You are on page 1of 56

CHEE2940: Particle Processing

Lecture 2: Particle Size and Shape

This lecture covers

Particle size and shape


Particle size analysis
Measurement techniques

Chee 3920: Particle Size and Shape


WHY IS PARTICLE SIZE ANALYSIS IMPORTANT?

• Determines the quality of final products


.

• Establishes performance of processing

• Determines the optimum size for separation

• Determines the size range of loses.

Chee 3920: Particle Size and Shape 1


2.1 PARTICLE SIZE AND SHAPE
• Particle size: refers to one particle.
• Precise particle size is difficult to obtain due to
the irregular shape of particles.

From M. Rhodes, Intro Part. Tech., Wiley, 1998

Chee 3920: Particle Size and Shape 2


• For spherical particles, defining particle size is
easy; it is simply the diameter of the particle.
• For non-spherical particles, the term
"diameter" is strictly inapplicable. For example,
what is the diameter of a flake or a fiber?
• Also, particles of identical shape can have
quite different chemical composition and,
therefore, have different densities.
• The differences in shape and density could
introduce considerable confusion in defining
particle size.

Chee 3920: Particle Size and Shape 3


• Equivalent diameter is often used.

- Equivalent volume diameter – diameter of a


sphere with the same volume (mass) as the
particle:

d v = 6V / π 3

V … real particle volume.

- Equivalent surface diameter - diameter of a


sphere with the same surface area as the
particle (BET isotherm):
Chee 3920: Particle Size and Shape 4
ds = A / π
A … real particle surface area.

- Equivalent volume-surface (Sauter)


diameter - diameter of a sphere with the
same volume to surface area ratio as the
particle.

d Sauter = 6V / A
V and A … real particle volume and surface
area.
Chee 3920: Particle Size and Shape 5
Example of equivalent diameters for a particle
with a shape of rectangular box

Dimension (mm) 20 x 30 x 40
2
Surface area (mm ) 5200
3
Volume (mm ) 24000
dv (mm) 35.8
ds (mm) 40.7
dSauter (mm) 27.7

Chee 3920: Particle Size and Shape 6


- Stokes (hydraulic) diameter – from settling
velocity (drag force and weight) – diameter
of a sphere with the same density and
terminal settling velocity (discussed later).

18µU
d Stokes =
g (ρ −δ )

U … real particle terminal settling velocity

Chee 3920: Particle Size and Shape 7


µ … liquid viscosity
µ = 0.001 Pa/s for water
µ = 0.00001 Pa/s for air
2
g … acceleration due to gravity (9.81 m/s )

ρ and δ … particle & liquid densities.


3
ρ = 2500 kg/m for quartz (SiO2)
3
δ = 1000 kg/m for water.

- Sieve diameter – The smallest dimension of


sieve aperture through which particles pass.
Chee 3920: Particle Size and Shape 8
• Microscopically Observed Shapes

Martin’s diameter – bisects the area of the


particle image –always taken in the same
direction.

Feret’s diameter – distance between parellel


tangents –always taken in the same direction.

Equivalent area – diameter of a circle with the


same area of the particle image.

Equivalent perimenter – diameter of a circle


with the same perimeter of the particle image.

From M Rhode, IPT, 1998.

Chee 3920: Particle Size and Shape 9


• Deviation of irregular shape from spheres.

Is described by sphericity.

- Volume sphericity, ψ V (the same volume)


ψ V = π ( dV ) / A
2

where dV is volume-equivalent diameter


A is the real surface area.

- Surface sphericity, ψ A (the same surface)


ψ A = π ( d A ) / ( 6V )
3

Chee 3920: Particle Size and Shape 10


where d A is surface-equivalent diameter
V is the real volume.

- Sauter-diameter sphericity, ψ VA and ψ AV

ψ VA = π ( d32 ) / A
2

ψ AV = π ( d32 ) / ( 6V )
3

where d32 is Sauter diameter.

Chee 3920: Particle Size and Shape 11


• Equivalent diameter of many particles
- Mean diameter, d
m
d = ∑ γ i di
i =1

where di is diameter of i-th size range


γ i is mass fraction of i-th size range.
- Volume equivalent diameter, dV
m
( dV ) = ∑ γ i di
3 3

i =1

Chee 3920: Particle Size and Shape 12


- Surface equivalent diameter, d A
m
(dA ) = ∑ γ i di
2 2

i =1

- Sauter diameter, d32

∑γ d
3
i i
d32 = i =1
m

∑γ d
2
i i
i =1

Chee 3920: Particle Size and Shape 13


2.2 METHODS OF PARTICLE SIZE ANALYSIS
Table 2.1 Some methods of particle size analysis
Method Equivalent size
Test sieving 100 mm – 10 microns
Elutriation 40 microns – 5 microns
Gravity sedimentation 40 microns – 1 microns
Centrifu. sedimentation 40 microns – 50 nano
Microscopy 50 microns – 10 nano
Ligth scattering 10 microns – 10 nano

Sieve Analysis
Chee 3920: Particle Size and Shape 14
- Good for particle >25 µm, cheap & easy.
- Carried out by passing sample via a series of
sieves (Fig 2.1)
- Weighing the amount collected on each sieve
- With wet or dry samples.
• Test sieves
• Designed by the norminal aperture size (Fig
2.2)
• Popular designs: BSS (British), Tyler series
(American), DIN (German).

Chee 3920: Particle Size and Shape 15


Largest
apertur

Smallest
aperture

Fig 2.1 Example of sieve arrangement (Wills)


Chee 3920: Particle Size and Shape 16
Mesh = number of apertures per inch

Table 2.2 BSS 410 wire-mesh sieves (Wills)


Chee 3920: Particle Size and Shape 17
Fig 2.2 Examples of aperture designs (Wills)

Chee 3920: Particle Size and Shape 18


Presentation of results for sieve analysis

Table 2.3 Example of size distribution


Size range Mid-point Mass Mass fraction Cumulative Cumulative
retained undersized Oversized
(micron) (micron) (g)
+200 200 0 0 1.000 0.000
200 - 150 175 10 0.111 1.000 0.000
150 - 100 125 40 0.444 0.889 0.111
100 - 50 75 30 0.333 0.444 0.556
50 - 0 25 10 0.111 0.111 0.889
0 0 0 0.000 1.000
sum 90

Chee 3920: Particle Size and Shape 19


Gravity sedimentation technique
• Uses the dependence of the settling velocity
on the particle size (the Stokes law)
du
mg − m ' g − F = m
dt
st nd
where the 1 term is the particle weight, the 2
rd
is the buoyancy, 3 is the drag force and the
last term is the inertial force. u is particle
velocity.

• Stokes law for drag: F = 3πµ du


Chee 3920: Particle Size and Shape 20
• (Terminal) settling velocity:

u=
( ρ − δ ) gd 2

18µ
where ρ and δ are particle and liquid density, g
is gravity acceleration and µ is liquid viscosity.
• Experimental steps:
- Sample is uniformly dispersed in water in a
beaker.
- A siphon tube is immersed into 90% of the
water depth.
Chee 3920: Particle Size and Shape 21
- Particle with size d is sucked from the beaker
at time interval t calculated from the
immersed depth and Stokes’ velocity: t = h / u .

Fig 2.3 Beaker


decantation
for gravity
sedimentation
size analysis
(Wills)
Chee 3920: Particle Size and Shape 22
Pipette filler to
collect the sample

Two-way stopcock

Fig 2.4 Andrean


pipette for
sedimentation size
analysis (Wills)
Chee 3920: Particle Size and Shape
Elutriation technique
• Uses an upward current of water or air for
sizing the sample.
• Is the reverse of gravity sedimentation and
Stokes’s law applies.
• Particles with lower settling velocity overflow
• Particles with greater velocity sink to under
flow.
• Sizing is achieved with a series of simple
elutriators (Fig 2.5).
Chee 3920: Particle Size and Shape
Fig 2.5 Simple
elutriator
(Wills)
Chee 3920: Particle Size and Shape
• For fine particles (<10 microns), cyclosizer is
usually used (Fig 2.6).

Fig 2.6 Warman cyclosizer (Wills)


Chee 3920: Particle Size and Shape
Microscopy techniques
• Used for small (dry) samples.
• Particle size is directly measured.
• Optical microscopes: 1 micron (wavelength of
light is ~ 100 microns)
• Electron (TEM and SEM): ~ 10 nm.

Chee 3920: Particle Size and Shape 20


Light scattering techniques
• Based on the capability of colloidal particles to
scatter light.
• Useful for colloidal particles.
• Static light scattering: Intensity ~ particle
volume and particle concentration.
• Dynamic light scattering measurements give
2
the r.m.s. of displacements, x .
• Brownian diffusivity, D, of particles is
determined from the Einstein-Smoluchowski
Chee 3920: Particle Size and Shape 21
equation
x 2
= 2 Dt

• Particle size is determined from Einstein’s


equation
3πµ d = k BT / D
where µ is liquid viscosity
kB is Boltzman’s constant
T is absolute temperature.

Chee 3920: Particle Size and Shape 22


2.3 ANALYSIS OF SIZE DISTRIBUTION
(Of many particles)
• Based on tabular results of size analysis
(Table 2.2)
• Characteristic parameters: mean diameter,
standard deviation, distribution functions, and
cumulative curves.
• Mean diameter (shown previously)
m
d = ∑ γ i di
i =1

Chee 3920: Particle Size and Shape 23


• Standard deviation, σ,
m m
σ = ∑ γ i ( di − d ) = ∑ γ i di − ( d ) = ( di ) − ( d )
2 2 2 2 2 2

i =1 i =1

• Frequency distribution
- Histogram: mass of size range versus size
range.
- Normalised histogram: mass fraction vs size
range.
- Continuous distribution function: mid-points of
mass fraction vs mid-points of size range
Chee 3920: Particle Size and Shape 24
40
Mass versus size range

30

Mass retained (g)


20

10

0
0 - 50 50 - 100 100 - 150 150 - 200
Size range (micron)

Histogram for data in Table 2.3

Chee 3920: Particle Size and Shape 25


0.5
Mass fraction versus size range
0.4

Mass fraction 0.3

0.2

0.1

0
0 - 50 50 - 100 100 - 150 150 - 200
Size range (micron)

Normalised histogram (Table 2.3)

Chee 3920: Particle Size and Shape 26


0.5

0.4
Midpoint of mass fraction versus
midpoint of size range
0.3
f(d)

0.2

0.1

0
0 50 100 150 200
d (microns)

Continuous distribution function (γ => f)


(Data in Table 2.3)
Chee 3920: Particle Size and Shape 27
- Theoretical distribution functions (taken from
theory on probability and statistics)

Nornal (Gaussian) distribution

1 
 1  d − d 
2


f (d ) = exp −   
σ 2π  2  σ  
σ … standard deviation of the distribution
d … mean (median) diameter

Property: ∫ f ( d ) dd = 1
−∞
or ∑ f ( d ) ∆d = 1 .
i

Chee 3920: Particle Size and Shape 28


2.5 
 1 d −d  
2
1 
f (d ) = exp −   
σ 2π 

2  σ  

2
Experiments
1.5
f(d)

0.5

0
0 50 100 150 200
d (microns)

Fig 2.7a Example of Gaussian (normal)


frequency distributions. d = 100 µ m & σ = 20
Chee 3920: Particle Size and Shape 29
Log-normal distribution

1 
 1  x − x  
2

f ( x) = exp −   
σ 2π  2  σ  
where x = log ( d ).
σ … standard deviation of the distribution
d … mean (median) diameter

Property: ∫ f ( x ) dx = 1
−∞
or ∑ f ( x ) ∆x = 1 .
i

Chee 3920: Particle Size and Shape 30


2.5 2.5
( )
  log ( d ) − log d  
2
1  1 
f (d ) = exp −  
σ 2π 2
   σ  

2 2

1.5 1.5
f(d)

f(d)
1 1

0.5 0.5

0 0
0 50 100 150 200 0.5 1.5 2.5
d (microns)
log(d/microns)

Fig 2.7b Example of log-normal frequency


distributions in the normal (left) and log-normal
(right) diagrams. log ( d / µm) = 1.6 & σ = 0.17 .
Chee 3920: Particle Size and Shape 31
Comments:
Many size distributions do not follow the
theoretical Gaussian and log-normal statistics.

The theoretical concepts remain valid for


describing the particle size distributions. We
need the mean (median) diameter and the
standard deviation.

A number of approximate equations have used


for the particle size distributions (shown later).
Chee 3920: Particle Size and Shape 32
• Viewing distributions

No details of fines
can be seen in the
normal-normal plot

The log-normal plot


gives more details
of fines

Chee 3920: Particle Size and Shape 33


• “Average” size of many particles
Mode – most frequent size occurring
Median – d50 (50% cumulative distribution)
Means – different types for different uses
- Arithmetic mean
- Quadratic mean
- Geometric mean
- Harmonic mean

Chee 3920: Particle Size and Shape 34


Graphical correlations (M Rhode, 1998)
Chee 3920: Particle Size and Shape 35
n

d + d 2 + ... + d n ∑d i

Arithmetic mean d= 1
n
= i =1

( d1 ) + ( d 2 ) + ... + ( d n )
2 2 2
1 n
Quadratic mean ( d ) ∴ ∑ i ( )
2
d=
2
= d
n n i =1

Geometric mean d = n d1 ⋅ d 2 ⋅ ... ⋅ d n = ( d1 ⋅ d 2 ⋅ ... ⋅ d n )


1/ n

(It presents the arithmetic mean of the lognormal distribution!)


1 1 1
+ + ... + n
1 d1 d 2 dn d=
Harmonic mean d
=
n ∴ n

∑ (1/ d ) i
i=1

Chee 3920: Particle Size and Shape 36


Modes of distributions
f(d)
- Mono disperse particles

- Mono modal distribution d

0.5

0.4

0.3
f(d)

0.2

0.1

0
0 50 100 150 200
d (microns)

Chee 3920: Particle Size and Shape 37


- Bimodal distributions (Fig 2.8 – solid line)
3

f(d) 2

0
0 50 100 150 200
d (microns)

Bimodal distribution occurs for


mixtures of two minerals.
Chee 3920: Particle Size and Shape 38
For analysis, bimodal distribution is separated
(using an appropriate mathematical technique
called deconvolution) into the Gaussian/log-
normal distributions.

• Cumulative distributions
- Undersized cumulative distribution
m
Q ( di ) = ∑ γ i
i =1

(Summing from the smallest size fraction)


Chee 3920: Particle Size and Shape 39
- Oversized cumulative distribution
1
P ( di ) = ∑ γ i = 1 − Q ( di ).
i =m

Example of determing cumulative distributions


Size range Mid-point Mass Mass fraction Cumulative Cumulative
retained undersized Oversized
(micron) (micron) (g)
+200 200 0 0 1.000 0.000
200 - 150 175 10 0.111 1.000 0.000
150 - 100 125 40 0.444 0.889 0.111
100 - 50 75 30 0.333 0.444 0.556
50 - 0 25 10 0.111 0.111 0.889
0 0 0 0.000 1.000
sum 90

Chee 3920: Particle Size and Shape 40


1.000
Oversized
Undersized
0.800

Cumulative mass fraction 0.600

0.400

0.200

0.000
0 50 100 150 200
d (microns)

Cumulative distribution curves (Table 2.3)


Chee 3920: Particle Size and Shape 41
- Many curves of cumulative oversized and
undersized distributions versus particle size
are S-shaped.
- Two approximations for cumulative
distributions are known, i.e., Rosin-Rammler
and Gates-Gaudin-Schuhmann distributions.

- Rosin-Rammler (RS) distribution


  d  n

P ( d ) = exp −   
  d '  
Chee 3920: Particle Size and Shape 42
where d ' and n are parameters.
d ' and n can be determined from the graph
of log {− ln ( P )} versus log ( d ) in the log-log
diagram which gives a straight line

{ }
log − ln  P ( d )  = n log ( d ) − n log ( d ')

n … the slope of the straight line.


-nlog(d’) … intercept of the straight line.

Chee 3920: Particle Size and Shape 43


1 1

0.8 0

log{-ln[P(d)]}
0.6 -1
P(d)

0.4 -2

0.2 -3

0 -4
0 50 100 150 200 0 1 2 3
d (microns) log(d/micron)

Example of data which can be described by the Rosin-Rammler distribution.


The slope of the log-log diagram gives n = 2.
The intercept is equal to –4 and gives d’ = 100 microns.

Chee 3920: Particle Size and Shape 44


- Gates-Gaudin-Schuhmann (GGS) distribution
Q ( d ) = const × ( d / d ')
n

n >1 represents samples with increasing coarse fractions, and


n < 1 represents samples with decreasing coarse fractions.
Q(d)
n<1

n =1

n>1

d
Chee 3920: Particle Size and Shape 45
• Relationship between frequency and
cumulative distributions
d d min

P(d ) = ∫ f ( x ) dx ; Q ( d ) = ∫ f ( x ) dx .
d max d

Differential relationships:
dP
= f ( d ) => f(d) is also called differential frequency distribution!
dd
dQ
= − f (d )
dd

Chee 3920: Particle Size and Shape 46


• Comparison of number, volume & surface
distributions
Many instruments measure number distribution but we want
surface area or volume distribution

(M Rhode, 1998)

Chee 3920: Particle Size and Shape 47


Conversions

Surface distribution: f s ( d ) = k S d f N ( d )
2

Volume (mass) distribution: f v ( d ) = kv d f N ( d )


3

And the condition of normalisation:


∫ f ( d ) dd = 1
0

We also have to assume constant shape and


density with size.

Chee 3920: Particle Size and Shape 48