Professional Documents
Culture Documents
a r t i c l e
i n f o
Article history:
Received 3 June 2008
Accepted 2 October 2008
Keywords:
Tungsten carbide
Cobalt
Nickel
Internal friction
Cermet
Cemented carbide
a b s t r a c t
Composites made of transition metal carbides, with a ferrous metal binder, commonly dened as hardmetals, are used for cutting tool applications. We investigate the mechanical behavior of the two main families
of hardmetals: the cemented carbides, constituted of WCCo, and the cermets made of a combination of
TiCTiN(Co,Ni). The comparative analysis of internal friction spectra of both hardmetal families as well
as the analysis of spectra of the ceramic phases alone (after binder etching) allows associating different
relaxation peaks with specic phases or grain boundaries. The mechanical properties of hardmetals are
controlled by the binder. Three temperature domains can be dened. Increasing the temperature binder
microplasticity occurs rst, then binder segregation at the ceramic grain boundaries prepares the way for
inltration and extensive deformation by grain boundary sliding.
2009 Elsevier B.V. All rights reserved.
1. Introduction
Composites made of transition metal carbides with a ferrous
metal binder are widely used for the manufacturing of cutting
tools. They are commonly dened as hardmetals. Among them, two
main families of materials can be distinguished: the cemented carbides, mainly constituted of WCCo, and the cermets made of a
combination of TiCTiN(Co,Ni) with other carbides [1]. The main
qualities required for a cutting tool are a good toughness and a
good resistance to deformation at high temperature. The cemented
carbides have a very high toughness and good hardness. However,
the performances of such materials are rapidly deteriorated when
high cutting speed and high temperatures are involved because
WCCo has a poor chemical resistance and undergoes plastic deformation at high temperature [2]. The cermets have in general a
better resistance to high temperature deformation but toughness
is lower than in cemented carbides. The morphology of hardmetals
is well described by a structure made of two interpenetrated skeletons: the ceramic and the metal ones. The ceramic skeleton is an
arrangement of joined single-crystal carbide or carbonitride grains.
In WCCo, grains are faceted with triangular or rectangular shape
revealing the hexagonal structure of WC. In TiCN-based composites,
the grains have a core-rim structure where the core corresponds
to pure TiCN and the rim to (Ti,M)CN precipitated during solidliquid phase sintering [35]. M represents a transition metal carbide
323
Table 1
Composition of the alloys presented in this work. Phase fractions are given in at%. The
grade denominations instead refer to vol%. In TiCN-based cermets, if not specied
the ratio C/N is 30/70.
Alloy
Co/Ni
Ti
Mo
WC18Co
WC10Co
TiCN6Mo18Co
TiCN13Mo7Ni
TiCN5WC10Co
(TiC0.5 N0.5 )6Mo6Co
TiCN6Mo6Co
16.93
9.71
9.13
6.15
9.13
10.47
5.80
41.53
45.15
0.00
0.00
3.92
0.00
0.00
0.00
0.00
38.96
41.37
42.08
45.39
45.21
0.00
0.00
3.65
7.40
0.00
1.72
1.70
0.00
0.00
11.69
12.41
16.41
22.82
13.76
41.53
45.15
26.99
32.66
28.46
24.22
33.53
()
1 + (2 2 )
0<1
(1)
(2)
Fig. 1. Grain structure of (a) WC18 vol%Co cemented carbide and (b) TiCN-5 vol%WC10 vol%Co cermet. The cobalt is bright in the cermet and darker in the WCCo. In the
cermet, the typical core-rim structure can be observed. The white regions near the core represent the inner rim formed during solid state sintering.
324
Fig. 4. Deconvolution of the internal friction spectrum (curve C in Fig. 2) after high
temperature background subtraction. A transition between two peaks PW1 at low
temperature and PW1 at high temperature can be observed. Another peak PC is tted
in correspondence with the magnetic transition.
Fig. 7. Deconvolution of the spectra of (curve 1) (TiC0.5 N0.5 )6Mo6Co and (curve 2)
(TiC0.3 N0.7 )6Mo6Co after background subtraction. The peak PT0 is much higher in
the high nitrogen alloy. After rst heating (curve 3) PT0 has disappeared; PT1 and
PT2 can be distinguished.
325
(TiC0.5 N0.5 )6Mo6Co and high carbon (TiC0.7 N0.3 )6Mo6Co cermet
are presented. Their comparison shows a strong dependence of PT0
and PT2 on nitrogen content in TiCN. Upon the rst heating, PT0 is
clearly higher in (TiC0.5 N0.5 )6Mo6Co. In the second heating, where
PT0 has disappeared, we can notice that also PT2 is much higher in
the high nitrogen sample.
3.3. Deconvolution of the high temperature part of the IF spectra:
frequency scans
The high temperature part of the spectra is not accessible by
temperature scans at frequencies in the range of 1 Hz typical of
free pendula. Frequency scans performed with forced pendula add
information about high temperature mechanisms. Fig. 8a shows the
deconvolution of the spectrum of WC10Co in isothermal frequency
scans. A low frequency the background is subtracted. Two relaxation peaks PW2 and PW3 are well dened. Their tted shape is
shown for one temperature. The amplitude of PW2 and PW3 varies
in parallel as a function of the temperature attaining a maximum at
about 1350 K (Fig. 8b). The same peaks also were found in the WC
skeleton at much lower amplitude. In reality, the low frequency
Fig. 9. (a) High temperature frequency scan in TiCN5WC10Co. At very low frequency,
a high temperature peak is outlined PTht. (b) After subtraction of the tted peak,
another peak PT4 is clearly appearing between 0.1 Hz and 1 Hz. The amplitude of
this peak is increasing as a function of temperature in a reversible way. For clarity,
only two curves measured on cooling at 1425 K and 1475 K are reported.
Fig. 8. Deconvolution of the frequency scan spectrum of WC10Co. (a) Two peaks
PW2 PW3 can be distinguished. They are thermally activated and (b) their amplitude
varies as a function of temperature.
326
Table 2
Fit parameters of the peaks found in TiCN-based cermets. Activation energies used in the t are taken from isothermal measurements.
/2
Temperature (1 Hz)
Broadening 1/
PT0
PT0 in skeleton
28 10
2 103
920
920
2.2
2.2
10
1013
1.3
1.9
PT1
24 103
980
1.6
109
1.8
13
PT2
PT2 in skeleton
48 10
1 103
1150
1150
2.1
1.7
10
109
1.6
1.6
PT3
PT3 in skeleton
1 103
0.5 103
1500
1500
2.7
2.3
1012
1010
1.3
1.2
PT4
PT4 in skeleton
060 103
3 103
1500
1400
4.0
3.4
1015
1013
1.3
1.0
PTht
0.8
>1700
5.3
1017
3.0
10
Table 3
Fit parameters of the peaks found in WC-based cemented carbides. Activation energies used in the t are taken from isothermal measurements.
/2
Temperature (1 Hz)
3
Broadening 1/
PW1
PW1
15 10
13 103
1080
1000
2.5
2.5
10
1012
1.3
2.5
PW2
PW2 in skeleton
120 103
4 103
1330
1330
2.0
2.6
1010
1010
1.4
1.5
PW3
PW3 in skeleton
115 103
5 103
1440
1440
4.5
4.9
1016
1018
1.2
1.2
12
large broadening of the peak PW1 can be justied by the fact that
in reality the spinodal decomposition has a quite irregular distribution. Moreover, there may be some regions where it is even not
present yielding dislocations with a larger free length. The hypothesis formulated in this work also justies the observation of a critical
behavior activation energy of the peak PW1 with a critical temperature estimated at 1073 K [19]. The model described in Ref. [19] is
essentially correct even if the phase transformation that should be
considered is not the hcp to fcc transformation but the solution
of the spinodal. The high temperature stable peak PW1 should be
a dragging peak than PW1 as conrmed by the activation energy
and attempt frequency. An important consequence of this analysis
is that the spinodal decomposition of the WCo alloy can give a substantial contribution to cobalt hardening in the cemented carbides
and therefore increase their mechanical resistance to deformation.
In the TiCNMoCo alloys, the peak PT1 has never shown a
behavior similar to PW1 even if from the phase diagram an equivalent spinodal decomposition could be expected [20]. This peak
should be attributed to Mo dragging by dislocations [10].
4.2. Towards grain boundary sliding: point defect relaxations at
grain boundaries
At intermediate temperatures, a number of relaxation peaks
appears both in cermets and in cemented carbides. At about 980 K,
the peak PT0 is observed in Co-based cermets. It characterizes the
as sintered structure and disappears after the rst heating above
1000 K. This peak is generally sensitive to material chemistry. It
increases with the cobalt content and, as shown in Fig. 7, it depends
strongly on the nitrogen content in the ceramic phase. It was shown
previously [10] that when the peak is present a thin cobalt lm
can be detected at grain boundaries. Therefore, the presence of PT0
witnesses the ability of cobalt to inltrate the grain boundaries
in a more or less massive way. On the other hand, Fig. 7 shows
that the peak PT2 is highly inuenced by nitrogen content and
in another paper [10] the dependence of PT2 PT3 on cobalt content was demonstrated. The similarity of the behavior of PT0 with
the PT2 PT3 pair suggests a different interpretation of these two
peaks with respect to former literature. The presence of PT2 PT3
has been associated with the movement of dislocations in the TiCN
grains controlled by a mechanism equivalent to the Bordoni peak
for the ceramic phase [6,21]. This interpretation was justied by
the observation of dislocation multiplication when cermets are
deformed in the temperature range where PT2 PT3 are observed at
1 Hz, and by the dependence of the peak amplitude on the binder
content. This last effect was supposed to be induced, as the Bordoni peak, by internal stresses [22]. In fact, such stresses increase in
the ceramic phase when the binder content increases. However, as
demonstrated by diffraction measurements, the residual stresses
are almost completely relaxed above 1200 K [23] where PT2 PT3
appear. As shown in [10], PT2 PT3 are also quite sensitive to the Mo
content in the ceramic. The sensitivity to the ceramic composition
and to the cobalt content suggests that the process involves both
constituents. The process must however be located in the ceramic
since it is present when the cobalt is removed or with a Ni binder.
The most likely hypothesis is that grain boundaries are involved
and that the process is related to cobalt (or nickel) segregation at
the grain boundaries. As shown in Fig. 7, also nitrogen has an important effect. Nitrogen is known to modify the Mo activity to favor its
dissolution in the binder [24]. Therefore, we propose that both PT0
and the peaks PT2 PT3 involve diffusion processes at the TiCN grain
boundaries enhanced by the solution of Mo or W into the binder.
It is tempting to draw a parallel with peaks PW2 and PW3 in the
cemented carbides. Also these two peaks are always detected in
parallel and have similar activation energies. They appear at higher
temperature though. The peaks PW2 PW3 show a varying ampli-
327
328
[3] D.I. Chun, D.Y. Kim, J. Am. Ceram. Soc. 76 (1993) 20492052.
[4] H.-O. Andrn, U. Rolander, P. Lindahl, Int. J. Refract. Metals Hard Mater. 12
(19931994) 107113.
[5] J. Zackrisson, U. Rolander, H.-O. Andrn, Metall. Mater. Trans. A 32 (2001) 8594.
[6] D. Mari, S. Bolognini, G. Feusier, T. Viatte, W. Benoit, Int. J. Refract. Metals Hard
Mater 17 (1999) 209225.
[7] D. Mari, S. Bolognini, T. Viatte, W. Benoit, Int. J. Refract. Hard Metals 19 (2001)
257265.
[8] R.M. Fuoss, J.G. Kirkwood, J. Am. Chem. Soc. 63 (1941) 385394.
[9] M. Gardon, Etude par Spectromtrie Mcanique de lEvolution Structurale
Haute Temprature du Composite WC-Co, Ph.D. Thesis INSA, Lyon 1993.
[10] S. Bolognini, G. Feusier, D. Mari, T. Viatte, W. Benoit, Int. J. Refract. Metals Hard
Mater. 16 (1998) 257268.
[11] D. Mari, K. Buss, W. Benoit, Mater. Sci. Eng. A 442 (2006) 179183.
[12] R. Schaller, J.J. Ammann, C. Bonjour, Mater. Sci. Eng. A 105106 (1988) 313321.
[13] R. Schaller, D. Mari, M. Maamouri, J.J. Ammann, J. Hard Mater. 3 (1992) 351362.
[14] H. Okamoto, J. Phase Equilib. Diffus. 29 (2008) 119.
[15] A. Fernndez Guillermet, Metall. Trans. A 20A (1989) 935956.
[16] G. stberg, B. Jansson, H.-O. Andrn, Scripta Mater. 54 (2006) 595598.
[17] A. Davin, V. Leroy, D. Coutsouradis, L. Habraken, Memoires Scientiques de la
Revue de Mtallurgie 60 (1963) 275283.
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]