International Journal of Mineral Processing 94 (2010) 180191

Contents lists available at ScienceDirect

International Journal of Mineral Processing

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j m i n p r o

Effect of comminution on particle shape and surface roughness and their relation to
otation process
Mahmoud M. Ahmed
Mining and Metallurgical Engineering Department, Faculty of Engineering, Assiut University, Assiut 71516, Egypt

a r t i c l e

i n f o

Article history:
Received 23 July 2008
Received in revised form 8 February 2010
Accepted 22 February 2010
Available online 1 March 2010
Keywords:
Particle shape
Surface roughness
Fractal
Comminution
Flotation
Wettability

a b s t r a c t
The essential physical characteristics of the distinction of particles are size, shape and surface roughness.
Particle shape and surface roughness are considered important parameters in the prediction of the behavior
of particles individually or collectively. These parameters are of great importance to industries employing
various materials in a powder form. These parameters have not been currently reviewed satisfactory in the
mineral processing eld. Therefore, this research is concerned with the different methods used to estimate
the particle shape and surface roughness and also to relate these parameters with the behavior of some
mineral processing operations, especially comminution and otation processes.
The surface roughness of mineral particle inuences the fundamental processes of particlebubble
attachment and the other sub-processes in froth otation. The contact angle is dependent on the surface
roughness. The modication of the wettability due to surface roughness can be greatly enhanced in the
fractal surface; that is the fractal surface will be superrepellen (superwettable) to a liquid when the contact
angle is greater (less) than 90. Correlations were found between the shape properties, surface roughness
values and wettability.
The dry grinding produces relatively rough particle surfaces with a high concentration of microstructural
defects while the wet grinding produces smoother cleaner surfaces. The dry ground samples exhibited more
stable, higher loaded froths and faster otation kinetics.
2010 Elsevier B.V. All rights reserved.

1. Introduction
In addition to process variables, the processing of powder depends
to a large extent on the materials (purity, structure, density, etc.), bulk
properties, and the morphological characteristics (size, shape, texture,
etc.). Material characteristics are well understood and almost
completely known because the chemistry and physics of the materials
are highly developed. The morphological characteristics are not yet
developed and explored completely. The denitions of some
morphological characteristics like particle shape, size, and surface
roughness are still also evolving (Singh and Ramakrishnan, 1996).
The geometry of particle has a pronounced inuence on the
physical and chemical actions occurring on the particle surface
during the technological processes. This illustrates the importance of
the exact geometrical evaluation of the particles by means of
quantied particle parameters. These parameters can be used to
differentiate between different materials and to correlate them with
the different processes. The essential characteristics of the distinction of particles are size, shape and surface roughness (Castellini

Tel.: + 20 012 3971453; fax: +20 088 2332553.

E-mail address: mamoah@aun.edu.eg.
0301-7516/$ see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.minpro.2010.02.007

et al., 1993). Particle shape is recognized as an important parameter

in the prediction of the behavior of particles and powders inuencing
the manner, in which they break, react, sinter, oat, agglomerate and
uidize (Clark, 1986). The distribution of particle sizes and shapes is
of great importance to industries employing various materials in a
powder form (Hurter and Clark, 1987). These include many of the
raw materials in civil and chemical engineering, pharmaceutical,
mining industries, pigments, metals, ceramics, pills, foods, and
industries interested with population of atmospheric dusts, smoke
and grit (Ulusoy et al., 2003).
The rst procedure used for describing the shape of a ne particle
was carried out by measuring simple linear parameters such as
the length, breadth and width. The ratios of these dimensions are
used as coefcients to describe the shape dependent property
(Sarkar and Chaudhuri, 1994). The methods used for estimating
the particle characteristics range from very simple such as the
determination of aspect ratio, elongation ratio or circularity to very
modern methods such as the use of Fourier analysis, delta analysis
and fractal geometry (Meloy and Williams, 1994; Singh and
Ramakrishnan, 1996; Singh and Ramakrishnan, 1996). The development of automated microscope techniques linked to computer dataprocessing systems (image analysis techniques) facilitated the
possibility of specifying the shape and surface roughness of a prole

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

by mathematical procedures (Meloy and Williams, 1994). Development of these methods may give a way to optimize the mineral
processing ow sheets not only on the basis of particle size
distribution but also by introducing the particle fractal properties
(Burstein et al., 1995).
Numerous factors inuence otation. Although particle hydrophobicity and size are important and have received much attention in
research, particle shape and roughness play a signicant role also, but
they have received much less attention.
The interaction between a solid particle and a gasliquid interface
controls the efciency of the otation process and has been
considered in detail (Ahmed, 1999; Ahmed et al., 1999).
The bubbleparticle interaction involves many stages, i.e. the
before-contact and after-contact interactions (Scheludko et al., 1968/
1969). The before-contact bubbleparticle interaction is strongly
affected by the long-range hydrodynamics when the bubble and the
particle are far from each other. It is controlled by the interfacial
physics e.g. interfacial dynamics, capillarity and colloidal interaction if
they are close enough.
The particle shape has a signicant inuence on otation behavior.
Many researchers have studied this relation qualitatively (Varbanov,
1984; Blake and Ralston, 1985; Ducker et al., 1989; Schmidt and Berg,
1996; Schmidt and Berg, 1997).
A few researchers have studied the effect of the surface roughness
on the otation process (Anfruns and Kitchener, 1997; Silverstein and
Breuer, 1993; Burstein et al., 1995; Ahmed, 1999; Ahmed et al., 1999;
Stechemesser and Ahmed, 1999).
The surface roughness of mineral particle inuences the fundamental processes of particlebubble attachment and the other subprocesses in froth otation.
Ahmed et al. (1998); Ahmed (1999); Ahmed et al. (1999) and
Stechemesser and Ahmed (1999) illustrated a modern and useful
approach, which was considered the rst attempt, to study the
quantitative effect of the surface roughness and shape of particle on
the detachment process of particle from the airliquid-interface by
means of a centrifuge technique and also on particle oatability
(under ideal conditions) using a modied Hallimond tube. It can be
supposed that the roughness of the solid surface has a decisive
inuence on otation effectiveness (Burstein et al., 1995). The
higher the surface roughness, the stronger is the detachment force
required to separate the particle from the bubble and hence a higher
otation recovery is obtained (Ahmed, 1999 and Ahmed et al.,
1999).
In froth otation, the probability of otation and hence the rate of
otation is directly determined by three microprocesses: collision,
adhesion and detachment (Trahar and Warren, 1976). To obtain an
improved understanding of the complete otation process, it is
necessary to analyze these three microprocesses quantitatively. The
ease with which the particle can be attached to the bubble and the
strength of detachment are considered very important when the
effectiveness of otation process is studied.
The differences in the values of detachment forces and hence the
otation effectiveness were markedly clear at the different degrees of
hydrophobicity for the same surface roughness (Ahmed, 1999 and
Ahmed et al., 1999). In addition, the difference in the values of
detachment forces increases with the increasing of the surface
roughness. At the lowest roughness, the difference in the value of
detachment force between the low and high hydrophobicity does not
exceed 1.5 * 105 N. This difference reached 7.1 * 105 N at the
highest roughness. This behavior may be attributed to the chemical
effects caused by adsorption of the silane lms on the surfaces having
different roughness. A surface with a lower roughness has a smaller
surface area compared to the surface of higher roughness and
therefore it reaches a maximum surface coverage at a lower
methylation time (lower contact angle) (Ahmed, 1999; Ahmed et al.,
1999).

181

The effect of dry and wet grinding on the otation of complex

sulde ores has been investigated (Feng and Aldrich, 2000). The dry
ground samples have relatively rough particle surfaces with a high
concentration of microstructural defects. The wet ground samples
have smoother and cleaner surfaces. The dry ground samples
exhibited more stable, higher loaded froths and faster otation
kinetics owing to the activated particle surfaces. Moreover, by
combining dry and wet grinding, the kinetics, the nal grades and
recoveries of the suldes were improved. In the otation of sand, the
dry ground sand showed signicantly faster otation kinetics and a
slightly higher otation yield than the wet ground sand. This
demonstrates that dry milling could improve otation kinetics
owing to the higher energy absorbed in the particles.

1.2. Aim of the work

Particle shape and surface roughness are considered important
parameters for predicting the behavior of particles individually or
collectively. Therefore, this research paper is interested in the
different methods used for estimating the particle shape and surface
roughness and relating them with the behavior of some mineral
processing operations, especially comminution and otation
processes.

2. Methods of estimating the shape and surface roughness

of particles
Particle shape is one of the most important measures and there is
no general shape factor available which clearly differentiates between
all possible kinds of shape (Podczeck, 1997). The development of the
automated microscope techniques linked to computer data-processing systems (image analysis techniques) facilitated the possibility of
using the methods of specifying the shape and surface roughness of a
particle prole by mathematical procedures (Meloy and Williams,
1994). A three-dimensional particle is projected in a plane to obtain
two-dimensional closed curve which is taken to estimate the particle
shape. The methods used to assess the particle shape and surface
roughness range from very simple methods such as the determination of aspect ratio, elongation ratio or circularity to very modern
methods such as the use of Fourier analysis, delta analysis and fractal
geometry (Meloy and Williams, 1994; Singh and Ramakrishnan, 1996
and Podczeck, 1997). These methods are reviewed in the following
sections:

2.1. Conception description

This is the simplest method used for characterizing the particle
shape. Words like spherical, disc, blade, rod, granular, rounded,
angular, isometric, irregular, nodular, crystalline, oblong, acicular,
brous, dendritic and aky have been used to classify the particles
into different shape groups. Although this identication corresponds well with the human imagination, it is however quite
subjective (Singh and Ramakrishnan, 1996 and Stechemesser et al.,
1996).
In many industries, an old system called the vibrating table was
used for separating the particles according to their shapes. In this
device at, elongated or needle particles may be separated from
rounded particles by owing the particles over a tilted table. The
eccentric vibrating motion of the table drives all particles laterally
along the table; however, rounded particles roll down the inclined side
of the table at right angles to the motion of the at particles. Parallel
groves in the table may keep acicular particles from rolling (Meloy and
Williams, 1994).

182

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

Jaggedness (Broyles et al., 1996):

2.2. Mathematical assessment of shape and surface roughness of a

particle
2.2.1. Shape factors
Flatness ratio and elongation ratio (Singh and Ramakrishnan,
1996; Burstein et al., 1995):
breadth B
thickness T

Flatness ratio =

Elongation ratio =

length L
:
breadth B

Circularity (Singh and Ramakrishnan, 1996):

Feret's diameter, D(Fmax.) is dened as the maximum static

length of particle projection derived from the co-ordinates of the
projected area in one specied direction of measurement.
Factor of circularity characterizes the deviation from the circular
shape. It has a value of 1.0 for the spherical particle and its value is less
than 1.0 for the other shapes.
Sphericity (Singh and Ramakrishnan, 1996):
Sphericity =

surface area of a sphere of the same volume as the particle

:
actual surface area of the particle
5

Circularity and sphericity are measures of the degree to which the

shape of a particle approaches to a circle and a sphere, respectively.
External compactness (Kaye, 1981):
External compactness =





diameter of a circle of equal area 2
DA 2
=
:
diameter of embracing circle
DEm

6
Chunkiness (Kaye, 1981 and Kaye et al., 1992):
Chunkiness =

Bm
:
Lm

avg

i=1

180

Factor of concavity (CONC) (IMAGE_C, 1993):

"
#
2DX + DY + DP + DM + 8
CONC =
1:
2nedg: ncor:

11

12

This measure has a zero value for the convex particle and increases
with the increase of fracturing.

FO =

LENG
:
LCNV

13

2.3. Fractal geometry

Factor of circularity (FK) (Stoyan, 1993):

DA
:
DFmax:

Jaggedness =

Factor of convexity (FO) (Stechemesser et al., 1996):

circumference of a circle of the same area as the particle

Circularity =
:
circumference of the particle
3

FK =


180 
1 RDi

In recent years, there has been a rapid development of two

innovative procedures for determining the shape and surface
roughness of ne particle proles. These are Fourier analysis and
the fractal geometry (Kaye et al., 1992). Fractal geometry was
introduced to the scientic community by Mandelbrot (1997). He
illustrated how natural or real physical boundaries, which cannot be
adequately described by traditional or Euclidean geometry, can be
characterized using fractal geometry. Fractal geometry can be applied
for specifying the real physical boundaries which have no tangent, i.e.
curves that have no differential functions (Brown et al., 1993). Fractal
dimension is an interesting feature proposed recently to characterize
the surface roughness and self-similarity of a picture (Sarkar and
Chaudhuri, 1994). The practical signicance of the fractal dimension is
more than the other measures where it can be quantied through a
range of measuring scales and is expressed with a simple scalar
number. It is practically well suited to estimate the highly re-entrant
particle, which has sharp corners and indentations (Graf, 1991).
The basic methodology of determination the fractal dimension of a
projected prole is to measure the perimeter many times with
different scales of measurements, i.e. at a series of step lengths or
resolutions. Smooth particles have constant lengths of perimeter
while the rough particles have larger perimeter lengths with smaller
step lengths (Meloy and Williams, 1994). Particles, which display
fractal behavior, observe the following exponential relationship
(Clark, 1987; Kaye, 1991; Brown et al., 1993; Sarkar and Chaudhuri,
1994; Meloy and Williams, 1994; Brown et al., 1994a and Lee and
Chou, 1994):
Lr = cr

1F

14

Aspect ratio (Kaye et al., 1992):

Aspect ratio =

Lm
:
Bm

2.2.2. Surface roughness factors

Roundness (Singh and Ramakrishnan, 1996):

With large step lengths, the particle is modeled as a polygon with

just a few sides of equal lengths. With small step lengths, the particle
is modeled as a polygon with several sides of equal lengths.
Eq. (14) can be rewritten as follows:
log Lr  = 1F log r  + log C :

15

Ri = ncor

Roundness =

i=1

Angularity (Singh and Ramakrishnan, 1996):

 g
3 ncor 
ij
Angularity = 180ij :
ri
i=1 j=1
The value of this factor ranges from 0 to .

10

By plotting log [L(r)] against log [r], the fractal dimension (F) can
be calculated from the absolute slope (1 F) of the best-tting line of
data in the plot. Such graph is known as Richardson plot and is
illustrated in Fig. 1 where 1.0 F 2.0. A fractal dimension value equal
to 1.0 is related to a Euclidean prole while the Brownian movement
will approximate to a fractal dimension value equal to 2.0. Large
slopes on the Richardson-graph correspond to a strong fracturing and
an irregular surface.

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

Fig. 1. Richardson plot of a particle prole (after Clark, 1987; Kaye, 1991; Brown et al.,
1993; Sarkar and Chaudhuri, 1994; Meloy and Williams, 1994; Brown et al., 1994a and
Lee and Chou, 1994).

It is usual to normalize the step length and the corresponding

estimated perimeter to the Feret's diameter of the particle as shown in
Fig. 2.
For natural systems, several dimensions may arise at various step
lengths. For mineral fragments, usually two linear portions with two
different slopes are observed on the Richardson plot through the step
length range used. This illustrates different irregularities at different
step lengths (different resolutions) as shown in Fig. 3. The region at the
smaller step lengths is known as the textural (surface roughness) region
while the structural (shape) region is at the coarser step lengths and
represents the gross boundary of the particle (Kaye et al., 1987; Kaye
and Clark, 1989; Graf, 1991; Kaye, 1991; Brown et al., 1993 and Brown
et al., 1994a). A sharp break point between the structure and the
texture is signicant where it is considered a rapid transition from one
fractal domain to the other. In many cases there is a gradual transition.
Graf (1991) concluded that the range, number and dimension of
fractal elements in a ne particle depend not only on the geometry of
the particle but also on the resolution limits (step lengths) of the
investigator. He listed four recommendations to remedy the problem
of scale sensitivity:
1. Scales of Richardson diagrams must be made by using both the
absolute and Feret-normalized measures. Absolute scales are

183

Fig. 3. Richardson plot showing the change in fractal dimension with the step length
where the perimeter estimated and step lengths have been normalized with respect to
Feret's diameter of the particle (after Kaye et al., 1987; Kaye and Clark, 1989; Graf,
1991; Kaye, 1991; Brown et al., 1993 and Brown et al., 1994a).

useful because the particle geometry can be linked to the physical

processes.
2. Select a constituent upper bound for the maximum step length.
When very large step lengths are used, changes in the perimeter
reect the trends of simple polygons rather than the geometry of
particle (Kaye, 1984).
3. Use the largest possible resolution. Poor resolution, which results
from the poor imaging or video limitations, can mask potentially
important textural elements.
4. Make interpretations with the help of physical models and physical
processes.
Various image analysis techniques have emerged over the last two
decades. In these techniques, the three dimensions of the particle (or
its projection in two-dimensional direction) are actually measured
(Hurter and Clark, 1987). The disadvantage of the two-dimensional
image analysis technique is that it gives no indication of the threedimensional geometry of the particle, where the technology is not
generally available for the three-dimensional digitization of the
particle surfaces (Clark, 1987). In earlier publications dealing with
the fractal structure of ne particle boundaries, the experimental
work was carried out manually. This was time consuming and the
amount of investigative work, which could be undertaken in a given
context, was limited (Kaye et al., 1987). Nowadays, several computer
aided image analysis procedures have been reported. The most
popular methods used for determination of the fractal dimension are
structured walk technique, equipaced polygon technique, dilatation
logic technique, the mosaic amalgamation technique and the line
scans triangulation technique (Schwarz and Exner, 1980; Kaye, 1984;
Clark, 1986; Kaye et al., 1987; Avnir, 1989 and Kaye et al., 1994).
2.4. Fourier analysis

Fig. 2. A particle prole with a maximum step length of 0.3 of Feret's diameter (after
Clark, 1987; Kaye, 1991; Brown et al., 1993; Sarkar and Chaudhuri, 1994; Meloy and
Williams, 1994; Brown et al., 1994a and Lee and Chou, 1994).

One of the most popular methods for shape analysis is the Fourier
analysis which determines the shape prole using Fourier coefcients
(Kaya et al., 1996). In Fourier analysis, rstly a reference point at the
center of prole is chosen. It is preferred to choose the centroid of the
prole treated as a useful reference point. Secondly, a point on the
periphery of the prole is chosen as the starting point for exploration of
the prole with a rotating vector. The maximum distance is then used as
a normalizing factor to summarize the magnitude of the vector as it
moves around the periphery of the prole with a uniform angular
velocity to generate what looks like a waveform (Schulze et al., 1989).
Two main procedures based on the Fourier analysis of a prole
may be applied to the morphologic characterization of a particle
(Singh and Ramakrishnan, 1996). The rst method is based on Fixed
Angular Boundary Sampling (FABS) (Luerkens et al., 1987) and the

184

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

second method is based on Fixed Boundary Segment Sampling (FBSS)

(Bonifazi and Massacci, 1996).
3. Application of atomic force microscopy in measurement of
surface roughness
The atomic force microscopy (AFM) was rst described in 1986 by
Binnig et al. (1986) as a new technique for imaging the topology of
surfaces to a high resolution. Since then AFM has been used in the study
of surface science, both as an imaging and surface characterization
technique. AFM has a number of advantages over electron microscopy
techniques, primarily the ability to make measurements in a uid
environment, as well as, the capability to directly measure interaction
forces between surfaces (Johnson et al., 2006). Atomic force microscopy
is a technique able to quantitatively measure interactions between
single particle and interfacial boundaries (Johnson et al., 2006).
Surface properties like roughness, adsorption layers and surface
chemistry strongly affect the adhesion forces. Measurements of
particleparticle and particlesurface interactions in the gas phase are
carried out with an atomic force microscope (AFM). For modeling of
particle adhesion, computer-assisted empirical potential methods are
used. Parameters like absorbed water and surface roughness are
considered. Parameters for weak interactions from the Lifshitz theory
and gas adsorption data are extracted. Absorbing molecules can be used
as probes to measure dispersive forces (Goetzinger and Peukert, 2003).
Many methods are used to characterize the surface roughness of solid
particle for quality control and for nding the correlations with other
properties. In this study, the fractal analysis was used to describe the
surface roughness. Atomic force microscopy (AFM) was used to obtain
three-dimensional surface proles. The variation method was applied to
calculate the fractal dimensions. Fractal dimensions of four granule
samples, four powders, and two freeze-dried powders were measured. A
computer-program was written to implement the variation method. The
implementation was veried using the model surfaces generated by
fractional Brownian motion. The fractal dimensions of most particles and
granules are between 2.1 and 2.2, and are independent of the scan size
measured. The freeze-dried samples however show wide variation in the
values of fractal dimension, which are dependent on the scan size. As
scan size increases, the fractal dimension also increases up to 2.5. This
variation method allows calculating reliable fractal dimensions of
surface proles obtained by AFM. The estimates are dependent on the
algorithm and the digitized model size (i.e., number of data points of the
measured surface prole) used. The fractal dimension is also a function
of the observation scale (i.e. the scan size) used in the prole
measurement. The multi-fractal features and the scale-dependency of
fractal dimension result from the articial processes controlling the
surface morphology (Li and Park, 1998).
To obtain additional information on the surface roughness of the
studied samples, AFM scans are performed and root mean square (r.m.s.)
roughness is determined at different lateral scales (Gladyszewski et al.,
1998 and Hazra et al., 2001). The dened sizes are corresponding to
specic physical properties measured on generally irregular particles,
while most sizing techniques assume that particles are spherical in
shape. Direct physical particle sizes can be determined with an atomic
force microscope (Gwaze et al., 2007).

A nickel sulde ore was subjected to two comminution events,

impact shattering and ball milling (Brown et al., 1993). The size
distributions of impact sample product were less than 2000 m
(median size = 250 m). The maximum size of ball milling sample
product was less than 3500 m (median size = 80 m). Fractal
analysis was performed on samples of the resulting comminution
products. The impact sample (formed by ballistic disintegration)
produces particles with higher boundary fractal dimensions than the
milled sample. The fractal dimension of the impact fragments
increases with decreasing size, whereas the milled particles displays
a more complex distribution. This behavior may be due to the
abrasion acting on fragmented particles in the ball mill which
smoothes off any protrusions. Also, it is possible that small amounts
of crushing (compressive failure) may have occurred from the
trapping of small particles between two larger balls. It is observed
that fragments produced by tensile fragmentation have a higher
fractal dimension than those produced by compressive fragmentation
(Brown et al., 1993).
The previous data was plotted on Gaussian probability graph
paper, as shown in Fig. 4, where the ordinate represents the
cumulative percentage of fractal dimension equal to or less than a
stated value. The fact that the fractal information gives straight line on
Gaussian probability paper is evidence that the fractal dimensions, in
a population of shattered rocks, are probably describable by a
Gaussian distribution function. Different comminution types occupy
also different regions on this type of chart, as indicated by Fig. 4 and
various charts could be constructed for different materials. The
implications are that if one requires a denite particle distribution, in
fractal terms, then the mechanism of size reduction can be selected
(Brown et al., 1993). There is clear evidence to support the concept
that the structural fractal dimension of a particle holds key
information upon its formation dynamics.
Another study was performed to investigate if the type of
pulverizer used to comminute limestone, a grade suitable for ue
gas desulphurization (FGD), inuences its behavior and use. At
present, more than 10 M tones of limestone are milled per year in
Europe for FGD purposes. The three pulverizers used are a laboratory
hammer mill, crushing rolls with smooth surfaces and a batch
laboratory ball mill. The failure mechanisms represented by these
devices can be broadly categorized as single and/or double impact
(hammer mill) and compression and shear failure (roll crusher). The
ball mill size reduction mainly takes place by the impact of balls on the

4. Relationship of the comminution process with the particle

shape and surface roughness
Although many aspects of comminution have been reported, there
have been relatively few papers devoted to the effects on particle
shape and surface roughness. This area of powder technology is
neglected partly due to the difculties in measuring the shape and
surface roughness of particles and also because there are no
universally accepted denitions of shape (Kaya et al., 1996).

Fig. 4. Fractal distribution data illustrated on Gaussian probability graph paper.

(a) Impact sample product (50 = 250 m). (b) Ball mill product ( 50 = 80 m) (after
Brown et al., 1993).

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

185

differences in the overall shape of the comminuted fragments.

Variations in the limestone utilization are noted for the samples
comminuted by the three different techniques. The hammer-milled
sample having the highest value because it has the highest boundary
fractal population (Brown et al., 1994a and Brown et al., 1994b).
Experimental investigations on the comminution of gibbsite in four
different types of mills are presented. These include tests of dry grinding
in a tumbling ball mill, a shaker bead mill and an air jet mill, as well as,
tests of wet grinding in a tumbling ball mill and in a stirred bead mill. A
quantitative characterization of the size and shape of the ground
product was made in order to analyze the effect of the grinding stress on
the properties of a complex agglomerate crystal such as gibbsite. The
analysis allowed us to determine the fragmentation route occurring in
the mills. This led to the identication of two fragmentation paths
imposed during comminution processes. These involved different
phenomena: rupture of joints and chipping and breakage according to
the nature of the main stress (attrition or impact) (Frances et al., 2001).

Fig. 5. Fractal distribution data, for the FGD limestone fragments, illustrated on a
Gaussian probability graph paper (size = 38 m) (after Brown et al., 1994b).

material, with either prior or subsequent abrasion (attrition) (Brown

et al., 1994b).
The limestone sample consists of 14.0 mm. It was pre-screened at
the quarry, from which a (4.0+1.7) mm size fraction was extracted to
produce an even feed for the pulverizing devices. Samples of this fraction
were comminuted by hammer milling (internal screen removed), roll
crushing and ball milling. The Raymond mill and roll crusher were
starvation fed, to reduce particleparticle interactions, and ball milling
was performed wet, with de-ionized water. In order to produce a very
narrow size range, to effectively negate size effects in acid digestion, the
resultant comminution products are screened at 38 m, and the undersize
classied using a Warman cyclosizer (elutriation technique). Cone ve on
the cyclosizer was chosen as the size band for investigation, (16.2+
12.7) m (0.5 m) (Brown et al., 1994b).
The populations of the structural boundary fractal dimensions of
the three sets of fragments are shown in Fig. 5. From this gure, it can
be shown that the particles are distributed according to Gaussian and
the three distributions are found to be signicantly different. This is
further evidence to support the concept that the structural boundary
fractal dimension of a particle holds key information upon its
formation dynamics. On the other hand, simple geometric shape
factors, i.e. circularity and aspect ratio did not detect any signicant

5. Effect of the grinding type on the otation behavior

As well known, otation differs from most of the separation
processes by the fact that the conditions of separation directly depend
on surface properties and therefore also on comminution process. The
deformation mechanism in comminution has a great inuence on the
properties of mineral surfaces formed during grinding because the
geometrical heterogeneity of the surface determines the area reacting
with otation reagents. Moreover, the character of the interactions
between the reagents and minerals is also dependent on the
conguration of the surface (Ocepek et al., 1990).
The effect of dry and wet grinding on the otation of complex
sulde ores from the Merensky Reef in South Africa was investigated.
The sample was crushed in a jaw crusher and a rotary cone crusher to
100% passing 4 mm. For comparative purposes, an oxide ore (sand)
with a particle size of 3 mm was also investigated. The rod mill was
run at a constant speed of 80 rpm. Three samples were ground to
approximately 60% passing 75 m under the conditions of dry milling,
as well as wet milling with pulp concentrations of 66% and 90% solids.
During dry milling, nitrogen was fed to the mill to drive off oxygen
before milling started. The grinding results are as follows: dry milling
took 40 min to result 58.0% of 75 m. The dry ground sample had a
specic surface area of 1.23 m2/g. Wet milling at a pulp concentration
of 66% solids took 30 min to ensure that 57.5% of the sample passed
75 m. This sample had a specic surface area of 1.29 m2/g. Wet

Fig. 6. Topographic of the sulde particle surfaces by SEM: (a) wet grinding with 66% solids (% 75 m = 58.0, surface area = 1.23 m2/g) (b) wet grinding with 90% solids
(% 75 m = 57.5, surface area = 1.29 m2/g) and (c) dry grinding (% 75 m = 58.5, surface area = 1.25 m2/g) (after Feng and Aldrich, 2000).

186

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

milling at a pulp concentration of 90% solids took 36 min to give 58.5%

of 75 m. This sample had a specic surface area of 1.25 m2/g.
Topographical examination of the ground particle surfaces by
scanning electron and atomic force microscopy showed that the dry
ground samples had relatively rough particle surfaces with a high
concentration of microstructural defects. The wet ground samples had
smoother, cleaner surfaces as shown in Fig. 6. Higher stresses were
included in the particles in the denser slurry (90% solids) and as a
consequence some defects also appeared on these particle surfaces.
When the same new surface formed, dry grinding consumed more
energy than wet grinding. This means that the particles in the dry
grinding conserved more energy, some of which existed in the form of
defects. These defects serve as active centers for spreading up particle
dissolution and reagent adsorption (Feng and Aldrich, 2000).
As a result, the activated particle surfaces from the dry ground ore
accelerate the dissolution of the particles, as well as, the adsorption of
reagents onto the particle surfaces. The dry ground samples exhibited
more stable, higher loaded froths and faster otation kinetics, owing
to the activated particle surfaces. High intensity conditioning of the
dry ground ores prior to otation could improve otation by cleaning
the particle surfaces through high shear force elds in the pulp.
Moreover, by combining dry and wet grinding, the kinetics, as well as,
the nal grades and recoveries of the suldes, could be improved as
shown in Figs. 7 and 8 (Feng and Aldrich, 2000).
Sand samples were ground under dry and wet milling conditions.
Dry milling took 50 min to ensure that 75% by mass of the sample
passed 75 m. The dry ground sample had a specic surface area of
1.68 m2/g. Wet milling at a pulp concentration of 66% solids took
37 min to ensure that 75% by mass of the sample passed 75 m. The
wet ground sample had a specic surface of 1.76 m2/g. Dry milling
consumed approximately 25% more energy than wet milling to
achieve more or less similar surface areas. Some of the energy
absorbed in the dry milling process was owed to particle deformation,
which was stored in the form of surface defects on the particles. In
addition, the particle surface was activated in the dry milling process.
The wet ground sample had a smooth particle surface and few loose
ne particles absorbed onto the surface, while the dry ground sample
had a rough particle surface and many small particles attached onto
the surface. Some defects appeared on the particle surface in the dry
ground sample, and these surface defects served as the active centers
during reagent adsorption. As illustrated in Fig. 9, the dry ground sand
showed signicantly faster otation kinetics and a slightly higher
otation yield than the wet ground sand. This demonstrates that dry
milling could improve otation kinetics, owing to the higher energy
absorbed in the particles (Feng and Aldrich, 2000).
In practice, dry milling of sulde ores is complicated by the
tendency of ne suldes to oxidize in air. This is not the case with

Fig. 7. Variations of sulfur recovery with otation time (after Feng and Aldrich, 2000).

Fig. 8. Variations of sulfur content with otation time (after Feng and Aldrich, 2000).

oxide ores, but the higher energy consumption has to be weighed

against the enhanced otation kinetics (Feng and Aldrich, 2000).

6. Relationship of the particle shape and surface roughness

with wettability
Spreading of liquids on rough surfaces was studied experimentally
and theoretically. Seven liquids, with viscosities spanning over a few
orders of magnitude, were used on glass surfaces of ve different
values of average roughness, ranging from about 0.07 to about 40 m.
It is shown that the experimental data can be empirically correlated
by a power law. In addition, a simple theoretical model was presented,
which took into account the capillary ow of liquid into the roughness
grooves. Results of the theoretical calculations were shown to be in
good agreement with the experimental data (Apel-Paz and Marmur,
1999). The study of spreading on practical rough surfaces is difcult,
because the problem of dening roughness is a complex one.
Therefore, much work was done on spreading on surfaces with
well-dened grooves or roughness. Nonetheless, some measurements
of the kinetics of spreading of liquids on practical rough surfaces were
also made. These studies revealed that spreading of the liquid occurs
by two modes: on top of the rough surface and inside the capillary
grooves (Apel-Paz and Marmur, 1999).
At any interface between two phases (e.g. liquidsolid, liquidgas,
oilwater), the adjacent interfacial multi-molecular layers are subjected to an unsymmetrical force and consequently the interfacial
tension is the net result of these forces (Hazlett, 1990). Quantication
of the effect of surface roughness is difcult and somewhat

Fig. 9. Flotation result of sand samples under different milling environments: (a) wet
milling with 66% solids (% 75 m = 75.0, surface area = 1.76 m2/g) and (b) dry
milling (% 75 m = 75.0, surface area = 1.68 m2/g) (after Feng and Aldrich, 2000).

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

controversial. Much effort was undertaken to study the equilibrium

contact angle which cannot be measured directly (Hazlett, 1993).
At sharp corners, the contact angle can be assumed as a range of
values. Wenzel recognized this difculty and approached the problem
in terms of surface energies. Wetting of a solid surface occurs only if
there is a resulting decrease in free energy. The apparent equilibrium
contact angle is then determined by an integrated surface of uid
uid interaction and not by the conditions existing only at the point of
contact (Hazlett, 1990).
It is stated that the contact angle is dependent on the surface
roughness (Onda et al., 1996). The apparent contact angle for rough
surfaces could be expressed in terms of smooth surfaces (Hazlett,
1990 and Onda et al., 1996). The modication of the wettability due to
surface roughness can be greatly enhanced in the fractal surface; that
is the fractal surface will be superrepellent (superwettable) to a liquid
when the contact angle is greater (less) than 90 (Onda et al., 1996).
In a paper published by Ulusoy et al. (2003), experimental studies
to determine the shape properties, surface roughness and wettability
of quartz mineral (a very common component of many raw materials)
were performed for the products of ball, rod and autogenous mills.
Shape properties have been stated in terms of shape descriptors such
as elongation, atness, roundness and relative width. The measurements were carried out on the projections of particles using the
scanning electron microscope (SEM). Surface roughness values have
been determined on the pelleted surfaces of the ground powders. The
wettability characteristics (critical surface tension of wetting) of
quartz mineral were determined by micro-otation technique using
the EMDEE Micro-FLOT test tube. Finally, some correlations were
found between the shape properties, surface roughness values and
wettability. The results have shown that the critical surface tension of
wetting values increases with increasing roughness and roundness,
i.e. elongated and smooth particles having lower critical surface
tension of wetting values. This means that these particles have higher
hydrophobicity (Ulusoy et al., 2003).
A new method for the measurement of apparent contact angles on
real surfaces was developed. The method consists of vibrating the
surface, taking top-view pictures of the drop, monitoring the drop
roundness, and calculating the contact angle from the drop diameter
and weight. The use of this new method was demonstrated for various
rough surfaces having the same surface chemistry. In order to
establish the optimal vibration conditions, the proper ranges for the
system parameters (i.e., drop volume, vibration time, frequency of
vibration, and amplitude of vibration) were determined. The
reliability of the method was demonstrated by the fact that the
ideal contact angles of all surfaces, as calculated from the Wenzel
equation using the measured apparent contact angles, came out to be
practically identical. This ideal contact angle was compared with three
methods of calculation from values of advancing and receding contact
angles (Meirona et al., 2004).
In many systems there exists a contact angle hysteresis between
an advancing and receding contact angle, i.e. a different contact angle
may be measured depending upon whether the three-phase contact
(TPC) line is advancing or receding over a particle surface. The reason
for this hysteresis is generally ascribed to roughness and heterogeneity of the surface chemistry, although other factors may be of
importance, such as polymer coatings, etc. (Johnson et al., 2006).
7. Effect of the particle shape and surface roughness on the
otation behavior
The particle shape and its surface roughness have signicant
effects on the otation behavior. Many researchers have studied these
phenomena qualitatively (Anfruns and Kitchener, 1997; Varbanov,
1984; Blake and Ralston, 1985; Ducker et al., 1989; Silverstein and
Breuer, 1993; Schmidt and Berg, 1996 and Schmidt and Berg, 1997),
but few researchers (Burstein et al., 1995) studied these phenomena

187

quantitatively. This may be due to the lack of techniques required to

measure them by quantied parameters. Ramesh and Somasundaran
(1990) decided that the shape of the particle has a substantial effect
on the force required to detach the particle from the gasliquid
interface. They found that homogeneous particles with a single
specic geometry would yield a single value for the detachment force.
On the other hand, a mixture of particles having the same contact
angle and a variety of shapes would yield a distribution of detachment
forces. Burstein et al. (1995) showed that the otation recoveries of
non-spherical particles are higher than that of spherical particles. A
possible explanation of this result might be attributed to the increase
of surface of non-spherical particles which leads to more adsorption of
surfactant and hence increasing otation recovery. This result is in a
conict with the result of Varbanov (1984) who concluded that, with
regard to practical otation, no major differences could be inferred
between the behaviors of spherical and non-spherical particles.
Schmidt and Berg (1996) investigated the role of particle shape on
otation using model particles of spherical and disc shapes. Their
results revealed that spherical particles oat better than disc or platy
particles. Spheres are deected away from bubble by ow but usually
attach if they contact the bubble surface. Discs often colloid with the
bubble edge-on and immediately bounce off and seldom attach to the
bubble due to the very short contact time. Alternatively, discs turn to
the side as they approach the bubble but seldom attach due to the
large thin lm drainage area. Schmidt and Berg (1997) in other
publication using a hydrodynamic model concluded that disc and
plate-shaped particles behave differently more than spherical
particles. Large disc-to-bubble ratios (N0.1) yield always greater
collision efciencies than those for equivalent spherical particles.
They found also that the large discs have always a lower probability of
attachment than equivalent spheres due to the higher tendency of
discs to bounce off the bubble surface after collision. For smaller discs,
the differences in the attachment efciency and otation behavior
between discs and spheres decrease and collision efciency become
the predominant step.
Wotruba (1991) in his Ph.D. thesis studied the inuence of corn
shape on the oatability of zirconium and microlith particles in a
pneumatic otation cell using prismatic, spherical and ellipsoidal
shapes. The most important conclusions of this thesis are that the corn
shape has an essential inuence on the oatability of zirconium.
Spherical and ellipsoidal particles have less oatability than prismatic
particles, independent of whether the spherical particles are rough or
not. The rst reason for this behavior is that the attachment process
with bubble is more difcult for spherical particles as compared to the
prismatic particles. The second reason is that the retreatment in the
liquid phase causes easier wetting on spherical particles than on
prismatic particles with edges. The strong inuence of the shape on
the oatability occurs at the range of (63 + 40) m, i.e. in a range
where the inuence of the turbulence on the stability of the particle/
bubble aggregate is not critical.
Few scientists studied the effect of surface roughness on the
otation process (Anfruns and Kitchener, 1997; Silverstein and
Breuer, 1993 and Burstein et al., 1995). Burstein et al. (1995)
concluded that the increase in the particle roughness (determined
in fractal terms) causes signicant intensication of the process of
collapse of lamellae near particle surfaces and reduction of froth
stability. Anfruns and Kitchener (1997) revealed that the particle
roughness ensures easy rupture of the wetting lm even when the
barrier is strengthened by the addition of low concentrations of
surfactant or when the contact angle is reduced from 87 to 30.
Ducker et al. (1989) have studied the otation of glass ballotini and
ground ballotini using a double-chained cationic surfactant. They
concluded that there are large differences in the otation properties of
these two materials. They interpreted these differences to the
difference in surface roughness between glass ballotini and ground
ballotini and not to the difference in their shapes. Anfruns and

188

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

Kitchener (1997) marked the differences in the behavior of collision

and capture between the angular and smooth particles. They showed
that the strongly hydrophobic angular quartz particles are captured by
bubbles with 100% efciency whereas the capture efciency of
strongly hydrophobic glass ballotini particles is very much lower.
The 100% capture efciency of angular particles is attributed to the
jagged projections or asperities on the particle surface which are
leading to a local thinning and rupture of the wetting lm, i.e. the
otation rate is enhanced compared to the smooth surface of glass
ballotini. Anfruns and Kitchener (1997) also observed, for smooth
particles, that although the static contact angle is 30, they are not
captured at all. They concluded, with smooth particles, that the
capture rate is more sensitive to the operation of electrical doublelayer forces during the thinning of the liquid lm.
Ahmed et al. (1998); Ahmed (1999); Ahmed et al. (1999) and
Stechemesser and Ahmed (1999) illustrated a modern and useful
approach, which is considered the rst attempt, to study the
quantitative effect of the surface roughness and shape of particles on
the detachment process from the airliquid-interface by means of a
centrifuge technique and also on their oatability (under ideal
conditions) using a modied Hallimond tube. His study was carried
out on four different materials (spherical soda-lime-glass ballotini,
ground ballotini, glass sand and quartzite) of the same particle size
fraction (315 + 250) m. For estimation of the surface roughness of
the particle, three parameters used are as follows: textural fractal
dimension (Ft), factor of convexity (FO) and factor of concavity (CONC).
The following parameters are of interest for identifying the shape of
particle: structural fractal dimension (Fs) and factor of circularity (FK).
The values of studied parameters are determined by means of the
software IMAGE_C. The estimation of particle geometry was made as a
two-dimensional analysis of the digitized particle projection on a slide
in the position of the greatest stability. According to Kaye (1994), the
average structure is described with a sufcient precision at the position
of the greatest stability of the particle. From the results, it was found that
the four materials investigated are different in surface roughness of
particle. On the contrary, the particle shape of ground ballotini and the
quartzite is the same in comparison to the other materials investigated.

7.1. Effect of the surface roughness of particle on oatability

Nguyen et al. (1997a) have made a detailed analysis of the particle/
bubble attachment. There are three microprocesses (or elementary

steps) for the before-contact stage which may occur for successful
particle/bubble attachment:
1. thinning of the intervening liquid lm to a critical thickness where
the lm rupture begins,
2. rupture of the intervening liquid lm, i.e. formation of the threephase contact of a critical wetting radius (a hole) and,
3. expansion of the three-phase contact line from the critical radius to
form a stable wetting perimeter.
Schulze et al. (1989) have decided that in the technological
otation process, the three microprocesses mentioned above occur
simultaneously. Therefore it is impossible to determine unambiguously from the temporal course and the efciency of the total process
which one of the microprocesses is signicant in the course of the
process. Consequently this problem must be solved by investigating
the individual microprocesses separately.
Fig. 10 illustrates the effect of the surface roughness of particle
(quantied by the textural fractal dimension) on otation recovery at
different degrees of hydrophobicity. From this gure, it can be seen
that as the textural fractal dimension increases, the otation recovery
increases linearly at different degrees of hydrophobicity. Similar
relationships between the surface roughness (quantied by the factor
of convexity and the factor of concavity) and otation recovery were
also obtained (Ahmed, 1999 and Stechemesser and Ahmed, 1999).
From Fig. 10, it can be seen that the differences in the values of
otation recovery between the roughest material (quartzite) and the
smoothest material (soda-lime-glass ballotini) range from 24% (at
contact angle = 71) to 33.5% (at contact angle=78). This means
that although the small differences in the values of textural fractal
dimension of materials were examined, very great signicant
differences in the values of otation recovery between the smoothest
and the roughest material were obtained. This assures that the effect
of the surface roughness on the otation recovery is signicantly
effective.
From the previous results, it can be concluded that the roughness has
a signicant effect on the otation recovery. This may be attributed to
the fact that larger detachment forces are required to separate the rough
particle from bubble than to separate the smoother particle. The larger
detachment forces will indicate a strong adhesion force for the bubble
particle aggregates, which could reserve these aggregates to reach
the oat receiver and hence higher recovery is obtained (Ahmed et al.,
1999; Ahmed, 1999 and Stechemesser and Ahmed, 1999).

Fig. 10. Effect of the surface roughness of particles (quantied by the textural fractal dimension) on the otation recovery at different degrees of hydrophobicity (after Ahmed, 1999).

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

189

Fig. 11. Effect of the shape of particles (quantied by the structural fractal dimension) on the otation recovery at different degrees of hydrophobicity (after Ahmed, 1999).

It was decided that the surface roughness has an important role in

the detachment process and then the effectiveness of otation
processes. This relationship is attributed to the larger values of
three-phase contact perimeter for rough particles more than the
smooth particles. Accordingly, a longer time, i.e. a larger detachment
force is required to separate the particles from the gasliquid interface
for rough materials more than that for the smooth materials (Ahmed
et al., 1999; Ahmed, 1999 and Stechemesser and Ahmed, 1999). This
interpretation is equivalent to the process of expansion of the threephase contact line on the surface of hydrophobic particle to attain the
dewetting case. In this process, Nguyen et al. (1997b) have observed
experimentally that the time of three-phase contact line expansion is
a function of the central angle, i.e. the radius of three-phase contact
line. From this result it was concluded that a particle with a larger
three-phase contact radius needs a longer time to arrive to the
dewetting case. This conclusion was exploited to support the results
mentioned above. In the detachment experiments (Ahmed et al.,
1999; Ahmed, 1999 and Stechemesser and Ahmed, 1999), a
contraction of three-phase contact line is occurring. This contraction
on the particle surface is leading to attain to the wetting case and

hence the particle is submerged into the liquid. This contraction needs
a longer time and then larger detachment forces (larger three-phase
contact perimeter) in the case of rough particles.
7.2. Effect of the particle shape on oatability
Fig. 11 shows the effect of the particle shape (quantied by the
structural fractal dimension) on otation recovery of the materials
examined at different degrees of hydrophobicity. From this gure, it
can be seen that as the structural fractal dimension increases from
1.020 to 1.043, the otation recovery slightly increases. Above this
value, any increase in the structural fractal dimension causes
signicant increase in otation recovery. This behavior is the same
at all the degrees of hydrophobicity (Ahmed, 1999 and Stechemesser
and Ahmed, 1999). Fig. 12 illustrates the effect of particle shape
(quantied by the factor of circularity) on otation recovery at
different degrees of hydrophobicity. From this gure, it can be seen
that as the particle shape increases (decreasing of the factor of
circularity), the otation recovery increases up to a certain point
above which an excessive increase in the particle shape will result in a

Fig. 12. Effect of the shape of particles (quantied by the factor of circularity) on the otation recovery at different degrees of hydrophobicity (after Ahmed, 1999).

190

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

decrease in the nal otation recovery. It can be also noticed that the
relationship between the particle shape and otation recovery has the
same behavior at all degrees of hydrophobicity (Ahmed, 1999 and
Stechemesser and Ahmed, 1999).
Comparison of the results shown in Figs. 11 and 12 displayed that
there is a clear difference in the behavior of otation recovery with the
particle shape (if this shape is quantied by the structural fractal
dimension or the factor of circularity). This means that the particle
shape is not as effective as the surface roughness in controlling the
otation recovery. This result may be ascertained also from Fig. 11. In
this gure, it can be observed that although the structural fractal
dimensions of ground ballotini and quartzite are almost the same
(these materials have the same shape), they have different otation
recoveries (Ahmed, 1999 and Stechemesser and Ahmed, 1999).
Figs. 11 and 12 showed also that the otation recoveries of ground
ballotini are higher than that of spherical ballotini at all degrees of
hydrophobicity. One of the possible explanations of this result may be
attributed to the increase of the particle surface with the rise of
structural fractal dimension of ground ballotini which leads to more
adsorption of surfactant (Burstein et al., 1995) and hence the increase
of otation recovery (Ahmed et al., 1999 and Ahmed, 1999). Other
reasons for this behavior are that the attachment with bubble is more
difcult for the spherical particles compared to the non-spherical
particles. The second reason is that the retreatment in the liquid phase
causes easier wetting on spherical particles than on non-spherical
particles (Wotruba, 1991). Ducker et al. (1989) interpreted these
differences to the difference in the surface roughness between glass
ballotini and ground ballotini and not to the difference in their shapes.
8. Conclusions
1. Particle shape and surface roughness are considered important
parameters in the prediction of the behavior of particle and powder.
2. The methods used to assess the particle shape and surface
roughness range from very simple methods such as the aspect
ratio, elongation ratio or circularity to very modern methods such
as Fourier analysis, delta analysis and fractal geometry.
3. The development of automated microscope techniques linked to
computer data-processing systems (image analysis techniques)
facilitate the possibility of specifying the particle shape and surface
roughness by mathematical procedures.
4. Particles of the impact crushing have higher boundary fractal
dimensions than the rotary milled particles.
5. The dry ground samples are more stable and have higher loaded
froths and faster otation kinetics. These samples have relatively
rough particle surfaces while the wet ground samples have
smoother and cleaner surfaces.
6. The contact angle is dependent on the surface roughness. At sharp
corners, the contact angle is assumed as a range of values. The
fractal surface will be superrepellent (superwettable) to a liquid
when the contact angle is greater (less) than 90.
7. There are large differences in the otation properties of glass
ballotini and ground ballotini due to the difference in surface
roughness.
8. The particle shape does not control clearly the otation recovery.
The surface roughness is responsible for the detachment process of
particle and the nal otation recovery more than its shape.
Nomenclature
maximum breadth of the prole projection on a direction at
BM
a right angle in which LM is measured
c
constant
DA
particle diameter of a circle of equal area
DEM
diameter of the embracing circle
D(Fmax.) Feret's diameter of the particle
Di
distance from the centroid of a particle to its edge

DX, DY, DP, DM particle diameters according to Feret in four different

directions
F
fractal dimension
distance of the corner from the center of the maximum
gij
inscribed circle of radius ri for each of the three sections
through the long, intermediate and short axes of the prole
LCNV
circumference of the convex envelope of the particle
LENG
circumference of the particle projection
maximum length of the prole
LM
L(r)
particle perimeter length at step length (r)
number of concave corners of the particle
ncor
number of edges of the particle
nedg
r
step length
R
maximum inscribed radius of the particle
average particle radius (average over 180 two-degree
Ravg
increments)
radii of the curvature of ncor. (corners) of a projection or
Ri
section of the particle
angle of each corner
ij
References
Ahmed, M.M., Stechemesser, H., Mabrouk, S.A., Ibrahim, G.A., Tarshan, M.M., 1998. The
relationship between the surface roughness, shape and detachment force of
particles from the liquid/gas interface using centrifuge method. 49 Berg-und
Huettenmaennischer Tag, Kolloquium 2, Partikeltechnologie, Freiberg, Germany,
18 und 19 Juni, pp. 207223.
Ahmed, M.M., Stechemesser, H., Mabrouk, S.A., Ibrahim, G.A., Tarshan, M.M., 1999.
Image analysis technique for determination of the size, structure and texture of ne
particle prole. The 6th International Conference on Mining, Petroleum and
Metallurgy, Vol. I-B, Mineral Processing, Fac. of Engng., Cairo Uni., Cairo, Egypt,
February 2024, pp. 4056.
Ahmed, M.M., 1999. Fractal surfaces of particles and their oatability. Ph.D. Thesis,
Assiut Univ., Assiut, Egypt.
Anfruns, J.F., Kitchener, J.A., 1997. Rate of capture of small particles in otation. Trans.
IMM, Sec. C 86, C9C15.
Apel-Paz, M., Marmur, A., 1999. Spreading of liquids on rough surfaces. Colloids and
Surfaces A: Physicochemical and Engineering Aspects 146, 273279.
Avnir, D., 1989. The Fractal Approach to Heterogeneous Chemistry: Surfaces, Colloids.
Polymers. John Wiley&Sons, New York.
Binnig, G., Quate, C.F., Geber, C., 1986. Atomic force microscope. Phys. Rev. Lett. 56 (9),
930933.
Blake, P., Ralston, J., 1985. Particle size surface coverage and otation response. Colloids
and Surfaces 16, 4153.
Bonifazi, G., Massacci, P., 1996. Particle identication by image processing. KONA No.14,
109129.
Brown, G.J., Miles, N.J., Hall, S.T., 1993. Fractal characterization of pulverized materials.
Part. Part. Syst. Charact. 10, 16.
Brown, G.J., Miles, N.J., Hall, S.T., 1994a. Characterization of mineral powders by applied
fractal geometry. Innovations in Mineral Processing, Sudbury, June, pp. 497502.
Brown, G.J., Miles, N.J., Hall, S.T., 1994b. Fractal assessment of nely ground limestone
for ue gas desulphurization. Minerals Engineering 7 (8), 10571067.
Broyles, D.A., Rimmer, H.W., Adel, G.T., 1996. Rapid shape analysis of crushed stone
using image analysis. KONA No. 14, 130137.
Burstein, M.A., Hall, S.T., Kornyshev, A.A., 1995. A study of the effect of particle fractal
geometry on otation performance. Proc. Int. Miner. Process. Congr., Soc. Min.
Metall. Explor., Littleton, Colorado, U.S.A. 3, 7780.
Castellini, C., Francini, C., Longobardi, G., Pampaloni, E., 1993. On-line characterization
of the shape and size of particles. Part. Part. Syst. Charact. 10, 710.
Clark, N.N., 1986. Three techniques for implementing digital fractal analysis of particle
shape. Powder Technology 46, 4552.
Clark, N.N., 1987. A new scheme for particle shape characterization based on fractal
harmonics and fractal dimensions. Powder Technology 51, 243249.
Ducker, W.A., Pashley, R.M., Ninham, B.W., 1989. The otation of quartz using a doublechained cationic surfactant. J. Colloid Interface Sci. 128 (1), 6675.
Feng, D., Aldrich, C., 2000. A comparison of the otation of ore from the Merensky Reef
after wet and dry grinding. Int. J. Mineral Process 60, 115129.
Frances, C., Bolay, N.L., Belaroui, K., Pons, M.N., 2001. Particle morphology of ground
gibbsite in different grinding environments. Int. J. Mineral Process 61, 4156.
Gladyszewski, G., Jaouen, C., Declemy, A., Girard, J.C., Guerin, P., 1998. Ion-assisted
deposition of Ag (001)/Fe (001) multilayers: interface roughness. Thin Solid Films
319, 4448.
Goetzinger, M., Peukert, W., 2003. Dispersive forces of particlesurface interactions:
direct AFM measurements and modeling. Powder Technology 130, 102109.
Graf, J.C., 1991. The importance of resolution limits to the interpretation of fractal
descriptions of ne particles. Powder Technology 67, 8385.
Gwaze, P., Annegarn, H.J., Huth, J., Helas, G., 2007. Comparison of particle sizes
determined with impactor. AFM and SEM. Atmospheric Research 86, 93104.
Hazlett, R.D., 1990. Fractal applications: wettability and contact angle. J. Colloid Interf.
Sci. 137 (No. 2), 527533 July.

M.M. Ahmed / International Journal of Mineral Processing 94 (2010) 180191

Hazlett, R.D., 1993. On surface roughness effects in wetting phenomena. In: Mittal, K.L.
(Ed.), Contact Angle, Wettability and Adhesion. VSP, pp. 173181.
Hazra, S., Pal, S., Kundu, S., Sanyal, M.K., 2001. Scanning probe microscopy and X-ray
studies of conned metal lms. Applied Surface Science 182, 244250.
Hurter, P., Clark, N.N., 1987. Persistence of polygonal harmonics as shape descriptors.
Part. Part. Syst. Charact. 4, 101105.
IMAGE_C: Software to Image Analyzing System 1993. IMTRON 2000. IMTRONIC GmbH,
Berlin.
Johnson, D.J., Miles, N.J., Hilal, N., 2006. Quantication of particlebubble interactions
using atomic force microscopy: a review. Adv. Colloid Interface Sci. 127, 6781.
Kaya, E., Kumar, S.R., Hogg, R., 1996. Particle shape characterization using an image
analysis technique. In: Kemal, Arslan, Akar, Canbazoglu (Eds.), Proc. of the 6th Int.
Mineral Process. Symposium, Changing Scopes in Mineral Process. Balkema,
Rotterdam, Turkey, pp. 4752. Sept. 2426.
Kaye, B.H., 1981. Fractal and shape index description of the shape and surface structure
of metal powders. In: Hausner, H.H. (Ed.), Modern Developments in Powder
Metallurgy, 12, pp. 6782.
Kaye, B.H., 1984. Fractal description of neparticle systems. Particle Characterization in
Technology, I. CRC Press, Boca Raton/Florida, pp. 81100.
Kaye, B.H., Clark, G.G., Leblanc, J.E., Trottier, R.A., 1987. Image analysis procedures for
characterizing the fractal dimension of neparticles. Part. Part. Syst. Charact. 4, 6366.
Kaye, B.H., Clark, G.G., 1989. Experimental characterization of neparticle proles
exhibiting regions of various ruggedness. Part. Part. Syst. Charact. 6, 112.
Kaye, B.H., 1991. The impact of fractal geometry on neparticle characterization. An
Invited Review at Particle Size Analysis Conference, Loughborough, pp. 300313.
Kaye, B.H., Clark, G.G., Liu, Y., 1992. Characterizing the structure of abrasive
neparticles. Part. Part. Syst. Charact. 9, 18.
Kaye, B.H., 1994. Katastrophen im Labor. Physik im Unserer Zeit 25, 2935.
Kaye, B.H., Clark, G.G., Kydar, Y., 1994. Strategies for evaluating boundary fractal
dimensions by computer aided image analysis. Part. Part. Syst. Charact. 11, 411417.
Lee, C.T., Chou, C.C., 1994. Application of fractal geometry in quantitative characterization of aerosol morphology. Part. Part. Syst. Charact. 11, 436441.
Li, T., Park, K., 1998. Fractal analysis of pharmaceutical particles by atomic force
microscopy. Pharm. Res. 15 (No. 8), 12221232 Aug.
Luerkens, D.W., Beddow, J.K., Vetter, A.F., 1987. Structure and morphology the science
of form applied to particle characterization. Powder Technology 50, 93101.
Mandelbrot, B.B., 1997. Fractals: Form, Chance and Dimension. W.H. Freeman, San
Francisco.
Meirona, T.S., Marmurb, A., Saguy, I.S., 2004. Contact angle measurement on rough
surfaces. J Colloid Interface Sci, 274, pp. 637644 June.
Meloy, T.P., Williams, M.C., 1994. Particle shape characterization, new and old methods:
does a particle have shape. Challenges in Mineral Processing, pp. 207221. Chapter 12.
Nguyen, A.V., Schulze, H.J., Ralston, J., 1997a. Elementary steps in particlebubble
attachment. Int. J. Miner. Process 51, 183195.
Nguyen, A.V., Stechemesser, H., Zobel, G., Schulze, H.J., 1997b. Order of three-phase (solid
liquidgas) contact line tension as probed by simulation of three-phase contact line
expansion on small hydrophobic spheres. J. Colloid Interface Sci. 187, 547550.

191

Ocepek, D., Strazisar, J., Kanduti-Sumej, B., 1990. Comminution and surface properties
of minerals. In: Schonert, K. (Ed.), 7th European Symposium Comminution:
Clausthal, Zellerfeld. BRD, pp. 273282 March.
Onda, T., Shibuichi, S., Satoh, N., Tsujii, K., 1996. Super-water-repellent fractal surfaces.
Langmuir 12 (No. 9), 21252127 May.
Podczeck, F., 1997. A shape factor to assess the shape of particles using image analysis.
Powder Technol. 93, 4753.
Ramesh, R., Somasundaran, P., 1990. Centrifugal immersion technique for characterizing the wettability of coal particles. J. Colloid Interface Sci. 139 (1), 291294.
Sarkar, N., Chaudhuri, B.B., 1994. An efcient differential box-counting approach to
compute fractal dimension of image. IEEE Transactions on Systems, Man. and
Cybernetics, 24 Jan.
Scheludko, A., Radoev, B., Fabrikant, A., 1968/1969. On the theory of otation: II.
Adhesion of particles to bubbles. Annuaire de l'universit de Soa, Kliment
Ochridski, Facult de Chemie, 63, pp. 4354.
Schmidt, D.C., Berg, J.C., 1996. The effect of particle shape on the otation of toner
particles. Progress in Paper Recycling, pp. 6777 February.
Schmidt, D.C., Berg, J.C., 1997. A preliminary hydrodynamic analysis of the otation of
disc-shaped toner particles. Progress in Paper Recycling 3849 February.
Schulze, H.J., Radoev, B., Geidel, Th., Stechemesser, H., Tpfer, E., 1989. Investigations of
the collision process between particles and gas bubbles in otation: a theoretical
analysis. Int. J. Miner. Process 27, 263278.
Schwarz, H., Exner, H.E., 1980. The implementation of the concept of fractal dimension
on a semi-automatic image analyzer. Powder Technology 27, 207213.
Silverstein, M.S., Breuer, O., 1993. Wettability and otation of etched ultra high
molecular weight polyethylene bers. Polymer 34 (16), 34213427.
Singh, P., Ramakrishnan, P., 1996. Powder characterization by particle shape
assessment. KONA No. 14, 1630.
Stechemesser, H., Partzscht, H., Zobel, G., 1996. A procedure for the characterization of
shape and roughness of particles by means of digitized image analysis.
Aufbereitungs Technik 37 (9), 422431.
Stechemesser, H., Ahmed, M.M., 1999. Flotation-shape and surface roughness. 50 Bergund Huettenmaennischer Tag, Kolloquium 2, Sortierung der Abfaelle und
Mineralischen Rohstoffe, Freiberg, Germany, 16 bis 18 Juni, pp. 120131.
Stoyan, D., 1993. Fraktale-Formen-Punktfelder: Methoden der Geometrie Statistik.
Akademie Verlag, Berlin.
Trahar, W.J., Warren, L.J., 1976. The oatability of very ne particles: a review. Int. J.
Miner. Process 3, 103131.
Ulusoy, U., Yekler, M., Hicyilmaz, C., 2003. Determination of the shape morphological and
wettability properties of quartz and their correlations. Miner. Eng. 16, 951964.
Varbanov, R., 1984. Flotation of spherical particles. Trans. IMM 93, C6C8 Sec. C.
Wotruba, H., 1991. Untersuchungen zur Flotativen Trennung von Mikrolith und Zirkon
unter besonderer Bercksichtigung des Kornformeinusses. Ph.D. Thesis, Fakultt
fr Bergbau, Aachen, Germany.