Carbon content dependence of grain growth mode in VC-doped WC-Co

hardmetals

I. Sugiyamaa,b*, Y. Mizumukaia, T. Taniuchic, K. Okadac, F. Shirasec, T. Tanased, Y.

Ikuharaa,b,e, T. Yamamotoa,e,f

a
Institute of Engineering Innovation, The University of Tokyo, 2-11-16, Yayoi, Bunkyo-ku,Tokyo 113-8656,
Japan

b
WPI Advanced Institute for Materials Research, Tohoku University, Katahira 2-1-1, Aoba-ku, Sendai-shi,
Miyagi, Japan, 980-8577
c
Tsukuba Plant, Mitsubishi Materials Corp., 1511 Furumagi, Joso-shi, Ibaraki 300-2795 Japan

d
Japan New Metals Co., Ltd., 1-6-64, Sen-nari-cho, Toyonaka-shi, Osaka, Japan

e
Japan Fine Ceramics Center, 2-4-1, Rokuno, Atsuta-ku, Nagoya 456-8587, Japan
f
Department of Quantum engineering, Nagoya University, Furou-cho, Chidane-ku, Nagoya-shi, Aichi 464-
8603, Japan

________________

*Corresponding author. Tel.:+81 3 5841 7689; fax:+81 3 5841 7694;

e-mail: sugiyama@wpi-aimr.tohoku.ac.jp

1
Abstract

Carbon content dependency of grain growth mechanism and that of grain growth

inhibition mechanism in VC-doped WC-Co hardmetals is investigated. VC-doped WC-

Co hardmetals with three different carbon contents were sintered with liquid phase and

then rapidly quenched to freeze up the structure at the sintering temperature. In these

samples, spatial distributions and atomic scale structures of V-rich phases are investigated

using transmission electron microscopy (TEM) and related techniques. In these

measurements, doped V are found in liquid phase as solute, in large (W,V)Cx precipitates

and in interface segregations. Further detailed observations and disucssions are carried

out for the (W,V)Cx segregated at the WC grains/Co phase interfaces. These (W,V)Cx

phases change their form from planar films to small islands depending on the carbon

content. The WC grains/Co phase interfaces are fully covered by planar (W,V)Cx in the

sample of low carbon content. On the other hand, the WC grains/Co phase interfaces

are partially covered by (W,V)Cx islands in the material of high carbon content. During

sintering, the WC grains in this sample grew much faster than those in the sample of low

carbon content. These structural difference are discussed in terms of WC grains/(W,V)Cx

interface energy.

Keywords: Hardmetals, Cermets, TEM, EDS, Grain growth, Liquid phase sintering

2
1. Introduction

WC-Co based hardmetals are widely used as bits for cutting, mining and many other

machining tools because of their remarkable hardness and wear resistance [1, 2]. They

are commonly prepared by liquid phase sintering at the temperature higher than the

liquidus line, where the WC-Co based hardmetals are mainly consist of Co-based liquid

phase and WC solid grains [3]. The mechanical properties of WC-Co hardmetals strongly

depend on the fineness of WC grains, so controlling WC grain growth during liquid phase

sintering is one of major research topics in the field of hardmetals. There are some

techniques to control the WC grain growth, for example, adding other carbides such as

VC and Cr2C3 [4-6], controlling carbon content in the hardmetals [7,8], controlling the

sintering temperature, and so on.

Doping other carbides, especially doping VC, is well known as one of the most

effective techniques to control a grain growth. Many studies have been devoted to

investigate the effect of other carbides doping on WC grain sizes [9,10], grain shapes

[11,12] and interface structures [13-15], however, the detail of grain growth inhibition

mechanism remains elusive. One reason is the cooling rate effect. The WC grains/Co

phase interface structures vary by changing cooling rate after sintering [16]. This was a

big problem because the grain growth is thought to be inhibited by the (W,V)Cx phase

formed at the WC grains/Co phase interfaces. The interface structures on the sintering

state should be observed to understand the authentic grain growth mechanism, but such

observation is impossible in slowly cooled specimens. Recently, we found the cooling

rate of oil quenching method is fast enough to freeze up the WC grains/Co phase interface

structures on the sintering state [17]. By using this oil quenching technique, we revealed

that the (W,V)Cx at the WC (0001)/Co forms nano-dots accompanied by planar (W,V)Cx

3
sheets at the sintering temperature [18]. Such structure is formed by lattice mismatch

between (W,V)Cx and WC (0001), so the interface structure can be optimized to improve

grain growth inhibition effect by controlling the lattice mismatch.

The lattice mismatch between (W,V)Cx and WC (0001) can be controlled by

controlling carbon content in the hardmetals. The lattice constant of (W,V)Cx depends on

the composition of itself, and the composition of (W,V)Cx depnds on the carbon content

in the hardmetals. In this study, we have precisely controlled the carbon content in the

VC-doped WC-Co hardmetals and the atomic structure at the WC grains/Co phase

interfaces are observed. The VC-doped WC-Co hardmetals with various carbon contents

were prepared by oil-quenching subsequently to the liquid phase sintering, and their

micro structures were investigated by high-resolution transmission electron microscopy

(HRTEM) and energy dispersive X-ray spectroscopy (EDS) by using transmission

electron microscope (TEM). From these observations, the proper tactics to improve grain

growth inhibition effect will be proposed.

2. Experimental procedure

The VC-doped WC-Co hardmetals were prepared by the ordinal liquid phase sintering

method. The base composition of the all hardmetals were WC-12 wt.% Co-0.5 wt.% VC.

The carbon content was varied from the lower limit of the three phase region of WC grains,

Co based liquid phase and (W,V)Cx to the higher limit in three steps. Each sample is

named low carbon (LC), middle carbon (MC) and high carbon (HC). Raw powders of

WC (Japan New Metals Co. Ltd., nominal grain size of 0.8 m), Co (Umicore Co. Ltd.,

1.4 m) and VC (Starck Co. Ltd., 1.6 m) were weighed to be the target composition,

and then they were attritor milled, dried and pressed into 16×16×5 mm 3 sized green

4
compacts under a pressure of 150 MPa. The carbon content was controlled by adding C

powders to the raw powders. The green compacts were divided into four pieces to

improve quenching rate. The divided green compacts were liquid phase sintered at 1653K

for 1 hour in vacuum better than 1×10-2 Pa. After soaking at the sintering temperature, the

samples were oil quenched by dropping them into oil. The quenching process was carried

out without gas introduction or any other special process and the quenching rate is high

enough for freezing up the interface structure at the sintering temperature [17]. The

samples were confirmed to be in target composition by saturated magnetization

measurements, scanning electron microscopy (SEM, H-S800, Hitchi Co. Ltd., Japan)

observations and TEM-EDS analysis. The results are summarized in Table 1. Both

saturated magnetizations and W contents in liquid phase correspond to the value in the

three phase region of W-V-C-Co quaternary phase diagram [3,19]. Carbon content is also

confirmed to be successfully varied.

The microstructures were investigated by conventional TEM (JEM-2010HC, JEOL

Ltd. Japan) and HRTEMs (JEM-4010, JEOL Ltd. Japan, and JEM-2010F, JEOL Ltd.,

Japan). The composition was analyzed by a twin EDS system (Voyager III, Noran, USA)

mounted on an analytical TEM (EM-002BF, Topcon Co., Japan). Thin foils for TEM

observations were prepared by mechanical thinning and ion milling with Ar ion beam.

3. Results

3.1 Structural and chemical compositional analysis in micro meter scale.

Typical SEM micrographs of LC, MC and HC are compared in Fig. 1. The

magnification of all three images are the same. Grey regions represent for WC grains and

black regions are Co-based liquid phase. No fourth phase such as Co3W3C nor free carbon

5
was observed in the images. WC grains in LC are smaller than those in HC. Sintering

conditions were the same for all samples, so the difference of grain size mean that the

grain growth rate is higher in HC than in LC.

Higher magnification images of the macroscopic structures were obtained by TEM.

Fig. 2 shows the bright field (BF) TEM images of LC, MC and HC at the same

magnification. The WC grains were observed as squares or triangles in TEM images

because the grains are triangular prismatic shapes [11,18]. The shapes of WC grains were
_ _
squares when they were observed from [1010] or [1210] and those were triangles from

[0001]. Some small steps were observed at the corners of WC grains as shown by arrows

in Fig. 2. This type of steps are frequently formed in VC [14,20,21] or Cr2C3 [22,23]

doped WC-Co. It is noted that the number of nano-steps were larger in LC than in MC

and that in HC was close to zero. Hence the formation of the nano-steps is closely related

to the carbon content. Here, according to the previous study, nano-steps were observed

when the WC grains with attached trapezoidal prisms on the prismatic planes were seen

from their rhombohedral axis or basal axis [18].

Spatial mapping of chemical composition was carried out in micro meter scale using

scanning TEM (STEM)-EDS. BF-STEM images, V, Co and W composition maps of LC,

MC and HC, respectively, are shown in Fig. 3. WC grains and Co based liquid phase can

be clearly identified in the Co and W maps. Some W solutions were observed in the Co

phase, but almost no Co and V solutions were observed in WC grains. The V maps show

that V exists in the Co phase evenly. Some hotspots were observed in the V maps as shown

by the black arrows in Fig. 3 and almost no Co was observed at the V hot spots although

W was detected. These V hotspots were observed at the points where two WC grains were

contacting with each other. Such features were often observed in VC doped WC-Co

6
[14,20,21]. According to these studies, cubic (W,V)Cx is formed at the hotspots. The

number of hotspots decreases as carbon content increases. Instead, the V concentration at

the hotspots increases and lines with high V concentration are formed as shown by the

white arrows in Fig. 3. V concentrated lines are formed at the interfaces of two WC grains.

This type of interfaces is formed by the grain growth and subsequent collision of two WC

grains [20].

3.2 Atomic scale structural analysis of WC grains/Co phase interfaces.

Atomic structural observations and analysis were carried out to reveal the interfacial

structure of WC grains and Co phase. The shape of WC grains is triangular prisms consist
_
of (0001) triangular basal plane and (10 1 0) rectangular prismatic plane, so all WC
_
(0001)/Co phase interfaces, WC (1010)/ Co phase interfaces and the corners consist of
_
WC (0001) and WC (1010) were investigated using HRTEM technique.

Fig. 4 shows the HRTEM images at the typical corners of WC grains in LC, MC and
_
HC viewed from [1210] of the WC grains. A number of small nano-steps were observed

at the corners of WC grains in LC. The number of nano-steps is smaller in MC than in

LC and the nano-steps were scarcely observed in HC. The corners of nano-steps in all LC,

MC and HC were sharp in atomic scale in contrast to the corners of WC grains in undoped

WC-Co were blunt [24]. In VC doped materials, owing to the sharp corners without any

diagonal habit plane, all WC grains/Co phase interfaces can be classified into WC
_
(0001)/Co phase and WC (1010)/Co phase interfaces. So we need to discuss only the

grain growth at these two types of interfaces. Here, some types of grain growth inhibitors

were observed in the HRTEM images of the WC grains/Co phase interfaces. On WC

(0001)/Co phase interfaces, two types of (W,V)Cx are formed. One is thin planar (W,V)Cx

7
films covering all of WC (0001)/Co phase interfaces evenly. The other is circular (W,V)Cx

as shown by arrows in Fig. 4. This circular (W,V)Cx exists only at the corners of nano-

steps. According to the previous study, the 3D form of these circular (W,V)Cx is

hemisphere and they are arraying along the corners of nano-stairs intermittently [18].

To reveal the detailed structure of grain growth inhibitors at the interfaces, higher
_
magnification HRTEM observations of WC (0001)/Co phase and WC (1010)/Co phase

interfaces in LC, MC and HC, respectively, were carried out as shown in Fig. 5. As for

the WC (0001)/Co phase interfaces, the interfaces in the all hardmetals are fully covered

by a few layers of the planar (W,V)Cx. The thickness of planar (W,V)Cx was about 1 layer

in LC, 2 layers in MC and 2 or 3 layers in HC. According to the W-V-C-Co quaternary

phase diagram [25], the difference of thickness can be explained by the difference of

(W,V)Cx amount: more (W,V)Cx is formed in the hardmetal of higher carbon content than
_
that of lower carbon content. On the other hand, WC (1010)/Co phase interfaces is not
_
fully covered in HC. Some small clusters were observed at the WC (1010)/Co phase
_
interfaces in HC. In contrast, the WC (1010)/Co interfaces in LC are fully covered by

planar (W,V)Cx films.

4. Discussions

4.1 Formation of micro meter scale V rich regions

The V hotspots observed in the STEM-EDS maps in Fig. 3 are formed as a result of

grain growth during liquid phase sintering. These V precipitates are estimated to be much

larger than the atomic scale (W,V)Cx structures observed at the WC (0001)/Co phase

interfaces because the atomic-scale planar or hemisphere (W,V)Cx observed in the

HRTEM images in Fig. 4 and 5 were not detected in the STEM-EDS maps in Fig. 3. All

8
of the V precipitates were observed at the pockets sandwiched by WC grains. These

precipitates are formed when two or more WC grains contact at a corner or a line. The

number of V precipitates is smaller in HC because the number of pockets in HC is smaller

than that in LC according to the difference in WC grain sizes.

The difference in the number of V precipitates is also showing that the grain growth

is slower in LC than in HC. If the WC grains grow after the pocket formation, the WC

grain will contact in face and result in formation of planar V precipitations as shown by

white arrows in Fig. 3. Faster grain growth of the WC grains in HC result in the formation

of more numbers of planar V precipitations than in LC and MC. High V concentration at

V precipitations in HC is according to the non-stoichiometry of (W,V)Cx. (W,V)Cx is

mixed crystal of cubic VC and cubic WC0.88, so higher carbon content result in the higher

V composition.

4.2 Atomic-scale grain growth inhibition mechanism

To control grain growth effectively, growth rate limiting step should be retarded.

Grain growth during liquid phase sintering can be classified into three steps [26]. The first

step is dissolution of W and C from WC grains into liquid phase. The solutes diffuse in

the liquid phase toward other WC grains as the second step. At the final step, diffused

solutes recrystallize on WC grains. The first dissolution and the third recrystallization

process are controlled by interfacial reactions and the second process is controlled by

diffusions in liquid phase. These two types of grain growth limiting steps can be

distinguished by the interface structures. The grains show flat-faced interfaces in the

interfacial reaction controlled material. On the other hand, the grains remain rounded with

rough interfaces in the materials of diffusion controlled growth [27]. In this study, the

9
grain growth limiting step can be determined to be the interfacial reaction process because

all WC grains/Co phase or WC grains/(W,V)Cx interfaces show flat-faced structures

regardless of the carbon content.

Hence the most effective way to inhibit grain growth is reducing reaction sites at the

WC grains/Co phase interfaces by grain growth inhibitors. In LC, both WC (0001)/Co

_
phase interfaces and WC (1010)/Co phase interfaces are fully covered by (W,V)Cx grain
_
growth inhibiting films as shown in Fig. 5. On the other hand, WC (1010)/Co phase

interfaces in HC are not fully covered, though there are a few layers of (W,V)Cx films at

the WC (0001)/Co phase interfaces. This carbon content dependency of the interface

structures results in the difference in grain growth rate. The fully covered WC grains in

LC grow slower than the partially covered WC grains in HC.

The step structures formed at the corners of WC grains are nonequilibrium structures.

The schematics of the corners of WC grains (a) with and (b) without nano-steps are

compared in Fig. 6. Although the WC grains/Co phase interface area in Fig. 6(a) and (b)

are exactly the same, the number of W and C atoms are smaller in (a) than in (b). In the

other words, if the number of W and C atoms in WC grains are constant irrespective of

the existence of nano-steps, the area of WC grains/Co phase interfaces becomes larger on

the WC grains with nano-steps than on that the WC grains without nano-steps. These facts

mean that the formation of nano-steps result in increase of interface area. Hence the nano-

steps do not exist at the equilibrium state.

This non-equilibrium step structures are formed during formation of (W,V)Cx phases.

Undoped WC-Co hardmetals show diagonal habit planes at the corner of WC (0001) and
_
WC (10 1 0) [24]. The truncated WC grains are already formed below the eutectic

temperature [28]. The WC grains in VC-doped hardmetals also form truncated corners

10
before reaching the temperature of (W,V)Cx phase formation. The WC grains surrounded

by planar (W,V)Cx prefer to form sharp corners than the corners with diagonal habit

planes because of the interface energy balance, so the truncated corner changes their

structure to nano-steps after forming (W,V)Cx layers. The nano-steps are formed as an

intermediate state between truncated corner and sharp corner as schematically shown in

Fig. 7. Although the structure is transient state, the structure is kept during sintering

because of quite low grain growth rate. More nano-steps are observed in LC than in HC

because the truncation of WC grains is larger in LC than in HC [24] and grain growth is

slower in LC than in HC.

4.3 Controlling interface structure by carbon content

The structure of the grain growth inhibitors varies by changing carbon content. The

origin of these structural variations must be revealed to precisely control the grain growth

by controlling interface structures. Here, the mechanism is discussed based on the lattice

mismatch between WC and (W,V)Cx. The composition of (W,V)Cx depends on the carbon

content because of its non-stoichiometry. The lattice constant of (W,V)Cx depends on the

composition, so the lattice mismatch also depends on the composition. The difference of

lattice mismatch results in the difference of (W,V)Cx film structure. When the same

thickness of crystalline thin films are deposited on crystalline grains, the films of small

lattice mismatch generally form flat-face surface and those of large lattice mismatch form

rough surface with nano-clusters or islands [29].

Fig. 8 compares HRTEM images at the corners of nano-steps in LC, MC and HC. The

planar (W,V)Cx on WC (0001) are thinner in LC than in HC and hemisphere (W,V)Cx is

larger in LC than in HC. So the lattice mismatch between WC (0001) and (W,V)C x is

11
expected to be larger in LC than in HC. On the other hand, the (W,V)C x forms planar
_
films on WC (1010) in LC and that forms islands in HC as shown in Fig. 5. The lattice
_
mismatch between WC (1010) and (W,V)Cx is expected to be smaller in LC than in HC.

To qualify this discussion, the lattice mismatch for each interfaces are measured in

HRTEM images. The lattice mismatch between (W,V)Cx (111)/WC (0001) and (W,V)Cx
_
(110)/WC (1010) in LC, MC and HC, respectively, are summarized in Table 2. This

result is consistent with the tendency of the lattice mismatch expected from the (W,V)Cx

morphology.

To confirm the discussion above, interface energy balance estimations were

performed using droplet wettability model. The model describes the wettability of the

liquid on the solid in terms of force equilibrium at the edges of droplets. The relationship

between surface tension and wetting angle is described by Young’s modulus:

 S −  S / D =  D cos (1)

Here, S, S/D and D are an interface energy between substrate (WC grains in this case)

and surrounding phase (liquid phase), an interface energy between substrate and droplet

((W,V)Cx), and an interface energy between droplet and surrounding phase, respectively.

 is wetting angle. Usually, the wetting angles of the solid droplets or islands grown on

solid substrate cannot be described by Young’s modulus because the solid islands

generally show clear habit planes [29,30]. On the other hand, (W,V)Cx grain growth

inhibition phase in VC-doped WC-Co hardmetals form hemisphere droplets without any

clear habit planes. In such system, the total interface energy is not dominated by habit

plane dependency, but by the interface area and wetting angle. Hence the (W,V)Cx

droplets can be dealt in the same way with a liquid droplet on a solid substrate.
_
The (W,V)Cx droplets are contacting with both WC (0001) and WC (101 0). Two

12
 _
wetting angle between (W,V)Cx droplets/WC (0001) and droplets/WC (1010) are defined

as 1 and 2, respectively, as shown in Fig. 8(d). The values of 1 and 2 in LC, MC and

HC are summarized in Table 2. 1 is smaller in HC than in LC and 2 is larger in HC.

These results mean that the wettability of (W,V)Cx droplets on WC (0001) is better in HC
_
than in LC and that of (W,V)Cx droplets on WC (1010) is worse in HC. In terms of the
_
interface energy, (W,V)Cx droplets/WC (1010) interface energy is higher in HC than in

LC, and the energy of (W,V)Cx droplets/WC (0001) interface is lower in HC than in LC.

These interface energy estimations are consitent with the lattice mismatch estimated from

HRTEM images in discussions above and compositions of (W,V)Cx in references [25].

The conclusions obtained from these discussions are summarized as schematics in Fig. 9.
_
The bad wettability of (W,V)Cx on WC (1010) in HC results in formation of (W,V)Cx
_
islands. The (W,V)Cx cannot cover the whole WC (1010)/Co phase interfaces in HC, and
_
hence, the grain growth is faster in HC than in LC. In contrast, in LC, WC (1010)/Co

phase interfaces are fully covered by (W,V)Cx according to the good wettability and result

in slow grain growth. As for the (W,V)Cx/WC (0001) interfaces, the wettability is good

enough to form planar (W,V)Cx layers in hardmetals of all carbon contents, so the

difference of wettability at this type of interface does not affect grain growth rate so much.

To obtain superior grain growth inhibition effect, dopants of the small lattice mismatch
_
against both WC(0001) and WC(1010) should be selected.

5. Conclusion

TEM observations, EDS mappings and atomic scale HRTEM observations were

carried out in quenched VC-doped WC-Co hardmetals with various carbon content. The

13
rapid quenching enabled us to observe microscopic features at the sintering state of

hardmetals with various carbon content. The combination of micro-meter scale

observations, chemical compositional analysis and atomic scale observations revealed the

reason for the difference of grain growth rate in the materials with various carbon content.

The difference of the grain size is according to both the grain growth before and after the

grain growth inhibition phase covers the WC grains.

(W,V)Cx grain growth inhibitors form two types of structures during sintering. One

type is large precipitations at the pockets in-between WC grains or at the WC/WC

interfaces. This type of (W,V)Cx retards only the grain growth of WC grains touching

with large (W,V)Cx precipitates. The other type is (W,V)Cx layers or islands formed at the

WC grains/Co phase interfaces. This type of (W,V)Cx covers whole WC grains and

strongly inhibiting grain growth. The structures of (W,V)Cx at the WC grains/Co phase

interfaces are depending on the carbon content. The difference of grain growth rate during

the liquid phase sintering is revealed to be depending on the WC grains/Co phase interface

structures, especially whether the (W,V)Cx forms planar layers or islands on the WC (10
_
10)/Co phase interfaces. The planar layers of (W,V)Cx are formed in hardmetals with low

carbon content and the formation of planar (W,V)Cx layers result in high grain growth

inhibition effect.

14
References
[1] A. Mukhopadhyay, B. Basu, J. Mater. Sci. 46 (2011) 571.
[2] Z.Z. Fang, X. Wang, T. Tyu, K.S. Hwang, H.Y. Sohn, Int. J. Refract. Metals Hard
Mater. 27 (2008) 288.
[3] C.M. Fernandes, A.M.R. Senos, Int. J. Refract. Metals Hard Mater. 29 (2011) 406.
[4] K. Hayashi, Y. Fuke, H. Suzuki, J. Jpn. Soc. Powder Powder Metall. 19 (1972) 658.
[5] W.D. Schubert, A. Bock, B. Lux, Int. J. Refract. Metals Hard Mater. 13 (1995) 281.
[6] A. Delanoe, S. Lay, Int. J. Refract Metals Hard Mater. 27 (2009) 189.
[7] H. Suzuki, M. Sugiyama, T. Umeda, J. Jpn. Inst. Met. Mater. 2 (1963) 55.
[8] A. Delanoe, S. Lay, Int. J. Refract. Metals Hard Mater. 27 (2009) 140.
[9] J. Gurland, Trans. Am. Inst. Mining Metall. Eng. 203 (1955) 311.
[10] H. Suzuki, Y. Fuke, K. Hayashi, J. Jpn. Soc. Powder Powder Metall. 19 (1972) 663.
[11] H.R. Lee, D.J. Kim, N.M. Hwang, D.Y. Kim, J. Am. Ceram. Soc. 86 (2003) 152.
[12] A.V. Shatov, S.S. Ponomarev, S.A. Firstov, R. Warren, Int. J. Refract. Metals Hard
Mater. 24 (2006) 61.
[13] T. Yamamoto, Y. Ikuhara, T. Sakuma, Sci. Tech. Adv. Mater. 1 (2000) 97.
[14] S. Lay, S. Hamar-Thibault, A. Lackner, Int. J. Refract. Metals Hard Mater. 20 (2002)
61.
[15] J. Weidow, H.O. Andren, Acta Mater. 58 (2010) 3888.
[16] M. Kawakami, O. Terada, K. Hayashi, J. Jpn. Soc. Powder Powder Metall. 51 (2004)
576.
[17] I. Sugiyama, Y. Mizumukai, T. Taniuchi, K. Okada, F. Shirase, T. Tanase, Y. Ikuhara,
T. Yamamoto, Int. J. Refract. Metals Hard Mater. 30 (2012) 185.
[18] I. Sugiyama, Y. Mizumukai, T. Taniuchi, K. Okada, F. Shirase, T. Tanase, Y. Ikuhara,
T. Yamamoto, Scripta Mater. 69 (2013) 473.
[19] S. Sundin, S. Haglund, Int. J. Refract. Metals Hard Mater. 18 (2000) 297.
[20] S. Lay, J. Thibault, S. Hamar-Thibault, Phil. Mag. 83 (2003) 1175.
[21] S. Lay, S. Hamar-Thibault, M. Loubradou, Interface Sci. 12 (2004) 187.
[22] T. Yamamoto, Y. Ikuhara, T. Watanabe, T. Sakuma, Y. Taniuchi, K. Okada, T. Tanase,
J. Mater. Sci. 36 (2001) 3885.
[23] A. Delanoe, M. Bacia, E. Pauty, S. Lay, C.H. Alibert, J. Crystal Growth 270 (2004)
219.
[24] I. Sugiyama, M. Goto, T. Taniuchi, F. Shirase, T. Tanase, K. Okada, Y. Ikuhara, T.
Yamamoto, J. Mater. Sci. 46 (2011) 4413.
[25] E.G. Obbard, S. Luyckx, S. Hamar-Thibault, C.H. Allibert, Int. J. Refract. Metals
Hard Mater. 19 (2001) 349.

15
[26] R.M. German, P. Suri, S.J. Park, J. Mater. Sci. 44 (2009) 1.
[27] W. Jo, D.Y. Kim, N.M. Hwang, J. Am. Ceram. Soc. 89 (2006) 2369.
[28] V. Bounhoure, J.M. Missiaen, S. Lay, E. Pauty, J. Am. Ceram. Soc. 92 (2009) 1396.
[29] I. Daruka, A.L. Barabasi, Phys. Rev. Lett. 79 (1997) 3708.
[30] K. Nakajima, J. Crystal Growth 203 (1999) 376.

Acknowledgement
This study was supported in part by the Grant-in-Aid for Scientific Research from Japan Society for the Promotion of

Science (JSPS) and Ministry of Education, Culture, Sports and Technology (MEXT). I. S. is supported by JSPS as
research fellow (JSPS-PD). This work is supported by a Grant-in-Aid for Scientific Research on Innovative Areas

"Nano Informatics" (Grant No. 25106003, 25106007) from JSPS and "Nanotechnology Platform" (Project No.

12024046) of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan.

16
Table and figure captions

Table 1. Magnetic saturation and W compositions in Co phase.

Table 2. Lattice mismatch and wetting angle between (W,V)Cx/WC(0001) and

_
(W,V)Cx/WC (1010) in LC, MC and HC.

Fig. 1. SEM images of (a) LC, (b) MC and (c) HC. The all images are shown in the same
scale. Grey regions are WC grains and black regions are Co phases.

Fig. 2. Typical bright field TEM images of (a) LC, (b) MC and (c) HC. The all images
are shown in the same scale. The arrows indicate the nano-steps.

Fig. 3. Bright field STEM, EDS W composition maps, Co composition maps and V
composition maps of (a) LC, (b) MC and (c) HC. Color scales are embedded in each maps.
The black arrows with white borders in V maps indicate the dots type V precipitates and
the white arrows show the lines type V precipitates.

Fig. 4. High resolution TEM images at the typical corners of WC grains in (a) LC, (b)
MC and (c) HC. The arrows indicate hemisphere (W,V)Cx.

Fig. 5. High resolution TEM images of (a), (c), (e) WC (0001)/Co phase interfaces and
_
(b), (d), (f) WC (1010)/Co phase interfaces of (a), (b) LC, (c), (d) MC and (e), (f) HC.

The bars and arrows in (a) – (e) indicate planar (W,V)Cx and the arrows in (f) indicate V
clusters.

Fig. 6. Schematics of the corners of WC grains (a) with and (b) without nano-steps. The
both schematics have the same interface area as shown by solid lines. The number of W
and C atoms are clearly smaller in (a).

Fig. 7. Schematics of the nano-steps formation process. The schematics are showing the
upper-right corner of WC grain. (a) The corner of WC grain before (W,V)Cx interfacial
phase are formed. (b) The corner of WC grain with nano-steps. The structure is formed
by the (W,V)Cx interfacial phase formation. (c) The corner of WC grain without nano-

17
steps. This type of structure is formed by grain growth after the (W,V)Cx interfacial phase
formation.

Fig. 8. High resolution TEM images at the corners of nano-steps in (a) LC, (b) MC and
(c) HC. (d) is a schematic of wetting angle between (W,V)Cx grain growth inhibitors, WC
grains and Co based liquid phase.

Fig. 9. Schematics of corners of nano-steps in (a)LC and (b) HC. WC (1010)/liquid phase
interfaces are fully covered by (W,V)Cx in LC and partially covered in HC. WC
(0001)/liquid phase interfaces are fully covered regardless of carbon contents. The
(W,V)Cx droplets are larger in LC than in HC. The arrows are schematically showing the
difference of grain growth rate.

Table 1
Low Carbon Middle Carbon High Carbon
4 (kgf) 179.6 201.9 227.7
W in Co (wt%) 26.33 20.18 12.02

Table 2
Low Carbon Middle Carbon High Carbon
Lattice mismatch
WC (0001) 3.90% 1.30% 0.60%
_
WC(1010) 5.70% 7.90% 9.60%

Wetting angle
WC (0001) 27.1 19.9 13.8
_
WC (1010) 148.1 155.7 162.4

18