avoid crackings in the future. In
most of the cases, cracks were not recognized in the copper layers
indicating higher fracture toughness of the copper layers than WCCo
as expected.
3.2. Mechanical properties of coatings
Examples of typical stressstrain curves obtained from the mono-
lithic coatings of WCCo (sample A) and copper (sample E) by Eqs. (1)
and (2), are shown in Fig. 4. In the case of WCCo coating, the stress
increased linearly with the increase of strain, and fracture occurred in a
brittle manner at
t
=433 MPa and
t
=0.19%. On the other hand, the
Cu coating exhibited signicant plastic elongation after elastic defor-
mation similar to bulk Cu. The 0.2% proof stress was 301 MPa. After
reaching anapparent ultimate strengthof 317 MPa at
t
=0.87%, ductile
failure initiated and the stress gradually decreased for further
deformation. Mechanical properties of the coatings were summarized
in Table 3. The reference data of bulk WCCo and copper materials
[28,29] are also presented in the table. The bending modulus and
strength of WS WCCo coating are about 40% and 20% of the bulk
counterpart with an average carbide size of 1 m. On the other hand,
mechanical properties of WS copper coatings are comparative to those
of rolled copper [29]. However, it should be noted that the mechanical
properties of materials strongly dependontheir microstructures suchas
carbide size, grain size, oxide content, degree of hardening, and so on.
Thus the reference values in the table should be interpreted as a rough
measure of bulk properties. The Vickers hardness on the surface of
samples is also shown in Table 3. The laminate coatings have even
higher values than the monolithic WC12Co coating. At the moment, it
is not clear whether this difference was originated from the scattering
due to the spray based process or the intrinsic properties due to
laminate structures. One possible explanation, for the latter case, might
be the introduction of higher compressive stress inWCCo layers due to
the existence of copper ones. Since the thermal expansion coefcient of
copper is much larger than that of WCCo, thus higher shrinkage can
occur in the copper layer during cooling stage after spraying. This
shrinkage can induce compressive stress in the intact WCCo layers.
However, the real residual stress states in the layered coatings are
unknown. This requires further investigation by measuring the residual
stress development through the deposition process. But it is expected
that the laminates have at least comparable wear properties to the
monolithic coatings until the top WCCo layer is worn out.
a
100 m 100 m 100 m
20 m 20 m 20 m
c e
b d f
Fig. 3. Cross-sectional image of laminated coatings deposited by warm spraying: (a) (b) sample B, (c)(d) sample C, (e)(f) sample D.
a) Sample A
b) Sample E
Fig. 4. Stressstrain curves of monolithic coatings of WCCo (sample A) and copper
(sample E) in three point bending tests.
5361 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368
Typical stressstrain curves of laminated coatings with different
copper layer thicknesses, samples B, C, and D, are given in Fig. 5. In
sample B (Fig. 5a), the load did not increase linearly with strain
increase but showed gradual reduction of the gradient resulted in the
non-linear increase of stress. This behavior was possibly caused by
cracking in the WCCo layer and following plastic deformation of Cu
layer on the tensile side. After several small dropping and recovering
as shown in the inset of Fig. 5a, the maximum stress of
t
=183 MPa
at
t
=0.17% was reached. Beyond the load maximum, the stress
strain curves exhibited large load drops. However it is worth noting
that load did not drop catastrophically like sample A (Fig. 4a) but
decreased in a step like manner with failure proceeding. Samples C
and D generally showed similar trend in stressstrain behaviors
(Fig. 5b,c). Both of the ultimate strength and the strain to failure
increased as the increase of volume fraction of copper. Interestingly,
the ultimate strength of sample D, which contained 62% volume
fraction of copper, was higher than sample E (monolithic copper) but
the strain to failure was only 20% of the sample E. This is perhaps
caused by restriction of plastic deformation in the copper layers due to
strong bonding with adjacent WCCo layers.
3.3. Fracture surface of monolithic coatings
SEM micrographs of the fracture surfaces of samples A and E are
presented in Fig. 6ac and df respectively. The fracture surface of
sample A is very smooth and at indicating brittle failure (Fig. 6a).
There exist many small pores, which were difcult to recognize in
Fig. 3b due to plastic deformation of binder phases during mechanical
polishing. These pores could be an origin of lower strength of sample
A than a sintered bulk WC12Co material. Although some cleavage
facets of carbides can be observed as indicated by the arrow in Fig. 6c,
fracture mainly occurred at the interfaces of WC/Co or WC/WC. In the
fracture surface of sample E, the deformed morphologies of individual
copper particles due to high velocity impact can be clearly recognized
in Fig. 3d. There are two fracture modes, one is brittle fracture at
particle boundaries as shown in Fig. 6e and the other is ductile
fracture with elongated dimple formation (Fig. 6f). These local plastic
deformations contribute the plastic elongation of sample E during the
bending test in Fig. 4b.
3.4. Fracture paths in laminate coatings
The typical cross-sectional images of crack paths at the center
regions in samples B, C, D are presented in Fig. 7ac, df, and gi
respectively. Please note that these images are not always taken from
the same test pieces in Fig. 5. In all pictures, the tension side is placed
at the bottom and thus cracking directions are from bottom to top. In
addition, in all cases, the order of spray was from the top to bottom,
and thus the bottom layer, which was sprayed at last, is WCCo, and
the top layer, which was sprayed at rst, is copper in all cases. In
sample B, two cracks were formed in the WCCo layer in tension side
and it seems that those cracks merged into one (Fig. 7a), propagated
in a straight line (Fig. 7b), and then deected at the interface between
WCCo and copper layers (Fig. 7c). The small load drops and non-
linear stress increase observed in the stressstrain curve (Fig. 5),
corresponds to the formation of cracks in the outer WCCo layer, and
the load reduction in a step like manner after maximum point can be
the results of the deections at the interface between layers. Since the
thickness of the copper was very thin and was not strong enough to
prevent crack propagation, the maximum load was the lowest
(Fig. 5a). In the case of sample C, the main crack did not go straight
but curved (Fig. 7d). The plastic elongation of a copper layer (Fig. 7e)
and bridging of the crack propagation (Fig. 7f) can be observed. From
these pictures, it is understood that the WCCo layer at the crack front
fractured before the copper layer was broken and that the main crack
propagated and curved so as to connect to the crack formed in WCCo
layer in front due to the stress concentration around the cracks. This
fracture behavior is more pronounced in sample D, which contained
the largest volume of copper (Fig. 7gi). The crack branching can be
observed in the second layer from the bottom. This probably occurred
due to the crack generations in the third WCCo layer prior to the
fracture of the second copper layer. Large deection in the fourth
copper layer was also due to cracking in WCCo layer in advance
implying the large fracture resistance in this sample (Fig. 7h and i). It
can be clearly said that the highest fracture strength and strain of
Table 3
Mechanical properties. *The values of bulk materials (ID: F and G) were taken from Refs.[28] and [29].
ID Sample type Bending modulus (GPa) Apparent Ultimate strength (MPa) Strain to failure (%) Hardness on surface (Hv)
A WC12Co monolithic coating 2349 40145 0.170.02 1409125
B Laminate coating 13356 17626 0.340.07 154262
C Laminate coating 17612 30173 0.310.06 1565128
D Laminate coating 16624 42839 0.550.04 158883
E Cu monolithic coating 657 31761 1.870.74 924
F* WC12Co bulk material [28] 570 1800 0.003 1300
G* Rolled copper [29] 3900.06 2.630.06
a) Sample B
b) Sample C
c) Sample D
Fig. 5. Stressstrain curves of laminated coatings in three point bending test.
5362 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368
sample D among all samples (Figs. 4, 5) is due to higher fracture
resistance from crack bridging, deection, and branching in copper
layers. Fig. 8a exhibits the microstructure in the Cu layer in front of
main crack in sample D. The main crack propagated from the bottom
and the vertical crack formed in the WCCo layer before the main
crack reached into it, and the main crack was deected into Cu layer to
the direction parallel to the interface of two layers. Large shear
deformation is manifest from the distorted pore shapes at particle
boundaries in the ligament (Fig. 8b). In addition, during crack
propagation in the Cu layer, it appears that the particle boundaries
were preferentially fractured and well bonded interfaces still
remained bonded and bridging the crack opening in the crack tip
(Fig. 8a). The delamination can be also observed at the interface
between the Cu and WCCo layers in front of the deected crack tip.
In Fig. 9, the SEMimages of the fracture surfaces of laminates sample
B (Fig. 9ac) and sample D (Fig. 9df) are shown. In sample B, the
fracture surface is smooth indicating the occurrence of brittle failure.
The thickness of the copper layer is about 20 minFig. 9b andthe plastic
deformation cannot be observed clearly (Fig. 9c). Since the Cu layers are
substantially thin compared with WCCo, their plastic deformation was
restricted by the neighboring WCCo layers, which resulted in the very
limitedelongation of those layers. Insample D, the fracture surface has a
stepdue to crack deection(Fig. 9d) andsubstantial plastic deformation
can be found in the Cu layers (Fig. 9e,f). This elongation is the reason of
high strain to failure of sample Dobserved in the bending test as shown
in Fig. 5. Fig. 10 is the sequential images of fracture propagation from
(a) to(f) during the threepoint bending test of sample D. Beforeloading,
pre-cracks can be observed on the right side of the fth and seventh
WCCo layers (dotted circle in Fig. 10a). As the load increased, a vertical
crack was initiated in the outer three layers of WCCo/Cu/ WCCo
(Fig. 10b). Right after the crack initiation, delamination occurred at the
interface of between the fourth and fth layers of Cu and WCCo
(Fig. 10b,c), and the fourth layer bridged and prevented the main crack
from propagating straightly (Fig. 10c). The delamination propagated
along the interface and connected with the pre-existing crack on the
right side in the fth WCCo layer (Fig. 10c). At the center region, a new
vertical crack also formed from the delamination (Fig. 10d). Finally, the
fourth Cu layer was fractured and then sixth Cu layer was broken
(Fig. 10e). The sample lost most of load sustainability (Fig. 10f). It is
concluded that the laminate coatings with higher volume fraction of Cu
exhibited better fracture strain because the WS Cu layers has better
resistance for crack propagation than WCCo due to higher elongation
to failure.
4. Discussions
4.1. Fracture strength
Measured values of the apparent bending strength, which is
dened as the peak value in the stressstrain curve, are compared
against Cu vol.% in Fig. 11a. Black symbols stand for the monolithic
coatings of WCCo and Cu, respectively and white circles for the
laminates. The error bars represent the standard deviation of the
obtained values. The laminate with the lowest metal fraction (19.4%)
exhibited the lowest strength, and as the volume fraction of metal
increased, the bending strength increased. The highest value was
obtained for the highest amount of Cu in laminates. Also it is
important to note that the highest value among the laminates is
almost comparable to the monolithic WCCo and higher than the
monolithic Cu coating. When a simple mixing law is considered, a
laminate with higher amount of WCCo, thus lower amount of Cu is
expected to have higher strength. This discrepancy may be caused by
two factors. One is the initial cracks in WCCo layer in the laminates.
As can be observed in Fig. 3, most of the as-deposited coatings contain
vertical cracks in the WCCo layers (Fig. 3a,c,e). It is apparent that
stress concentration occurs during loading at the tip of such initial
cracks and the strength of entire body is reduced signicantly. That is
why the fracture strength of sample B (v
m
=19.4%) is substantially
lower than the monolithic sample A. The other factor is the constraint
of plastic deformation in Cu layer by the intact WCCo layers. These
laminates showed much lower strain to failure (Table 3) indicating
the suppression of plastic deformation in the laminate samples. In
Fig. 11, the linear least square t for the laminates is also shown and
the bending strength at v
m
=100% can reach to 658 MPa. This value
a
10 m
20 m
1 m 250 nm
5 m
b c
d e f
10 m
Fig. 6. Fracture surface of: (a)(b)(c) sample A, (d)(e)(f) sample E.
5363 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368
can be considered as the effective constrained strength
e
of the
copper layer. By dividing it by the strength of the monolithic Cu
coating
m
,
e
/
m
=2.1 is obtained. Hwu et al. [30,31] carried out the
w-notched tensile tests of the constrained metal mono-layer in the
metal/ceramic laminates fabricated by diffusion bonding processes.
They reported the ratio of
e
/
m
=1.81, 1.86, and 1.72 for Al/Al
2
O
3
,
Cu/Al
2
O
3
, Ni/,Al
2
O
3
, where
e
and
m
corresponds to the effective
yield strength of the constrained metal layer and the metal foil yield
strength (unconstrained), respectively. The constrained factor
obtained in the present study has similar values to them. Thus it
could be concluded that the high bending strength obtained in sample
D (v
m
=62.0%) is attributed to the plastic constraint of the Cu layer by
the intact WCCo layers.
Failure strain is also plotted in Fig. 11b as a function of Cu vol.%. In
all laminates, the strain to failure is larger than the monolithic WCCo
sample. As the volume fraction of copper increases, the value tends to
increase. Although the sample C has higher volume fraction and
thicker copper layers than sample B, it has almost same value with
sample B. As can be seen in Fig. 5, samples B and C sometimes
exhibited the step like reduction of load after the maximumpeak. This
implies that a main crack was bridged and deected so as to connect
the pre-existing cracks in WCCo layers when it propagated, which
resulted in larger strain to failure. But it is not always the case and
sometimes load quickly dropped like brittle failure. In addition, the
metal layer thickness is not perfectly uniform in one sample, and thus
the thinner metal regions might be preferentially fractured at the
propagation of a main crack due to limited plastic deformation leading
to lower strain to failure. Perhaps these scatterings in fracture
behaviors would be the reason for similar values in strain to failure
between samples B and C.
4.2. Work of fracture
The work of fracture is plotted in Fig. 12 for the volume fraction of
Cu layer. When the volume fraction of Cu is small, is lower than the
monolithic WCCo. As the volume fraction increases, monotonically
increases, and sample D, which contains thickest Cu layers, exhibited
2.53 times higher than the monolithic WCCo. Thus signicant
improvement of toughness can be achieved by laminate structures.
This higher is attributed to the ductile elongation of Cu layers and
bridging at the main crack as observed in Figs. 7 and 10. However, it
appears that the size of bridging zone is limited in the present study
due to relatively low ductility of WS Cu layers as can be seen in Figs. 7
and 10. Hence small scale birding conditions is applicable and the
incremental toughening of the laminates for a metal layer can be
related to the work of stretch and fracture of the bridging layer [32
a
100 m
200 m
200 m
50 m
50 m
100 m
50 m
50 m
100 m
b c
g h i
d e f
Fig. 7. Cross section after three point bend test of (a) (b)(c) sample B, (d)(e)(f) sample C, (g)(h)(i) sample D.
5364 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368
34]. The total fracture energy of the laminates G
lam
can be estimated
as [32,33]
G
lam
= G
cer
1v
m
+ v
m

0
u du 4
where (u) is the nominal stress required to stretch a bridging metal
layer by u, and u* is its failure stretch. G
cer
and v
m
stand for the fracture
energy of WCCo layer and the volume fraction of metal layer,
respectively. The integration termcorresponds to the energy required
to fracture a metal layer G
met
for the given thickness. The degree of
a
20 m
5 m
b
Fig. 8. Crack bridging and deection in Cu layer (a) and higher magnication image of circle region (b).
a
100 m
10 m 5 m
200 m
4 m 50 m
WC-Co
WC-Co
WC-Co
Cu
Cu
Cu
WC-Co
Cu
Cu
WC-Co
b c
d e f
Fig. 9. Fracture surfaces of (a)(b)(c) sample B, (d)(e)(f) sample D.
5365 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368
plastic constraint and the resulted G
met
can be largely varied
depending on its layer thickness, and thus the G
met
is not constant
for the samples with different layer thickness. By following Mataga
[35] and Marakaki et al. [36], Eq. (4) can be rewritten as,
G
lam
= G
cer
1v
m
+ v
m
h
m
2

Y
w 5
w =
u

h
m
0

Y
d
u
h
m

6
where h
m
is the thickness of metal layer and
Y
is metal yield strength.
w is the work of fracture parameter which can be interpreted as the
normalized-fracture energy of a constraint metal layer [35]. By
substituting h
m
=(v
m
/1v
m
)h
c
into Eq. (5), G
lam
can be expressed
as [36],
G
lam
= G
cer
1v
m
+
w
2
h
c

Y
v
2
m
1v
m
!
: 7
where hc is the thickness of WCCo layer. By substituting G
cer
=400 J/
m
2
, h
c
=90 m (Table 3), and
Y
=300 MPa (Fig. 4) into Eq. (7), the
variation of G
lam
was predicted as a function of v
m
for various w(Fig. 13).
G
cer
=400 J/m
2
was chosen to be a half of the value for the monolithic
WCCo coating (817 J/m
2
) because of the existence of crackings in the
WCCo layers in the laminates. Since Eq. (7) diverges at v
m
=1, the
predicted value at higher metal fraction such as v
m
=0.9 are not realistic,
but the trend of the predicted G
lam
appears to have a good agreement
with the experimental data especially for w=2. From Eqs. (4) and (5),
the fracture energy of the metal layer G
met
can be written as,
G
met
=
u

0
u du =
h
m
2

Y
w 8
For w=0.2, G
met
can be predicted as 711, 1308, and 3630 J/m
2
for
samples B, C, and D indicating that thicker metal layers result in
tougher laminate coatings. In addition, it is manifest that larger u* is
very effective to increase fracture energy and thus it would be highly
effective to improve the ductility of Cu layers by changing spray
conditions or by applying post heat treatments.
a
400 m
400 m
400 m
400 m 400 m
400 m
b c
d e f
Fig. 10. Crack propagation behavior from side views during three point bending test of WCCo/Cu layered coatings (sample D). Dotted circle in (a) shows the pre-crack.
S
t
r
a
i
n

t
o

f
a
i
l
u
r
e

(
%
)
a
b
Fig. 11. (a) Variation of bending strength as a function of Cu volume fraction and
(b) variation of failure strain as a function of Cu volume fraction. Fig. 12. Variation of work of fracture as a function of Cu volume fraction.
5366 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368
4.3. Property map
The variations of work of fracture were re-plotted for the fracture
strength of samples BE in Fig. 14. As elaborated so far, sample D
(v
m
=62%) has far better work of fracture and moderately improved
strength over sample E (monolithic WCCo). Thus it clearly demon-
strates the advantages of laminate WCCo/metal coatings. On the other
hand, the laminates which contain the lower amounts of Cu (samples B
and C) did not exhibit any advantages over the monolithic WCCo due
tothe existenceof pre-cracks inthe WCColayer inas-sprayedstate and
the limited plastic deformation of Cu layer. Even for no degradation in
the WCCo layers, it has been reported that the crack extension
resistance of laminates is enhanced as the metal layer thickness
increases and that there is no benet for thinner metal layers when
the volume fraction of metals is xed [14,37,38]. However, for coatings
for wear resistance, it is obvious that thicker metal layers lead to
reduction of total wear resistance of the coating because of poor wear
resistance of soft metals. Thus there should be an optimized volume
fraction of metal layers and thickness ratio of metal and cermet layers in
terms of overall performance of the coating.
In the present study, the interfaces between Cu and WCCo have a
strong bonding and the delamination between layers was only
observed near the neutral axis. There are two most effective
toughening mechanisms in laminates. One is absorption of the
fracture energy with highly ductile layers. In the present study, this
mechanism improved the mechanical properties of the coatings. The
other is the introduction of debonding and/or sliding at the interfaces
among layers. Clegg [39,40] developed ceramic multilayers with very
weak interfaces and demonstrated their extremely high toughness.
Folsom et al. [41] provided a theoretical explanation of the
mechanisms and the experimental demonstrations [41,42]. It is of
great challenge for thermal spray process to introduce such extremely
weak interfaces because moderate bondings are required to pile up
sprayed particles. But if such weak interface were introduced
successfully, it should be possible to form very tough coatings against
the large impact loadings. On the other hand, such weak interfaces
may cause the spallation of layers and thus the interfacial strength
would need to be optimized for practical applications.
5. Conclusions
WCCo/copper multilayer coatings containing 8 layers have been
developed by warm spray deposition. While the coatings with lower
volume fraction of copper showed no benecial feature in mechanical
performance, the coating with higher volume fraction of copper (62%)
exhibited more than two times higher work of fracture and
moderately better bending strength than the monolithic WCCo
coatings. The mechanisms of the improved mechanical properties
were attributed to the ductility of WS copper layers and the plastic
constraint by the intact WCCo layers. The total fracture energy of the
laminates was predicted by the theoretical model for the variation of
work of fracture parameter w. The experimental data showed good
agreement with the prediction for w=0.2. The analysis also
suggested the effectiveness of thicker metal layers to improve the
toughness of the laminate coatings. Poor work of fracture observed in
the laminates with lower volume fraction of copper, was possibly due
to the existence of initial crackings in WCCo layers which was
introduced during spraying. Although the mechanisms to form such
initial crackings remain to be understood for further improvements, it
has been concluded that cermet/metal laminate coatings can be one
alternative approach to enhance the mechanical properties of thermal
sprayed cermet coatings, especially for the spray processes which
have a capability to deposit ductile coatings such as warm spraying.
Acknowledgements
The authors greatly acknowledge Fujimi Incorporated (Japan) for
providing WCCo feedstock powder. This work was supported as a
part of Fail-Safe Hybrid Materials project (20062010) in National
Institute for Materials Science Japan.
References
[1] G. Barbezat, A.R. Nicoll, A. Sickinger, Wear 162 (1993) 529537.
[2] C. Verdon, A. Karimi, J.L. Martin, Mater. Sci. Eng., A: Struct. Mater. Prop.
Microstruct. Process. 246 (1998) 1124.
[3] D.A. Stewart, P.H. Shipway, D.G. McCartney, Wear 225 (1999) 789798.
[4] L. Jacobs, M.M. Hyland, M. De Bonte, J. Therm. Spray Technol. 7 (1998) 213218.
[5] L. Jacobs, M.M. Hyland, M. De Bonte, J. Therm. Spray Technol. 8 (1999) 125132.
[6] H.J. Kim, C.H. Lee, S.Y. Hwang, Mater. Sci. Eng., A: Struct. Mater. Prop. Microstruct.
Process. 391 (2005) 243248.
[7] M. Watanabe, C. Pornthep, S. Kuroda, J. Kawakita, J. Kitamura, K. Sato, J. Jpn. Inst.
Met. 71 (2007) 853859.
[8] P. Chivavibul, M. Watanabe, S. Kuroda, J. Kawakita, M. Komatsu, K. Sato, J.
Kitamura, J. Therm. Spray Technol. 17 (2008) 750756.
[9] P. Chivavibul, M. Watanabe, S. Kuroda, J. Kawakita, M. Komatsu, K. Sato, J.
Kitamura, J. Therm. Spray Technol. 19 (2010) 8188.
[10] S. Osawa, T. Itsukaichi, R. Ahmed, J. Therm. Spray Technol. 14 (2005) 495501.
[11] M. Hadad, M. Hockauf, L.W. Meyer, G. Marot, J. Lesage, R. Hitzek, S. Siegmann, Surf.
Coat. Technol. 202 (2008) 43994405.
[12] A. Valarezo, G. Bolelli, W.B. Choi, S. Sampath, V. Cannillo, L. Lusvarghi, R. Rosa, Surf.
Coat. Technol. 205 (2010) 21972208.
[13] F. Zok, C.L. Hom, Acta Metall. Mater. 38 (1990) 18951904.
[14] H.C. Cao, A.G. Evans, Acta Metall. Mater. 39 (1991) 29973005.
[15] A.G. Evans, B.J. Dalgleish, Acta Metall. Mater. 40 (1992) S295S306.
[16] K.H. Baik, P.S. Grant, J. Therm. Spray Technol. 10 (2001) 584591.
[17] K.H. Baik, P.S. Grant, Mater. Sci. Eng., A: Struct. Mater. Prop. Microstruct. Process.
265 (1999) 7786.
[18] K.S. Ravichandran, K. An, R.E. Dutton, S.L. Semiatin, in, 1999, pp. 673682.
w=0.3
w=0.05
w=0.2
w=0.1
Experiment
Fig. 13. Variation of predicted work of fracture G
lam
as a function of Cu volume fraction
and w.
Fig. 14. Distribution of work of fracture and fracture strength.
5367 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368
[19] S. Kuroda, M. Watanabe, K. Kim, H. Katanoda, J. Therm. Spray Technol. (2011) 124.
[20] S. Kuroda, M. Watanabe, K.H. Kim, H. Katanoda, J. Therm, Spray Technol. 20 (2011)
653676.
[21] J. Kawakita, S. Kuroda, T. Fukushima, H. Katanoda, K. Matsuo, H. Fukanuma, Surf.
Coat. Technol. 201 (2006) 12501255.
[22] S. Kuroda, J. Kawakita, M. Watanabe, H. Katanoda, Sci. Technol. Adv. Mater. 9
(2008) 033002.
[23] S.P. Timoshenko, J.N. Goodier, Theory of Elasticity, second ed. McGraw-Hill Book
Company, Inc., 1951.
[24] H. Tomaszewski, H. Weglarz, M. Boniecki, W.M. Recko, J. Mater. Sci. 35 (2000)
41654176.
[25] H. Tomaszewski, H. Weglarz, A. Wajler, M. Boniecki, D. Kalinski, J. Eur. Ceram. Soc.
27 (2007) 13731377.
[26] ASM Handbook, Mechanical Testing and Evaluation, ASM International, Materials
Park, Ohio, 2000.
[27] M. Toparli, F. Sen, O. Culha, E. Celik, J. Mater. Process. Technol. 190 (2007) 2632.
[28] G.S. Upadhyaya, Cemented Tungsten Carbides. Production, Properties, and
Testing, Noyes, New Jersey, 1998.
[29] F. Gartner, T. Stoltenhoff, J. Voyer, H. Kreye, S. Riekehr, M. Kocak, Surf. Coat.
Technol. 200 (2006) 67706782.
[30] K.L. Hwu, B. Derby, Acta Mater. 47 (1999) 529543.
[31] K.L. Hwu, B. Derby, Acta Mater. 47 (1999) 545563.
[32] A.G. Evans, R.M. Mcmeeking, Acta Metall. 34 (1986) 24352441.
[33] S.J. Howard, S.K. Pateras, T.W. Clyne, Mater. Sci. Technol. 14 (1998) 535541.
[34] B. Budiansky, J.W. Hutchinson, A.G. Evans, J. Mech. Phys. Solids 34 (1986)
167189.
[35] P.A. Mataga, Acta Metall. 37 (1989) 33493359.
[36] A.E. Markaki, T.W. Clyne, Mater. Sci. Eng., A: Struct. Mater. Prop. Microstruct.
Process. 323 (2002) 260269.
[37] Q. Ma, M.C. Shaw, M.Y. He, B.J. Dalgleish, D.R. Clarke, A.G. Evans, Acta Metall.
Mater. 43 (1995) 21372142.
[38] M.Y. He, F.E. Heredia, D.J. Wissuchek, M.C. Shaw, A.G. Evans, Acta Metall. Mater. 41
(1993) 12231228.
[39] W.J. Clegg, Acta Metall. Mater. 40 (1992) 30853093.
[40] W.J. Clegg, K. Kendall, N.M. Alford, T.W. Button, J.D. Birchall, Nature 347 (1990)
455457.
[41] C.A. Folsom, F.W. Zok, F.F. Lange, J. Am. Ceram. Soc. 77 (1994) 689696.
[42] C.A. Folsom, F.W. Zok, F.F. Lange, J. Am. Ceram. Soc. 77 (1994) 20812087.
5368 M. Watanabe et al. / Surface & Coatings Technology 205 (2011) 53585368