You are on page 1of 13

Available online at www.sciencedirect.

com

Surface & Coatings Technology 202 (2007) 509 – 521


www.elsevier.com/locate/surfcoat

Effects of carbide size and Co content on the microstructure and mechanical


properties of HVOF-sprayed WC–Co coatings
Pornthep Chivavibul ⁎, Makoto Watanabe, Seiji Kuroda, Kentaro Shinoda
Composites and Coatings Center, National Institute for Materials Science, Ibaraki, Japan
Received 5 February 2007; accepted in revised form 12 June 2007
Available online 27 June 2007

Abstract

Twelve commercially available WC–Co powders with different average WC grain sizes (0.2, 2, and 6 μm) and cobalt contents (8, 12, 17 and
25 wt.%) were sprayed on carbon steel substrates using High Velocity Oxy-Fuel (HVOF) spraying process. Hardness, Young's modulus, and
fracture toughness of the coatings were measured. While the hardness and Young's modulus decreased with increasing cobalt content from 1600
to 1100 Hv and from 400 to 300 GPa respectively, the fracture toughness remained in the range from 4 to 6 MPam1/2. The coatings with 2 μm
carbide showed lower hardness than those deposited from 0.2 and 6 μm carbide. These measured mechanical properties were discussed with the
help of microstructures of the coatings investigated by scanning electron microscopy, X-ray diffraction and chemical analysis. Finally, the
hardness of the binder phase in these coatings was estimated to range from 1000 to 1300 Hv by applying the mixture rule for composites to the
experimental data, demonstrating that such hardening of the binder phase is a key factor affecting the mechanical properties of the coatings.
© 2007 Elsevier B.V. All rights reserved.

Keywords: WC–Co; Carbide size; Co content; Mechanical properties; Binder phase

1. Introduction sintering atmosphere, temperature and time have been carefully


controlled, HVOF-sprayed WC–Co coatings still suffer from
Sintered WC–Co materials have a good combination of high decomposition and decarburization during spraying process
modulus, high wear resistance, and adequate fracture toughness, leading to formation of detrimental phases such as W2C, W, and
making them the choice material for a wide range of industrial amorphous or nanocrystalline Co–W–C phase [3,4]. Many
applications such as metal-cutting tools and wear resistant efforts to reduce the degree of decomposition and decarburiza-
components [1]. Recently, WC–Co cermet surface coatings have tion of WC–Co powder have been made such as one using a gas
been used to enhance the wear resistance of various engineering shroud to introduce an inert gas in HVOF spraying [5].
components. Many thermal spraying techniques such as air In-flight phenomena in HVOF spray of WC–Co and coating
plasma spraying (APS) and high velocity oxy fuel (HVOF) formation have been studied by Li et al. [6,7]. Their works
spraying can be applied to deposit the WC–Co coatings, how- focused on the modeling and feedback control of the HVOF
ever, the properties of coatings strongly depend on spraying process. The simulation results demonstrated that the particle-
technique. Compared to other spraying techniques, HVOF melting degree and the particle velocity strongly influence
spraying is one of the best methods for depositing conventional coating microstructure formation and they can be adjusted by
WC–Co cermet powder because the higher velocities and lower the chamber pressure and fuel/oxygen ratio. When the particle-
temperatures experienced by the powder result in less decom- melting ratio is not high, many large particles are bounced off as
position of the WC during spraying process [2]. Consequently, they hit the substrate and the deposited coatings tend to have a
coatings with higher amount of retained WC and less porosity are high porosity. Therefore, a higher melting degree is considered
obtained. However, compared to sintered WC–Co, for which the to be desirable to obtain dense coatings. Furthermore, it is
generally accepted that the higher particle velocity results in
⁎ Corresponding author. denser coating. Besides these process parameters, the particle-
E-mail address: chivavibul.pornthep@nims.go.jp (P. Chivavibul). melting degree and particle velocity depend on the morphology
0257-8972/$ - see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.surfcoat.2007.06.026
510 P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521

of the feedstock powder. It was shown that the larger the particle Table 1
size, the higher the particle density and heat capacity, the lower Powder properties
the highest particle temperature and velocity that are reached Type Powder Composition Apparent density
during flight in HVOF spraying. (wt.%) (g/cm3)
It is well known that the microstructure of sintered WC–Co Carbide Co Content W Co C
affect their mechanical properties. The hardness increases with Size (μm) (wt.%)
decreasing of carbide size and Co content, while the fracture F8 0.2 8 86.5 7.7 5.8 5.63
toughness shows an opposite trend [1]. The reduction of carbide F12 0.2 12 83.5 11.0 5.5 4.97
F17 0.2 17 77.5 17.4 5.1 4.48
size to the nanometer range is of interest recently. Sintered
F25 0.2 25 70.4 24.9 4.7 4.13
nanostructured WC–Co materials have been reported to exhibit M8 2 8 86.5 7.6 5.8 4.43
enhanced performance in both sliding and abrasive wear and M12 2 12 82.0 12.4 5.6 4.50
substantially higher hardness than their conventional counter- M17 2 17 77.8 17.0 5.2 3.94
parts [8,9]. These property improvements originate from the fact M25 2 25 71.7 23.4 4.8 3.65
C8 6 8 86.0 8.2 5.8 4.51
that as the carbide size is decreased (for a given cobalt content)
C12 6 12 81.5 13.0 5.4 4.09
the mean free path of the matrix is reduced resulting in greater C17 6 17 76.9 17.7 5.3 3.71
constraint against deformation, increased hardness and a C25 6 25 71.0 24.2 4.7 3.61
reduced tendency for binder phase extrusion. The same
approach of using nanosize carbide to improve the performance
of HVOF-sprayed WC–Co coatings has been pursued by many showing the Co content. These will also be used to designate the
researchers [10–20]. The nanostructured WC–Co coatings coatings. Backscattered electron (BSE) photographs of F8, M8,
usually showed higher hardness but lower wear resistance than and C8 powders and their polished cross-sections are shown in
conventional coatings [11,15–17]. This disappointing perfor- Fig. 1. All powders have typically spherical shape. The F8
mance of HVOF-sprayed nanostructured coatings has been powder shows dense packing of fine carbide, while M8 and C8
attributed to their higher tendency to decarburization, as the powders show more loosely packed structure with some amount
surface-to-volume ratio of the nanosized WC is much higher of pores. A large scattering of carbide size in all the powders
than that of their conventional counterparts. One exception of is clearly observed in the polished cross-sections. Since it
improvement in wear resistance with the nanostructured was not possible to conduct reliable image analysis of the
coatings was reported by Yang et al. [21], which suggests that carbide particles, a rough size range of carbide particles for each
optimization of spray parameter is necessary to improve the feedstock powder is given in Table 4, where they will be
properties. There have been no general guiding principles compared with the size of carbide particles observed in sprayed
proposed to achieve this, however. An important step to solve coatings in Section 3.2.4. A commercial HVOF spraying
this problem is to understand more precisely how the micro- equipment (JP5000, Praxair Technology Inc., USA), which uses
structure of the coating affects the coating properties. For kerosene and oxygen to generate a supersonic combustion
example, while the effect of carbide size has been greatly flame, was used to spray these powders onto carbon steel
studied, the effect of variations in Co content on the properties (0.45% C) substrates. The substrates had dimensions of
of such coatings is still unclear. 50 × 100 × 5 mm and their surfaces were blasted by alumina
In the present investigation, the effects of both carbide size grit and degreased by ultrasonic cleaning in acetone before
and Co content on the microstructure and mechanical properties spraying. The spraying conditions are listed in Table 2. The
of HVOF-sprayed WC–Co coatings are studied systematically. mass powder feed rate ranged from 74 to 107 g/min reflecting
The relation between microstructural features and mechanical the difference in the density of the powders. The substrates were
properties are discussed based on experimental results. air cooled from the backside during spraying. The thickness of
coatings was approximately 300 μm but coatings with a
2. Experimental procedure thickness of 1000 μm were also produced for elastic modulus
measurement. In-flight temperature and velocity of the sprayed
2.1. Materials powders were also measured using the in-flight thermal sprayed
particle analyzer (DPV-2000, TECNAR Co., Canada). The
Twelve commercially available WC–Co powders with measuring volume of the analyzer was set at the center of the
different average carbide sizes (0.2, 2, and 6 μm) and cobalt particles flux at the position of the substrate surface, which is
contents (8, 12, 17 and 25 wt.%) were used in this work. The 380 mm downstream from the outlet of the spray gun.
powders were manufactured by spray drying of slurries
containing WC and Co particles, followed by light sintering, 2.2. Coating characterization
crushing and classification. The diameter of the powders ranged
from 15 to 45 μm. The reported composition and apparent X-ray diffraction (XRD) was conducted for the powders
density of the powders provided by the manufacturer (Fujimi and as-fabricated coatings with Cu Kα radiation at 40 kV and
Inc., Japan) are given in Table 1. For convenience, the powders 300 mA. The scanning speed and scanning step were 2°/min
were labeled F, M and C signifying fine, medium and coarse and 0.02°, respectively. Chemical analysis was performed on
carbide size, respectively, with the number after the character some coatings to determine the composition. The amounts of
P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521 511

Fig. 1. BSE photographs of WC–Co powders showing the external and cross-section structures: (a), (b) F8, (c), (d) M8, and (e), (f) C8.

W and Co in the coatings were determined by inductively the coatings was determined from the polished cross-sectional
coupled plasma atomic emission spectrometry (ICP-AES) after photographs observed by an optical microscope with 500×
acid dissolution. The amount of C was traced by the infrared magnification. The mean free path of the binder (L) was also
absorption method after combustion in high-frequency induc- measured using a formula L = (1 − f )/NL [22], where NL is the
tion furnace. The amount of oxygen was determined by the inert number of non-continuous carbides intersected on a metallo-
gas fusion (IGF) method. The cross-sections of the coatings graphic plane by a line of unit length and f is the volume
were obtained by embedding specimens in cold mounting resin fraction of dispersed phase determined from the cross-section
followed by grinding and polishing to a 1 μm finish and photograph. At least 10 measurements were made to obtain an
examined by secondary (SEM) and backscattered (BSE) average value for each coating.
electron microscopy. To evaluate the carbide distribution and Microhardness tests were carried out with a 300 g load
porosity in the coatings, cross-sectional images were digitized and dwell time of 15 s. At least ten measurements were made
and analyzed using an image analysis software. The porosity of for each sample. The fracture toughness of the coatings was
512 P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521

Table 2
HVOF spraying condition
Barrel length (mm) 152
Spraying distance (mm) 380
Kerosene (l/min) 0.38
Oxygen (l/min) 944
Torch velocity (mm/min) 700
Powder feed rate (g/min) 74–107
Powder feed gas Nitrogen

determined by the indentation method with a 10 kg load


and dwell time of 15 s. Both the crack lengths and Vickers
diagonals were measured by an optical microscope. The fracture
toughness, KIC, was determined under an assumption that
the cracks generated from an indentation were radial cracks
possessing the Palmqvist geometry by the following equation
[23]:
 2=5
E
KIC ¼ 0:0193ð Hv DÞ ðaÞ1=2 ; ð1Þ
Hv

where Hv is the Vickers hardness, E the Young's modulus, D the


half-diagonal of the Vickers indentation and a the indentation
crack length. At least five indentations were made for each
sample. Both microhardness and fracture toughness tests were
carried on the polished cross-section of the coatings. It should
be noted that since thermally spray coatings generally have an
anisotropic microstructure, typically characterized by the so-
called “lamellar structure”, cracks produced under indentation
Fig. 3. Thermal spray particles' in-flight velocity and temperature of (a) M-series
test mainly develop in the direction parallel to lamellae because powders and (b) F12, M12, and C12 powder.
the lamellar interface is prone to decohesion. The absolute
values of fracture toughness obtained by this technique,
therefore, may not be accurate but they should provide a useful (1 − v)− 1ρV2, where ρ is the density, V is the wave velocity in
means to quantify differences among the specimens. the coating, and v is the Poisson's ratio, respectively. The
Young's modulus of the coating was measured by an Possion's ratio was assumed to be equal to 0.3 for all speci-
ultrasonic technique as shown in Fig. 2. The longitudinal- mens [28]. Since the measurement was made in the thickness
wave transducer with the center frequency of 10 MHz (M112, direction, the Young's modulus in the direction of spraying
Panametrics, USA) was placed on the as-sprayed coating surface (perpendicular to the coating surface) was obtained.
to generate and detect ultrasonic waves propagating in the
specimen. The first and third bottom echoes were recorded by an 3. Results and discussion
oscilloscope and transferred to a computer for further analysis.
The wave velocity was calculated by the difference of arrival 3.1. In-flight temperature and velocity of sprayed particles
time of the wave signals in coating with different thickness.
The Young's modulus was determined using E = (1 + v)(1 − 2v) Fig. 3a and b show the spray particles' velocity and
temperature in flight for the powders with different Co contents
and carbide sizes, respectively. These data represent the average
and standard deviation measured at the position of the substrate
surface. Particles' surface temperature decreased from 1680°
to 1580 °C with increasing Co content, while the velocity
almost remained constant around 800 m/s (Fig. 3a). The
lower temperatures of powders with higher Co contents were
considered to be due to the higher specific heat value of Co
(0.42 J/g K) compared to that of WC (0.05 J/g K). Furthermore,
since the powder temperature upon impact was higher than the
melting point of Co (1495 °C), the latent heat required to melt
the Co is higher when Co content is higher. The powders with
Fig. 2. Experiment set-up for Young's modulus measurement of the coating by fine carbides have a lower particle velocity than those obtained
ultrasonic technique. from the powders with medium and coarse carbides (Fig. 3b).
P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521 513

3.2. Microstructural identification of the coatings the coatings with higher Co content and coarser carbide. It
could be inferred that increasing Co content and reducing
3.2.1. SEM observations carbide size increases a degree of dissolution of WC into the Co
Backscattered electron (BSE) photographs of the F, M, matrix. More quantitative analysis of this point will be given in
and C-series coatings with different Co contents are shown in the Section 3.2.4.
Figs. 4, 5, and 6, respectively. In Fig. 4a, BSE image of F8
coating shows densely packed fine carbide particles in the 3.2.2. X-ray diffraction
binder phase. A number of large pores were also observed. Fig. 7 shows examples of XRD patterns of the powder and
Increasing the Co content, BSE images of F12, F17, and coatings. The effect of carbide size (F12, M12, C12) and Co
F25 revealed a more widely-spaced dispersion of fine carbide content (M-series) are shown in Fig. 7a and b, respectively. In
particles and binder layers with a stronger contrast. The dark spite of the fact that XRD spectra of the powders used in this
and bright regions in the binder correspond to regions of lower study show that they are composed only of WC and Co as
and higher mean atomic number, respectively. These bright and exemplified by M12 powder in Fig. 7a , the coatings showed the
dark binder layers are known to be W-rich and Co-rich regions, crystalline peaks of WC, W2C, and W. Moreover, there was a
respectively [18]. These inhomogeneous composition of the broad peak corresponding to an amorphous or nanocrystalline
binder resulted from the dissolution of WC during spraying phase detected at 2θ angles between 40° and 48°. The ratios of
process and rapid cooling of the powders when they reached the main peak height of W2C at 2θ = 39.57 and the amorphous
the substrate. As compared to F-series, microstructures of phase (or nanocrystalline) to the WC peak were plotted as a
M and C-series show much greater variation in the size of function of Co content and shown in Fig. 7c. There is a
carbide particles distributed in the matrix. There are two im- tendency for the amorphous peak to increase as the Co content
portant observations that could be made from these photo- increases for all the carbide size, while W2C tended to decrease
graphs. Firstly, blocky and angular WC particles were observed with the Co content except for F8 and C12. It is well-known that
more in the coatings with lower Co content, while rounded the formation of W2C is largely dependent on the heating degree
WC particles were observed more in the coatings with higher of the powders [24]. Higher heating degree leads to more W2C
Co content. Secondly, dark binder layers were observed more in phase. Increasing Co content of the powder reduces the in-flight

Fig. 4. BSE photographs of coatings with 0.2 μm carbide at low and high magnifications: (a) and (e) F8, (b) and (f) F12, (c) and (g) F17 and (d) and (h) F25.
514 P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521

Fig. 5. BSE photographs of coatings with 2 μm carbide: (a) M8, (b) M12, (c) M17 and (d) M25.

particle temperature as shown in Fig. 3b and so, lowers the the concentration of W in the coating and the feedstock powder
heating degree. respectively. They clearly indicate that Co content in all the
Another point noted was that the WC peaks of the coatings coatings increased by 10 to 30%, while W and C contents
shifted to higher angles as compared to that of the powder. decreased. A possible reason is that Co has a higher tendency to
To investigate the origin of the shift, XRD measurements were be deposited on the surface than WC because it can be fully
performed on the identical M12 coatings after removal from melted. Li et al. [25] studied the effect of carbide rebounding
the substrate by machining and polishing. The prepared free- during HVOF spraying of Cr3C2–NiCr and concluded that
standing M12 coating showed a shift of WC peak (100) to larger carbide particles are likely to rebound during spraying
the angle close to the standard value (provided by the Powder process by observing the splat morphology produced by several
Diffraction File (PDF) 00-051-0939). Therefore, the shift starting powders with different carbide sizes. This preferential
should be related to the relief of residual stress when the rebound-off of large solid carbide particles should decrease both
substrate was removed. The strain, e, normal to the coat- the mean particle size of carbide and the W and C contents in the
ing surface was calculated from the interplanar spacing by e = coatings. The fact that carbon decreased more than tungsten
(d − d0) / d0, where d and d0 are the spacings of the planes in implies, however, other factors such as oxidation of carbon
the presence and absence of stress, respectively. When d0 from during spraying also affected the material in HVOF spraying.
PDF 00-051-0939 was used, the strains determined from the The table shows that 1) the degree of decarburization
interplanar spacing of the coating before and after removing the increases with increase of carbide size (F12 b M12 b C12), 2)
substrate were − 0.61 and − 0.04%. The large difference of e there is not a strong dependence of decarburization on the Co
between the two conditions suggests the existence of significant content (M8 ≈ M12 ≈ M17 ≈ M25). It is of great interest to
residual stress in the as-sprayed coating due to the constraint discuss how this is related to the formation of W2C phase as it is
by the substrate. generally regarded as a product of decarburization in thermal
spraying. Fig. 7c shows a significant decrease of W2C peak
3.2.3. Decarburization and formation of W2C height with Co content in the M-series, it appears that one
Table 3 shows the chemical compositions of the selected cannot use the W2C peak height as a direct index of decar-
seven coatings. The composition changes shown in the burization. Guilemany et al. [3] reported that W2C can be
right hand side of the table were calculated by a formula such formed by three different mechanisms. Firstly, WC directly
as (Cw − Cw0) / Cw0 for W for example, where Cw and Cw0 are reacts with oxygen resulting in the loss of carbon in a solid-state
P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521 515

Fig. 6. BSE photographs of coatings with 6 μm carbide: (a) C8, (b) C12, (c) C17 and (d) C25.

reaction. This commonly occurs at the free surface of the section should largely dependent on the heating degree of the
powder. Secondly, WC thermally decomposes itself to W2C by powders [24]. Higher heating degree should promote the
a solid-state reaction in the regions of the powder where oxygen reactions and lead to severe decarburization. The heating degree
is unavailable and the released carbon is dissolved into the of the powders depends on many factors such as fuel/oxygen
binder. The third mechanism is that during spraying process ratio, powder morphology, etc. Since the spraying condition
WC in contact with the binder is dissolved into the binder and C was kept constant for all the powder types in the present study,
is then oxidized and converted to CO/CO2. When the liquid the heating degree should depend largely on the powder
cools, W2C is precipitated from the W-rich liquid phase binder. morphology. As for the F8 powder, there are a couple of reasons
If most of W2C were formed by the first mechanism, the amount to believe that the heating degree of this powder was the lowest
of W2C phase should be proportional to the degree of among the tested powders: 1) this powder has the highest
decarburization. On the other hand, if most of W2C was formed density as shown in Table 1, which means higher heat capacity
by the second and/or the third mechanism, the amount of W2C and less surface area for the gas–solid reaction, 2) it produced
phase should not be regarded as an index of decarburization the coating with the highest porosity as shown in Fig. 8, which
because the released C could remain trapped in the binder is another indication that the melting degree of this powder was
without being oxidized till the time when the coating cools less as compared to the powders with the same composition
down. At this stage, one can only state that the first mechanism such as M8 and C8.
of W2C formation was not the dominant one in the present
study. 3.2.4. Microstructural characteristics
Such reactions are affected by the morphology of the Table 4 gives a summary of microstructural characteristics of
feedstock powders. Previous studies [18–20,22] showed that a the coatings obtained by the image analysis of polished cross-
decrease in carbide size in the feedstock powder led to a higher sections. Note that the properties of F8 are not given here
amount of W2C in the coatings due to the larger surface-to- because the finer structure caused difficulty in image analysis.
volume ratio. The results in Fig. 7c also showed the same trend Also it was not possible to identify W2C under the SEM, which
as the previous studies for F12 and F17, however, the opposite should be dispersed finely in the binder phase. Therefore, when
tendency was observed clearly for F8. The kinetics of the term “carbide” is used in this section, no distinction is made
decarburization of WC to W2C as described in the previous between the two types of carbides but from the XRD results it is
516 P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521

Table 3
Chemical analysis of the coatings
Type Composition (wt.%) Composition change
compared to the starting
powders (wt.%)
W Co C O W Co C
F12 80.3 12.7 4.7 0.18 − 3.8 +15.5 − 14.6
M8 85.1 9.1 4.3 0.23 − 1.7 +19.7 − 25.9
M12 81.5 14.0 3.9 0.13 − 0.6 +12.9 − 30.4
M17 76.6 19.2 3.6 0.12 − 1.5 +12.9 − 30.2
M25 69.0 26.2 3.4 0.16 − 3.9 +12.0 − 29.4
C12 78.5 16.5 3.6 0.16 − 3.8 +26.9 − 33.0

coatings with different carbide sizes by focused ion beam (FIB).


Their results showed that the deformation and the compression
of substrate microstructure by large carbides (5–7 μm) was
observed, indicating the high impact pressure of carbides. In
contrast, no such carbide intrusion into the substrate was
observed in the splat deposited from fine carbide (0.2 μm).
Their result indicates that the larger carbide particles in HVOF-
sprayed powders of WC–Co possess a strong peening effect,
which can densify the coating surface in the middle of coating
formation period. One more reason, the density of the starting
powders influences the melting degree of binder phase and
subsequently the density of the coating. The higher the powder
density, the lower the highest particle temperature is achieved
[6]. Also, the particle's velocity in Fig. 3b shows that the impact
velocity of powders with the fine carbide is slightly lower
than the other two. This should be an additional reason for the
higher porosity with the feedstock powders with fine carbide.
Increasing Co contents also tended to reduce porosity of the
coatings because compared to WC, Co has lower melting
point and easier to be melted, which can fill the pores in the
coating [6].
Carbide volume fractions of the coatings were significantly
lower than those of the feedstock powders. The theoretical
carbide volume fractions of the feedstock powders with 8, 12,
17, and 25 wt% Co content were 86, 81, 73 and 63%,

Fig. 7. (a) XRD results of the M12 powder and F12, M12 and C12 coatings,
(b) XRD results of the M-series coatings and (c) plots of WC and amorphous
peaks as a function of Co content.

obvious that the amount of W2C is not so important as


compared to WC. Porosity, carbide size and volume fraction are
discussed in the following.
The coatings deposited from the powders with medium
and coarse carbides generally had lower porosity than those
deposited from the powder with fine carbide as shown in Fig. 8.
Watanabe et al. [20] examined the deformation of substrate
surface caused by the formation of WC–12Co splats and Fig. 8. Porosity of the coatings as a function of Co content.
P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521 517

Table 4 2) The effect of carbide size on the microstructure of the coating


Microstructural characteristics of coatings was quite complex. The finer carbide yielded the coatings
Type WC volume Carbide size Mean carbide Mean free path with high porosity.
fraction (%) range (μm) size (μm) of binder (μm) 3) The morphology of the feedstock powder was likely to give a
F8 – – – – strong effect on the degree of decarburization, which affects
F12 53 0.1–0.8 0.34 ± 0.11 0.21 ± 0.04 the amount of W2C and W in the coatings.
F17 50 0.1–0.8 0.3 ± 0.12 0.26 ± 0.04
F25 37 0.1–0.8 0.3 ± 0.88 0.46 ± 0.15
M8 68 0.3–4 0.75 ± 0.34 0.38 ± 0.04 3.3. Mechanical properties of the coatings
M12 53 0.3–4 0.88 ± 0.50 0.74 ± 0.22
M17 37 0.3–4 0.91 ± 0.47 1.40 ± 0.31 3.3.1. Hardness and Young's modulus
M25 27 0.3–4 0.72 ± 0.44 2.53 ± 1.26 Plots of microhardness, Young's modulus and fracture
C8 68 0.3–5 2.11 ± 0.88 0.66 ± 0.08
toughness values of the coatings as a function of Co content
C12 48 0.3–5 1.87 ± 0.70 2.44 ± 0.94
C17 36 0.3–5 1.46 ± 0.69 2.96 ± 0.66 are shown in Fig. 9. The hardness of the coatings in the
C25 30 0.3–5 2.11 ± 0.73 2.52 ± 0.73 present study were generally higher than those reported by
other researchers [11,19,22]. Generally, the hardness of sintered

respectively. As compared to these values, the carbide volume


fractions of the coatings were approximately 20% and 50%
lower for the 8% and 25%Co content coatings, respectively. The
dramatic loss of carbide should be due to the rebounding
of solid carbide particles as mentioned previously and the
dissolution of WC during spraying process. As shown in
Figs. 4, 5 and 6, most of the binders consist predominantly of
bright W-rich regions, which means that the amount of WC
dissolved into the Co content.
Mean carbide sizes of the coatings determined from image
analysis were different from the nominal carbide size of the
feedstock powders, i.e., 0.2 μm for F, 2 μm for M, and 6 μm for
C powder. According to the powder manufacturer, the nominal
carbide powder size is determined by the Fisher sub-sieve sizer.
Actually as shown in Fig. 1, carbide particles of various sizes
were observed in the feedstock powders, which are roughly
shown as the carbide size range in Table 4. Especially, a large
number of small carbide observed in M and C-series powders
yielded lower average values than the nominal ones reported by
the powder manufacturer. Other reasons for the discrepancy
could be the rebounding of large carbide, dissolution of WC
and the measurement method. When looking at the cross-
section, one is more likely to see a dimension smaller than the
actual diameter of the particle, because the chances that the
particle is seen in the cross-section at its actual diameter are
small [26].
The mean free path of binder of F and M-series increased
with increasing Co content, while those of C-series did not
show any significant change because a number of the small
carbide distributing overly in the binder (Fig. 6) affected the
determination of the mean free path. For a given Co content,
smaller carbide gives lower mean free path of binder.
Before going on to the mechanical properties of the coating,
it is useful to make a summary of the microstructure evaluation.

1) Increasing Co content resulted in increases of Co-rich binder


layer, amorphous phase and degree of dissolution (more
rounded carbides and reduction of carbide volume fraction),
while the porosity was reduced and a degree of decarburi- Fig. 9. Hardness, Young's modulus, and fracture toughness as a function of Co
zation was unchanged. content.
518 P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521

(125–225 GPa) directions and the former agreed well with


those measured via laser-ultrasonics. This showed that the
WC–Co coatings exhibit the anisotropic behavior. The Young's
moduli obtained from the present study was in the different
direction from those reported by Lima et al. and cannot be
directly compared, however, the values were reasonable.
A plot of hardness with carbide volume fraction is shown in
Fig. 10. For comparison, the hardness of sintered WC–Co with
different carbide size is also given [1]. The hardness of both
coating and sintered WC–Co increased with increasing carbide
fraction. For a given carbide content, the hardness of the coating
was much higher than that of the sintered one. The slopes of the
fitted lines for the coating data points are approximately 50%
lower than those of the sintered. This means that while the
hardness of sintered WC–Co strongly depends on the WC
volume fraction, the hardness of the coating depends more on
the properties of binder because there are new phases such as
Fig. 10. Hardness of the coating as a function of carbide volume fraction. The
hardness of the sintered WC–Co are also indicated as the dash lines.
W2C and W distributed in the binder. The hardness of pure,
fully annealed polycrystalline cobalt has been reported to lie
between 140 and 210 kg/mm2 [29]. The hardness of the binder
WC–Co decreases with increasing Co content and carbide size. phase in sintered WC–Co, which is a solid solution of cobalt
The previous studies on HVOF-sprayed coatings also reported with 2–10% tungsten, is reported as 490 kg/mm2 in a 50 wt.%
the same trend as the sintered WC–Co [13,14,19,22]. The Co and 660 kg/mm2 in a 25 wt% Co [29]. Unfortunately, there
hardness in the present study decreased also with increasing Co is no report about the hardness of the binder phase in HVOF-
content for all the carbide sizes. However, the effect of carbide sprayed coating.
size on hardness behavior was different from the previous
reports. Except for 8%Co series, the hardness of F and C-series 3.3.2. Fracture toughness
was approximately the same, while M-series showed the lowest The fracture toughness ranged from 3 to 5 MPam1/2 and
hardness value. If the effect of the carbide size on the hardness slightly increased with increasing Co content for M and C-series
of the coating has a similar behavior as the sintered material, coatings. The fracture toughness of the coatings in the present
the hardness of the coating of the F-series should be the study were higher than those reported by other researchers
highest followed by the M and C-series, respectively. Possible [11,19,22]. However, these values were much lower than those
explanations for this behavior are as followings. The hard- of sintered WC–Co (9–20 MPam1/2) [1]. In general, toughness
ness of the coating is a result of a balance of microstructure, of such composites should increase as the binder content
phase contribution (amounts of W2C, amorphous phase), increases, because of crack blunting afforded by plastic defor-
porosity, peening effect, etc. It is reasonable that the hardness mation. However, the increase of toughness in the coatings in
increases with increasing content of W2C in the coatings, this study was very small as compared with sintered materials.
because W2C phase (Hv = 3000) is harder than the WC phase The decarburization and dissolution of WC in the coatings
(Hv = 1300–2300) [27]. The higher hardness of the fine carbide during spraying process causing the formation of amorphous or
is a result of nanostructure. The lowest hardness of the F8 nanocrystalline phase are considered to affect the mechanical
coating can be explained by the high porosity and the XRD
results that it contained lower amount of W2C phase than those
deposited from coarse and medium carbides. Furthermore, since
the pore distribution in the coating is not uniform, the large
scattering of hardness value with fine carbide is also considered
to be a result of porosity. The high hardness value of the coating
deposited with coarse carbide can be considered from the
peening effect as described in the Section 3.2.4.
Young's moduli of the coatings with fine and medium
carbides decreased with increasing Co content, while those of
the coatings deposited from coarse carbide were almost the
same. The Young's moduli obtained from present studies
ranged from 200 to 350 GPa. Lima et al. [28] measured
the Young's modulus of the WC–Co coating using Knoop
indentation and laser-ultrasonic techniques. They found that
the Young's moduli obtained by Knoop indentation were
different in the cross-section (300–350 GPa) and top-surface Fig. 11. Fracture toughness as a function of decarburization.
P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521 519

of perfect bonding at the interface and negligible interaction


between particles. Considering the nature of the coatings, which
are composed of splats or lamellae with some amounts of
defects and pores, these lead to difficulty of developing a
suitable model for coating. In the present study, we tried to
predict the hardness and elastic properties of the binder phase by
using the rule of mixture. It is important to point out that this
model was developed for isotropic materials. Consequently, the
real values of hardness and modulus for anisotropic materials as
coatings may be different from those obtained from the model.
Also, the binder phase is not homogenous and its composition
as well as the mechanical properties should vary significantly
from place to place. However, it is expected that such estimation
provides some useful information of binder phase.
In the case of hardness, the relation of the hardness of each
constituent and the coating can be expressed as:

Hc ¼ HWC VWC þ Hb ð1  VWC Þ ð2Þ

where c, WC, and b denote coating, tungsten carbide, and


binder, respectively. H and V are hardness and volume fraction,
respectively [30]. Considering the Young's modulus, the
experimental Young's modulus of the coating obtained from
the ultrasonic test was the modulus in the direction of spraying
(perpendicular to the coating surface). In this direction, the

Fig. 12. BSE photographs showing crack path generated from indentation
testing in: (a) F8 and (b) F25.

properties of the binder phase. For a given Co content, the


fracture toughness of sintered WC–Co increases with increas-
ing carbide size because an increase in carbide size leads to
increased mean free path. Although the mean free paths also
increased with increasing carbide size in this study, no
significant change of fracture toughness of the coatings with
different carbide sizes was observed. A plot of facture toughness
as a function of decarburization is shown in Fig. 11. It clearly
shows that the fracture toughness decreased with increased
decarburization, suggesting that the degree of decarburization
is a more dominating factor on fracture toughness than the
carbide size. SEM observations of crack paths generated
from indentation test shown in Fig. 12b indicate that cracks
preferentially propagated through the W-rich binder regions
(white colored area) indicating lower fracture toughness of these
phases. Also, it appeared in Fig. 12a that cracking occurred at
interfaces between WC and the binder.

3.3.3. Estimation of binder's properties


The mechanical properties of the binder phase in the coat-
ings such as hardness and elastic modulus are very important
because these values provide useful information for under-
standing the mechanical behavior of the coatings such as crack
propagation and toughness. There are many theoretical models
available to predict the hardness and elastic constants of two- Fig. 13. Estimated binder hardness (a) and Young's modulus (b) as a function of
phase materials. Most of them are made under the assumptions carbide volume fraction.
520 P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521

coating microstructure is lamellar structure. Therefore, the coating properties. The hardening of binder phase due to the
elastic properties of the coating can be derived by using the formation of amorphous or nanocrystalline phase leads to a
basic relationships of elastic properties of parallel arrangements great increase of coating hardness and a reduction of fracture
of two-phases (WC and Co) loaded in iso-stress configuration toughness. If such distribution of various phases in the binder
and the elastic modulus can be expressed as [31] phase is controlled, a great improvement of toughness of the
coating can be expected.
Eb EWC
Ec ¼ ð3Þ
Eb VWC þ EWC ð1  VWC Þ 4. Conclusions

where E is Young's modulus. The microstructure of the coating 1) It has been shown that increasing the Co content reduces the
is characterized by the existence of various pores, microcracks, hardness of the coating because a reduction of WC and W2C
splat boundaries, and unmelted particles. These attributes phase. This is consistent with a general trend observed for
greatly influence the overall mechanical properties of coatings HVOF deposited coatings and sintered materials.
and should be considered in the calculation. In this respect, the 2) Reducing carbide size of starting powders has not shown an
mechanical properties (hardness and Young's modulus) of the increase of coating hardness. The peening effect and the
coating containing porosity may be calculated using the em- microstructure properties such as porosity and amount of
pirical expression developed for ceramics [32] W2C phase, which are strongly related to the density of
the starting powder show a great influence on the coating
M ¼ Mo eðBpÞ ð4Þ hardness.
3) The binder hardness of the coating was estimated using a
where Mo and M represent the property of the fully dense rule of mixing and showed quite higher values than those
materials and porous materials, respectively, p is volume reported for sintered materials. The hardening of binder
fraction of the porosity, and B is an empirical constant. Because phase increases the coating hardness and reduces the fracture
there is no reported data of B for WC–Co, the calculation was toughness.
made with varying B from 4 to 8. This range was chosen
because previous studies on Y2O3-stabilized tetragonal zirconia Acknowledgements
polycrystals (Y-TZP) showed that B were approximately 2 to 7
[30,33]. The hardness of WC is approximately 1800 kg/mm2. This work was supported by the National Institute for
This value was obtained by averaging the hardness of WC on Materials Science. We wish to express appreciation to M.
the prismatic planes (1300 kg/mm2) and on the basal plane Komatsu and N. Kakeya for supporting fabrication of the
(2300 kg/mm2) [27]. The Young's modulus of WC is 700 GPa coatings and H. Yamaguchi for chemical analysis.
[1]. By substituting the coating hardness, Young's modulus,
carbide volume fraction, and volume fraction of the porosity
References
obtained from the experiment into Eqs. (2), (3), and (4) the
hardness and Young's modulus of binder phase of the coating [1] G.S. Upadhyaya, Cemented Tungsten Carbides, Noyes, New Jersey, 1998.
were determined as shown in Fig. 13(a) and (b), respectively. [2] V.V. Sobolev, J.M. Guilemany, J. Nutting, High Velocity Oxy-Fuel
The predicted values obtained only from Eq. (2) or (3) were Spraying, Maney Publishing, UK, 2004.
plotted with open symbols, while those with a consideration of [3] J.M. Guilemany, J.M. de Paco, J. Nutting, J.R. Miguel, Metal. Mater.
Trans. A 30 (1999) 1913.
porosity effect using Eq. (4) with B = 6 were plotted with closed
[4] C. Verdon, A. Karimi, J.L. Martin, Mater. Sci. Eng., A Struct. Mater.: Prop.
symbols. The values calculated with B = 4 and 8 are shown with Microstruct. Process. 246 (1998) 11.
the lower and upper bars, respectively. The estimated binder [5] Y. Ishikawa, J. Kawakita, S. Sawa, T. Itsukaichi, Y. Sakamoto, M. Takaya,
hardness ranged from 740 to 1312 kg/mm2, which was quite S. Kuroda, J. Therm. Spray Technol. 14 (2005) 384.
higher than those of the sintered one [29]. The effect of porosity [6] M.H. Li, D. Shi, P.D. Christofides, Powder Technol. 156 (2005) 177.
on predicted binder hardness becomes more pronounced with [7] M.H. Li, P.D. Christofides, Chem. Eng. Sci. 60 (2005) 3649.
[8] K. Jia, T.E. Fisher, Wear 200 (1996) 206.
increasing WC volume fraction because porosity tends to [9] K. Jia, T.E. Fisher, Wear 203–204 (1997) 310.
increase with higher WC content as shown in Fig. 8 (higher [10] C. Bartuli, T. Valente, F. Cipri, E. Bemporad, M. Tului, J. Therm. Spray
porosity). The estimated Young's modulus of binder varied Technol. 14 (2005) 187.
from 100 to 200 GPa. The effect of porosity on predicted values [11] A.H. Dent, S. DePalo, S. Sampath, J. Therm. Spray Technol. 11 (2002) 551.
was quite low compared to that on hardness. It should be noted [12] J.M. Guilemany, S. Dosta, J. Nin, J.R. Miguel, J. Therm. Spray Technol.
14 (2005) 405.
that the estimated hardness and Young's modulus in this study [13] J.H. He, J.M. Schoenung, Surf. Coat. Technol. 157 (2002) 72.
includes the effect of other phases such as W2C, nanocrystalline [14] B.R. Marple, R.S. Lima, J. Therm. Spray Technol. 14 (2005) 67.
and amorphous Co–W–C distributing in all of the binder, which [15] Y.F. Qiao, T.E. Fischer, A. Dent, Surf. Coat. Technol. 172 (2003) 24.
are harder than WC. Also the peening by impinging particles [16] P.H. Shipway, D.G. McCartney, T. Sudaprasert, Wear 259 (2005) 820.
during spraying as discussed in Section 3.2.4 should cause [17] D.A. Stewart, P.H. Shipway, D.G. McCartney, Wear 229 (1999) 789.
[18] D.A. Stewart, P.H. Shipway, D.G. McCartney, Acta Mater. 48 (2000) 1593.
significant work-hardening of the binder phase. Although these [19] Q.Q. Yang, T. Senda, A. Ohmori, Wear 254 (2003) 23.
estimated values need a proof by experiment, they give an [20] M. Watanabe, A. Owada, S. Kuroda, Y. Gotoh, Surf. Coat. Technol. 201
important hint of how strongly the binder properties affect the (2006) 619.
P. Chivavibul et al. / Surface & Coatings Technology 202 (2007) 509–521 521

[21] Q. Yang, T. Senda, A. Hirose, Surf. Coat. Technol. 200 (2006) 4208. [29] H.C. Lee, J. Gurland, Mater. Sci. Eng., A Struct. Mater.: Prop. Microstruct.
[22] S. Usmani, S. Sampath, D.L. Houck, D. Lee, Tribol. Trans. 40 (1997) 470. Process. 33 (1978) 125.
[23] M.M. Lima, C. Godoy, J.C. Avelar-Batista, P.J. Modenesi, Mater. Sci. [30] R. Chaim, M. Hefetz, J. Mater. Sci. 39 (2004) 3057.
Eng., A Struct. Mater.: Prop. Microstruct. Process. 357 (2003) 337. [31] C.L. Hsieh, W.H. Tuan, Mater. Sci. Eng., A Struct. Mater.: Prop.
[24] C.J. Li, A. Ohmori, Y. Harada, J. Mater. Sci. 31 (1996) 785. Microstruct. Process. 393 (2005) 133.
[25] C.J. Li, G.C. Ji, Y.Y. Wang, K. Sonoya, Thin Solid Films 419 (2002) 137. [32] R.W. Rice, Porosity of Ceramics, Marcel Dekker, New York, 1998.
[26] H.L. de Villiers Lovelock, J. Therm. Spray Technol. 7 (1998) 357. [33] J. Luo, R. Stevens, Ceram. Int. 25 (1999) 281.
[27] H. Engqvist, S. Ederyd, N. Axen, S. Hogmark, Wear 230 (1999) 165.
[28] R.S. Lima, S.E. Kruger, G. Lamouche, B.R. Marple, J. Therm. Spray
Technol. 14 (2005) 52.

You might also like