Cemented Carbide Sintering:

Constitutive Relations and

Microstructural Evolution
Anders Petersson

Stockholm 2004
Doctoral Dissertation
Royal Institute of Technology
Department of Materials Science and Engineering

Akademisk avhandling som med tillstnd av Kungliga Tekniska Hgskolan framlgges till offentlig granskning fr avlggande av teknologie doktorsexamen onsdagen
den 28 april 2004 kl 10.00 i Kollegiesalen, Administrationsbyggnaden, Kungliga
Tekniska Hgskolan, Valhallavgen 79, Stockholm.
Fakultetsopponent r Professor Colette Allibert, Institut National Polytechnique
de Grenoble, Frankrike.
Typsatt i LATEX.
ISBN 91-7283-739-x
ISRN KTH/MSE/--04/21--SE+METO/AVH
c Anders Petersson, March 2004

Hgskoletryckeriet, Stockholm 2004

Abstract
Cemented carbides based on tungsten carbide and cobalt are commonly produced
by a powder metallurgy route including liquid phase sintering. The pressed compact
densifies to almost half its volume during sintering due to pore elimination. The
sintering behaviour changes with material composition, such as carbide grain size,
binder fraction, carbon content and addition of cubic carbides.
This thesis is devoted to the study of constitutive behaviour, in particular densification, and the microstructural evolution during cemented carbide sintering.
Dimensional changes are monitored using dilatometry with and without applied
external load. The microstructural evolution is investigated with light optical microscopy and scanning electron microscopy. Thermodynamic calculations are used
as reference.
Constitutive relations are derived for uniaxial viscosity, viscous equivalent of
Poissons ratio and sintering stress based on relative density and temperature. The
relations are extended to a model describing sintering shrinkage with explicit dependencies on carbide grain size and binder content. The model is divided in three
stages of which two pertain to the solid state and the third to liquid phase sintering.
Solid state shrinkage is suppressed in a material with coarse carbides and in the
stage of liquid phase sintering grain size strongly influences the uniaxial viscosity.
The binder content affects primarily the later densification.
The effects of carbon content and grain size distribution on shrinkage have been
studied. High carbon content enhances shrinkage rate, but the effect of grain size
distribution is rather small. The mean carbide grain size is insufficient to describe
densification for very broad distributions only.
Shrinkage occurs through rearrangement and solution-reprecipitation. Rearrangement is studied through the evolution of the pore size distribution and simulated generically using a discrete element method.
Keywords: Cemented carbides, Sintering, Constitutive relations, Microstructure,
Densification, Modelling
ISBN 91-7283-739-x ISRN KTH/MSE/--04/21--SE+METO/AVH

iii

iv

Preface
This thesis is based on the following appended papers that will be referred to in
the text by their Roman numerals.
I. A. Petersson and J. gren, Constitutive behaviour of WC-Co materials with
different grain size sintered under load, Acta Mater 52(2004):7, 18471858
II. A. Petersson and J. gren, Numerical simulation of shape changes during
cemented carbide sintering, In: F.D.S. Marquis (ed.) Powder Materials:
Current Research and Industrial Practices III, TMS, Warrendale, 2003, 65
75
III. A. Petersson and J. gren, Modelling WC-Co sintering shrinkageEffects
of carbide grain size and cobalt content, submitted manuscript
IV. A. Petersson, Sintering shrinkage of WC-CO and WC-(Ti,W)C-Co materials
with different carbon contents, in manuscript
V. A. Petersson and J. gren, Sintering shrinkage of WC-Co materials with
bimodal grain size distributions, submitted manuscript
VI. A. Petersson and J. gren, Rearrangement and pore size evolution during
WC-Co sintering below the eutectic temperature, submitted manuscript
VII. A. Petersson, B. Jansson, J. Qvick and J. Zackrisson, M6 C formation during
sintering of cemented carbides containing (Ti,W)C, Int J Refract Met Hard
Mat 22(2004), 2126
The authors contributions to the co-authored papers, besides being the principal writer, planning most of the experiments and performing all calculations, are
participation in experimental work for Paper I, parts of experimental work for Papers II, III and VII and all experimental work for Paper VI. The author performed
microscopy for Papers II, V and VI and the main part of computer coding and simulations for Paper VI. The author outlined the main part of analysis and evaluation
for Papers IIII and contributed to the evaluation for Papers VVII.

vi

Contents
Introduction

Objectives

Liquid phase sintering

Surface energy relations . . . . . . . . . . . . . . . . . . . . . . . . . . .
Sintering under applied stress . . . . . . . . . . . . . . . . . . . . . . . .

5
6
8

Cemented carbide sintering

Thermodynamic aspects . . . . . . .
Binder spreading and rearrangement
Solution-reprecipitation and the final
Microstructural evolution . . . . . .

. . . .
. . . .
stage
. . . .

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

Topics of present work

Constitutive modelling . . . . . . . . . . . . . . . . .
Outlining the parameters: Paper I . . . . . . .
Composition and sintering shrinkage: Paper II
A shrinkage model for binder content and grain
Densification . . . . . . . . . . . . . . . . . . . . . .
Influence of chemical composition: Paper IV . .
Effect of particle size distribution: Paper V . .
Microstructural evolution . . . . . . . . . . . . . . .
Rearrangement: Paper VI . . . . . . . . . . . .
M6 C formation and dissolution: Paper VII . . .
Concluding remarks

.
.
.
.

.
.
.
.

.
.
.
.

9
10
11
13
14

. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
size: Paper III
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .
. . . . . . . . .

.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.

17
17
17
19
20
22
22
23
24
24
27

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

.
.
.
.

29

vii

viii

Introduction
Cemented carbides is a class of composite materials in which a carbide, often tungsten carbide (WC), provides hardness and wear resistance and a metal binder,
almost exclusively cobalt (Co), provides toughness [1]. These materials have long
been used in applications such as cutting, grinding, drilling and wear-resistant mechanical components. A wide range of material properties can be obtained by
changing the material composition. Although the most often used carbide is the
hexagonal WC, atlternatives are available. With cubic carbides such as TiC, NbC
and TaC the high temperature properties are improved [2]. Minor additions, typically less than a few percent by mass, of for instance Cr2 C3 or VC, are often used
as grain growth inhibitors [2].
Cemented carbides are commonly produced using a powder metallurgy route.
Powders with the proper compositions are mixed and milled with additives, spraydried to granules and compacted to green compacts. The added pressing aid is
removed in a de-waxing step and the compacts sintered to eliminate porosity and
increase strength.
The microstructural events occuring during sintering are driven by the lowering
of excess energy attributed to surfaces and the pressure that develops under a
curved interface. The surface area and energy reduction may be accomplished
in different ways. One major division in sintering theory is between solid state
sintering and liquid phase sintering, respectively, depending on whether or not any
liquid phase is present. Densification during solid state sintering is assumed to
be due to vacancy flux from pores to grain boundary sinks driven by the vacancy
concentration difference [3, 4]. Some possible mechanisms of matter transport are
viscous or plastic flow, evaporation and condensation, volume diffusion and surface
diffusion [3].
The presence of a small amount of liquid phase is often beneficial for sintering
and three stages are indentified in classical liquid phase sintering theory [5], distinguished by the microstructure evolution. Firstly, as the liquid forms, it flows out in
the microstructure and causes the solid particles to rearrange into denser packing.
Secondly, the particle centres approach and density increases by means of solution
and reprecipitation. Thirdly, the solid particles can sinter to a rigid skeleton.
Cemented carbide sintering is in some respects in between solid state sintering
and liquid phase sintering, but should for two reasons rather be characterised as
1

Introduction

a liquid phase sintering process. Sintering is in practice almost always performed

at temperatures where a quite significant volume fraction liquid exists. However,
for finer grained cemented carbides much of the shrinkage, if not the most, occurs
during heating to the sintering temperature and prior to liquid formation. This
shrinkage shares however most typical features with the first stage of classical liquid
phase sintering rather than solid state sintering processes.
The particle bonding and increased strength during sintering is often, but not
always, accompanied by pore elimination and densification. The term sintering is
therefore frequently used synonymously with densification. Since sintering can be
performed without shrinkage, or even with swelling, the rate of increasing density
will in the following be referred to as densification rate, shrinkage rate or strain
rate rather than sintering rate.
A vast amount of literature is available in the technically important field of
cemented carbides and their sintering. Extensive reviews are given with emphasis
on fundamental physical and chemical properties in Reference [1], industrial aspects
in Reference [2], mechanical properties in Reference [6], microstructural aspects in
Reference [7] and sintering in Reference [8].

Objectives
Present work is intended to improve knowledge of sintering in cemented carbide systems and propose models that can describe shape changes during cemented carbide
sintering. The work is founded on two pillars, namely the constitutive behaviour
during sintering and the concurrent microstructural evolution. These cannot be
separated since it is through the microstructural changes the macroscopical dimensions of the compacts change.
This is largely an experimental work aiming for a broad description of cemented
carbide sintering, but limited to mediumcoarse carbide grain sizes. The main tool
for studying densification is dilatometry, which is a thermo-analytical (TA) tool,
in which temperature and length of the sample are simultaneously recorded. By
taking the time derivative of the length measurement it is possible to obtain the
instantaneous shrinkage rate that is modelled on a physically sound basis. The
microstructural evolution accompanying the shape changes is investigated mainly
using scanning electron microscopy (SEM). Thermodynamic constraints are crucial
for sintering and thermodynamic calculations are therefore an integral part of the
sintering studies.
First and foremost, a dense compact is required since retained porosity is detrimental to most of the mechanical properties [6]. The main focus of the constitutive
behaviour is therefore on densification. However, friction between powder and die
wall, as well as within the powder, may lead to a heterogenous packing in the
green compact. This will in turn lead to heterogenous shrinkage during sintering. Shrinkage calculations from given green densities should therefore be useful
in compaction-tool design aiming for near net shaping. Compaction and sintering densification of a WC-Co material were successfully simulated in the works
of Brandt [9] and Haglund [10]. The composition of a cemented carbide dictates
extensively its shrinkage behaviour and the aim of present work is to extend the
constitutive modelling to enable simulations for alloys covering different grain sizes,
varying carbon contents and addition of cubic carbides.
Due to the complexity of cemented carbide sintering, the aim of a reasonably
short-term research cannot be to fully describe the parameters with bearing on
densification. That would be a tremendous task. This work rather strives to isolate
the most crucial aspects of sintering and characterise their variaton with main
compositional parameters. Furthermore, experimental information on some of the
3

Objectives

key parameters such as interfacial energies and contact angles with their variations
with sintering conditions are still largely lacking. This necessitates a fair amount
of parameter fitting to study sintering of a given alloy once constitutive models are
obtained. To predict, within narrow tolerances, the outcome of sintering one will
probably still have to perform a number of experiments, but it is in the scope of
present work to limit the number of such experiments by giving a sense of directions.
The ability to simulate processes and to predict properties, and ultimately to
virtually design materials, give promises for a faster and more cost efficient materials
development. The desire for sintering models that can be used to simulate sintering
via computer calculations is founded in this context. With such simulations it
should be possible to explore effects of various compositional aspects and process
parameters under different sintering conditions.

Liquid phase sintering

During sintering the excess surface energy attributed to porosity is decreased by
means of matter transport that becomes kinetically possible at high temperature.
Thus, sintering deals with a central issue of materials science, i.e. how much is
the total energy of the system lowered and in what way. When penetrating the
process in order to understand or predict its course, one inevitably finds several
possible reaction steps, making sintering a quite complicated process. Not only is
the total energy reduced by pore elimination, but also by microstructural coarsening
and in many cases through phase transformations. Under such circumstances the
description of sintering is further complicated.
For complete densification by liquid phase sintering three things are usually
required, namely (i) an appreciable amount of liquid, (ii) solubility of the solid in
the liquid and (iii) complete wetting [1115]. Moreover, densification in a typical
liquid phase sintering system is normally divided in three stages: (i) rearrangement,
(ii) solution-reprecipitation and (iii) solid phase sintering based on the respective
dominant process.
When the liquid phase forms it wets the solid particles resulting in capillary
forces rearranging the particles into a denser packing. The powder compact responds to the capillary pressure as a viscous solid and viscosity increases with
decreasing porosity [5]. Rearrangement is retarded by friction between solid particles [16], but is usually so fast that all other events are overshadowed and full
density can be reached by rearrangement if around 35 volume percent liquid is
present [5].
In the second stage densification occurs with shape accomodation of particles
separated by liquid films through solution and reprecipitation of solids leading to
closer packing [16]. Material is dissolved at solid-liquid interfaces with high chemical potential and reprecipitated at other sites with lower chemical potential. A
number of solution-reprecipitation mechanisms have been identified, e.g. Ostwald
ripening, Ostwald ripening with shape accomodation, contact flattening, particle
disintegration, coalescence and pore elimination by liquid flow [17]. The solubility
of the solid in the liquid is inversely related to the particle size. The resulting
concentration gradient causes material flow from finer particles to coarser particles. Thus, the second stage is usually marked by microstructural coarsening [5].
One possible process is illustrated with the initial configuration of two particles
5

Liquid phase sintering

separated by a liquid film in Figure 1 and after some material from the particle
contact has dissolved in the liquid and precipitated elsewhere leading to a denser
configuration in Figure 2. In the case of equally sized spherical particles the stress
gradient driving dissolution in the contact zone stems from the capillary pressure
provided by the binder. Both figures are redrawn from Reference [11].

Liquid

Figure 1. Two particle held together by capillary pressure. After Reference [11].

x
r

Figure 2. Centre-to-centre approach due to solution-reprecipitaion. After Reference [11].

In the third stage the solid phase may sinter into a rigid skeleton. The densification is then described with typical solid state sintering models [16].

Surface energy relations

Surface energy is the driving force for the events that leads to a minimum-energy
configuration of the microstructure during liquid phase sintering [5].
Wetting is important for the sinterability of a powder system and characterised
by the contact angle, given by the solid-liquid-vapour equilibrium. This is demonstrated in Figure 3 drawn after Reference [5].

Surface energy relations

vapour
liquid

SL

LV

SV

solid

Figure 3. Interfacial energies and the definition of a contact angle. After Reference [5].

Interfacial energies and the wetting angle are related by means of Youngs equation [5]
SV = SL + LV cos ,

(1)

where the respective interfacial energies are solid-vapour, SV , solid-liquid, SL ,

and liquid-vapour, LV . The contact angle is denoted . Good wetting is characterised by a small contact angle and conversely poor wetting by a large contact
angle. Wetting is promoted by solubility of the solid in the liquid, formation of
intermediate compounds and interdiffusion [5]. If the sum of the liquid-vapour
and solid-liquid interfacial energies is lower than the solid-vapour interfacial energy
the net interfacial energy is reduced by spreading the liquid over the solid surface
as demonstrated in Figure 4 after Reference [5]. Chemical affinity and solubility
between the phases generally contribute to efficient spreading [5].
LV-SV+SL< 0
vapour

liquid
solid

Figure 4. Spreading of a liquid film over a solid surface driven by a reduction of

net interfacial energy. After Reference [5].

If a volume element dV is removed or added from a surface of principal radii r1

and r2 the energy change becomes [16]
dE
dA
=
=
dV
dV

1
1
+
r1
r2


,

(2)

Liquid phase sintering

where is the surface energy and dA the change in surface area. Identifying the
term dE/dV as a stress , the Laplace equation


1
1
=
+
(3)
r1
r2
is obtained. If the surface is convex the stress is tensile and if the surface is concave
the stress is compressive. It is this stress gradient that drives material transport
during sintering [16].

Sintering under applied stress

It is self evident that densification during sintering is enhanced by an applied pressure if it is sufficiently large. On a micro-level such a pressure enhances not only
rearrangement, but also solution at particle contacts and plastic flow in the solid
particles [18]. By applying an external pressure less sinterable powders can be sintered to high density and strength [16]. The applied pressure is often modelled
as additive to the capillary pressure [5, 18]. For an extensive review on sintering
modelling, including application of load, see Reference [19].

Cemented carbide sintering

In industrial practice cemented carbides are sintered by liquid phase sintering at
temperatures in the range 13501650 C leading to virtually pore-free compacts [2].
The foundation of the process in terms of thermodynamic and kinetic aspects
of the alloy systems has been studied extensively, as well as their interaction in the
developing microstructures, see for instance References [1, 2, 68] mentioned in the
introduction.
Freytag and Exner [20] studied sintering through changes in the magnetic properties of a WC-12 mass-% Co material. It was found that below 800 C practically
no shrinkage is observed but stresses introduced by milling are relieved and cobalt
oxide on surfaces are reduced. In the range 8001100 C shrinkage is limited, tungsten carbide partially dissolves in the binder that spreads over the carbide surfaces.
Further heating, but still below the temperature of liquid formation, results in a
large portion of the total shrinkage. It is stressed that enhancement of diffusion in
the cobalt-saturated carbide grain boundaries should have a decisive role in solid
state sintering. When the binder melts a substantial amount of carbide is dissolved
in the liquid. Rearrangement and pore filling eventually lead to full density.
Surface oxides, practically always present in a powder, are reduced in a wide
temperature range 650850 C with a maximum around 750 C with an additional
peak in CO/CO2 emission at 950 C for doped WC materials [21]. Freytag and
Exner [20] relate the main carbon loss during sintering to reaction of oxygen at the
carbide surfaces in the temperature range 7001000 C.
Dilatometry coupled with thermogravimetry, calorimetry and mass spectrometry has further shown that the shrinkage onset occurs after formation of CO and
CO2 from adhesive oxygen and oxides with carbon from WC, which leads to the
conclusion that shrinkage starts when all surface oxides are reduced [22]. This was
however recently questioned by de Macedo et al. [23] who studied wetting of Co, Ni
and Fe on W and WC. It was found that wetting occurs also on oxidised surfaces,
and it is concluded that the onset of shrinkage is not related to oxide reduction
although they appear in sequence. Gthelid et al. [24] studied Co spreading on WC
substrates with scanning tunneling microscopy (STM). It was found that at first
the substrate is partially covered by a so called 22 structure on top of which Co
then spreads rather easily. Clusters of cobalt oxide form in the presence of oxygen
rather than the 22 structure. It appears as spreading is impeded until oxygen is
9

10

Cemented carbide sintering

removed and the 22 structure restored. This was found to occur at a temperature
of 800 C, which is the temperature where shrinkage typically starts during WCCo sintering. Although sintering conditions differ substantially from the ultra-high
vacuum employed in STM studies this points to a correlation between an existing
oxide film and poor wetting conditions.

Thermodynamic aspects
The simplest cemented carbide and the base for most alloys is the mixture of WC
and Co. An isothermal section of the ternary phase diagram Co-W-C at 1350 C,
calculated with the Thermo-Calc software [25] and a database for cemented carbides [26], is shown in Figure 5. Compositions are given in atomic fractions. It is
apparent that the carbon content is important for the material constitution and for
only small deviations additional phases will form. For low carbon contents the M6 C
may become stable and for high carbon contents graphite. Neither one is usually
desired in the finished product. It is also demonstrated that the two-phase regions
{fcc-Co+WC} and {liq-Co+WC} narrow with decreasing Co content. Within the
narrow compositional range for the two-phase field the carbon activity increases
from 0.3 to 1 with respect to graphite.
graphite

1.0
0.8
0.6

xC

WC

0.4

W2C

0.2

M6C

liq-Co

M12C

fcc-Co

0.2

0.4

Co7W6

0.6

0.8

1.0

bcc-W

xW
Figure 5. Isothermal section of the ternary phase diagram Co-W-C at 1350 C.

The narrow range of carbon content permissible if only WC and Co are desired
in the sintered structure is further demonstrated in Figure 6. This is a calculated
vertical section through the Co-W-C system at 10 mass-% Co. Here the stable

Binder spreading and rearrangement

11

phases can be viewed as function of temperature. A stoichiometric mixture of WC10 mass-% Co corresponds to 5.52 mass-% C. The two-phase regions {fcc-Co+WC}
and {liquid Co+WC} are limited to the carbon range 5.385.54 mass-%. Since
some carbon inevitably reacts with oxygen during sintering it is in practice often
necessary to add carbon to the WC-Co mixture to obtain the desired composition
of the finished product [2].

1500

WC
liquid Co

temperature, C

1400

1300

M6C
WC
fcc-Co

1200

graphite
WC
fcc-Co

WC
fcc-Co

1100
5.2

5.4

5.6

5.8

carbon content, mass-%

Figure 6. Vertical section of the ternary phase diagram Co-W-C at 10 mass-% Co.

With the addition of other carbides such as TiC, NbC, TaC, VC and Cr2 C3 the
cobalt-rich liquid becomes stable at lower temperature [26].
It is believed that conditions close to equilibrium prevail at temperatures typically above 1000 C owing to the short diffusion distances involved in cemented
carbide sintering [27].
It should perhaps be stated that porosity present in the compact is not thermodynamically stable. Porosity contributes excess energy that is the driving force
for sintering. Indeed, all interfaces in the compact contribute to the total energy.
By reducing the interfacial area the total energy is lowered. This is the cause of
coarsening when small constituents shrink on behalf of larger ones that grow.

Binder spreading and rearrangement

Densification during cemented carbide sintering occurs initially through rearrangement followed by solution-reprecipitation [28,29]. With higher amount of liquid and
smaller carbide particle size most densification is achieved rapidly [28] and the high

12

Cemented carbide sintering

density obtained in the very first stage of sintering even before the liquid is formed
distinguishes the WC-Co system from other liquid phase sintering systems [29].
Froschauer and Fulrath [30] studied the microstructural evolution in situ and
confirmed that rearrangement is the main densification process. Haglund and coworkers [31] showed for a WC-10 mass-% Co material that densification in solid
state can be treated as a rearrangement process like in the first stage of liquid
phase sintering. No drastic change in densification rate occurred as the binder
melted.
During ball milling of the powder the carbide particles become partially covered
by cobalt and small carbide fragments are embedded in Co, which is important for
the initial sintering [32]. Milling and cold pressing also create highly deformed contact areas with high internal stresses that may deform by creep or other transport
mechanisms at elevated temperature relaxing the defect structure [21, 33]. Snowball and Milner [34] showed that co-milled WC-Co densifies to a higher extent than
WC-Co milled separately and mixed. Densification started locally in cobalt-rich
areas packed with carbides from milling.
At temperatures above one half of the melting temperature, 650710 C for fine
grained cemented carbides, the dispersed binder particles behave like a viscous mass
spreading by wetting onto the surrounding carbide surfaces [33,3537]. The spreading of the interfacial zones immediately after surface oxide reduction constitutes the
first stage of densification [21].
The surface free energy is lowered by the cobalt spreading [38] and exchanging
WC-gas and Co-gas interfaces for WC-Co lowers the interfacial energy to approximately one fifth [35]. Solid state spreading of cobalt is driven by the surface energy
relations in a similar way as desribed by the Youngs equation, Equation 1, although the contact angle between carbide and binder is more complex to define.
Very high spreading rates are reported for Co on WC substrates and it was observed
that it is only a film that spreads over the surface and not the bulk metal. The
mechanism behind binder spreading is believed to be viscous flow of a defect-rich
contact zone [23]. Meredith and Milner [36] suggested that the viscosity in the
carbide-metal interface is further reduced by carbide dissolution owing to enhanced
diffusion.
The rate of binder spreading is limited by factors such as area, curvature and
chemistry of the carbide particles and content, dispersion, chemistry and physical
properties of the binder. Binder spreading is connected to the Laplace forces acting
along the wetting front line between carbide and binder. These microscopic forces
rearrange the carbide particles and reduce the mean distance between neighboured
particles, resulting in densification [35].
The smaller the grain size is the more interfacial area is contained within the
compact and the higher is the stored interfacial energy. As a result sintering starts
earlier for fine-grained cemented carbides and the share of solid state shrinkage
increases [35]. Moreover, milling and mixing introduce more defects for finer WC
materials, which results in higher internal stress, shorter distance between source
and sinks of diffusion and thus lower viscosity [39]. The maximum in shrinkage rate

Solution-reprecipitation and the final stage

13

occurs at lower temperature and the shrinkage rate curve is wider but flatter [21,22].
Densification behaviour also becomes increasingly sensitive to processing steps prior
to sintering. For ultrafine cemented carbides the solid state sintering can contribute
with 90 percent of the total densification [35]. Earlier shrinkage is observed for
attritor-milled powders compared to balled-milled powders [37].
Grain growth inhibitors such as VC and Cr2 C3 in small amounts shift shrinkage
onset to higher temperature, decrease the solid state shrinkage rate and lower the
melting point of the binder [21, 35]. Addition of either carbon or tungsten to a
fine-grained WC-Co material decreases the shrinkage rate with stronger effect of
tungsten. The influence on a coarse-grained material is smaller [20].
Except for the ultrafine materials solid state shrinkage can kinetically be divided
in two stages. The first stage of densification, typically below 11001200 C, is
believed to be homogeneous shrinkage due to binder spreading and creep, while
in the second stage unsymmetical forces become strong enough to introduce twist,
rotation and shift among particles, although still governed by binder creep [21].
In the first stage intrinsic stresses of the binder contribute to the capillary force
in driving densification, while in the second stage the binder is annealed and its
intrinsic stresses neglible compared to the capillary stress. Moreover, in the first
stage the binder is nearly free to creep, while in the second stage creep is restricted
by the large number of carbide-binder interfaces [35]. Densification in the first stage
depends strongly on processing prior to sintering and only weakly on grain size,
while in the second stage a weak dependence on processing but strong dependence
on grain size are observed. The solid state shrinkage appears more complex for
ultrafine materials for which it is not possible to divide it in only two stages [35].

Solution-reprecipitation and the final stage

Densification in the intermediate stage of liquid phase sintering is accomplished
through grain shape accomodation and for cemented carbides with contributions
from contact flattening where carbide particles are in contact, dissolution of small
grains and reprecipitation on large grains and by coalescence via grain boundary
migration and solution-reprecipitation [29].
Microstructural coarsening leads to a prismatic grain shape. However, the prismatic shape is not necessarily compatible with the aggregated structure formed in
the initial stage. With a sufficient amount of binder this may lead to additional
rearrangement [36].
Practically full density is normally reached for most WC-Co materials by rearrangement and shape accomodation. Sintering of the solid phase skeleton, i.e.
the last stage of classical liquid phase sintering is usually not observed in cemented
carbide sintering. Other microstructural changes may however occur, for instance
grain size and grain size distribution, grain shape and binder phase distribution
may all change [29].

14

Cemented carbide sintering

Since full density is reached through continuous repacking of clusters leaving

cobalt free to flow further into the structure Snowball and Milner [34] argue that it
is not necessary to postulate a third stage of shrinkage, but with low cobalt content
a slow final densification pertaining to solid state sintering of the carbides may be
observed [38].
The actual existence of a continuous solid skeleton in WC-Co has been debated.
For instance, Warren and Waldron [40] suggested that there is a skeleton during
and after sintering, while Nelson and Milner [38] claimed that there is a cobalt film
between carbide particles for short sintering times. This would allow for particle
sliding during densification, but a rigid carbide network will eventually form during
prolonged sintering. Gurland and Norton [14] ruled out the possibilty of a carbide
network based on infiltration experiments. Andrn [41] refers to measurements
with atom probe field ion microscopy (APFIM) and analytical electron microscopy
(AEM) showing approximately half a monolayer Co is present between carbide
grains. Accordingly, there is a continous carbide skeleton, but half the contact
surface is covered by Co.

Microstructural evolution
Many key properties of cemented carbides are determined by the microstructure.
Carbide grain size and shape, binder fraction, composition and distribution in addition to retained porosity are a few of the more important microstructure characteristics that influence the properties such as transverse rupture strength and
hardness [6]. Primary interest here is, however, the microstructural evolution concurrent with densification.
Along with dimensional change, i.e. shrinkage and creep deformation, the microstructure changes too. If it was not for the microstructural change, compact
shape would not actually change since it is through the rearrangement of carbide
particles and grain shape accomodation that densification occurs.
The carbides are rearranged from a rather homogenous spatial distribution in
the green body to an aggregated structure during the initial stage of shrinkage, and
Meredith and Milner [36] attributed the rapid initial densification to the formation
of close-packed boundaries between carbides when cobalt diffuses into the carbide
interfaces and bringing the particles closer. In a second stage of densification the
binder bridges can rearrange the carbide particles [21]. During the intense shrinkage
in solid state WC-Co clusters that have formed upon cobalt spreading behave like
a viscous mass, and the clusters creeps into the surrounding void area. Smaller
Co particle size and better dispersion leads to more WC-Co clusters formed during
binder spreading, shorter diffusion distances and more uniform capillary forces. The
result is a more homogeneous void evolution that will influence the later stages of
densification [37].
The early rearrangement into locally dense areas occurs at the expense of creating micropores due to inhomogeneities in the green compact [35]. With binder

Microstructural evolution

15

flow into the surrounding matrix more and more carbide particles are cemented in
a growing agglomerate [34]. Later densification proceeds by creep of the WC-Co
material to fill the micropores [35]. Spreading of Co along carbide surfaces and rearrangement into an aggregated structure is shown in Figure 7 which was obtained
in SEM with secondary electron mode after heating a WC-Co material with 3 m
WC and 10 mass-% Co to 1280 C and cooling.

Figure 7. Micrograph obtained after heating a WC-Co material to 1280 C and

cooling. WC appears light grey, Co dark grey and pores black.

The carbide particles are typically rounded after milling the starting powders.
The wide extent of the solution-reprecipitation process during sintering is evidenced
by the prismatic grain shape typical for WC-Co materials after prolonged sintering [36]. The preference for prismatic growth stems from the WC crystal structure
and interfacial energy anisotropy [1, 42]. Prismatic carbide shapes develop faster
in fine-grained materials than coarse-grained materials [38] and the shape of the
facetted WC grain changes with the carbon content [4244]. The carbide facetting
is evident in Figure 8 which was obtained in SEM with secondary electron mode
after sintering the same material as in Figure 7 at 1430 C for 60 min. This sintering
cycle is similar to the one used in industrial practice.
The grain size of the sintered structure basically depends on the initial size
distribution and thus the milling conditions [45]. Large grains may act as starting
point for rapid discontinuous grain growth widening the grain size distribution.
One trend in more recent materials development is towards finer carbide grain
sizes, since this improves hardness and wear resistance [46]. This has naturally
led to studies devoted to sintering mechanisms in submicron and nanocrystalline

16

Cemented carbide sintering

Figure 8. Micrograph obtained after sintering a WC-Co material at 1430 C for

60 min. WC appears light grey, Co dark grey and pores black.

cemented carbide materials. Since this study is limited to mediumcoarse grain

sizes the reader is referred to for instance References [35, 47, 48].

Topics of present work

Constitutive modelling
Outlining the parameters: Paper I
gren and co-workers [49, 50] devised a model for the simulation of shape changes.
The model was based on rearrangement of the carbide particles under capillary
action of the binder. Rearrangement is driven, in the absence of an applied stress,
by the sintering stress inversely proportional to grain size and linearly propotional
to the pore fraction. Binder spreading was considered practically instantaneous
and diffusional creep of the binder rate limiting. A linear combination of surface
diffusion and volume diffusion was introduced. Since a strong influence of heating
rate on shrinkage rate had been observed [31] the effect of heating was taken into
account explicitly.
The first attempt to extend this model to describe shrinkage of materials with
different grain sizes failed, and the study reported in Paper I was initiated. A
slightly different approach, used for instance by Gillia et al. [51] for WC-Co, was
chosen in which the uniaxial viscosity and the viscous equivalent of Poissons ratio
are derived from sintering under applied axial load. The sintering stress can be
calculated from these two parameters and the shrinkage rate under free-sintering
conditions, i.e. without the application of load. The evaluation of the constitutive
parameters followed in much the approach by Bordia and Scherer [52, 53].
The constituive parameters were modelled as functions of temperature and relative density assuming that the material is Newtonian, i.e. the strain rate is linearly
related to stress. No direct relation to heating rate was assumed.
The calculated uniaxial viscosity showed three stages as demonstrated in Figure 9. Of these are the first two before liquid formation. The third stage is during
liquid phase sintering and marked by an initially decreasing viscosity. Note then
the drastic increase as the materials approaches full density during liquid phase
sintering and the relatively large difference towards the end of sintering for fine
(1.7 m) and coarse (5.4 m) carbide grain sizes.
Solid state sintering was modelled with diffusional viscosity similarly as Herring [54]. The apparent activation energies in the two stages differed and were
17

18

Topics of present work

roughly twice as high in the second stage. For liquid phase sintering a little different approach was chosen than usual. With the analogy of free volume in glass
transition and percolation theory [55, 56] a description of rearrangement facilitated
by the present porosity was suggested. Combining the deformation pertaining to
the dense material and the rearrangement an effective viscosity was obtained with
the desired properties of a fast increase with relative density, but still allowing a
finite deformation rate of the dense material. It should be pointed out that the
suggested model perhaps does not give the easiest numerical treatment. If the
viscosity of the dense material is neglected an exponential function describing the
proper increase with relative density should be suitable.
1400

10

1300

1200
temperature, C

1100

10

coarse
1000
10

uniaxial viscosity, MPa s

10
fine

900

800

500

1000

1500

2000

2500

time, s

Figure 9. Uniaxial viscosity calculated from sintering under axial load.

The experimentally determined viscous Poissons ratio was unfortunately burdened with large scatter. Two modelling approaches were tested. In addition to an
exponential relation, as used by Gillia et al. [51], Scherers model [57]
1
p
2

,
3 2

(4)

was used. This was derived for a particular pore arrangement for a different material, but captured the essence of derived data in Paper 2. Moreover, it has the
benefits of a rather simple relation to relative density as sole parameter and the
correct limits of p 0 for 0 and p 0.5 for 1. Although a satisfactory
representation is found with this model it is naturally a source of uncertainty in
the analysis.
The sintering stress is the thermodynamic force associated with sintering. It
can be viewed as the collective action of interfacial tension throughout the compact.
One common definition seemingly derives from the work of Gregg and Rhines [58]
who measured the tensile force that was necessary to counteract shrinkage. In this

Constitutive modelling

19

view the sintering stress is a pressure of equal magnitude as the hydrostatic tensile
stress that would stop shrinkage. The calculation of sintering stress in Paper I
showed a value slightly increasing with decreasing porosity, which contrasts the
assumptions in References [49, 50]. The sintering stress was described by a power
function of relative density.
In Paper I the influence of an applied stress on densification was studied as
well. It is a commonly held opinion that the hydrostatic component of an applied
stress enhances densification [5, 16]. Though the applied uniaxial load increases
axial shrinkage and decreases radial shrinkage the total densification is expected to
increase. The first series of experiments with axial stresses up to 0.7 MPa influenced
densification rate only marginally, and a second series was performed under much
higher stress of up to 6 MPa for which a marked effect on densification was found.
Under all sintering conditions a densification-rate maximum was found near a
relative density of 0.8 such as shown in Figure 10, although temperature differed.
1 N is believed to be close to free sintering. 15 N load resulted in a maximum stress
of 0.7 MPa and 150 N in 6 MPa at most. The carbide grain size was 1.7 m. Similar
diagrams obtained under free sintering for a variety of compositions together with
calculated solidus and liquidus temperatures indicate that the first local maximum
is related to solid state shrinkage and the second to liquid phase shrinkage.
1.2

x 10

-3

1N

densification rate

0.8

15 N
150 N

0.6

0.4

0.2

0
0.5

0.6

0.7
0.8
relative density

0.9

Figure 10. Densification rate versus relative density for three levels of uniaxial
load.

Composition and sintering shrinkage: Paper II

The general constitutive behaviour was outlined in Paper I, although some parameters scattered. Moreover, a model of more practical interest should account
explicitly for parameteters of the material composition. The aim of Paper II was
therefore twofold. Firstly, the suggested models were used to describe shrinkage

20

Topics of present work

data from dilatometry where experimental certainty normally is rather high. Secondly, results from sintering a number of different materials should point to the
general influence of respective parameter on shrinkage. Microstructures were examined at various stages of sintering before and after liquid formation using scanning
electron microscopy.
The microstructural evolution was found generally coherent with findings in for
instance References [31,36,59], with the main contribution to densification from rearrangement, although leading to clustering, and subsequent solution-reprecipitation.
Different carbide grain sizes were shown to still result in similar temperatures
of shrinkage onset. However, shrinkage rate during solid state sintering decreases
with increasing carbide grain size, and for a material with approximately 10 m
WC the main solid state densification was suppressed. The latter can be linked to
the microstructural evolution showing that although the binder spreading onto the
neighbouring carbide particles is fast the continuous linking of more carbides into an
agglomerate is impeded. Following the decrease in shrinkage rate during solid state
sintering the relative amount of shrinkage during liquid phase sintering increases
with carbide grain size. With a medium carbide grain size, typically 2 m, only a
fraction of the densification is due to liquid phase sintering. For coarse materials,
with carbide grain sizes around 5 m, densification before and after liquid formation
are practically equal. In extra-coarse materials, up to 10 m WC, only a fraction of
the densification is caused by solid state sintering. For the terminology of carbide
grain sizes, see for instance the discussion by Roebuck [60].
Cobalt content mainly influences the later stages of shrinkage from a relative
density of approximately 0.7 where shrinkage rate increases with cobalt content,
although microstructural evolution occured in similar ways.
It was shown that the constitutive models with adjustable parameters outlined
in Paper I can be combined to successfully describe free sintering shrinkage rate
in three stages as shown in Figure 11 for a WC-Co material with 2 m WC and
10 mass-% Co. At the transition from the first stage to the second stage the
modelled curves overlap and cannot be distinguished. It was also shown how these
models can be used to calculate sintering shrinkage by numerical integration for a
given sintering cycle.

A shrinkage model for binder content and grain size: Paper III
The constitutive relations outlined in Paper I were shown suitable for describing
shrinkage in Paper II. With phenomenological relations for the effects of cobalt
content and carbide grain size on the respective parameters uniaxial viscosity, viscous equivalent of Poissons ration and sintering stress the three stages of shrinkage
can be described with no less than 13 adjustable parameters. In Paper III these
13 parameters were reduced to only four by fitting model parameters to dilatometry
data for three cobalt contents and three carbide grain sizes. The remaining four
parameters are enough to reasonably well represent shrinkage for other materials.

Constitutive modelling

21

x 10

-4

free sintering strain rate, 1/s

-0.5

-1

-1.5

-2

-2.5
0

1000

2000

3000
time, s

4000

5000

Figure 11. Shrinkage rate () measured and (---) described by the shrinakage
model. After Paper II.

The obtained model of linear shrinkage with adjusted parameters reads

1
2
3

A2.1 d0.063 exp (Q1 /RT ) (1 2p )

RT


B2.1 d0.035 exp (Q2 /RT ) (1 2p ) 1 fWC
=
RT
1 + 3fWC


0.96
exp
(0.096/
(1 ))
4.1 0.69
= C fCo (1 2p ) 1 +
0.18
d0.45 fCo
=

(5)
(6)
(7)

for the respective stages. is the relative density, R the gas constant, T absolute
temperature, p the viscous equivalent of Poissons ratio described by Equation 4,
d carbide grain size, k the Boltzmann constant, Q2 apparent activation energy of
typically 250 kJ/mol, fWC volume fraction carbide and fCo volume fraction cobalt
in the dense material. The change from the first to the second stage occured
around 1100 C. In the solidus-liquidus interval the shrinkage rate is expected to be
a combination of the solid fraction of the material shrinking according to Equation 6
and the liquid fraction shrinking according to Equation 7.
Two adjustable parameters, A and Q1 refer to the first stage of shrinkage. The
apparent activation energy Q1 was found in the range 60140 kJ/mol. In this stage
of sintering Gille et al. [35] observed activation energies of densification ranging
from 50 kJ/mol for attritor-milled powders to 100 kJ/mol for ball-milled powders.
Apparent activation energies of 70 kJ/mol and 120 kJ/mol was found for uniaxial
viscosity in Paper I. All these are about the same order of magnitude. The enhanced

22

Topics of present work

diffusion of Co in the WC-Co interfaces discussed in the literature [20, 36] may be
responsible for the very low activation energies observed. In this stage of sintering
deformation is most likely governed by surface diffusion of Co over WC surfaces.
In the second stage all materials showed practically the same apparent activation
energy of 250 kJ/mol in Reference [35], which was also observed for most materials
in Papers I and only a little higher for uniaxial viscosity in Paper III. This value
is typical for volume diffusion of cobalt [61], and it is reasonable to assume that
deformation is governed by diffusional creep pertaining to volume diffusion of the
cobalt-rich binder. Missiaen and Roure [62] analysed WC-Co shrinkage with a
morphological description of shrinkage. They conclude that densification and binder
spreading are coupled and that volume diffusion in the cobalt phase is the governing
mechanism.
As discussed in Paper III an accurate model of densification would preferably
describe shrinkage for a material with only a few, if any, adjustable parameters. It
was also pointed out, as well as in Reference [35], that the early shrinkage is sensitive
to processing prior to sintering, and accordingly the first stage of shrinkage differs
between different materials, which then must be reflected by model parameters.
Moreover, Gille et al. [22] investigated a new ultrafine WC material where the
carbide had a relatively small lattice distortion compared to other ultrafine grades.
Given the lower solution pressure for the less distorted lattice the solution and
reprecipitation was slower resulting in a more stable sintering, less coarsening and
a more homogenous microstructure [22]. Accordingly, processing influences also the
later stages of densification and it seems difficult to realise one common description
of sintering shrinkage for WC-Co materials.
Though the suggested shrinkage models in Paper III use the mean grain size,
microstructural coarsening can be accounted for in simulations by solving the equations of shrinkage and grain growth simultaneously, should the latter be available.

Densification
Influence of chemical composition: Paper IV
Paper IV aimed at investigating the influence of chemical composition on shrinkage. Three WC-Co and and three WC-(Ti,W)C-Co materials with different carbon
contents were investigated.
Additions of the cubic carbide (Ti,W)C delays shrinkage to higher temperature
and high carbon content above the limit of graphite formation enhances shrinkage
for both WC-Co and WC-(Ti,W)C-Co materials. A low carbon content towards
the limit of M6 C formation delays shrinkage for a material with (Ti,W)C but not
for WC-Co.

Densification

23

The densification model of Paper III with four adjustable parameters described
shrinkage reasonably well for WC-Co materials with low and medium carbon content and for WC-(Ti,W)C-Co with high carbon content. The model failed to capture the enhanced densification in WC-Co with high carbon content and the solid
state shrinkage of WC-(Ti,W)C-Co with low and medium carbon contents. For
WC-(Ti,W)C-Co the three stages of shrinkage is reduced to two stages where the
first stage of lower activation energy in WC-Co is absent. The observed variation
of shrinkage rate with temperature points to a much higher apparent activation
energy than the 250 kJ/mol observed in Papers I and III and Reference [35].
Freytag and Exner [20] related the cobalt spreading and thereby increasing
specific surface in the initial stage of WC-Co sintering to the change in coercive
force occuring in the temperature range 8001100 C. Since continued heating does
not change the magnetic properties much [20], it may be concluded that cobalt
spreading has been completed at this stage. At this temperature densification rate
starts to increase considerably and here was also the transition from the first stage
of shrinkage to the second stage found in Paper III and Reference [35].
However, with a significant amount of WC exchanged for the mixed cubic carbide (Ti,W)C almost no shrinkage is yet observed. Although de Macedo et al. [23]
recently reported that spreading can occur also on an oxidised substrate this is
in opposition to the prevailing view that wetting is significantly poorer on oxides
than on carbides. It is plausible that relatively stable surface oxides present on
the carbide surfaces effectively prevent binder spreading. With an oxide stability
increasing with decreasing carbon content oxide reduction and onset of shrinkage
would then be delayed to higher temperature as observed.
This leads to the suggestion that the two stages of surface diffusion and volume
diffusion are sequential and not simultaneous. It was observed by de Macedo et
al. [23] that Co spreads on a WC as a very thin layer, and by Gthelid et al. [24] that
a pre-cursor layer of Co must form on top of the WC surface before spreading can
occur. It seems reasonable that surface diffusion in the first stage of shrinkage forms
a Co film over the carbide with spreading completed around 1100 C. This leads also
to a little rearrangement and densification. When temperature is high enough for
volume diffusion to dominate further rearrangement occurs through binder creep.
If, like it was suggested for WC-(Ti,W)C-Co, the surface diffusion is limited in the
presence of oxides also the volume diffusion is hindered since the necessary film
is lacking. Thus, shrinkage behaviour of the two less sinterable types of materials
studied, WC-Co with coarse carbide and WC-(Ti,W)C-Co, are different. In the
former case the second stage is suppressed as discussed above and in the latter case
the first stage.

Effect of particle size distribution: Paper V

This work was initiated to investigate how the width of a particle size distribution
affects shrinkage and if the mean carbide grain size is sufficient in models such as
those devised in Papers IIII.

24

Topics of present work

Shrinkage of materials with different unimodal and bimodal particle size distributions were compared. It was found that the mean grain size, indeed, is sufficient
to describe shrinkage for a bimodal distribution with initially 75 percent fine particles and 25 percent coarse by mass. With 50 percent each of fine and coarse
particles, however, shrinkage was initially dominated by the fine fraction and subsequently by the coarse.
The extent of solution-reprecipitation during liquid phase sintering was clearly
demonstrated. With bimodal particle size distribution an extreme coarsening was
observed during liquid phase sintering in addition to marked facetting. A unimodal
particle size distribution showed much less coarsening but still exclusively facetted
carbides.

Microstructural evolution
Rearrangement: Paper VI
This study was started in light of the observation by Haglund [10] that the average
pore size in WC-Co seems to shrink during heating but grow during isothermal
holding although macroscopic shrinkage is observed.
The experimental basis for the work was measurements of the pore size distribution for various stages of a heating cycle with and without imposed isothermal
holding. Sintering was performed below the temperature of liquid formation. The
pore size distribution, if plotted with number frequencies on a logarithmic scale
changed in a nearly self-similar way as indicated by Figure 12 where Gaussian distribution curves have been adjusted to measured data for the ease of comparison.
With isothermal holding time bimodal pore size distributions developed as shown in
Figure 13 for the holding times 60, 180 and 480 min. Isothermal holding after slow
heating resulted in a pore size distribution with less pronounced bimodal character
than after fast heating.
The formation of bimodal pore size distributions was viewed in light of particle
rearrangement into clusters where pores inside clusters shrink and pores between
clusters grow. The direct cause of clustering in liquid phase sintering is that the
inter-particle attraction force increases with decreasing particle distance, which enhances irregularities in particle packing [63]. The formation of clusters was verified
in simulations using a Discrete Element Method (DEM), where the capillary attraction between particles connected by a liquid bridge was modelled. The particle
attraction forces, contact forces and viscous drag forces were used to solve the equation of motion and simulate the microstructural evolution during rearrangement.
The absence of clustering during heating was interpreted as a result of different particle attraction attributed to the carbide solubility in the binder. With
increasing temperature the solubility increases monotonously, while at isothermal
conditions the equilibrium solute content is rather soon established and a more rigid

Microstructural evolution

25
5

x 10

1150oC

-2

Number frequency (mm )

1200oC
3
1250oC
2

0
-2

-1
0
logarithmic pore size

Figure 12. Pore size distributions measured after heating.

structure forms. Moreover, DEM simulations of isothermal holding following heating showed that the microstructural evolution retards considerably simply by the
realisation of constant viscosity. The DEM simulations failed however to account
for solubility effects leading to significant densification.
There has been a question about a possible influence of heating rate on shrinkage
rate. Haglund et al. [31] noted that shrinkage rate for WC-Co decreases drastically
as isothermal conditions are reached. This has also been observed by Schubert et
al. [37]. Roura et al. [64] investigated theoretically commonly used sintering models
and came to the conclusion that a decreasing shrinkage rate during isothermal
holding is inherent in standard theory of sintering. In Paper VI it was shown with
DEM simulations in Paper VI that the establishment of a constant viscosity results
in a retarding rate of rearrangment. These are certainly parts of the explanation
to slow densification during isothermal holding. However, in experiments on Co
spreading over WC surfaces de Macedo et al. [23] found very high spreading rates
and that spreading rate is related to heating rate. During fast heating the metal
spread on the carbide surface as much as in longer time [23]. Spreading was believed
to be caused by viscous flow of a defect-rich contact zone of the metal and carbide.
Defects are created in the surface zone as discussed by Schatt [33], but may also
annihilate. With high heating rate the formation of defects is faster than the
annihilation resulting in an increased defect concentration, lower viscosity of the
contact zone and faster spreading of the binder [23].
The same argument may hold also for shrinkage. Mixing, milling and pressing
of powders before sintering introduce a defect structure in the cobalt binder. According to Gille et al. [35] this is important for the early stage of sintering, although
Fischmeister and Exner [65] did not observe any drastic influence of milling, but related shrinkage behaviour solely to carbide grain size and surface activity. However,
if a correlation between milling, defect concentration and shrinkage exist, a lower

26

Topics of present work

x 10

-2

Number frequency (mm )

2
1250oC, 0 min
60 min
1

180 min
480 min

0
-2

-1
0
logarithmic pore size

Figure 13. Pore size distributions measured after heating and isothermal holding.

heating rate would allow for a longer annealing time during heating to the sintering
time, which in turn would result in slower sintering kinetics. With a higher heating
rate the material would reach the sintering temperature with higher defect concentration in the binder and higher sinterability. This would point to the existence of
a thermal-history effect.
In Figure 14 the simultaneous densification and microstructural evolution is
shown for a WC-10 mass-% Co material with 3 m carbide grain size. The micrographs were obtained after cooling from 1150 C, 1200 C, 1280 C and 1380 C,
respectively. WC appears light grey, Co dark grey and pores black. Some of the
features discussed above are demonstrated:
Shrinkage starts around 800 C,
rearrangement initially leads to an aggregated WC-Co structure,
solution-reprecipitation and facetting is limited during solid state sintering,
i.e. below 1310 C, but clearly visible after liquid phase sintering, and
formation of the liquid phase leads to further rearrangement and a more
homogeneous binder distribution.
Although high density is achieved in solid state for the more sinterable materials,
liquid phase sintering still brings changes to the microstructure. Cobalt is present
in the compacts as quite large entities all to the melting temperature. Accordingly,
it is only with liquid phase sintering one can expect a fairly homogenous cobalt
distribution.

Microstructural evolution

27

1500

1400
0.9

0.8

1200
1100

0.7

relative density

temperature, C

1300

1000
0.6
900
800

2000

4000

6000 8000
time, s

10000 12000

0.5

Figure 14. Densification and microstructural evolution.

M6 C formation and dissolution: Paper VII

The work reported in Paper VII started after finding M6 C in a WC-(Ti,W)C-Co
material with carbon content above the limit of M6 C stability after interrupting
the heating cycle and cooling. Heating to a higher temperature and then cooling
did not show any M6 C.
The explanation was found in the (Ti,W)C used for the material. (Ti,W)C has,
unlike WC, a compositional range for carbon. As manufactured the carbide has
a carbon content above the equilibrium content at the sintering temperature, but
this excess carbon and the free carbon seems kinetically trapped to the carbide in
the early stage of sintering. Accordingly, the material behaves as if it is below the
M6 C limit until carbon mobility is higher in the later stages of heating and the
M6 C can dissolve.
Since the M6 C carbide was dissolved only at 1350 C the sintering conditions constitute one major difference compared to WC-Co materials. Haglund and gren [27]
found that conditions close to thermodynamic equilibrium are established at temperatures above 1000 C, but this is evidently not true for WC-(Ti,W)C-Co materials.

28

Concluding remarks
This work has explored compositional effects of cemented carbide sintering. The
main processes of densification have been confirmed rearrangement under capillary
pressure provided by the binder and solution-reprecipitation. For more sinterable
materials, such as WC-Co with fine grain size, rearrangement in solid state provides
the most densification. Less sinterable materials, e.g. WC-Co with coarse carbides
or containing cubic carbides, show also a considerable densification during liquid
phase sintering.
A continuum-type of model has been devised to describe shape changes during
sintering based on the constitutive parameters uniaxial viscosity, viscous Poissons
ratio and sintering stress. Explicit relations to cobalt content and carbide grain
size have been derived.
The driving force for shrinkage is provided by the interfacial energy relations. It
is therefore troublesome that quantitative relations for these as function of sintering
conditions are unavailable. It was shown that high carbon content has a positive
effect on shrinkage rate. Solid state spreading and liquid wetting with their dependencies on carbon activity should therefore preferably be addressed in future
research.
The results obtained in this work supports the common view, see for instance
Reference [29], of rearrangement of carbide grains and grain shape accomodation as
the two sources of densification in cemented carbide sintering. Solid state sintering
at finite sintering times (up to a few hours) leads in practice only to rearrangement.
This is evident from microstructures obtained after sintering just below the solidus
temperature. Such micrographs display only limited carbide facetting, which is an
indication of little solution-reprecipitation. On the other hand, heating above the
liquidus temperature leads rather quickly to an extensive facetting.
It seems to exist a limit of solid state rearrangement at a relative density
slightly above 0.8. Finer grained materials show a local densification rate maximium around this value, which is typically reached some ten degrees below the
binder melting temperature. A decreasing densification rate is then measured for
a monotonously increasing temperature. It is reasonable to assume that this is
caused by the two partly overlapping densification mechanisms, rearrangement and
solution-reprecipitation, with different kinetic characteristics. The possibility that
the event is related to liquid formation can be ruled out since the maximum occurs
29

30

Concluding remarks

at different temperatures for different applied loads. Since solution-reprecipitation

is rather slow in solid state the densification rate decreases until the liquid forms and
solution-reprecipitation can occur significantly faster. In addition, the rearrangement process is facilitated by a more fluid binder and carbide contacts are dissolved
due to the higher solubility. Coarse-grained materials show no minimum in densification rate during continuous heating. These materials reach the solidus-liquidus
interval with a relative density below 0.8 and rearrangement occurs simultaneously
as shape accomodation.
It is known that powder metallurgy processing, cemented carbide sintering included, carries a certain degree of sensitivity towards changed conditions. One
example is how a barely detectable amount of impurity can bring the outcome of
sintering far from the expected [31]. Thus, truly predictive models and simulations
seem difficult to realise. Nevertheless, it is the authors hope that present work, or
at least parts therefrom, will prove useful for the powder metallurgy community.
One suggestion for future research has already been mentioned, i.e. the effect
of sintering conditions on the wetting and spreading of Co on WC. Regarding densification mechanisms there are still open questions regarding the more complex
sintering behaviour of materials based on ultrafine and nanocrystalline WC. Another interesting field to explore further would be the carbide grain growth that
perhaps does not have a decisive role in densification, but is increasingly important
for finer-grained materials and not the least important for the mechanical properties.

Acknowledgements
This work has been performed within the Brinell Centre Inorganic Interfacial Engineering (BRIIE), supported by the Swedish Agency for Innovation Systems (VINNOVA) and the following industrial partners: Erasteel Kloster AB, Hgans AB,
AB Sandvik Coromant, Seco Tools AB and Atlas Copco Secoroc AB. The author
is grateful for their support enabling this work.
The author also wishes to extend his sincere appreciation to a number of people
who contributed in one way or another. This applies first and foremost to Professor
John gren for his professional guideance, co-operation, encouragement and trust
in the direction of the work.
Within the BRIIE sintering group the author is particularly grateful to Bo
Jansson and Jan Qvick at Secotools AB for invaluable feedback, powder preparation and co-authorship. Anders Olsson at Atlas Copco Secoroc AB ably provided
powders. Bjrn Uhrenius formerly at AB Sandvik Hard Materials contributed at
various stages of the work. Jenni Zackrisson at Secotools AB is thanked for help in
powder manufacturing and co-authoring one paper. Sven Haglund at the Swedish
Institute for Metals Research is acknowledged for support during the initial stage.
Contributions by Per Lindskog and Lars Tholander at AB Sandvik Coromant made
the experiments with sintering under load possible. Stefan Wikman at the Swedish
Institute for Metals Research is thanked for help with the dilatometer and Kjell
Jansson at Stockholm University for performing the last series of dilatometry.
Some present and former colleges deserve to be mentioned. Joakim Odqvist is
acknowledged in particular for coping with numerous interruptions for questions
and discussions. The author is thankful to Lu Xiaogang for advices regarding
thermodynamic calculations. Martin Schwind contributed in numerous ways, for
instance with valuable pointers in scanning electron microscopy. Henrik Strandlund is acknowledged particularly for discussions relating to numerical methods
and Henrik Larsson for general discussions. Lars Hglund merits a special thanks
for help with computers and software. The technical and administrative staff at
the Department of Materials Science and Engineering deserve recognition for their
attempts to make practical matters run smoothely.
My parents, brother, sisters and friends, thank you for being there. Finally, my
wife and daughters are true sources of inspiration. Gali, Eden and Yuvalthis is
to you with love.
31

32

Bibliography
[1] H.E. Exner, Physical and chemical nature of cemented carbides, Inter Mater
Rev 4(1979), 149173
[2] K.J.A. Brookes, World Directory and Handbook of Hardmetals, Fourth edition,
International Carbide Data, East Barnet, 1987
[3] G.C. Kuczynski, Self-diffusion in sintering of metallic particles, Trans Am
Inst Min Met Eng 85(1949), 169178
[4] R.L. Coble, Sintering crystalline solids. I. Intermediate and final state diffusion
models, J Appl Phys 32(1961):5, 787792
[5] R.M. German, Liquid Phase Sintering, Plenum Press, New York, 1985
[6] H.E. Exner, and J. Gurland, A review of parameters influencing some mechanical properties of tungsten carbide-cobalt alloys, Powder Metall, 13(1970):25,
1331
[7] H.E. Exner, Qualitative and quantitative interpretation of microstructures in
cemented carbides, in: R.K. Viswanadham, D.J. Rowcliffe and J. Gurland
(eds.), Science of Hard Materials, Plenum Press, New York, 1983, 233262
[8] C.H. Allibert, Sintering features of cemented carbides WC-Co processed from
fine powders, Int J Refract Met Hard Mat 19(2001), 5361
[9] J. Brandt, On constitutive modelling of the compaction and sintering of cemented carbides, Dissertation, Linkping University, Linkping, 1998
[10] S. Haglund, Sintering of cemented carbides experiments and modeling,
Dissertation, The Royal Institute of Technology, Stockholm, 1998
[11] W.D. Kingery, Densification during sintering in the presence of a liquid phase.
I. Theory, J Appl Phys 30(1959), 301306
[12] G.H.S. Price, C.J. Smithells and S.V. Williams, Sintered alloys. Part I
Copper-nickel-tungsten alloys sintered with a liquid phase present, J Inst Metals 62(1938), 239264
33

34

Bibliography

[13] F.V. Lenel, Sintering in the presence of a liquid phase, Trans Am Inst Min
Met Eng 175(1948), 878896
[14] J. Gurland and J.T. Norton, Role of the binder phase in cemented tungsten
carbide-cobalt alloys, Trans Am Inst Min Met Eng 194(1952), 10511056
[15] H.S. Cannon and F.V. Lenel, Some observations on the mechanism of liquid phase sintering, In: F. Benesovsky (ed.), Pulvermetallurgie: Vortrge
gehalten auf dem 1. Plansee Seminar De Re Metallic 22. bis 26. Juni 1952,
Reutte/Tirol, Metallwerk Plansee, Reutte, 1953, 106121
[16] H.E. Exner and E. Artz, Sintering processes, in: R.W. Cahn and P. Haasan
(eds.), Physical Metallurgy, Third edition, Elsevier Science Publishers, Amsterdam, 1983, 18851912
[17] G. Petzow and W.A. Kaysser, Basic mechanisms of liquid phase sintering,
In: G.S. Upadhyaya (ed.), Sintered metalceramic composites, Elsevier Science
Publishers, Amsterdam, 1984, 5170
[18] W.D. Kingery, J.M. Woulbroun and F.R. Charvat, Effects of applied pressure
on densification during sintering in the presence of liquid phase, J Am Ceram
Soc 46(1963):8, 391395
[19] E.A. Olevsky, Theory of sintering: from discrete to continuum, Mater Sci
Eng R23(1998), 41100
[20] J. Freytag and H.E. Exner, The influence of tungsten and carbon additions on
the sintering and the magnetic properties of WC-12Co cemented carbides, In:
H.H. Hausner and P.V. Taubenblat (eds.), Modern Developments in Powder
Metallurgy, vol. 10 Ferrous and Non-Ferrous P/M Materials, MPIF-APMI,
Princeton, 1977, 511523
[21] G. Gille, B. Szesny and G. Leitner, A new 0.4 m WC powder as well as
powder-related properties and sintering behaviour of 0.6 to 30 m WC-Co
hardmetals, J Adv Mater 31(1999):2, 922
[22] G. Gille, J. Bredthauer, B. Gries, B. Mende and W. Heinrich, Advanced and
new grades of WC and binder powder their properties and application, Int
J Refract Met Hard Mat 18(2000), 87102
[23] H.R. de Macedo, A.G.P. da Silva and D.M.A. de Melo, The spreading of
cobalt, nickel and iron on tungsten carbide and the first stage of hard metal
sintering, Mater Lett 57(2003), 39243932
[24] M. Gthelid, S. Haglund and J. gren, Influence of O and Co on the early
stages of sintering of WC-Co: A surface study by AES and STM Acta Mater
48(2000), 43574362

Bibliography

35

[25] B. Sundman, B. Jansson and J.-O. Andersson, The Thermo-Calc Databank

System, Calphad 9(1985), 153190
[26] K. Frisk, L. Dumitrescu, M. Ekroth, B. Jansson, O. Kruse and B. Sundman,
Development of a database for cemented carbides: Thermodynamic modeling
and experiments, J Phase Equilib 22(2001), 645655
[27] S. Haglund and J. gren, W content in Co binder during sintering of WC-Co,
Acta Mater 46(1998), 28012807
[28] W.D. Kingery, E. Niki and M.D. Narasimhan, Sintering of oxide and carbide
metal compositions in presence of a liquid phase, J Am Ceram Soc 44(1961):1,
2935
[29] G.S. Upadhyaya, Materials science of cemented carbides an overview, Mat
Des 22(2001), 483489
[30] L. Froschauer and R.M. Fulrath, Direct observation of liquid-phase sintering
in the system tungsten carbide-cobalt, J Mater Sci 11(1976), 142149
[31] S. Haglund, J. gren and B. Uhrenius, Solid state sintering of cemented carbides an experimental study, Z Metallkd 89(1998), 316322
[32] B. Lehtinen and K.E. Easterling, The effect of mixed milling on the sintering
of WC-Co hardmetals, Planseeber Pulvermet 20(1972), 141151
[33] W. Schatt, Sintervorgnge, VDI-Verlag, Dsseldorf, 1992
[34] R.F. Snowball and D.R. Milner, Densification processes in the tungsten
carbide-cobalt system, Powder Metall 11(1968):21, 2340
[35] G. Gille, B. Szesny, G. Dreyer, H. van den Berg, J. Schmidt, T. Gestrich and
G. Leitner, Submicron and ultrafine grained hardmetals for microdrills and
metal cutting inserts, Int J Refract Met Hard Mat 20(2002), 322
[36] B. Meredith and D.R. Milner, Densification mechanisms in the tungsten
carbide-cobalt system, Powder Metall 19(1976):1, 3845
[37] W.D. Schubert, A. Bock and B. Lux, Aspects of ultrafine hardmetal sintering,
In: C.L. Rose and M.H. Thibodeau (eds.), Advances in Powder Metallurgy and
Particulate Materials, vol. 3, Princeton University Press, Princeton, 1999, 102310-36
[38] R.J. Nelson and D.R. Milner, Densification processes in the tungsten carbidecobalt system, Powder Metall 15(1972):30, 346363

36

Bibliography

[39] G. Gille, G. Leitner and B. Roebuck, Sintering behaviour and properties of

WC-Co hardmetals in relation to the WC powder properties, In: Advances
in Hard Materials Production, Proceedings of the 1996 European Conference
on Advances in Hard Materials Production, Stockholm, Sweden, May 2729,
1996, EPMA, Shrewsbury, 1996, 195210
[40] R. Warren and M.B. Milner, Microstructural development during the liquidphase sintering of cemented carbides I. Wettability and grain contact, Powder
Metall 15(1972):30, 166201
[41] H.-O. Andrn, Microstructures of cemented carbides, Mat Des 22, 491498
[42] Y. Wang, M. Heusch, S. Lay and C.H. Allibert, Microstructure evolution in
the cemented carbides WC-Co I. Effect of the C/W ratio on the morphology
and defects of the WC grains, Phys Status Solidi A 193(2002), 271283
[43] S. Kim, S.-H. Han, J.-K. Park and H.-E. Kim, Variation of WC grain shape
with carbon content in the WC-Co alloys during liquid-phase sintering, Scripta
Mater 48(2003), 635639
[44] S. Lay, Characterisation from m to nm scale for interface analysis, textitMaterials Science Forum 426432(2003), 42034208
[45] H.E. Exner, A. Walter, P. Walter and G. Petzow, Auswirkung
der Wolframkarbid-Ausgangspulver auf gesinterte Wolframkarbid-KobaltHartmetalle, Metallwissenschaft und Technik 32(1978):5, 443448
[46] K. Brookes, Round tools rule in hardmetal powders, 57 (2002):10, 2026
[47] S. Berger, R. Porat and R. Rosen, Nanocrystalline materials: A study of
WC-based hard metals, Prog Mater Sci 42(1997), 311-320
[48] D.F. Carroll, Sintering and microstructural development in WC/Co-based
alloys made with superfine WC powder, Int J Refract Met Hard Mat 99(1999),
123-132
[49] J. gren, J. Brandt, S. Haglund and B. Uhrenius, Modeling of solid state sintering of cemented carbides, In: R.M. German, G.L. Messing and R.G. Cornwall (eds.), Sintering Technology, Marcel Decker, New York, 1996, 141156
[50] J. gren, Modelling of sintering of cemented carbides, In: Proceedings of
the 1998 Powder Metallurgy World Congress & Exhibition, Special Interest
Seminar Fundamentals of Sintering, Granada, Spain, October 1822, 1998,
EPMA, Shrewsbury, 1998
[51] O. Gillia, C. Josserond and D. Bouvard, Viscosity of WC-Co compacts during
sintering, Acta Mater 49(2001), 14131420

Bibliography

37

[52] R.K. Bordia and G.W. Scherer, On constrained sinteringI. Constitutive

model for a sintering body, Acta Metall 36(1988):9, 23932397
[53] R.K. Bordia and G.W. Scherer, On constrained sinteringII. Comparison of
constitutive models, Acta Metall 36(1988):9, 23992409
[54] C. Herring, Diffusional viscosity of a polycrystalline solid, J Appl Phys
21(1950), 437445
[55] M.H. Cohen and D. Turnbull, Molecular transport in liquids and glasses, J
Chem Phys 11(1959), 11641169
[56] G.A. Campbell and G. Forgacs, Viscosity of concentrated suspensions: An
approach based on percolation theory, Phys Rev A41(1990):8, 45704573
[57] G.W. Scherer, Sintering inhomogeneous glasses: Application to optical waveguides, J Non-Cryst Solids, 34(1979), 239256
[58] R.A. Gregg and F.N. Rhines, Surface tension and the sintering force in copper, Met Trans 4(1973), 13651374
[59] A.G.P. da Silva, W.D. Schubert and B. Lux, The role of the binder phase in
the WC-Co sintering, Materials Research 4(2001):2, 5962
[60] B. Roebuck, Terminology, testing, properties, imaging and models for fine
grained hardmetals, Int J Refract Met Hard Mat 13(1995), 265279
[61] K.-I. Hirano, R.P. Agarwala, B.L. Averbach and M. Cohen, Diffusion in cobaltnickel alloys, J Appl Phys 33(1962):10, 30493054
[62] J.-M. Missiaen and S. Roure, A general morphological approach of sintering kinetics: Application to WC-Co solid phase sintering, Acta Mater 46(1998):11,
39853993
[63] J.-M. Chaix, Liquid phase sintering: the rearrangement stage results in clusters and large pores CR Acad Sci II 315, 17251730
[64] P. Roura, J. Costa and J. Farjas, Is sintering enhanced under non-isothermal
conditions?, Mater Sci Eng A337(2002), 248253
[65] H.F. Fischmeister and H.E. Exner, Beobachtungen ber den Mahlvorgang bei
Hartmetallpulvern, Planseeber Pulvermet 13(1965), 178198