Instrumentation

A Guide to Characterizing
Particle Size and Shape
Remi Trottier Knowledge of particle properties is essential for
Shrikant Dhodapkar
Dow Chemical Co. understanding how the particles will impact your
process. Learn how to identify the relevant properties
for a particular application, select the best
measurement technique, and analyze the data
to extract meaningful information.

P
article science and technology is relevant to a broad The many ways to define particle size
range of chemical processes, including those for The size of a spherical particle is unambiguously defined
manufacturing paints and coatings, pharmaceuticals, by its diameter. This diameter, which can be considered the
catalysts, fertilizers, food products, and cosmetics. For particle’s characteristic dimension, can be directly measured.
processes that have particles as a raw material, intermedi- However, particles encountered in most industrial processes
ate, or final product, the ability to measure, monitor, and are rarely spheres, and very rarely of the same size.
characterize fundamental particle properties, such as size, For particles that deviate from the spherical shape, rel-
size distribution, shape, density, and surface texture, is evant characteristic dimensions (e.g., length, width, diameter)
essential. or derived dimensions (e.g., aspect ratio) that relate to a
A particle can be defined as a solid with well-defined size-dependent property must be identified. These so-called
boundaries; its linear dimensions can range from nano- equivalent spherical diameters are often used to translate a
meters (colloids) to millions of meters (extraterrestrial certain property (e.g., surface area, volume, perimeter) of an
objects). Whether for an industrial process or a natural irregular particle to a spherical dimension (Figure 1, Table 1).
process, measuring particle properties (e.g., size and shape)
is a means to an end — the ultimate objective is to under- Sphere with Same
stand the effect of morphological parameters on particle Maximum Length Sphere with Same
Projected Area
behavior. Particle properties can affect absorption, agglom-
eration, aggregation, bioavailability, compressibility, com- Sphere with Same
bustibility, entrainment, fluid-particle interactions, packing, Surface Area
particle trajectories, permeability, reactivity, segregation,
separation, settling, and toxicity.
This article reviews different ways to define particle
size and explains how to decide which size variable is
appropriate for a particular process and objective. It Sphere Passing
Through Same
describes techniques for measuring particle size and pro- Sieve Aperture
vides key advantages and disadvantages of each. Finally, Sphere with
Same Volume
it discusses particle shape, as it is difficult to separate the
effect of particle size and particle shape on the macro- p Figure 1. Most particles of interest in industrial processes are not spherical.
scopic response of a complex particulate process. Equivalent diameters can be defined to translate a measured parameter of the
particle (e.g., surface area, volume, perimeter) to a spherical dimension.

36  www.aiche.org/cep  July 2014  CEP Copyright © 2014 American Institute of Chemical Engineers (AIChE)
 Each derived equivalent diameter represents a mechanism size of that interval) is not constant across the entire size dis-
or characteristic relevant to the process of interest. Therefore, tribution. For example, if the particles are classified into ten
it is critically important to understand the relevance of these intervals from 0 μm to 10 μm with each interval 1 μm wide,
definitions, and select wisely. the resolution of the first interval (0−1 μm) is 2, while for the
last interval (9−10 μm) the resolution is 0.1.
Particle size data The data can also be organized on a geometric basis,
If every particle in the population were identical in size with the interval widths increasing by the same factor
and shape, only one particle would need to be analyzed. (e.g., 1−2 μm, 2−4 μm, 4−8 μm, 8−16 μm). Grouping the
However, that is extremely rare, so to get a full, accurate, particles in this way maintains a constant resolution over
and statistically representative profile of the size or shape the entire distribution.
distribution of a sample, a large number of particles, often Table 2 summarizes a population of particles whose size
as many as hundreds of thousands, must be analyzed. The intervals are a geometric progression increasing by a factor
data are then typically sorted into a series of successive size of 1.5. The fourth column shows the number of particles
intervals characterized by the number of particles, surface in each interval. The remaining columns show the propor-
area, or mass of each interval. The entire size range, which tion of the total population in each interval, which can be
can span up to several orders of magnitude, can be covered expressed in terms of the number of particles, surface area,
with a relatively small number of intervals. volume, or any other basis upon which data are acquired.
For samples with a narrow size distribution, it may be Once the data have been collected and grouped into
appropriate to group the data into linear intervals, such as the appropriate size intervals, they are usually tabulated
0−1 μm, 1−2 μm, 2−3 μm, etc., and express the contents of and then transformed into a graphical representation. The
each interval as a percentage of the whole population. One size data can be plotted as a differential frequency plot (the
disadvantage of grouping into linear intervals is that the percentage of the total relevant quantity, e.g., number of
resolution (i.e., the ratio of the interval width to the mean particles, surface area, or mass, in each interval) or a cumu-
lative frequency plot, as illustrated in Figure 2.
Table 1. Equivalent diameters can be used to characterize Differential frequency plots allow for the direct com-
the size of nonspherical particles.
parison of different distributions, provided the intervals
Symbol Name Definition of the distributions are identical. Cumulative distributions
dv Volume Diameter Diameter of a sphere with the are useful for comparing several distributions that have
same volume as the particle intervals of different widths.
ds Surface Area Diameter of a sphere with the
Diameter same external surface area as What does “mean diameter” mean?
the particle
Size distributions can be reduced to a single average
dsv Surface-Area-to- Diameter of a sphere with the diameter, such as the mean, median, or mode. Distribution
Volume Diameter same external surface-area-to- averages are often a source of confusion when data from
volume ratio as the particle
different instruments that may have calculated the aver-
DSt Stokes Diameter Diameter of a sphere with the age diameter differently are being compared. An average
same terminal velocity and density
as the particle (in laminar flow)
particle diameter is meaningless without specifying how it
was calculated.
da Projected Area Diameter of a circle with the same
Diameter area as the projected area of the
The mean diameter is a calculated statistic that rep-
particle (in random orientation) resents the size of the entire particle population. Several
definitions of the mean are commonly used for various
dc Perimeter Diameter of a circle with the same
Diameter perimeter as the projected outline types of comparisons. The most appropriate mean diameter
of the particle for a particular application is the one that corresponds most
dA Sieve Diameter Width of the minimum square closely to the relevant property of the particle system.
aperture through which the The simplest mean is the arithmetic mean, which is the
particle will pass sum of all of the diameters in the population divided by the
dF Feret Diameter Distance between a pair of total number of particles:
parallel lines tangent to the ∑ ni di
projected outline of the particle dn =
∑ ni
(1)
dM Martin’s Length of the chord dividing the
Diameter projected outline of the particle in where ni is the number of particles in group i and di is the
two equal halves
midpoint diameter of the particles in group i. This diameter

Copyright © 2014 American Institute of Chemical Engineers (AIChE) CEP  July 2014  www.aiche.org/cep  37
Instrumentation

Table 2. Particle size data should be grouped into appropriate intervals.

The sample data shown here are grouped into geometric intervals with widths that increase by a factor of 1.5.
Diameter Particles Number Basis Surface-Area Basis Volume Basis
Min Max Mid Count Diff % Cum % Diff % Cum % Diff % Cum %
1.00 1.50 1.25 22 0.044 0.044 0.00 0.00 0.00 0.00
1.50 2.25 1.88 710 1.42 1.464 0.03 0.03 0.00 0.00
2.25 3.38 2.81 2,520 5.04 6.504 0.27 0.31 0.03 0.04
3.38 5.06 4.22 5,500 11 17.504 1.33 1.63 0.26 0.29
5.06 7.59 6.33 10,000 20 37.504 5.44 7.07 1.58 1.87
7.59 11.4 9.49 16,000 32 69.504 19.57 26.65 8.51 10.38
11.4 17.1 14.2 11,000 22 91.504 30.28 56.92 19.74 30.12
17.1 25.6 21.4 3,000 6 97.504 18.58 75.50 18.17 48.29
25.6 38.4 32.0 1,000 2 99.504 13.94 89.44 20.44 68.73
38.4 57.7 48.1 2,00 0.4 99.904 6.27 95.71 13.80 82.53
57.7 86.5 72.1 40 0.08 99.984 2.82 98.53 9.31 91.85
86.5 130 108 7 0.014 99.998 1.11 99.64 5.50 97.35
130 195 162 1 0.002 100 0.36 100.00 2.65 100.00

is commonly referred to as D[1,0] because the diameter of the particle volumes or masses within the population:
term in the numerator of the equation is of the power 1 and 1/3
the diameter term in the denominator is of the power 0. ∑ ni di3 ⎞ ⎛
The mean surface-area diameter and volume (or mass) d m = dv = ⎜ ⎟ (3)
⎝ ∑ ni ⎠
diameter reflect the fact that surface area is proportional
to the square of the diameter and the volume and mass are It is not unusual for a population with a wide size distri-
proportional to the cube of the diameter. bution to have a mean mass diameter that is one or two
The mean surface-area diameter is used when the par- orders of magnitude larger than the arithmetic mean diam-
ticle surface behavior is important. It is referred to as D[2,0] eter based on the number of particles. In any polydisperse
and is the arithmetic mean of the particle surface areas: system, the diameters always follow this relationship:
1/ 2
∑ni di2 ⎞ ⎛ d n < ds < dv ( 4)
ds = ⎜ ⎟ ( 2)
⎝ ∑ ni ⎠
Equations 1–3 are based on the number of particles
Similarly, the mean volume diameter (or mean mass being measured, and therefore are limited to particle-sizing
diameter) is used when the volume (or mass) of the particles techniques that provide information on individual particles.
is important. Referred to as D[3,0], it is the mean diameter The most common techniques measure particle parameters
40 40

Surface Area
Cumulative Frequency, %

Number
Differential Frequency, %

ber

rea
Num

ce A

20 20
e

Volume
um
Surfa

Vol

0 0
1 10 100 1,000 1 10 100 1,000
Particle Diameter, μm Particle Diameter, μm

p Figure 2. The particle size data of Table 2 can be plotted in terms of differential frequency (left) and cumulative frequency (right).

38  www.aiche.org/cep  July 2014  CEP Copyright © 2014 American Institute of Chemical Engineers (AIChE)
 40
related to bulk quantities, such as surface area, volume, or
mass, and do not provide information on the number of par- Number
ticles. Thus, other ways to calculate averages must be used.

Differential Frequency, %
The two most important means calculated from bulk
Volume
quantities are the surface-to-volume mean (i.e., the Sauter Surface Area
20
mean diameter, D[3,2]), and the volume mean diameter
(i.e., the De Broukere mean diameter, D[4,3]).
The Sauter mean diameter is:
∑ si di ∑ ni di
3

dsv =
S
=
∑ ni di2
(5)
0
1 10 100 1,000 10,000
where si is the total surface area of the particles in group i Particle Diameter, μm
(si = niπdi2) and S is the total surface area of the entire popula- p Figure 3. It is important to choose the appropriate basis (e.g., number,
tion. The Sauter mean diameter can be defined as the diameter surface area, volume) for measuring particle size. Adding a single 820-μm
of a sphere having the same surface-area-to-volume ratio as particle to the population in Table 2 skews the volume-based data but does not
the entire particle population. affect the number-based data.
In the calculation of the mean, most particle-size ana- However, the surface-area-based particle-size distribu-
lyzers assume that the particles are smooth spheres, which tion, as well as the volume-based curve (Figure 2), indicate
may lead to significant errors for populations of highly that 100% of the particles have diameters larger than
irregular particles. In such cases, the Brunauer-Emmett- 2 µm — which cannot be correct since there is a foul-
Tellet (BET) gas adsorption technique, which estimates ing problem. Instruments that measure particle size on a
the total surface area from the number of gas molecules number basis are needed to detect those smaller particles.
required to cover the surface of the particles, can be used to As a general rule of thumb, when a small percentage of
measure the surface area; this surface area can then be used small particles in a distribution is of interest, single-particle
to calculate the Sauter mean diameter: counting methods should be used.
When a number-based distribution is needed, laser dif-
⎛6 ⎞M
fraction and dynamic light scattering should not be used.
dsv = ⎜
ρ
⎟ (6) They should only be used for surface-area- or volume
⎠S
⎜ ⎟

⎝ p
(mass)-based measurements.
where ρp is the density of the particles, M is the total mass, Consider another example in which one 820-µm par-
and S is the surface area measured by the BET technique. ticle is added to the population in Table 2, which is now
The De Brouckere diameter (also referred to as the represented by Figure 3. This one particle accounts for
mass mean diameter) is: roughly 30% of the volume distribution of the population,
but does not effect the number-based distribution.
∑ mi di ∑ ni di
4
Surface-area-based measurements should be consid-
d mm =
M
=
∑ ni di3
(7 ) ered when the surface area of the particulate system plays
an important role, such as in catalysis, fluidization, dust
where mi is the total mass of the particles in group i and explosion hazard assessment, or droplet formation. In cases
M is the total mass of the entire population. where the particle shapes are very irregular, the BET tech-
nique may be required to accurately determine the surface
Instrument selection area of the particles.
To obtain the information required to calculate these The most common size-measurement basis in the
mean diameters, the proper instrument must be selected. chemical process industries is mass (or volume). For
When choosing an instrument, consider the type of data example, a wastewater treatment plant is subject to regula-
needed, including the required accuracy and precision. tions that limit the total mass of suspended solids in the
Consider a fouling problem in a heat exchanger han- effluent water, so a distribution or average based on mass
dling the stream characterized by Table 2. Since small- or volume is the relevant quantity of interest.
particle behavior tends to be dominated by surface forces,
fouling is typically associated with small particles sticking Particle-sizing instruments
to the surface of the heat exchanger. In this example, the With the exception of imaging technologies, every par-
experimental data show that only the particles with a diam- ticle characterization technology provides the measurement
eter of <2 µm are causing the fouling. of an equivalent spherical diameter. This equivalent spheri-

Copyright © 2014 American Institute of Chemical Engineers (AIChE) CEP  July 2014  www.aiche.org/cep  39
Instrumentation

cal diameter is deduced indirectly from the behavior of the scattered at the same angle to converge at specific locations
particles as they pass through restricted volumes or channels on a series of photodetectors. The detectors are strategi-
under the influence of gravity or centrifugal force fields, or cally positioned to reveal the angular diffraction pattern,
from interactions with some form of radiation or ultrasonic which corresponds to the size of the particles (larger par-
waves (1, 2). Therefore, data for irregular-shaped particles ticles diffract light at low angles, smaller particles at higher
obtained from different instruments are typically not in good angles). While the position of the diffraction pattern pro-
agreement. Table 3 summarizes some of the advantages and vides information on the size of the particles, the intensity
disadvantages of each of the technologies described in the of the diffracted light relates to the total volume of particles
following sections. of a particular size. A mathematical model then converts
Particle characterization technologies in use today fall the diffraction pattern into a particle-size distribution.
into one of three categories: ensemble techniques, fraction- The mathematical models are based on the Fraunhofer
ation techniques, and single-particle counting techniques. and the Mie theories. The Fraunhofer model is a simple
approximation that accounts for the diffraction phenom-
Ensemble techniques enon, but ignores reflection, transmission, and refraction,
Ensemble techniques, which include laser diffraction, which may become significant for small particles (<20 µm).
dynamic light scattering, and ultrasonic spectroscopy, can The more-sophisticated Mie scattering model takes into
measure large numbers of particles simultaneously. account all light-particle interactions. To use the Mie model,
Laser diffraction. These instruments became commer- the refractive index of the particles and of the suspension
cially available in the 1970s, and soon were considered the medium must be known. The Fraunhofer approximation is
workhorse particle analyzer for industrial applications. typically used when these optical parameters are not known
As illustrated in Figure 4, light from a laser source is and lower accuracy is acceptable.
collimated (i.e., its rays are made parrallel) and transmitted Over the years, instrument manufacturers have sought
through the particle dispersion. As the light passes through to provide a wider analytical range, the ability to switch
the dispersion, it is diffracted by the particles. The dif- from liquid dispersions to dry-powder dispersions, and
fracted light passes through a Fourier lens that causes rays the ability to measure high-concentration dispersions. The

Table 3. Each type of particle-sizing instrument has advantages and limitations.

Analysis Capital Skill
Technique Range, µm Basis Time Precision Accuracy Resolution Throughput Cost Level
Ensemble Techniques
Laser Diffraction 0.04–2,000 Mass Short High High Low High High Low
Dynamic Light 0.003–1 Intensity Short High High Low High Medium Low
Scattering
Ultrasonic 0.01–1,000 Volume Short High High Low High High Low
Spectroscopy
Fractionation Techniques
Wet Sieving >5 Mass Medium High High Low Medium Low Low
Dry Sieving >45 Mass Medium High High Low Medium Low Low
Sedimentation 0.05–100 Mass Medium High High High Medium Medium Medium

Field-Flow 0.003–3 Mass Medium High High High Medium Medium High
Fractionation
Hydrodynamic 0.02–1 Mass Medium High High High Medium Medium High
Chromatography (HDC)
and Capillary HDC

Single-Particle Counting Techniques

Dynamic Image >5 Number Short High High High High Medium Low
Analysis
Optical Particle 0.2–2,500 Number Short High High High High Medium Medium
Counting
Electrozone Counter 0.4–1,200 Number Short High High High High Medium Medium

40  www.aiche.org/cep  July 2014  CEP Copyright © 2014 American Institute of Chemical Engineers (AIChE)
 Particles Multi-Element Focusing Lens
Detector Scattered
Laser Light

Laser
Light Source
Photon
Particle Detector
Dispersion
Obscuration
Detector
Collimator
Fourier Lens

Intensity

Intensity
p Figure 4. In laser diffraction, light from a laser source is collimated and
transmitted through the particle dispersion. Light diffracted by the particles Diameter Time

passes through a Fourier lens and is collected on a series of photodetectors. Particle Size Autocorrelator
The Fourier lens causes all rays scattered at the same angle to converge at Distribution
specific locations on the detection plane.
p Figure 5. Dynamic light scattering also uses a laser to illuminate particles
in a suspension. With this technique, the size of the particles is deduced from
combination of the Fraunhofer model with Mie theory has the random motions of the particles undergoing Browning diffusion.
expanded the analytical range to roughly 0.04 µm (some
vendors claim as low as 0.02 µm) to several millimeters, extract the particle-size distribution. The measurement basis
although the reliability at both ends of the range is some- is intensity weighted, which makes the measurement more
what questionable. Even with the combination of these two sensitive to the larger particles in the population.
models, the validity of any multimodal distribution with This type of instrument is ideal for quality control
particle sizes below 2 µm should be carefully verified. applications involving unimodal colloidal dispersions. The
Several vendors offer dry-powder dispersion instruments analytical range is about 3 nm–1 µm. The lower size limit
capable of dispersing particles as small as 10 µm (and in is reached as the intensity of scattered light gets below the
some cases, smaller). It is good practice to validate measure- detection limit, and the upper size limit is reached when the
ments based on dry dispersion by comparing them to data particles start to settle.
obtained by the more-traditional wet-dispersion method. Ultrasonic spectroscopy. This technique uses ultrasound
Some laser-diffraction instruments designed for high- to probe particles that are homogeneously suspended in a
concentration measurements have been used for online liquid, thereby exploiting the inherent advantage that sound
applications. Online applications should be thoroughly waves (unlike light waves) can propagate through opaque
validated. It is also important to note that the measurement dispersions. It can therefore handle suspensions with much
basis for laser diffraction is mass (or volume) and conver- higher particle concentrations than optical techniques.
sions to number-based distributions should never be used In ultrasonic spectroscopy, sound waves are passed
even when offered by instrument vendor software. through a suspension. As they travel through the sample and
Dynamic light scattering. The early particle-size instru- interact with the particles, the sound waves lose energy via
ments from the 1960s and 70s had limited capabilities and scattering and absorption. Mathematical models based on
required expert operators. It took almost two decades of fundamental theory convert this energy loss (or attenuation)
technical improvements, including major advances in digi- as a function of sound frequency into a size distribution.
tal autocorrelators, microprocessors, and lasers, to bring A key advantage of ultrasonic spectroscopy over optical
dynamic light-scattering technology to a viable existence in particle analysis is its ability to analyze high-concentration
the marketplace. Today, dynamic light-scattering instru- samples. However, it is generally more difficult to imple-
ments require a low skill level for operation. ment because, for both phases, numerous physical con-
In dynamic light scattering, the size distribution of stants — speed of sound, density, thermal coefficient of
sub-micrometer particles dispersed in a liquid medium is expansion, heat capacity, thermal conductivity, viscosity of
deduced from the random movements of particles under­ the fluid phase, shear rigidity of the solid phase — must be
going Brownian diffusion. Unlike laser diffraction, where known for accurate size determination. Thus, this method
the static intensity of light is measured, a dynamic light- is best suited for online applications involving the analysis
scattering instrument records the fluctuations in scattered of high-concentration samples where other technologies
light and processes these data using an autocorrelator, as are not practical.
illustrated in Figure 5. The smaller particles have a higher The analytical range of ultrasonic spectroscopy is
diffusion constant and generate a higher-frequency signal 0.01–1,000 μm, with concentrations ranging from a few
than the lower-frequency signal of the larger particles. The volume percent to 80 vol% for emulsions. The concentra-
autocorrelator analyzes the periodicity of these signals to tion limits are highly dependent on the material.
Article continues on next page

Copyright © 2014 American Institute of Chemical Engineers (AIChE) CEP  July 2014  www.aiche.org/cep  41
Instrumentation

Fractionation techniques just passes (i.e., the particle gets caught by the next sieve,
Fractionation methods — which include sieving, sedi- which has a smaller aperture). Since irregular particles tend
mentation, field-flow fractionation, hydrodynamic chroma­ to orient themselves along their smallest dimension to pass
tography, and capillary hydrodynamic chromatography through a two-dimesional aperture, the sieve diameter is
— separate particles by size prior to detection and measure- typically related to the second-smallest dimension of the
ment. Some fractionation techniques are based on physical particle. The data from sieving are mass-based measure-
phenomena related to first principles and do not require ments — i.e., the mass of the material retained on each
calibration (others are not and do need calibration). Both the sieve is measured to obtain a mass fraction, which is then
analytical range and the resolution of these methods are gov- presented in tabular or graphical form.
erned by the efficiency of the separation process they utilize. Sedimentation. This mass-based method measures the set-
Sieving. This technique, which has become the workhorse tling rate of particles dispersed in a liquid that is subjected to
method for analyzing coarse particles, is straightforward and gravitational or centrifugal forces. At the start of the measure-
the equipment is simple. There are two types of sieving — ment, the particles may be uniformly distributed through-
dry sieving and wet sieving — both of which filter particles out the liquid (homogeneous start) or concentrated within
through a series of sieves of different mesh sizes. Dry sieving a narrow band at the liquid’s surface (line start). Particle
is typically carried out by shaking a stack of sieves using a movement can be monitored using optical or X-ray detectors.
mechanical vibrator, air pulses of sonic frequency, or rotating Regardless of the system configuration (line/homogeneous
air jets. Dry sieving works well for particles with diameters start, gravitational/centrifugal force, optical/X-ray detector),
larger than about 45 µm. Wet sieving, with water or another the Stokes equation applies to all sedimentation equipment:
liquid facilitating particle passage through the sieves, is used
for finer powders, down to roughly 5 µm. v=
(
gd 2 ρp − ρl ) (8 )
The equivalent diameter obtained from this method 18η
is the sieve diameter, which is defined as the size of the where ν is the terminal velocity of a particle, g is the
sieve aperture through which the particle being measured gravitational constant, d is the particle diameter, ρp is the
particle density, ρl is the liquid density, and η is the liquid
viscosity. For centrifugal sedimentation, the constant g is
Nomenclature replaced with the actual acceleration of the system.
Sedimentation instruments have high resolution
A = projected area of the particle’s outline
d = mean diameter
because the separation technique is based on terminal
di = midpoint diameter of the particles in group i velocity, which is proportional to the square of the particle
dmm = De Brouckere diameter diameter. However, this high resolution comes with a dis-
ds = surface-area equivalent diameter advantage: obtaining results for a sample with a wide size
dS = diameter of a sphere circumscribing the particle distribution generally takes a long time, thereby reducing
dsv = surface-area-to-volume equivalent diameter the throughput of the technique.
dv = volume equivalent diameter Field-flow fractionation (FFF). Although field-flow
g = gravitational constant
mi = total mass of particles in group i fractionation methods were first described in 1967, the
M = total mass of the entire population of particles technology transfer from an academic laboratory technique
ni = number of particles in group i to a viable commercial instrument took more than 20 years,
P = perimeter of a particle’s outline after which rapid proliferation took place.
r = radius of curvature of a corner of a particle’s profile The term field-flow fractionation refers to a family of
R = maximum inscribed radius of a particle mass-based techniques in which particles dispersed in a liq-
si = total surface area of particles in group i
S = total surface area of the entire particle population Applied Force
v = terminal velocity of a particle
V = volume of a particle Particle
Dispersion
Greek Letters Flow
αs,a = surface-area-projected-area shape coefficient Diffusion
αv,a = volume-area shape coefficient
η = viscosity of the liquid p Figure 6. Field-flow fractionation subjects particles dispersed in a flow-
ρp = density of the particles ing liquid to forces perpendicular to the flow. The applied force causes the
ρl = density of the liquid particles to move toward the bottom of the channel, while the diffusion of
ϕsf = circularity shape factor the particles causes them to move against the applied force. Since a para-
Ψ = sphericity bolic flow profile exists in the channel, the rate at which the particles travel
through the channel depends on their position, and therefore their size.

42  www.aiche.org/cep  July 2014  CEP Copyright © 2014 American Institute of Chemical Engineers (AIChE)
uid flow through a narrow channel that is subjected to forces count particles one at a time. These methods are not based
perpendicular to the flow (Figure 6). The applied forces on first principles and therefore require calibration.
(e.g., cross-flow stream, temperature gradient, centrifugal Particle counters offer the ultimate in resolution, but
force) cause the particles to move toward the bottom of the can suffer from poor statistical accuracy when the data are
channel, where diffusion opposes the applied forces. The converted from a number-based distribution to a mass- or
opposing forces confine particles of different sizes to specific volume-based one, especially when the size range is larger
regions of the channel — smaller particles have higher dif- than two orders of magnitude. Because of the inherently
fusion coefficients and thus flow near the center of the chan- narrow analytical range of these techniques, most count-
nel, while larger particles, with lower diffusion coefficients, ing devices come equipped with several (typically two
travel near the edges. Since a parabolic flow profile exists in or three) distinct sensors or sets of magnifying optics to
the channel, particles of different sizes are transported at dif- increase their analytical ranges.
ferent rates, allowing them to be separated according to size. Dynamic image analysis. The pioneering studies of
These techniques provide a high-resolution separation over particle characterization by imaging technologies were
the size range of 3 nm–3 μm. carried out in the late 1960s and early 1970s. Increasing data-
Hydrodynamic chromatography (HDC) and capillary processing power coupled with the high performance and
hydrodynamic chromatography (CHDC). In both HDC and falling costs of video cameras and high-speed frame grabbers
CHDC (Figure 7), the particle-containing liquid travels have enabled the development of sophisticated and powerful
through microchannels; in HDC, the microchannels are in image processing and analysis systems. Modern instruments
the interstices between beads packed in a column, while are capable of acquiring tens or even hundreds of frames per
CHDC utilizes straight microchannels. Under laminar condi- second with simultaneous image processing to extract size
tions, a parabolic flow profile arises, in which the smaller and shape information. This new breed of instruments has
particles travel closer to the walls in the lower-velocity zones gained considerable popularity within the past decade.
and the larger particles travel farther away from the walls at As illustrated in Figure 8, particles flow through the
higher velocities. HDC and CHDC provide a high-resolution sensing zone, where they are illuminated by a light source
separation over the analytical size range of 0.02–1 μm. (e.g., strobe, laser, or incandescent), magnified, and then
The first CHDC instruments became commercially avail- detected by a charge-coupled device (CCD) camera. The
able around 1990, about 30 years after the discovery of this particles are dynamically presented to the instrument — as
fractionation technique. Meanwhile, HDC was developed dry particles carried in a gas stream or falling from a vibra-
in an industrial laboratory to fulfill a sub-micron analytical tory feeder, or as a liquid dispersion moving through thin
need; although it provides high-quality sizing data, HDC has rectangular glass cells.
seen limited success in the market, mainly because a high Dynamic image analysis systems are normally used for
skill level is required to operate the instrument and the packed particles larger than roughly 5 μm in dia. Several manufactur-
column has a tendency to retain particles and plug over time. ers have exaggerated claims regarding the lower limit of the
analytical range for these instruments. The smallest particle
Single-particle counting methods measured should be represented by at least 7–10 pixels —
Particle-counting instruments (also referred to as not a single pixel, which would imply a square or rectangular
stream-counting methods), including dynamic image analy- particle. The measurement basis is the number of particles,
sis, optical counters, and electrozone counters, detect and but the data can be converted to a volume-based distribution.
Capillary Hydrodynamic Hydrodynamic Optical particle counting. The first optical particle
Chromatography Chromatography

deff
Zoom Lens
v
h
Strobe Light
Flow
vmax

Flow

Sample Flow
p Figure 7. In hydrodynamic chromatography and capillary hydrodynamic Camera
chromatography, a particle-containing liquid travels through microchannels
— in the interstices between beads packed in a column in HDC, and through p Figure 8. Dynamic image analyzers consist of a light source to illuminate
straight capillary tubes in CHDC. A parabolic flow profile forms in which the particles in the sensing zone, a zoom lens, a camera to capture images of
particle velocity is a function of particle size. the particles, and a processor to quantify the size and/or shape of the particles.

Copyright © 2014 American Institute of Chemical Engineers (AIChE) CEP  July 2014  www.aiche.org/cep  43
Instrumentation

counters were developed in the late 1950s and early 1960s

to measure the size distribution of aerosols. Instruments for
liquid-based systems were subsequently developed and used
extensively to measure the cleanliness of hydraulic fluids.
Optical particle counters (Figure 9) use the principle
Anode
of light blockage to count particles and measure their size. Cathode
As a particle passes through the illuminated sensing zone,
it casts a shadow onto the photodetector, which converts
Aperture
the shadow into an electrical signal, or pulse. The photo­
detector sends this pulse to an analyzer, where the height
of the pulse, which is proportional to the cross-section p Figure 10. In electrozone counters, particles suspended in an electrolyte
of the particle, is converted into an equivalent diameter, pass through an aperture positioned between two electrodes. The change in
the resistance of the electrolyte is proportional to particle volume.
referred to as the projected-area diameter (da).
The analytical range of optical particle counting is Particle shape
roughly 0.2–2,500 μm, depending on the manufacturer; Irregularly shaped particles are much more common than
two or three sensors, each with a limited analytical range, spherical particles. Two irregular (nonspherical) particles
are required to cover the full analytical range. The number- with the same equivalent spherical diameter can have very
based measurement can be converted to a volume-based different shapes. British Standard BS 2955 (3) includes a
distribution. list of terms related to particle shape, some of which are
Electrozone counters. The first electrozone counter was shown in Table 4. While these terms may be useful descrip-
introduced by Wallace Coulter in 1954 to count blood cells. tors, their lack of quantification does not lend them to use in
Since then, this technology has found a wide range of applica- mathematical equations.
tions, and after more than half a century, the Coulter counter is Shape factors and shape coefficients provide quantita-
still the preferred technique for obtaining blood counts. tive information about particle shape and can serve many
Particles suspended in an electrolyte are drawn through purposes, such as:
an orifice situated between two electrodes (Figure 10), • relating various definitions of derived particle-size
altering the electrical resistance of the electrolyte. The measurements
change in the resistance experienced by the electrolyte is • relating particle volume or surface area to a character-
proportional to the volume of the particle (i.e., the volume istic dimension (e.g., length, diameter, etc.)
of electrolyte displaced). These instruments measure the • generating a 2D outline of the original particle.
volume equivalent diameter (dv) and provide a number- The shape of a large particle can be physically measured
based measurement that can be converted to a volume- and expressed as linear dimensions on orthogonal axes. For
based distribution. They have an analytical range of example, geologists routinely take direct measurements to
0.4–1,200 μm using multiple sensors. define the shapes of rocks and pebbles. However, the study of
microscopic particles must rely on silhouettes or outlines. The
Particle
Flow Table 4. These adjectives describe particle shape
Laser
in qualitative terms.
Sensing
Zone Term Description
Acicular Needle-shaped
Angular Sharp-edged or having roughly polyhedral shape
Dendritic Branched crystalline shape
Fibrous Thread-like
Granular Approximately equidimensional but irregular
Lamellar Plate-like
Particle Diameter Pulse Height Crystalline Freely developed in a medium of geometric shape
Distribution Distribution
Spherical Globular-shaped
p Figure 9. Optical particle counters measure a voltage drop propor-
tional to the cross-section of the particle. As particles pass through the Irregular Lacking any symmetry
illuminated sensing zone, they cast shadows onto the photodetector and Modular Having rounded irregular shape
reduce the voltage of the detector. The photodetector sends the pulse to an
Source: (3).
analyzer, which converts the height of the pulse into an equivalent diameter.

44  www.aiche.org/cep  July 2014  CEP Copyright © 2014 American Institute of Chemical Engineers (AIChE)
orientation of the particles is usually random, so particles of in terms of ratios of length (L), breadth (B), and thickness (T).
identical form often have different outlines. Thus, it is neces- As shown in Figure 11, T is the minimum distance between
sary to use statistics to obtain a representative characterization two parallel planes that are tangential to the opposite surfaces
of particle shape. of the particle when the particle is in its most stable orienta-
tion; B is the minimum distance between two parallel planes
Common shape factors that are tangential to the particle surface and perpendicular
References 4–6 provide excellent reviews on shape mea- to the planes defining particle thickness; and L is the distance
surement. However, clear guidance on the selection of suitable between two parallel planes that are perpendicular to the
shape factors for various applications is not available. planes defining T and B.
The underlying physics of the macroscopic behavior of These three dimensions can be expressed as ratios:
interest dictates the appropriate morphological properties for • elongation ratio = L/B (ranges from 1 to infinity)
consideration. Before selecting a shape factor, ask the follow- • flakiness ratio = B/T (ranges from 1 to infinity)
ing questions to narrow the choices: • chunkiness ratio = B/L (ranges from 0 to 1).
• Is the shape factor independent of size, magnification, For a 2D particle outline, the aspect ratio (AR) may be
orientation, and rotation? defined in terms of the minimum distance (Fmin) and the
• How does the measurement method affect the shape maximum distance (Fmax) between pairs of tangents to the
factor or shape coefficient? particle profile: AR = Fmin/Fmax (ranges from 0 to 1).
• Does the resolution of the instrument affect the Sphericity. This property indicates how closely the
measurement, especially at the ends of the particle-size particle resembles a sphere. Wadell defined sphericity (Ψ) as
distribution? the surface area of a sphere having the same volume as the
• Is the measurement based on a 3D measurement or a particle divided by the actual surface area of the particle.
2D profile or outline? Owing to the difficulty in measuring the surface area
• Is the shape factor relevant to the macroscopic prop- of an irregular particle, Wadell proposed an approxima-
erty under investigation? tion, in which Ψ is the square of the diameter of a sphere
Broadly speaking, the quantification of particle shape having the same volume as the particle (dv) divided by the
falls into four categories (5): diameter of a circumscribing sphere (dS):
1. dimensional ratios 2
2. sphericity, which depicts form or overall shape ⎛d ⎞
3. roundness, or circularity, which indicates angularity or Ψ =⎜ v ⎟ ( 9)
⎝ dS ⎠
the sharpness of corners
4. roughness, which characterizes surface texture. Riley proposed a quicker method to estimate Ψ as the
Heywood (6) recognized that the definition of shape must square of the diameter of the largest inscribed circle divided
include its form and the relative proportions of its dimensions. by the diameter of the smallest circle circumscribing the par-
Form refers to the particle's resemblance to a defined geo­ ticle outline.
metry (e.g., sphere, cube, tetrahedron). He proposed the use of Circularity. Circularity (ϕ) is often used as the shape
an elongation ratio and a flakiness ratio as shape factors. factor for 2D planar profiles. The circularity of a particle
Dimensional ratios. The form of a particle can be defined can be defined as the circumference of a circle whose area
is equal to that of the particle's projected area divided by
Roundness the perimeter of the actual particle.
T The circularity shape factor (ϕsf) is defined as:
Sphericity
B 4πA
φ sf =
P2
(10)
where A is the projection area of the particle outline and P
is the perimeter of the particle outline. Note that the circu-
L
larity shape factor and circularity are different.
Roundness. Roundness is a measure of the angularity of
a particle profile, and an indicator of the presence of sharp
Roughness corners:
N
p Figure 11. The shape of a particle can be characterized in terms of ∑ i=1 ri R
sphericity, roundness, roughness, and dimensional ratios of length (L), Roundness =
N
(11)
breadth (B), and thickness (T).

Copyright © 2014 American Institute of Chemical Engineers (AIChE) CEP  July 2014  www.aiche.org/cep  45
Instrumentation

where ri is the radius of curvature of a corner, N is the total In the same way, the surface-area-projected-area shape
number of round corners in the particle profile, and R is the coefficient (αs,a) relates the surface-area diameter (ds ) to the
maximum inscribed radius of the particle. projected-area diameter:
Convexity. The convex hull perimeter of a particle can
be visualized as the length of an elastic rubber band that fits
S = πd a2 = α s,a d a2 (13)
around the particle. Convexity is the convex hull perimeter where S is the surface area of the particle.
divided by the perimeter of the particle excluding roughness.
It is a measure of the compactness of a particle profile, and is Closing thoughts
affected by surface irregularities, spikes, and concave regions Knowledge of particle size, size distribution, and shape
of the perimeter. Convexity is sometimes measured as the is critical for designing and operating complex particulate
projected area of the particle divided by the area of the convex processes. While the modern age of electronics and comput-
hull, which are easier to measure. ers made the measurement techniques more robust and less
Surface texture. The surface texture of the particle can onerous, the amount of data that can be generated has also
be quantified by rugosity, which is the perimeter of the par- proliferated. Such data can result in misleading conclusions
ticle outline including roughness divided by the perimeter unless you clearly understand the fundamentals. CEP

of a smooth curve circumscribing the particle profile.

Shape coefficients Literature Cited

Surface area and volume are related to the square and the 1. Allen, T., “Powder Sampling and Particle Size Determination,”
cube of the characteristic linear dimension of the particle, Elsevier, Philadelphia, PA (2003).
respectively. The shape coefficient expresses this relationship. 2. Dalla Valle, J. M., “Micromeritics — The Technology of Fine
The value of the shape coefficient depends on the Particles,” Pitman Publishing Corp., New York, NY (1943).
definition of particle size (e.g., projected-area diameter, 3. British Standards Institution, “Glossary of Terms Relating to
volume-equivalent diameter, Stokes diameter) and the bulk Particle Technology,” BS 2955, BSI, London, U.K. (1993).
property of interest (i.e., volume or surface area). The mea- 4. Hawkins, A. E., “The Shape of Powder-Particle Outlines,” John
Wiley and Sons, New York, NY (1993).
surement method must be reported with a shape coefficient.
5. Rodriguez, J. M., “Particle Shape Quantities and Measurement
For example, the volume-area shape coefficient (αv,a) Techniques — A Review,” Electronic Journal of Geotechnical
relates the volume diameter to the projected-area diameter: Engineering, 18 (A), pp. 169–198 (2013).
6. Singh, P., and P. Ramakrishnan, “Powder Characterization
π 3 by Particle Shape Assessment,” KONA Powder and Particle
V= d = α v,a d a3
6 v
(12) Journal, 14, pp. 16–30 (1996).
where V is the volume of the particle, dv is the volume-
equivalant diameter, and da is the projected-area diameter.
Additional Resources
Davies, R., “A Simple Feature-Space Representation of
Remi Trottier is a research scientist in the solids processing discipline of Particle Shape,” Powder Technology, 12 (2), pp. 111–124
engineering sciences at Dow Chemical (Email: [email protected]).
He has over 20 years of industrial experience in particle characteriza- (Sept–Oct 1975).
tion, aerosol science, air filtration, and solids processing technology. Gy, P., “Sampling Theory and Sampling Practice: Heterogeneity,
He has authored roughly 20 papers and has served as an instructor for Sampling Correctness, and Statistical Process Control,” 2nd ed.,
several short courses on particle characterization. He received BS and
CRC Press, Boca Raton, FL (1993).
MS degrees in applied physics from Laurentian Univ., Sudbury, ON,
Canada, and his PhD in chemical engineering from Loughborough Univ. Heywood, H., “Symposium on Particle Size Analysis,” Institution of
of Technology, U.K. Chemical Engineers, London, U.K. (Feb. 4, 1947).
Shrikant Dhodapkar is a Fellow in the Process Fundamentals Group, Jilavenkatesa, A., et al., “NIST Recommended Practice Guide:
Performance Plastics Process R&D at Dow Chemical (Email: Particle Size Characterization,” U.S. Government Printing Office,
[email protected]). He has been with Dow for 22 years and has Washington, DC (Jan. 2001).
worked closely with the elastomers business for the past 15 years
to develop, design, and implement technologies related to material Kaye, B., and R. A. Trottier, “The Many Measures of Fine Par-
handling. He is an expert in the field of solids processing and bulk ticles,” Chem. Eng., 102 (4), pp. 78–86 (Apr. 1995).
solids handling and has extensive industrial experience in powder Leschonski, K., “Sieve Analysis, the Cinderella of Particle Size
characterization, fluidization, pneumatic conveying, silo design, gas-
Analysis Methods?,” Powder Technology, 24 (2), pp. 115–124
solid separation, mixing, coating, computer modeling, and scaleup.
He received his BTech from Indian Institute of Technology in chemical (Nov–Dec 1979).
engineering, and an MS and PhD from the Univ. of Pittsburgh, both in McGlinchey, D., ed., “Characterisation of Bulk Solids,” Wiley-
chemical engineering. He is a senior member of AIChE and past chair Blackwell, Hoboken, NJ (2005).
of the Particle Technology Forum. He has authored or co-authored over
35 external publications and four chapters in handbooks. He is also an Trottier, R. A., and S. Dhodapkar, “Sampling Particulate Materials
adjunct professor of chemical engineering at the Univ. of Pittsburgh. the Right Way,” Chem. Eng., 119 (4), pp. 42–49 (Apr. 2012).

46  www.aiche.org/cep  July 2014  CEP Copyright © 2014 American Institute of Chemical Engineers (AIChE)