ceramics

Paper No. : ceramics:processing of ceramics

Module : particle size

Development Team
Prof. Vinay Gupta, Department of Physics and Astrophysics,
Principal Investigator University of Delhi, Delhi

Prof. V. K. Gupta, Department of Physics and

Paper Coordinator
Astrophysics,University of Delhi, Delhi

Content Writer Prof. V. K. Gupta, Department of Physics and

Astrophysics,University of Delhi, Delhi

Prof. Pankaj Sharan, Physics department from JMI, New Delhi

Content Reviewer
Description of Module
Subject Name Material Science
Paper Name Ceramics:processing of ceramics
Module Name/Title Particle size
Module Id M 2.1

Unit 2

Processing of Ceramics

Module 1: Particle Size

Introduction

The history of ceramics dates back to at least 24000 years BC. Animal and
human figurines made of clay dating back to this period have been
discovered. Tiles made of clay have been found in India and Mesopotamia
dating back to 14000 BC. Clay is the most abundantly occurring raw
material on the surface of earth. The ceramic industry was born once the
humans learned how to mix it with water, shape it and fire it in kilns. No
wonder that the work “ceramics” itself is believed to be derived from
keramos, the Greek word for clay. Even today the more familiar ceramics
such as tiles, porcelain, sanitary ware, bricks, etc are all based on the
potter's clay. However, in addition to these so called "traditional
ceramics", we have a new class of ceramics, called the "advanced
ceramics" which came on the scene in the 20th century as the materials
systems became more refined and special compounds and processes were
developed for structural and electronic applications. These advanced
ceramics are distinguished by their high chemical purity, careful
processing and high values of the useful properties. They comprise
broadly of two classes – the functional ceramics and the structural
ceramics. The functional ceramics are the advanced ceramics used for
electrical, electronic, magnetic, optical and related applications while the
structural ceramics are those in which the mechanical properties such as
strength, modulus, toughness, wear resistance, hardness, etc. are of
primary interest. Almost 70 % of the advanced ceramics manufactured
today belong to the functional ceramics category. In addition to these two
main types of ceramics, small volume of specialty ceramics for bio, thermal and chemical applications
are also manufactured.

Ceramics are hard and brittle materials and have a high melting point. The techniques which are common for
shaping of metals and polymers can not be used for them. It is difficult and not economical to cast them into
shape after melting because of their very high melting points, in most cases more than 2000 o C. Because of their
brittleness they can not be shaped by deformation processing techniques such as rolling and forging. By far the
most used technique, technique which is also used for metals to some extent and also for polymers is powder
processing.
In powder processing, one starts with a powder of the material to be shaped. Since the ceramics are brittle
materials, it is easy to crush and grind them to convert into powders. In the most common technique of ceramic
processing, the powders are filled in a die and pressed to the desired shape. Usually some binders and additives
may also be added to the powder mixture. In some cases, a slurry or paste of the powder is prepared and
techniques such as slip casting or extrusion are used to form them into a shape. The process of slip casting is
commonly used for the preparation of sanitary wares, but can also be used to advanced components.
The body produced using the above shaping steps is called a “green body”. The adjective green comes from
the fact that in most cases the body contains some moisture. In the dry pressed articles, the moisture content is
low, but in products made using slip casting or other techniques which start with a slurry or paste, the moisture
content is very high. A drying step is to be used in the latte case to remove the moisture. During drying,
considerable shrinkage of the body may occur. If care is not taken, the high shrinkage produces stresses which
can cause the cracking of the body. After drying, the green body is very fragile with a very low strength and a
low relative density. The relative density is defined as the ratio of the actual density to theoretical density of the
material. For example, if the green body is made of alumina, Al2O3, the theoretical density is 3.9 g.cm-3. The
relative density of a green body usually is of the order of 0.6 or less. It is useful to maximize the green density,
as a high green density leads to a high final density. Often special techniques such as isostatic pressing and hot
isostatic pressing are used to maximize the green and the final densities..
As mentioned above, the green body has a low density, low strength and is very fragile. To develop the
desired properties, it is subjected to a firing operation in which the green body is heated to a high temperature,
usually of the order of 0.7 to 0.8 of the melting point of the material. During this step, a phenomenon called
sintering occurs. Some grain growth also occurs during firing. A properly sintered body has a high relative
density, which may be more than 0.95, high strength and other desired properties. As the relative density is less
than 1, there is always some porosity in the powder processed ceramics.
The final step in the processing may be some grinding and polishing of the sintered body to get the desired
dimensions and finish. In some cases, special techniques are used to machine the body to achieve the desired
shape.
The steps described above are summarized in the diagram shown in Fig. 2.1.1

Ceramics:processing of ceramics
Physics
Particle size
 Fig. 2.1.1 The steps involved in the processing of ceramics

The properties of the final product depend on several factors. These include the characteristics of the starting
powder, additives mixed in the powder, green body preparation technique used and the firing conditions. In this
topic, we will first discuss some characteristics of the powder. This will be followed by the techniques of cold
pressing, slip casting and some others to produce the green body. Finally, the processes occurring during
sintering and how they control the properties will be discussed. The emphasis throughout will be on the science
of ceramic processing.
Amongst the characteristics of the starting powder, the particle size is one of the most important
characteristics. It has a great influence on powder flow and handling, powder compaction and green density,
sintering temperature, the sintered density and the properties of the sintered product. In general, finer powder
needs a low sintering temperature and leads to higher sintered densities, a finer grain size and an improvement in
properties. However, a fine powder is also more difficult to process and may require special processing.
The range of particle sizes used in ceramic processing is very large, Usually the particle sizes range between
50 nm (50 Χ 10-9 m) to 1 mm but in some cases they can be much larger ( a few cm as in refractory castables) or
much smaller (a few nm for nanocrystalline ceramics). It may sometime require more than one measurement
technique to cover full range of sizes in a given powder.
In this chapter we first make some comments on the determination of the particle size and its distribution.
Then we discuss the statistics of particle size distribution.

Particle size measurement techniques

Several techniques are available for measurement of the particle size. Choice of a technique depends on the
expected range of particle size, cost of the measuring instrument, time available to perform the measurement,
6

Ceramics:processing of ceramics
Physics
Particle size
etc. Sometimes two or more techniques may be required to cover the whole range of sizes in a sample. The
common techniques are sieving, microscopy, instruments using the sedimentation of particles and light
scattering techniques. For determination of the particle sizes in the nanometer range, the technique of photon
correlation spectroscopy or imaging techniques such as electron microscopy or atomic force microscopy are
used. The details of these techniques are covered in several books.[1-3]

Analysis and presentation of the particle size data

Statistics of particle size data
The particles in a given sample of any powder in general have sizes lying in a range i.e. they have a “size
distribution”. The simplest way to present the particle size data is to count (or weigh) the particles lying in
different size intervals and give the result in a table or as a figure.

An illustrative example is given in Table 2.1.1 Here the sample powder contains 500 particles. The number
of particles lying in different size ranges is

Table 2.1.1 : An example of particle size data

Size interval MidpointNumber of particlesRelative Cumulative

End point value In the interval ni Frequency Frequency
value ai* (m) ai (m) (frequency) (ni/ni) = ni/N  ni/N)
1
3 2 7 0.014 0.014
5 4 130 0.26 0.274
7 6 249 0.498 0.772
9 8 110 0.22 0.992
11 10 4 0.008 1.000
Total 500 1.0

tabulated (column 1 and 3). In column 2, the midpoints of the size ranges are given. In column 4, the “relative
frequency” of a particular size is tabulated, while in the last column the cumulative frequency is given. In Fig.
2.2 the data in the table is drawn as a histogram (Fig 2.1.2a), a relative frequency plot (Fig. 2.2b) and a
cumulative frequency plot (Fig 2.2c).

Ceramics:processing of ceramics
Physics
Particle size
Fig. 2.1.2 Plots of the data in Table 2.1.1 (a) Histogram (b) Relative frequency f vs. particle size (c) Cumulative frequency F vs. particle
size.

From a particle size distribution, various parameters of the distribution can be calculated. These parameters
are defined below:

n a i i
Mean particle size   i 0

where N  ni
i  Total number of particles

∑ n( a -  )2
I

i i
i =1
Variance =  2 =
N -1
I

 n a  
3
i i
Skewness =   i 1

N
Mode: The size having the highest frequency

Ceramics:processing of ceramics
Physics
Particle size
Median: The size having 50% cumulative frequency (i.e. half the particles are below and half above the
median).

As an illustration, the values of  and  2 for the data in Table 2.1 are calculated below
I

n a i i
7  2  130  4  249  6  110  8  4  10
 i 1
 m  5.9m
N 500

∑ n( a -  )2
I

i i
i =1
2 =
N -1
= 7(2 - 5.9) 2 + 13(4 5.9) 2 + 249(6 5.9) 2 + 110(8 5.9) 2 + 4(10 5.9) 2 499
= 2.266 m 2
Square root of variance,  , is called the standard deviation of the distribution. Thus for the data in Table 2.1.1,
the standard deviation  = 1.505 m.
The most common method of presenting the particle size data is as a cumulative frequency plot (Fig 2.1.2c);
if the ordinate of this plot is multiplied by 100, it becomes a cumulative percent. This plot gives the fraction or
percentage of particles finer than a given size and is called a CNPF (cumulative number percent finer than) plot.
If the ordinate is subtracted from 100, then it becomes CNPL (cumulative number percent larger than) plot.

Depending on the technique used for the determination of the particle size, the data may be available as
particle diameter vs. number of particles, mass of particles or volume of particles. If the density of all the
particles is the same, then the data should be identical whether presented as diameter vs. mass or diameter vs.
volume. When the data is presented as diameter vs. mass of particles, (say the mass of powder retained on a
sieve) then the resulting plots will be CMPF (cumulative mass percent finer than) or a CMPL (cumulative mass
percent larger than) plots.
As we will see below, the particle size data can often be fitted to a standard statistical distribution function.
In that case, the complete information about the particle size distribution can be specified by giving just the
parameters of the distribution which are usually two in number. In case it is not possible to fit the data to a
function, then the particle sizes corresponding to CMPL 90%, 50% and 10 % are usually given to give an idea of
the distribution of particle sizes. These are represented by D90, D50 and D10 respectively.
If the particle size data can be fitted to a standard function as discussed below, then the average size can be
calculated from this function as follows.

a max
i= N c

aL = ∫af N (a )da or a L = ∑a f
i =1
i N ( ai )
a min

Ceramics:processing of ceramics
Physics
Particle size
 Sometimes it is required to calculate the average diameter of the particle which has the average surface area
or average volume or average volume. These quantities can also be calculated as follows.
a max
i= N c

Area average mean size, aA = ∫a 2

f N (a )da or a A = ∑a
i =1
2
i
f N ( ai )
a min

a max
i= N c

Volume average mean size aV = 3 ∫a 3

f N (a )da or aV = 3
∑a
i =1
3
i
f N ( ai )
a min

Here a is the measured particle size and a is the mean size in the interval as discussed earlier
If the size data is measured on the mass basis, then the function available would be fM(a). The function fN(a)
may be calculated from this as follows

FM(a)=(a3)fN(a)

Here  is the density of the particles and  is a shape function which relates the volume to a3 .

Fitting of the particle size data to mathematical functions

Very often the particle size data can be represented by a mathematical function. The most common such
function is the lognormal distribution. This distribution is usually followed by powders produced by milling of
fine-grained minerals and calcined aggregates. Other functions to which the particle size data can in some
cases be fitted are the Rosin-Ramler equation and the Gaudian Schuhmann equation.

The log-normal distribution:

As mentioned above, in many cases the particle size data is found to follow a log normal distribution. To
understand this distribution function, it is useful to first review the well known normal distribution.
A parameter ‘x’ is said to be normally distributed if its relative frequency is given by (Fig 2.3a)
 x   x 2 
f x  
1
exp  
2  x  2 x2 

where  x is the mean and  x the variance (  x , the standard deviation) of the distribution. A normal
2

distribution is symmetric and for it the mean, mode and the median are all equal.

Mean  x = mode = median for the normal distribution. These quantities are also shown on the cumulative
curve for the normal distribution (Fig 2.3b).
Another important property of the normal distribution is that
[x for F(x) = 0.8413] – [x for F(x) = 0.5] =  x

10

Ceramics:processing of ceramics
Physics
Particle size
Here F(x) is the cumulative frequency for x. This implies that a fraction equal to 0.6826 of the sample lies within
one standard deviation of the mean size. The value of the standard deviation is thus easily found by
subtracting the value of x for F(x) =0.5 from the value of x for F(x) =0.8413, using the cumulative plot.

Now we come to the log normal (LN) distribution. In a LN distribution the natural logarithm of the
parameter, lnx, is normally distributed i.e. if y=lnx, y is normally distributed. The distribution is given by
 ln x  ln  g 2 
f ln x  
1
exp  
2 ln  g 2ln  g  
2


The parameters μg and g are called the geometric mean and the geometric standard deviation. A log normal
distribution is positively skewed when plotted as particle size vs. relative frequency. The normal and the log
normal distributions are compared in Figure 2.1.3 . Note that the log normal distribution is skewed to the right
i.e. to larger sizes.

Fig. 2.1.3 . Schematics of (a) the normal distribution and (b) the log normal distribution

If the cumulative frequency of a LN distribution F(lnx), or equivalently, F(x), is plotted against ‘x’ (particle
size), then the geometric standard deviation is obtained. An example of such calculations is given below:

Example: Particle size data of a milled bauxite powder is as follows:

Size (m) 9.0 7.0 5.0 3.0 2.0 1.5 1.0 0.6 0.4 0.2

CNPF(%) 100 97.2 92.5 86.3 70.7 61.0 44.3 18.9 11.0 1.6

Assuming that the powder follows a log normal distribution, find the geometric mean and the geometric
standard deviation of the distribution.
11

Ceramics:processing of ceramics
Physics
Particle size
Solution: The plot of particle size d vs CNPF (as fraction) is given in Fig. 2.1.4.

Fig. 2.1.4. Plot of the data as particle size vs. CNPF (as fraction)

The geometric mean g is the value of d corresponding to the 50% value of the CNPF. From Fig. 2.1.4, this is
1.15 m.

The value of the geometric standard deviation g is obtained by taking the ratio of the values of d
corresponding to CNPF = 84.13 % and the g. Thus, from Figure 7,

g = 2.85/1.15 = 2.48 m.

Hence for the above particle size distribution, the geometric mean g is 1.15 m and the geometric standard
deviation g is 2.48 m



Other particle size distributions
(a) Rosin-Rammler equation

The particle size data of some materials follows this distribution. It is given by

FM(a) = 1- exp – (a/ao)m

12

Ceramics:processing of ceramics
Physics
Particle size
Here ao is a characteristic parameter for the distribution for which FM(a)= 0.37 and m is the shape parameter.
Reciprocal of m is a measure of the width of the distribution i.e. if m is large, the distribution is narrow. While
discussing the mechanical properties we will again come across this parameter where it is called the Weibull
distribution.

The parameters ao and m can be obtained graphically by plotting the quantity ln[– ln(1-FM(a))] vs.ln a . On
this plot, the slope gives the value of m and ao is the size for which the ordinate, i.e., ln[– ln(1-FM(a))] is zero.
(b) Gaudin-Schuhmann equation. The particle size data of coarse industrial minerals produced by crushing and
grinding often follow this distribution. It is given by the following equation.

FM(a) = (a/amax)n

Here n is a constant indicating the range of sizes and amax is the apparent maximum size obtained by
extrapolation. For this distribution a straight line would be obtained when it is plotted on a log-log paper.

Summary
Processing of ceramics is mostly carried out using powder processing. The powder is first compacted into a
desired shape by a technique such as dry pressing. The piece thus obtained is called a green body. The green
body is first dried and then fired to obtain full density and other properties.
The characteristics of the starting powder have a great influence on the final properties. One of the most
important characteristics of the ceramic powder is its particle size distribution. It can be measured using several
techniques. From the data obtained, several parameters of the particle size distribution such as the average
particle size and the particle sizes at given mass percent can be specified. Often it is possible to fit the particle
size data to a standard statistical distribution function. In that case, the complete information about the particle
size is specified by just giving the parameters of the distribution which are usually two in number One of the
most common of these functions is the log normal distribution function. From a plot of the cumulative number
fraction or the cumulative mass fraction, the parameters of the distribution like the mean and the standard
deviation can be deduced which give complete information about the particle size distribution. Several other
statistical distributions are also used to fit to the particle size distribution data.

13

Ceramics:processing of ceramics
Physics
Particle size