Sintering:

Fundamentals
INTRODUCTION

The heat treatment process in which a powder or porous material, already formed
into a required shape, is converted to a useful solid is referred to as sintering.

Sintering processes are commonly divided into several categories, depending on the
type of system. One category is that for a pure, single-phase, polycrystalline
material, such as pure Aluminum. Sintering is achieved by heating the consolidated
mass of particles, referred to as the green body or powder compact, to a temperature
that is in the range between approximately 60% and 80% of the melting temperature.
For Al with a melting temperature of 660°C, the sintering temperature is commonly
between 400°C and 550°C. The powder does not melt; instead, the joining together
of the particles and the reduction in the porosity (i.e., densification) of the body, as
required in the fabrication process, occur by atomic diffusion in the solid state.
This type of sintering is referred to as solid-state sintering. For many
polycrystalline ceramics, the required density or microstructure of the final article is
difficult to achieve by solid-state sintering, or the solid-state sintering temperature
is too high for the intended fabrication process. One solution is the use of an additive
that forms a small amount of liquid phase between the particles or grains at the
sintering temperature. This method is referred to as liquid-phase sintering. The
liquid phase provides a high-diffusivity path for transport of matter into the pores to
produce densification, but it is insufficient, by itself, to fill up the porosity.

Another solution to the difficulty of inadequate densification is the application of an

external pressure to the body during heating in either solid-state or liquid-phase
sintering. This method is referred to as pressure-assisted sintering or pressure
sintering; hot pressing and hot isostatic pressing are well-known examples.

1
Viscous sintering refers to the sintering process in amorphous materials, such
as glasses, in which matter transport occurs predominantly by viscous flow.

The heating schedule in the sintering of industrial ceramics can have several

temperature–time stages, but in laboratory-scale studies, a simple schedule involving

isothermal sintering or constant heating rate sintering is often used. In isothermal

sintering, the temperature is increased rapidly to a fixed (isothermal) sintering
temperature, maintained at this temperature for the required time, and finally
lowered to room temperature.

DRIVING FORCE FOR SINTERING

As with all other irreversible processes, sintering is accompanied by a lowering of

the free energy of the system. The sources that give rise to this lowering of the free
energy are commonly referred to as the driving forces for sintering. Three possible
driving forces are: the curvature of the particle surfaces, an externally applied
pressure, and a chemical reaction. Schematic diagram of the mentioned three main
driving forces for sintering is illustrated in Fig. 1.

2
 FIGURE 1, Schematic diagram illustrating three main driving forces for sintering: surface free
energy, applied pressure, and chemical reaction [1].

The macroscopic driving force operative during sintering is the reduction of the
excess energy associated with surfaces. This can happen by (1) reduction of the total
surface area by an increase in the average size of the particles, which leads to
coarsening (Fig. 2), and/or (2) the elimination of solid/vapor interfaces and the
creation of grain boundary area, followed by grain growth, which leads to
densification (Fig. 2).

3
The total interfacial energy of a powder compact is expressed as A, where is the
specific surface (interface) energy and A the total surface (interface) area of the
compact. The reduction of the total energy can be expressed as

Here, the change in interfacial energy () is due to densification and the change in
interfacial area is due to grain coarsening. For solid state sintering,  is related to
the replacement of solid/vapour interfaces (surface) by solid/solid interfaces. As
schematically shown in Figure 2, the reduction in total interfacial energy occurs via
densification and grain growth, the basic phenomena of sintering.

Figure 2. Basic phenomena occurring during sintering under the

driving force for sintering, (A) [2].

4
Let us consider a powder compact in which the particles, assumed to be spherical,
are in point contact (Figure 3a). The upper limit of the free-energy reduction during
sintering can be expressed as

max = - As sv

where As is the total free surface area of the powder. When the powder compact
densifies, the free surfaces are replaced by grain boundaries, as shown in Figure 3b.
If full density is achieved, then the change in free energy associated with the
densification is

Ed = - (svAsv -gbAgb)

where Agb is the total grain boundary area of the dense solid. During densification,
approximately two free surfaces coalesce into one grain boundary, as shown
schematically in Figure 3, so Agb ≈ As/2. The total change of free energy resulting
from densification is, therefore,

(1)

The free energy reduction given by Equation 1 drives the densification process.
The remaining part of the free energy reduction, given by

can be considered to drive the grain growth process. For most materials, γsv is
larger than γgb, so ΔEd is normally negative, and there is always a
thermodynamic driving force for elimination of the pores [1].

One of the most important consequences of the presence of the grain boundaries is
that grain growth provides an alternative process by which the powder system can

5
decrease its free energy. In sintering, grain growth is normally accompanied by pore
growth, a process described as coarsening. Coarsening, therefore, occurs
concurrently with sintering [1].

FIGURE 3, Schematic diagram illustrating the replacement of free surfaces (a) by grain
boundaries (b) during the densification of polycrystalline systems [1].

These two mechanisms are usually in competition. If the atomic processes that lead
to densification dominate, the pores get smaller and disappear with time and the
compact shrinks. But if the atomic processes that lead to coarsening are faster, both
the pores and grains get larger with time. A necessary condition for
densification to occur is that the grain boundary energy gb be less than twice
the solid/vapor surface energy SV. This implies that the equilibrium dihedral angle
 shown in Fig. 4a and defined as

(2)

6
has to be less than 180°. For many oxide systems, the dihedral angle is around 120°,
implying that gb/sv ~ 1.0, in contrast to metallic systems where that ratio is closer
to between 0.25 and 0.5.

Figure 4 (a) Equilibrium dihedral angle between grain boundary and solid/vapor
interfaces, (b) Equilibrium dihedral angle between grain boundary and liquid phase [3].

WORKED EXAMPLE 1

(a) Calculate the enthalpy change for an oxide as the average particle diameter
increases from 0.5 to 10 m. Assume the molar volume of the oxide to be 10
cm3/mol and a surface energy of 1 J/m2. (b) Recalculate the enthalpy change
if, instead of coarsening, the 0.5 m spheres are sintered together as cubes,
given that the dihedral angle for this system was measured to be 100°.

Answer

(a) Take 1 mol or 10 cm3 to be the basis and assume monosize spheres of 0.5 m
diameter. Their number N is given by

7
Their corresponding surface area S is

S = (4πr2)(1.5 x 1014) = 120 m2

The total surface energy of the 0.5 m spheres:

0.5 = S x N = (120 m2 )(l J/m2) = 120 J.

Similarly, it can be shown that the total surface energy of the 10 m spheres is ~ 6
J. The change in enthalpy associated with the coarsening is

6–120 = -114 J/mol

In other words, the process is exothermic.

(b) Given that the dihedral angle is 100 o . applying Eq. (2). one obtains

gb = 2SVcos  = 2 x 1 x cos 50 = 1.28 J/m2

Mass conservation requires that a3 = 4/3πr3, or a ~ 0.4 m, where a is the length of
the side of the cubes. The total grain boundary area (neglecting free surface) is

Sgb ~ 6/2(0.4 x 10- 6)2(1.5 x 1014) = 72 m2

Thus the energy of the system after sintering is 1.28 x 72 ~ 92.16 J. which is less
than the original of 120 J. The difference between the two is the driving force for
densification.

In general, the size of powders for sintering is in the range between 0.1 and 100 m;
the total surface energy of the powder is 500–0.5 J/mole. This energy is
inconsiderably small, compared with the energy change in oxide formation which is
usually in the range between 300 and 1500 kJ/mole. If the desired microstructure
8
of the sintered body is to be achieved by the use of such a very small amount of
energy, it is necessary to understand and control the variables involved in the
sintering processes.

CHARACTERIZATION OF SINTERING

While several measurements can be used to characterize the sintering and

microstructural evolution of a compacted mass of particles, by far the most widely
used measurements are the density (or shrinkage) and the grain size. The progress
of sintering is often determined from the density or the linear shrinkage of the
powder compact as a function of time or temperature during the heat
treatment. The bulk density is defined as the mass divided by the external volume
of the body. A better parameter is the relative density ρ, defined as the bulk density
divided by the theoretical density of the solid. Relative density and porosity  of a
body are related by ρ = 1 − 

The linear shrinkage is defined as ΔL/Lo, where Lo is the original length, L is the

length at a given time or temperature, and ΔL = L – Lo. If the shrinkage is isotropic,

the relationship between the relative density and the shrinkage is

where ρo is the initial relative density and ΔL/Lo is negative as defined. (please prove
this equation).

Measurements of ρ or ΔL/Lo are easy to perform and provide substantial information

about the rate of sintering. It is often required to determine the densification rate,

9
defined as (1/ρ)(dρ/dt), where t is time, which is equivalent to a volumetric strain
rate.

In Figure 5, a component is shown that shrinks during sintering process. In this

photograph, the upper component is shown after sintering and although the same
shape, it is significantly smaller due to shrinkage.

Figure 5, An example of sintering shrinkage in a stainless steel component, where the lower
piece is prior to sintering (50% dense) and the upper pieces is after sintering (100% dense) [5].

If a powder compact shrinks, its density will increase with time. Hence, densification
is best followed by measuring the density of the compact (almost always reported as
a percentage of the theoretical density) as a function of sintering time.

Density is best measured by direct determination of the mass and volume. For simple
geometries, the volume is calculated from the dimensions. However, if the shape is
complex, then the immersion density is obtained by using the Archimedes technique.

Before immersion, the open pores are filled with a fluid, such as water, mineral oil,
silicone oil, or paraffin oil under vacuum. The Archimedes technique requires a
10
series of weight determinations. In the test, the sample is first weighed dry (W 1),
then weighed after fluid impregnation (W2), and finally weighed while immersed in
water (W3). A wire is used to suspend the sample in the water, and its weight W w is
also measured in water. The density ρ is calculated from the weight determinations
as:

where ρw is the density of water (which is slightly temperature dependent), for

example:

with T being the water temperature in K.

Densification coefficient Ψ indicates the relative change in density normalized to the

density change needed to reach full density:

Where f is fractional density and defined as f=ρ/ρth, where ρth is the theoretical
density, and fO is the initial density. At the start of sintering the densification
coefficient is zero. If the material is sintered to full density (f=1), then densification
is unity. Like an exhaustion parameter, densification measures how much of the
potential change has been achieved. Figure 6 is an example of sintering densification
for a mixture of tungsten and copper sintered at 1150 and 1250 oC in hydrogen for
120 min [5].

11
Figure 6 Densification versus sintering time at two temperatures for the liquid phase sintering of
W-32Cu compacts formed from 3.5 μm tungsten powder starting with 58% green density [5].

Change in density also can be followed by measuring linear shrinkage. This is

usually carried out dilatometrically, where the length of a powder compact is
measured as a function of time at a given temperature. Figure 6 diagrams a vertical
arrangement of a recording dilatometer.

12
 Figure 6 A recording dilatometer (dilation meter) for sintering studies [5].

Typical shrinkage curves are shown in Fig. 7 for two different temperatures T2 >
T1. For reasons that will become clear shortly, the densification rate is a strong
function of temperature, as shown in the figure.

Figure 7, Typical axial shrinkage curves during sintering as a function of temperature,

where T2 > T1 [3].

13
CONDUCTIVITY

Thermal and electrical properties are important for applications in computers,

electronics, and electric power. Pores are detrimental to conductivity.

Conductivity decreases as the porosity increases. Early in sintering, during

neck growth prior to significant densification, the conductivity K varies with
the square of the neck size ratio X/D (shown in Fig. 8) as follows:

where KO is the bulk conductivity at the measurement temperature. The parameter g

depends on the grain size and initial packing density, since grain boundaries and
pores both lower conductivity.

Figure 8, The interparticle neck size X is a common feature used to monitor sintering, shown
here for a two sphere contact. The grain or particle size is denoted as D, so the neck size ratio is
X/D.

As sintering progresses, the necks enlarge to a limit near X/D = 0.5. Further
conductivity arises with densification, giving

(3)

14
where is an empirical coefficient expresses the pore sensitivity and f is fractional
density. Analysis of several materials, representing a variety of pore sizes and
shapes, gives a best fit value of 11 for χ. As full density is attained the pores are
spherical and the conductivity depends on fractional density:

where ω is between 1 and 2.

Thus, prior to significant densification the conductivity increases with neck

size, and as densification progresses it depends on density, and as full density is
approached conductivity is a linear function of density.

The behavior of sintered copper, alumina, and stainless steel based on relative
conductivity (K/KO) is illustrated in Figure 9.

Figure 9 Relative conductivity versus fractional sintered density for several materials, showing
the nearly linear behavior at high densities [5].

15
The solid line is Equation 3. Strength evolution during sintering is similar, so
conductivity is also means for assessing strength [5].

Microstructural parameters

The average grain size and its standard deviation are often the microstructural
parameters of most interest, but a distribution of the grain sizes can also be shown,
if necessary, as a histogram or a continuous function. The mean grain intercept
length determined from two-dimensional sections is conventionally taken as a
measure of the average grain size. Measurement of the average pore size and the
pore size distribution of the partially sintered samples using mercury porosimetry
provides additional information about the characteristics of the open pores and the
homogeneity of the microstructure.

If a powder compact coarsens, no shrinkage is expected in a dilatometric experiment.

In that case, the coarsening kinetics are best followed by measuring the average
particle size as a function of time via optical or scanning electron microscopy. It is
useful to plot the resultant behavior in what is known as grain size versus density
trajectories, such as those shown in Fig. 10a and b. Typically, a material will follow
the path denoted by curve y, where both densification and coarsening occur
simultaneously. However, to obtain near-theoretical densities, coarsening has to be
suppressed until most of the shrinkage has

occurred; i.e., the system should follow the trajectory denoted by curve z. A powder
that follows trajectory x, however, is doomed to remain porous — the free energy
has been expended, large grains have formed, but more importantly so have large
pores. Once formed, these pores are kinetically very difficult to remove, and as
discussed below, they may even be thermodynamically stable, in which case they

16
would be impossible to remove. An alternate method of presenting the sintering
data is shown in Fig. 10b, where the time evolution of the grain and pore sizes
is plotted; coarsening leads to an increase in both, whereas densification
eliminates pores.

Figure 10 (a) Grain size versus density trajectories for densification (curve z) and coarsening
(curve x). Curve y shows a powder for which both coarsening and densification are occurring
simultaneously. (b) Alternate scheme to represent data in terms of grain and pore size trajectories
[3].

17
SINTERING VARIABLES

The major variables which determine sinterability and the sintered microstructure of
a powder compact may be divided into two categories: material variables and
process variables (Table 1.1). The variables related to raw materials (material
variables) include chemical composition of powder compact, powder size, powder
shape, powder size distribution, degree of powder agglomeration, etc. These
variables influence the powder compressibility and sinterability (densification and
grain growth). In particular, for compacts containing more than two kinds of
powders, the homogeneity of the powder mixture is of prime importance.

The other variables involved in sintering are mostly thermodynamic variables, such
as temperature, time, atmosphere, pressure, heating and cooling rate. Many previous
sintering studies have examined the effects of sintering temperature and time on
sinterability of powder compacts. It appears, however, that in real processing, the
effects of sintering atmosphere and pressure are much more complicated and
important [4].

18
Local Driving Force

Pressure Due to Curved Surfaces

The snapping back into a planar form of a curved soap film illustrated in Fig. 11,
results from the fact that the surface energy causes a pressure difference across a
curved surface. This is the basis for methods of measuring surface tension of liquids
by the height of rise in a capillary, or by the pressure required to blow a spherical
bubble at the end of a submerged capillary. As illustrated in Fig. 7, the work of
expanding a spherical surface P.dv must equal the increase in surface energy dA.
For a sphere, dv and dA are given by

and so the pressure exerted by the spherical surface is

In general, when the surface is not spherical, similar analysis gives

where r1 and r2 are the principal radii of curvature [6].

19
Fig. 11, Determination of the pressure at equilibrium to maintain a spherical surface of radius R
[6].

As mentioned earlier, the global driving force operating during sintering is the
reduction in surface energy, which manifests itself locally as curvature
differences. From the Gibbs-Thompson equation;

“The Gibbs free energy change is given by dG= V dP - S dT For an isothermal

process dT = 0 and dG = V dP. Integrating yields G = VP. Substituting for the
value of P given above and assuming 1 mol, one obtains

where Vm is the molar volume, which is related to the atomic volume by Ω =

Vm/Nav.”

20
, it can be shown that the chemical potential difference per formula unit  between
atoms on a flat surface and under a surface of curvature  is

(4)

where ΩMX is the volume of a formula unit. For simplicity in the following
discussion, it will be assumed that one is dealing with an MX compound.

Curvature  depends on geometry; e.g., for a sphere of radius r,  = 2/r. Equation

(4) has two very important ramifications that are critical to understanding the
sintering process. The first is related to the partial pressure of a material above a
curved surface, and the second involves the effect of curvature of vacancy
concentration [3].

Effect of curvature on partial pressure

At equilibrium, this chemical potential difference translates to a difference in partial

pressure above the curved surface, i.e.,

(5)

Combining the two equations reveals that

(6)

If Pcurv ~ Pflat then we have

Or Pcurv / Pflat = exp { (ΩMXsv / kT)} and if  (ΩMXsv / kT) << 1 then

Pcurv / Pflat = [ 1 +  (ΩMXsv / kT)]

21
 If =2/r then Pcurv / Pflat = [ 1 + (2ΩMXsv / rkT)]

Given that the radius of curvature is defined as negative for a concave surface and
positive for a convex surface, this expression is of fundamental importance because
it predicts that the pressure of a material above a convex surface is greater than that
over a flat surface, and vice versa for a concave surface. For example, the pressure
inside a pore of radius r would be less than that over a flat surface; conversely, the
pressure surrounding a collection of fine spherical particles will be greater than that
over a flat surface [3].

It is only by appreciating this fact that sintering can be understood. Given the
importance of this conclusion, it is instructive to explore what occurs on the atomic
level that allows this to happen. To do so, consider the following thought
experiment: Place each of three different-shaped surfaces of the same solid in a
sealed and evacuated chamber, as shown in Fig. 12, and heat until an equilibrium
vapor pressure is established. Examining the figures shows that PI < P2 < P3,, since
on average the atoms on a convex surface are less tightly bound to their neighbors
than atoms on a concave surface and will thus more likely escape into the gas phase,
resulting in a higher partial pressure [3].

Figure 12 Effect of surface curvature on equilibrium pressure. At this scale it is easy to

appreciate why P1, < P2 < P3 [3] .

22
Effect of curvature on vacancy concentrations

The other important ramification of Eq. (4) is that the equilibrium vacancy
concentration is also a function of curvature. The relationship between the
equilibrium concentration of vacancies Co and their enthalpy of formation Q, and
temperature is given by

In this equation, the entropy of formation and all preexponential terms are included
in the constant K'. An implicit assumption made to derive this expression was that
the vacancies formed under a flat, stress-free surface. Since the chemical potential
of an atom under a curved surface is either greater or smaller than that over a
flat surface by , this energy has to be accounted for when one is considering
the formation of a vacancy. Hence it follows that

And since for the most part SVΩMX << kT, then with little loss in accuracy

which is identical to

23
It is possible to show that the vacancy concentration under a concave surface is
greater than that under a flat surface, which in turn is greater than that under a
convex surface.

The physics can be explained as follows: given that a good measure of the enthalpy
of formation of a vacancy is the difference in bonding between an atom in the bulk
of the solid versus when it is on the surface, referring once again to Fig. 12 and
focusing on the shaded atoms will make it obvious why it costs less energy to create
a vacancy in the vicinity of a concave surface than a convex one.

To recap: curvature causes local variations in partial pressures and vacancy

concentrations. The partial pressure over a convex surface is higher than that over
a concave surface. Conversely, the vacancy concentration under a concave
surface is higher than that below a convex surface. In either case, a driving force
is present that induces the atoms to migrate from the convex to the concave areas,
i.e., from the mountaintops to the valleys. Given these conclusions, it is now
possible to explore the various atomic mechanisms taking place during sintering
[3].

Sintering stages

Coble described a sintering stage as an "Interval of geometric change in which pore

shape is totally defined (such as rounding of necks during the initial stage sintering)
or an interval of time during which the pore remains constant in shape while
decreasing in size." Based on that definition, three stages have been identified: an
initial, an intermediate, and a final stage.

During the initial stage, the interparticle contact area increases by neck growth (Fig.
9b) from 0 to ~0.2, and the relative density increases from about 60 to 65 percent.

24
The intermediate stage is characterized by continuous pore channels that are
coincident with three-grain edges (Fig. 9c). During this stage, the relative density
increases from 65 to about 90 percent by having matter diffuse toward, and vacancies
away from the long cylindrical channels.

Pore rounding and pore coarsening are other important events in this stage.

The final stage begins when the pore phase is eventually pinched off and is
characterized by the absence of a continuous pore channel (Fig. 9d). Individual pores
are either of lenticular shape, if they reside on the grain boundaries, or rounded, if
they reside within a grain. An important characteristic of this stage is the increase
in pore and grain boundary mobilities, which have to be controlled if the
theoretical density is to be achieved. Three stage of sintering is shown in Fig, 10
for Ni0.5Zn0.5Fe2O4 compound.

Fig. 10, Plot of log D versus the reciprocal of absolute temperature (1/T) of Ni0.5Zn0.5Fe2O4
showing three stages of sintering.

25
Figure 9 (a) Initial stage of sintering model represented by spheres in tangential contact, (b) Near
end of initial stage; spheres have begun to coalesce, (c) Intermediate stage; grains adopted shape
of dodecahedra, enclosing pore channels at grain edges, (d) Final stage; pores are tetrahedral
inclusions at corners where four dodecahedra meet [3].

26
Clearly, the sintering kinetics will be different during each of the aforementioned
stages, and depends on the material transport mechanisms in that stage [3].

Initial-Stage Sintering

Given the multiplicity of paths available to a powder compact during this stage, it is
impossible to address them all in detail and just the rate of neck growth by
evaporation condensation (path 1 in Fig. 11a) is worked out in detail here;

Figure 11 Basic atomic mechanisms that can lead to (a) coarsening and change in pore shape and
(b) densification. (c) Thought experiment illustrating how removal of material from the area
between particles into the pore leads to shrinkage and densification [3].

27
Evaporation-condensation model (path 1 in Fig. 11a)

In this mechanism, the pressure differential between the surface of the particle and
the neck area results in a net matter transport, via the gas phase from the surface to
the neck. The evaporation rate (in molecules of MX per square meter per second),
is given by the Langmuir expression;

(7)

where  and mMX are, respectively, the evaporation coefficient and the mass of the
evaporating gas molecules; P is the pressure differential between the surface
and neck areas. By applying Eq. (6), it can be easily shown that the pressure
differential between these two regions is

(8)

where  and r, defined in Fig. 12a, are, respectively, the radius of curvature of the
neck area and the sphere radius. Furthermore, according to this figure.

(9)

where x is the neck radius. For x << r, this equation simplifies to

(10)

Multiplying the flux of material arriving to the neck area by ΩMX yields the rate at
which the neck will grow, or

28
 (11)

Combining Eqs. (7) to (11) and integrating yields

(12)

This equation predicts that the rate of growth of the neck region (1) is initially
quite rapid but then flattens out, (2) is a strong function of initial particle size,
and (3) is a function of the partial pressure Pflat of the compound, which in turn
depends exponentially on temperature [3].

Figure 12 (a) Sphere tangency construction used for initial-stage sintering modeling.

(b) Spherical particles held together by liquid capillary pressure, (c) Sphere tangency

construction during hot pressing [3].

29
Atomic Mechanisms Occurring during Sintering

There are basically five atomic mechanisms by which mass can be transferred in a
powder compact:

1. Evaporation—condensation, depicted as path 1 in Fig. 11a.

2. Surface diffusion, or path 2 in Fig. 11a.

3. Volume diffusion. Here there are two paths. The mass can be transferred from
the surface to the neck area — path 3 in Fig. 11a — or from the grain boundary area
to the neck area — path 5 in Fig. 11b.

4. Grain boundary diffusion from the grain boundary area to the neck area — path
4 in Fig. 11b.

5. Viscous or creep flow. This mechanism entails either the plastic deformation or
viscous flow of particles from areas of high stress to low stress and can lead to
densification [3].

Consider now which of these mechanisms leads to coarsening and which to

densification.

Coarsening

At the outset, it is important to appreciate that any mechanism in which the

source of material is the surface of the particles and the sink is the neck area
cannot lead to densification, because such a mechanism does not allow the particle
centers to move closer together. Consequently, evaporation—condensation,
surface diffusion, and lattice diffusion from the surface to the neck area cannot
lead to densification. They do, however, result in a change in the shape of the pores,

30
a growth in the neck size, and a concomitant increase in compact strength. Moreover,
the smaller grains, with their smaller radii of curvature, will tend to "evaporate" away
and plate out on the larger particles, resulting in a coarsening of the microstructure.

The driving force in all cases is the partial pressure differential associated with
the local variations in curvature. For instance, the partial pressure at point s in Fig.
11a is greater than that at point n. which in turn results in mass transfer from the
convex to the concave surfaces. The actual path taken will depend on the kinetics of
the various paths, a topic that will be dealt with shortly. At this point, it suffices to
say that since the atomic processes are occurring in parallel, at any given
temperature, it is the fastest mechanism that will dominate [3].

Densification

If mass transfer from the surface to the neck area or from the surface of smaller to
larger grains does not lead to densification, other mechanisms have to be invoked to
explain the latter. For densification to occur, the source of material has to be the
grain boundary or region between powder particles, and the sink has to be the
neck or pore region. To illustrate why this is the case, consider the thought
experiment illustrated in Fig. 11c: Cut a volume (shaded area in Fig. 11c) from
between two spheres, bring the two spheres closer together, and then place the extra
volume removed in the pore area. Clearly such a process leads to shrinkage and the
elimination of pores. Consequently, the only mechanisms, apart from viscous or
plastic deformation, that can lead to densification are grain boundary diffusion
and bulk diffusion from the grain boundary area to the neck area (Fig. 11b).
Atomistically, both mechanisms entail the diffusion of ions from the grain boundary
region toward the neck area, for which the driving force is the curvature-induced

31
vacancy concentration. Because there are more vacancies in the neck area than in
the region between the grains, a vacancy flux develops away from the pore surface
into the grain boundary area, where the vacancies are eventually annihilated.
Needless to say, an equal atomic flux will diffuse in the opposite direction, filling
the pores [3].

Thermodynamics of pore elimination

It should be clear by now that pore shape and volume fraction continually evolve
during sintering, and understanding that evolution is critical to understanding how
high theoretical densities can be achieved. An implicit and fundamental assumption
made in the foregoing analysis is the existence of a driving force to shrink the pores
at all times — an assumption that is not always valid. As discussed below, under
some conditions, the pores can be thermodynamically stable.

To demonstrate the conditions for which this is the case, consider the four grains that
intersect as shown in Fig. 13a. Referring to the figure, if the pore is allowed to shrink
by an amount equal to the shaded area, the excess energy eliminated is proportional
to 2SV while the excess energy gained will be proportional to gb, where  and  are
defined in the figure. It follows that the ratio of the energy gained to that lost is

Energy Gained / Energy lost = gb / 2sv.

Combining this result with Eq. ( ) and the fact that cos (/2) =
/(see Fig. 13a), one can easily show that the right-hand side of above equation
equals 1.0, when  = . In other words, when the grains around a pore meet such

32
that gb = 2sv cos (/2), the driving force for grain boundary migration and pore
shrinkage goes to zero.

This is an important conclusion since it implies that if pores are to be completely

eliminated, their coordination number has be less than a critical value nc.

If > cos < cos 

gb/2sv means 2sv.gb < 1 that means Energy lost / Energy gained < 1

It can be further shown (Prove by yourself) that nc is related to the dihedral angle
by nc < 360/( 180 - ). It is left as an exercise for the reader to determine which of
the three pores in Fig. 13 is stable and which is not. Based on these results, one may
conclude that increasing the dihedral angle should, in principle, aid in the later
stages of sintering. The situation is not so simple, however, since it can also be
shown that the attachment of pores to boundaries is stronger for lower 's. Given
that in order to eliminate pores, they have to remain attached to the grain
boundary,

This latter property would tend to suggest that low dihedral angles would aid in the
prevention of boundary-pore breakaway and thus be beneficial. Finally, note that
channel breakup at the end of the intermediate-stage sintering occurs at smaller
volume fractions of pores as the dihedral angle decreases, which again is beneficial.

33
 Figure 13, Effect of pore coordination on pore shrinkage for a system for which the dihedral
angle  = π/2. (a) Intersection of four grains around a pore, (h) Intersection of six grains; that is,
= 120. Note that to maintain the equilibrium dihedral angle, the surfaces of the grains
surrounding the pore have to be convex, (c) Same system with = 60 . Here the pore surface has
to be concave in order to maintain the equilibrium dihedral angle [3].

34
Kinetics of Pore Elimination

As described before, densification occurs by the flux of matter from the grain
boundaries (the source) to the pores (the sink). For sintering by diffusion
mechanisms, at a given temperature and density, the dependence of the densification
rate on the grain size G can be written

where K is a temperature-dependent constant and the exponent m = 3 for lattice

diffusion and m = 4 for grain boundary diffusion. Rapid densification requires that
the diffusion distance between the source of matter and the sink be kept small;
that is, the grain size must remain small (Figure 14a).

FIGURE 14 (a) Densification mechanisms for porosity attached to a grain boundary. The arrows
indicate paths for atom diffusion. (b) Densification mechanisms for porosity separated from a
grain boundary. The solid arrows indicate paths for atom diffusion, and the dashed arrows
indicate the direction of boundary migration [1].

35
According to above equation, rapid grain growth causes a drastic reduction in
the densification rate, so prolonged sintering times are needed to achieve the
required density, which increases the tendency for abnormal grain growth to
occur. When abnormal grain growth occurs, the pores become trapped inside the
grains and become difficult or almost impossible to remove because the transport
paths become large, in the case of lattice diffusion, or are eliminated, in the case of
grain boundary diffusion (Figure 14b). The attainment of high density therefore
requires the control of normal grain growth as well as the avoidance of
abnormal grain growth [1].

Effect of microstructure and grain boundary chemistry on boundary mobility

The grain boundaries usually are included pore, inclusion, or solutes. the presence
of "second phases" or solutes at the boundaries can have a dramatic effect on their
mobility, and from a practical point of view it is usually the mobility of these phases
that is rate-limiting. Mobility of grain boundaries can be controlled by:

1- Intrinsic grain boundary mobility.

2- Extrinsic or solute drag. If the diffusion of the solute segregated at the grain
boundaries is slower than the intrinsic grain boundary mobility, it becomes rate-
limiting. In other words, if the moving grain boundary must drag the solute
along, that tends to slow it down.
3- The presence of inclusions (basically second phases) at the grain boundaries. It
can be shown that larger inclusions have lower mobilities than smaller ones,
and that the higher the volume fraction of a given inclusion, the larger the
resistance to boundary migration.

36
In addition to these, the following interactions, between pores and grain boundaries
can occur:

1. What is true of second phases is also true of pores. Pores cannot enhance boundary
mobility; they only leave it unaffected or reduce it. During the final stages of
sintering as the pores shrink, the mobility of the boundaries will increase.

2. The pores do not always shrink — they can also coarsen as they move along or
intersect a moving grain boundary.

3. The pores can grow by the pore coarsening mechanism.

To discuss even a fraction of these possibilities in any detail is clearly not within the
scope of this book. What is attempted instead is to consider in some detail one of the
more important grain boundary interactions namely, that between the grain boundary
and pore.

As the grains get larger and the pores fewer, the grain mobility increases
accordingly. In some cases, at a combination of grain size and density, the
mobility of the grain boundaries becomes large enough that the pores can no
longer keep up with them; the boundaries simply move too fast for the pores to
follow and consequently unpin themselves. This region is depicted on the grain size
versus density trajectory in the upper right-hand corner of Fig. 15 a.

If theoretical density is to be achieved, it is important that the grain boundary

versus density trajectory not intersect this separation region. The importance of
having the pores near grain boundaries is illustrated in Fig. 15b; the migration of the
boundary downwards has swept and eliminated all the pores in its wake. Pores that
are trapped within the grains will remain there because the diffusion distances
between sources and sinks become too large.

37
 Figure 15 (a) Grain size versus densification trajectory including region where separation of
boundaries and pores will occur. In that region, the grain boundaries will break away from the
pores, entrapping them within the grains. To achieve full density, path 1 has to followed. Path 2
will result in entrapped pores, (b) Micrograph of the sweeping out of pores by the migration of
grain boundaries. The original position of the boundary is depicted by the dotted line.

38
There are essentially two strategies that can be employed to prevent pore breakaway,
namely, reduce grain boundary mobility and/or enhance pore mobility. An
example of how slowing grain boundary mobility enhances the final density is
shown in Fig. 16, where the grain size versus density trajectories for two aluminas,
one pure and the other doped with 250 ppm MgO, are compared. It is obvious from
the results that the doped alumina achieves higher density — the reason is believed
to be the result of impurity drag on the boundary by the MgO [3].

Figure 16, Grain size trajectory for undoped and 250 ppm MgO-doped alumina sintered in air at
1600°C [3].

Kinetics of Pore–Boundary Interactions [1]

The definition of the pore mobility allows us to go on to analyze how the interactions
between the pores and the grain boundaries influence the kinetics of grain growth.
There are two cases that can be considered: (1) the pore becomes separated from the

39
boundary and (2) the pore remains attached to the boundary. Pore separation will
occur when vp < vb, and this condition can also be written as

(13)

where F is the effective driving force on the boundary. If Fd is the drag force
exerted by a pore, then a balance of forces requires that Fd is equal and opposite
to Fp. Considering unit area of the boundary in which there are NA pores, Equation
13 can be written (14)

where Fb is the driving force on the pore-free boundary due to its curvature.
Rearranging Equation 14, the condition for pore separation can be expressed
as

(15)

Putting vp = FpMp = vb and rearranging gives (just you need to extract F p and then
put it again in eq.(15), while the left side is supposed to be vb)

(16)

Two limiting conditions can be defined. When NAMb >> Mp, then

(17)

The effective driving force on the boundary is F = Fb – NAFp, and using Equation 15

gives F = vp/Mb. Putting vp = vb = FbMp/NA gives F = FbMp/NAMb << Fb. The driving
force on the boundary is nearly balanced by the drag of the pores, and the boundary
motion is limited by the pore mobility. This condition is referred to as pore control.
The other limiting condition is when NAMb << Mp, in which case
40
 (18)

The drag exerted by the pores is NAFp = NA(FbMb/Mp) << Fb. The presence of
the pores has almost no effect on the boundary velocity, a condition referred
to as boundary control [1].

Factors Affecting Solid State Sintering

Typically, a solid-state sintered ceramic is an opaque material containing some

residual porosity and grains that are much larger than the starting particle sizes. On
the basis of the discussion and models just presented, it is useful to summarize the
more important factors that control sintering. Implicit in the following arguments is
that theoretical density is desired.

1. Temperature. Since diffusion is responsible for sintering, clearly increasing

temperature will greatly enhance the sintering kinetics, because D is thermally
activated. As noted earlier, the activation energies for bulk diffusion are usually
higher than those for surface and grain boundary diffusion. Therefore, increasing the
temperature usually enhances the bulk diffusion mechanisms which lead to
densification.

2. Green density. Usually a correlation exists between the green (prior to sintering)
density and the final density, since the higher the green density, less pore volume
has to be eliminated.

3. Uniformity of green microstructure. More important than the green density is the
uniformity of the green microstructure and the lack of agglomerates (see Fig. 17).

41
Figure 17, Temperature dependence of sintered density for an agglomerated or "as received" and
agglomerate-free yttria-stabilized zirconia powder (1 h). Eliminating the agglomerates in the
green body resulted in a powder compact that densified much more readily [3].

4. Atmosphere. The effect of atmosphere can be critical to the densification of a

powder compact. In some cases, the atmosphere can enhance the diffusivity of a
rate-controlling species, e.g., by influencing the defect structure. In other cases, the
presence of a certain gas can promote coarsening by enhancing the vapor pressure
and totally suppressing densification. An excellent example of the effect of
atmosphere was shown in Fig. 18: Fe2O3 will readily densify in air but not in HCl-
containing atmospheres. Another important consideration is the solubility of the
gas in the solid. Because the gas pressure within the pores increases as they shrink,
it is important to choose a sintering atmosphere gas that readily dissolves in the solid.

42
 Figure 18, Effect of atmosphere on relative density versus time for FeO sintered at 1000°C.

5. Impurities. The role of impurities cannot be overemphasized. The key to many

successful commercial products has been the identification of the right pinch of
magic dust. The role of impurities has been extensively studied, and to date their
effect can be summarized as follows:

a. Sintering aids. They are purposefully added to form a liquid phase (discussed in
the next section). It is also important to note that the role of impurities is not always
appreciated. The presence of impurities can form low-temperature eutectics and
result in enhanced sintering kinetics, even in very small concentrations.

43
b. Suppress coarsening by reducing the evaporation rate and lowering surface
diffusion. A classic example is boron additions to SiC. Without which SiC will not
densify.

c. Suppress grain growth and lower grain boundary mobility (Fig. 16).

d. Enhance diffusion rate. Once the rate-limiting ion during sintering is identified,
the addition of the proper dopant that will go into solution and create vacancies on
that sublattice should, in principle, enhance the densification kinetics.

6. Size distribution. Narrow grain size distributions will decrease the propensity for
abnormal grain growth.

7. Particle size. Since the driving force for densification is the reduction in surface
area, the larger the initial surface area, the greater the driving force. Thus it would
seem that one should use the finest initial particle size possible, and while in
principle this is good advice, in practice very fine particles pose serious problems.
As the surface/volume ratio of the particles increases, electrostatic and other surface
forces become dominant, which leads to agglomeration. Upon heating, the
agglomerates have a tendency to sinter together into larger particles, which not only
dissipates the driving force for densification but also creates large pores between
the partially sintered agglomerates which are subsequently difficult to eliminate.
The dramatic effect of ridding a powder of agglomerates on the densification kinetics
is well illustrated in Fig. 17 [3].

References;

1- Sinterng of Ceramics by Mohamed N. Rahaman, 2007 by Taylor & Francis.

44
2- Sintering, Densification, Grain Growth, and Microstructure by Suk-Joong
L.Kang, 2005 by Elsevier Butterworth-Heinemann.
3- Fundamentals of Ceramics, by Michel W Barsoum, 2003 by IOP Publishing
Ltd.
4- Sintering Densification, Grain Growth, and Microstructure, by Suk-Joong
L.Kang, 2005 by Elsevier Butterworth-Heinemann.
5- Sintering from Emprical Observations to Scientific Principles, by Randall
M. German, 2014 Elsevier Inc. Oxford UK.
6- Physical Ceramics; Principles for Ceramic Science and Engineering by Yet-
Ming Chiang, Dunbar P. Birnie, III, W. David Kingery, John Wiley & Sons,
Inc. 1997.

45