You are on page 1of 24

This article was downloaded by: [University of Toronto Libraries]

On: 24 December 2014, At: 03:53


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK

Mineral Processing and Extractive Metallurgy Review:


An International Journal
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/gmpr20

Flotation Kinetics
a a
N. AHMED & G. J. JAMESON
a
Department of Chemical Engineering , University of Newcastle , N.S.W, 2308, Australia
Published online: 06 Apr 2007.

To cite this article: N. AHMED & G. J. JAMESON (1989) Flotation Kinetics, Mineral Processing and Extractive Metallurgy
Review: An International Journal, 5:1-4, 77-99, DOI: 10.1080/08827508908952645

To link to this article: http://dx.doi.org/10.1080/08827508908952645

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the
Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and
are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and
should be independently verified with primary sources of information. Taylor and Francis shall not be liable for
any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever
or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of
the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
Mineral Processing and Extractive Metallurgy Review, 1989, Vol. 5, pp. 77-99
Reprints available directly from the publisher
Photocopying permitted by license only
© 1989 Gordon and Breach, Science Publishers, Inc.
Printed in Great Britain

Flotation Kinetics
N. AHMED and G. J. JAMESON
Department of Chemical Engineering, University of Newcastle, N.S. W. 2308 Australia
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

The physical variables that influence the rate of flotation are examined. The probabilistic model of
flotation is used to establish the effect of the particle size and density, bubble size and agitation on the
rate of flotarion.
In quiescent flotation, it appears that the flotation rate is limited by the particle-bubble collision and
subsequent attachment of the particle to the bubble. For fine «20 pm) or low density particles the
remedy for low recovery rates would be to either use small bubbles of the order of 100 pm. or to use
moderate to high agitation with larger bubbles.
In the usual turbulent conditions. the limit is set by the destruction of the bubble-particle aggregates.
Broadly speaking, the same parameters favour both attachment and detachment so that the ultimate
flotation rate is a compromise between these two competing mechanisms.
The bounds which define the best agitation level and bubble size to use are strong functions of the
particle size and density. This results in conflicting requirements for the optimum flotation of the fine
and the coarse particles. Best conditions for the flotation of each are indicated.

INTRODUCTION

For a reasonable understanding of any process the quantities that must be evaluated
may be divided into three groups: (i) experimental methods determining the rate of
the process; (ii) effects of the process variables; and (iii) the mechanism or the
equation denoting the rate. For flotation however, a conclusive determination of
these quantities remains an elusive goal after more than half a century of "more
systematic" investigations. It would appear that even today the flotation cell, the
heart of the process, remains what Kitchener! terms a "magic box", with a number
of variables being studied empirically in both theory and practice. Consequently,
flotation cells are invented, not designed.
The basics of flotation kinetics and the problems and methods in flotation
modelling and simulation have been the subject of a number of reviews in recent
yearsv". The fundamental mechanisms involved in flotation are yet to be fully
established. The complex nature of the process, where a number of separate effects
take place, some competitively and some consecutively, precludes the evaluation of
a suitable working model from first principles. This area of flotation has been
discussed by Kitchener ', Barbery' and Inoue et ai.6 , and the inference is that there
is no prospect of predicting theoretically the performance of a flotation cell unless
more is known about the characteristics of three-phase turbulence.
77
78 N. AHMED AND G. J. JAMESON

An alternative method, is to study the effect of various parameters on the


flotation rate in controlled experiments. The performance may then be linked to
these changes to develop optimum operating conditions, or control strategy, on one
hand and search for clues to the actual mechanics on the other. This article adopts
this approach to establish the effects of the physical variables like the particle size
and density, the bubble size and the system turbulence. The probabilistic or micro-
kinetics approach is found to be convenient in explaining the role of these para-.
meters in determining the overall flotation rates. Best values for these variables
may be suggested for optimum flotation performance.
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

FLOTATION SUB-PROCESSES

A large number of variables, both chemical and mechanical, may affect the ulti-
mate performance or rate of flotation. Although the physical variables are under
general discussion here, it is the chemical and thermodynamic criteria which deter-
mine the hydrophobicity of a particle, a prerequisite for flotation. It is the
ingenuity of the physical and surface chemists who have made the selective bene-
ficiation of minerals possible. Professor Jan Leja's impressive tome? is a testimony
to that effort, as is his personal contribution to the field over the past several
decades.
The flotation process, although complex, may be conceptualized in terms of a
number of sub-processes. A reasonable break up of the sub-processes would be:
i) The introduction of feed materials.
ii) The attachment of particles to bubbles.
iii) The transport between pulp and froth.
iv) The removal of flotation product and tailing.
Each of these sub-processes may again be divided into microprocesses and in
each a number of separate effects take place. One conclusion is that it would be
more profitable to study the different sub-processes separately, determining the
effects of the important variables in each case, in order to better understand the
overall operation i.e., combining the micro-level interactions to develop a macro-
level model. Recent investigations suggest a general trend in that direction. This
development seems to be the most logical, if a complete description of the flotation
process is to be ever developed from the fundamentals.
There is a general concensus that flotation rate is strongly influenced by the sub-
processes (ii) and (iii).

Attachment of Particles to Bubbles


As a bubble rises through the pulp it encounters particles of ore or gangue. Pro-
vided the ore particle is hydrophobic and sufficiently near the bubble, coalescence
FLOTATION KINETICS 79

will occur. If the adhesion is strong enough, the aggregate will rise to the top with
the bubble. There is photographic evidence of particle-bubble collision and at-
tachment in the aqueous phase and this is assumed to be the rate controlling step
in flotation":".

Transport Between Pulp and Froth


When a bubble reaches the pulp-froth interface it remains beneath the froth, while
the liquid layer, separating the two, gradually drains away. New bubbles arriving
behind push the ones in question further into the froth zone with its load of ore.
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

The liquid film between bubbles in the froth consists of values as well as some
gangue that may have been entrained. Attempts are made to sweep away the
gangue by washing or irrigating the froth. The froth on the top of the cell is swept
away with its load of particles. It is recognised that the froth phase interactions play
a significant role in the flotation efficiency, especially in relation to the product
grade. The various aspects of froth phase transparent processes are the subject of a
number of articles in the present volume and outside the scope of this paper. Here
we will make the usual assumption that the rate controlling parameters in flotation
are determined by the effects occurring in the pulp phase, i.e., once a hydrophobic
particle enters the froth zone, it does not return to the pulp.

THE RATE OF FLOTATION

There are no standard procedures for determining the rate of flotation. From a
chemical kinetics analogy, the kinetics of bubble-particle collision and coalescence
in the pulp may be defined by the ordinary differential equation:
(1)
where Cp and Cb are the concentrations of the particles and bubbles respectively;
t is the flotation time; nand m are the respective orders and k is a pseudo rate
constant, a complex function, dependent on the various parameters governing the
flotation process. If the air supply is constant, any tendency for a change in the
bubble concentration is small. In such a situaiton the rate equation converts to:
-dCp/dt = kC; (2)
Using first order kinetics, i.e., n = 1
-dCp/dt = kC p (3)
Using the boundary conditions
C = Co at t =0
C = C, at t =t
we obtain:
80 N. AHMED AND G. J. JAMESON

In(CrICo) = -kt
or Cr=Coexp(-kt) (4)

The assumption of first-order kinetics is convenient and simple and has been
used satisfactorily for decades since it was first suggested by Zuniga 10 and used by
Gaudin et al. II However, the controversy regarding the order of the process has
never been resolved. The main debate is between the first and second-order,
although Bogdanov et al. 12 report orders which varied from one to six in their
experiments. Klassen and Mokrousov':' are of the opinion that the parameter "n"
should most frequently equal one, less frequently two and seldom three or more.
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

Arbiter!" used a graphical test to support the validity of second-order kinetics, but
warned that the method was not reliable. De Bruyn and Modi'" found the rate to
be first-order for particle sizes below 65,um provided the solids content of the pulp
was less than 5.2%. Tomlinson and Fleming found the order to be one when the
solid concentration in the pulp was small (uninhibited flotation), and zero when the
concentration was high. Arbiter and Harris? and Klassen and Mokrousov':' discuss
the various defects of assuming a chemical kinetics analogy and using these
equations essentially derived for a batch process for steady state systems. Some
authors have argued that the vlaue of "k" changes with time and as such is a
variable function!". In general, a survey by Inoue" shows the use of the first-order
equation in the majority of cases. More recently, Dowling et al. 17 report flotation
tests with a copper ore using various collectors and frothers to produce different
time-recovery profiles. These were then used to test thirteen rate models from
the classical first-order chemical analogy model to the three-parameter gamma
distribution of Huber-Panu et al. 18 The flotation of the copper ore was shown to be
essentially a first-order process, and all the models tested were found to give a
reasonably good fit to the experimental data, though some models were clearly
better than others. The best overall agreement was obtained for the first-order
model with a rectangular distribution of particle floatabilities!". Further details on
the nature of flotation rate constant and various manipulations which have been
tried to obtain a reasonably accurate picture of a real cell in flotation modelling,
process simulation and plant design may be obtained from the literature. 1-5.20-22

THEORETICAL DEVELOPMENTS

The ultimate success of the flotation process depends on the capture of hydro-
phobic mineral particles by bubbles in the pulp phase and the successful transfer of
the aggregate to the froth phase. In the earlier days of flotation, there was con-
siderable controversy regarding the mechanism of particle-bubble attachmenr'". To
rival the collision theory was the belief that bubbles grew by preciptation from
solution on the surface of the particles at low pressure regions behind the agitation
impellers. However, the high-speed cine-photographic evidence of Bogdanov and
Filanovskr" and others 2S-26 has proven beyond doubt that discrete particle-bubble
collision is a prerequisite for the attachment of particles to bubbles except in
FLOTATION KINETICS 81

systems that are designed specifically to form bubbles from super-saturated solu-
tions (dissolved air flotation).
The removal rate of particles or the rate of flotation from the pulp therefore, is
the consequence of:
i) collision between particles and bubbles;
ii) adhesion of particles to bubbles; and
iii) detachment or otherwise of particles from bubbles.
Putting it differently,
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

chance of recovery (R) = probability of bubble-particle collision (Pc)


x probability of attachment (P a )
x probability of detachment (P d)
i.e., R = P; . Pa • (1 - Pd ) (5)
Thus (i)-(iii) may be viewed as the micro-processes in the pulp phase and form
the basis of the probabilistic model for the determination of the flotation rate,
which views the recovery in terms of the probability of success (P) of a sequence of
events occurring in the cell. The rate is then open to mathematical analysis on the
basis of the factors entering into the evaluation of each probability. This method of
analysis, which is now well established, was first put forward by Schuhmann?" who
considered P; and Pa . Sutherland/" took into consideration the term Pd while
Tomlinson and Flerning/" also introduced the probability of a particle being
retained in the froth. Each of the micro-processes above will have its own con-
trolling factor or factors and the summation of these factors will result in an
observable overall effect on the flotation rate.

Particle-Bubble Collision
The motion between particles and bubbles in a conventional agitated flotation
machine is extremely complex and few mechanisms of collision for such systems
have been proposed. The main difficulty is the determination of the relative motion
between the particles and bubbles. A convenient first step is to model the approach
of a particle to a single bubble rising in an infinite quiescent body of liquid; an
analogous situation being the particle deposition on fabric filters". To collide with
a bubble, a solid particle must have sufficient momentum to resist the tendency to
follow the streamlines around the bubble. Gaudirr" considered the particles to
have either a negligible settling velocity or one in the Stoke's regime, but the
chance of encounter with a bubble was found to be zero because of the erroneous
assumption that the two bodies move independently in a fluid. Sutherland'"
introduced the concept of a "collision efficiency". Imagine a bubble of radius 'b
rising vertically through the pulp (Figure 1). Sutherland showed that a stream of
collision radius R can be defined, such that all particles within the tube will
eventually be captured, and related to the bubble radius, as
82 N. AHMED AND G. J. JAMESON
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

FIGURE 1 Definition sketch for the collection efficiency, E = ("R2)/("r~) for a bubble moving
relative to the liquid.

Rlr.; = (3rplrb)"2 (6)


Since according to Sutherland all particles lying within a distance R from the line of
motion will collide with it, a collection efficiency E may be defined as the ratio of
the area of the collision tube (nR 2 ) to the projected area of the bubble (nr~), so
that
E = R21r~ = 3rplrh (7)
Remembering that the volume swept out by a bubble is proportional to r~, the
average number of particles picked up by a bubble (assuming constant E) should
be independent of bubble size but proportional to the bubble frequency i.e., to r/;3,
at constant gas flow rate. The collision or flotation rate should therefore vary as
dpld~.
Derjaguin and Dukhirr'? made the first attempt to present a unified concept of
FLOTATION KINETICS 83

the flotation kinetics of fine particles but their hydrodynamics was oversimplified in
neglecting the gravitational effects, which implied that very small particles,
somewhere less than 10 .urn, would not be collected.
Flint and HowartlrP, taking the gravitational effect into consideration, char-
acterised particle behaviour by two dimensionless groups:
K = !2pd~Ub/18/ldb' and G = (!2p - !2t)d~g/18/lUb (8)
where (lp and (It are the particle and fluid density respectively, Vb is the bubble
velocity and u the fluid viscosity. For fine particles (K < 0.1) the collision efficiency
is independent of K but strongly dependent on G, so that collision efficiency in-
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

creases with decreasing bubble size. For larger particles (K > 1) collision efficiency
is strongly dependent on the particle inertial effects and increases with increased
bubble size. Modifications to the above theory were made on the basis of particle-
bubble trajectory by Reay and Ratcliff" and on the basis of hydrodynamic drag by
Collins'".
Flint and Howarth and Reay and Ratcliff obtained E oc (dp/d h )" , where n
approximates 2 while Collins' analysis results in E cc db·3/d~ i.e., the flotation rate
oc d b3 in all cases.
Reay and Ratcliff34 also investigated the collision efficiency for particles so small
that Brownian diffusion becomes the primary capture mechanism. They deduced
that:
(9)
and the predicted collision efficiency shows a minimum at 0.6 .urn.
Equations have been solved for streamlines around larger bubbles to predict the
collision effici ency36-38 . However, the basic assumptions remain the same and the
predicted behaviour fall within the range determined by Flint and Howarth and
Reay and Ratcliff. The topic is discussed comprehensively by Yoon and Luttrell in
the next chapter.
Fundamentally the collision models discussed above depend on the solution of
equation of motion around two bodies, which is analytically possible in few special
cases and far removed from actual mineral flotation conditions. However, they
retain their importance as the few attempts to explain bubble particle collisions,
especially for small bubbles and particles.
Flotation cells operate under intense agitation and theories of collision which
assume a viscous flow of fluid past bubbles are not applicable. Levich'" proposed a
model for collision which assumes that the deformation of streamlines by intense
turbulence is complete and encounters on the basis of a straight line motion,
would be realised. The final result shows a collision rate which is proportional to
id, + d j )3, where d, and d, are the diameters of two species.lt is interesting to note
that when applied to particles and bubbles where db is expected to be much larger
than dp , a significant collision rate is predicted for even small particles.
Further suggested expressions have been adopted from models developed in
other contexts, for example, for rain drop collisions?", for coagulation'", etc. and
applying them to particle-bubble collision in stirred tanks. As the knowledge of
84 N. AHMED AND G, J. JAMESON

three-phase turbulence is in the early stages of development, Kolmogoroff's


dimensional analysisf for isotropic turbulence, as applied to stirred tank, seems
to have been used to evaluate the micro-level interactions. This is usually ac-
complished by using a macro-level and measurable input into a stirred system,
usually the power per unit volume (PIV) or power per unit mas (Ph!! V) denoted by
10, in the line suggested first for flotation by Mika and Puerstenau'". Thus, Schubert
and Bischofberger'" proposed an expression for collision between particles and
bubbles, given by:
z = 5NhNpdf(VV~ + VV~) (10)
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

where d, = O.5(dp + db) and VV~. VV~ are the effective r.m.s. values of relative
velocity between the particle and bubble respectively, and the fluid. Assuming
isotropic turbulence, these velocities were calculated in terms of the energy
dissipated per unit mass, e. This expression is an adaptation of Abrahamson's
analysis'" for collision of small particles in a turbulent fluid.
Another mechanism which may be of interest has been applied to colloid col-
lision in turbulent flow in flocculators by Camp and Stein'", where the number of
collisions is given by:
z= 4/3NiNjd~(dv)112 (11)

where d ij = O.5(d; + d j ) and v is the kinematic viscosity of the fluid. The expression
is again essentially that of the Smoluchowski'" mechanism for gradient coagula-
tion in laminar flow, the gradient being expressed as (flV)I/2 from Kolmogoroff's
dimensional analysis'".
Nonaka et al. 47 predicted the turbulent collision between monodispersed
particles and a single bubble based on a diffusion model, giving the number of
collisions as:

N; = 4nR 2[2D tanpla']'~R (12)


where R = (db + dp ) and D, is the turbulent diffusion coefficient. Assuming
isotropic turbulence and substituting Komogoroff's diffusion coefficient they found
the collision rate to be proportional to eu.5N~/5, where Nafis the air flow number.
Interestingly, the collision rate constants calculated from the model correlate well
with the flotation rate constants, measured from a series of tests with quartz
particles, showing a direct link between the micro and macro-kinetics. This area of
flotation has been reviewed by Inoue et al.6

Bubble-Particle AttachmentlAdhesion
A sequential mechanism for bubble-particle adhesion involves (i) approach of a
particle to bubble, (ii) thinning of the water film between particle and bubble to
rupture thickness; and (iii) receding of the residual film to give an air-solid
interface. It is believed that step (ii) is the controlling mechanism for adhesion.
Sven Nilsson'" postulated the existence of an induction time, a finite time of contact
during which the above steps take place. During this period the particle will be
FLOTATION KINETICS 85

swept around the surface of the bubble and into the fluid and will fail to adhere if
the induction time is larger than this contact time. This concept is discussed in the
next chapter by Yoon and Luttrell.
Sutherland" derived an expression for the induction time which is dependent
only on the angle of collision from the centre of the line of motion of the bubble and
is independent of the particle size. Philippoff" and Evans'" developed expressions
for the time of contact (tc) on the basis that a particle hits a bubble, deforms it
elastically and then bounces off as the bubble regains its shape. Both derive
t c ex m 1/2 , where m is the particle mass; the induction time therefore is expected to
increase as (!p and d~·5. These equations have been discussed by Mackenzie and
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

Mathesorr", Arbiter and Harris", and Klassen and Mokrousov':', Mackenzie and
Matheson show that the film thickness between the particle and bubble may be
thinned by increasing the particle velocity, i.e .. increase the agitation in the system.
This would probably decrease the contact time but that point has not been
considered.
Meloy'" has severely criticised the use of static analysis for what essentially is a
dynamic process and points out that the contact times evaluated from the above
equations are orders of magnitude smaller than actually established experimentally
by Glernbotskii'".
Although the contact time hypothesis is not favourable for very fine particles,
other forces come into play in that situaiton. According to Derjaguin and Dukhirr"
once a small particle is close to the surface, diffusio-phoretic forces (a combination
of surface forces and ion diffusion) could be sufficiently strong to effect the final
thinning and rupturing of the film. Collins and Jamesonvhave demonstrated the
effect of electrostatic forces. By reducing the zeta potential of neutrally buoyant
particles (4-20 .urn), substantial increases in flotation rates were observed with
53.um diameter bubbles.
Jowen'" attempted to solve the liquid thinning problem by considering the case
of drainage of liquid from between two solid discs. The ultimate contact time is
dependent on the particle terminal velocity and is given as:

(,(laminar) = constant (13)


(,(turbulent) ex d~·5 (14)

(,(transitional) ex dp (15)

The induction time is proportional to the bubble size implying that smaller
bubbles would promote faster adhesion. This conclusion also follows from the
capillary theory of adhesion V,
In turbulent flow the bubbles deform continuously. The pulsating bubbles will
behave differently than those in a quiescent system and the effects cannot be
generalized. However, the higher momentum of collision is expected to increase
the chance of thinning and rupturing the disjoining film. Inoue et al. 6 are of the
opionion that a model for describing the attachment step as a rate process is one of
the most essential problems in flotation micro-kinetics research.
86 N. AHMED AND G. J. JAMESON

Particle-Bubble Detachment
To ensure the stability of a particle-bubble aggregate in a static system, the distinc-
tive forces associated with the weight of the particle must be balanced by the
restoring force of surface tension. The balance between these forces determines the
maximum size floatable. Morris'" derived a static relationship linking the surface
tension and the bubble and particle size. The chances of retention improves with
the hydrophobicity of the surface, increasing bubble size and decreasing particle
size. Early investigations on this topic have been discussed by Gaudirr" and
Klassen and Mokrousov':', Obviously, for detachment the static situation is not
appropriate. Mika and Fuerstenau'" provide a critical review of development to
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

date and, based on the assumption of isotropic turbulence, find the detachment rate
to be proportional to dJ:3.
The concept was further developed by Schulze 59-6 1 by arguing that, once a stable
three-phase contact is formed, the particle can only be pulled away by some
external factor, in this case the energy of the turbulent field.
According to Schulze a particle-bubble couplet caught in a turbulent eddy will
rotate with a frequency appropriate to the eddy size, and if the kinetic energy of the
particle exceeds the work of rupture, the particle will detach (Figure 2). Assuming
isotropic turbulence, the stress on the couplet was found to be proportional to
[e(dl' + db Wl3 . The maximum size flotable was a function of the pH, through its
influence on the contact angle, a measure of the tenacity of attachment.
Jowett'" in a similar argument considers the rupture force in terms of the centri-
fugal force, a "g" factor, developed by the gyration of turbulent eddies in the
system. The very nature of the argument suggests that the detachment force will
increase with decreasing bubble diameter and increasing particle sizes.

PRESSURE. OF LIQIJIU

P\{SSSll~E OF CAS

a b

FIGURE 2 Mechanism of detachment of a particle from a bubble in a turbulent eddy. (a) The bubble
rotating around its centre. (b) The forces acting on the particle.
FLOTATION KINETICS 87

EXPERIMENTAL

The Effectof Particle Size


Gaudin and co-workers 11.62 appear to have initiated the study of flotation rate as a
function of the discrete particle sizes, analysing the recovery of lead, zinc and
copper sulphide particles up to 50,um in diameter. They obtained a linear relation-
ship down to 4,um below which the flotation rate was constant. Much of the earlier
work on the particle size effect has been reviewed by Sutherland and Wark 63 ,
Klassen and Mokrousov':' and Arbiter and Harris", Further experimental evidence
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

was gathered in flotation cells among others, by de Bruyn and Modi'", Morris'",
Bushell?", Mackenzie and Matheson'S and reviewed by Jameson et al.", Trahar'"
and Ahmed?". Further work has been reported since then and they all follow the
general pattern of the form shown in Figure 3. Recovery is small for fine particles,
followed by a maximum and then again a decrease for larger particles. Another
common factor is the flotation rate constant-particle size relationship. The

100

Qo

Z
0 80
j::::
oc(
a:
u,
70
0 GALENA
W
N 0 SPHALERiTE

••
fi) 60 PYRITE
:t PENTLANDITE
0
a:
u,
50

>-
a:
w 40
>
0
oW
a: 30

20

10

0
tooo
AVERAGE PARTICLE SIZE (MICRO METRES)

FIGURE 3 Size-by-size recovery of some sulphide minerals afler 60 seconds flotation in timed batch
tests. [Adapted from Jowetr", with the permission of AIME. Inc.]
88 N. AHMED AND G. J. JAMESON

10r----r-----,.---,.---,.----..,----,

t-
Z
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

0(
t-
rJl
Z
o()
W
t-
o( o. o SPHALERITE pHS
II:
o GALENA pHB

~ CHALCOC11'E pH8

\l PYRITE pHS
o CHALCOPYR 1TE pHR

10Om~ KEX

o.tL- L- .L- .L- .L- ...L_ _----J


I 10 20 50 100
AVERAGE PARTICLE SIZE (MICROMETRES)

FIGURE 4 Variation in flotation rate constant with particle size for the balch flotation of several
sulphide minerals. [From Trahar'", by courtesy of Elsevier Science Publishers].

relationship is approximately linear of the form given by Trahar'" for various


minerals (Figure 4).
This behaviour is in sharp contrast to information available for quiescent
systems 29 •67-69 which have been reviewed by Jameson et al." and Ahrned'". The
trend is unmistakable. Expressing the effect of particle size on the flotation rate
constant as k oc d~, for a variety of bubble sizes (40-1000 ,urn diameter), the value
of n is found to be greater than one and in most cases close to two. Literature to
date, therefore, suggests a particle diameter dependence of the form
(16)
where

n> 1 in quiescent conditions


n < 1 in turbulent conditions
FLOTATION KINETICS 89

Moreover, in non-agitated flotation, the upper limit to the particle size that may
be floated is much higher than in turbulent conditions and is not a strong function of
the density. Thus, Tomlinson and Fleming" show increasing rates for apatite
particles up to 300,um. Most probably, in such cases, the upper limit is set by the
buoyancy of the bubble-particle aggregate. On the other end of the scale, it is very
difficult to float low density fines «20 ,urn) and indeed it is only made possible by
the use of bubbles less than 100,um in diameter67 .68 . This agrees well with the
theoretical predictions for quiescent systems.
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

The Effect of Bubble Size and Turbulence


The importance of bubble size in flotation was noted by Cassell et at.!" Bennett
et al," and De Vivo and Karger ? found smaller bubbles to be more efficient.
Reay and Ratcliff 67 using only two bubble sizes td; < 100,llm) in a quiescent
system found the collison efficiency, E ex d/;1.9 (i.e., k ex d/;2.9) while Anfruns
and Kitchener''" using bubbles between 500 and 1000 Jim in diameter found the
collision efficiency to be proportional to d/; 1.69 Ahmed'" summarised the available
data as:
(17)
k ex: d;;2.6; 100 f'm < d6 < IOOOllm (18)

Since then Yoon and Luttrell" have reported


300 lim < d; < 450 .um (19)
also for a quiescent system.
lt should be noted that bubble size and turbulence are intimately connected in
stirred cells. Accordingly, inferences from results obtained by monitoring only one
parameter, especially in the case of turbulence 52 .74 , must be viewed with caution.
This has been demonstrated by the authors" in experiments in which all the
variables, including the bubble size, could be varied independently of each other.
Essentially, the flotation apparatus consisted of a standard stirred tank with a
specially designed aerator at the bottom. Sintered discs of various porosities were
installed in the aerator to generate bubbles with mean diameters of 75, 165.360 and
655 Jim in the presence of 150 ppm of the frother (MIBe). There was no statistical
difference in the bubble size for agitation up to 600 rpm and this represented the
upper limit of agitation. Three solids, with particle sizes ranging from 4 to 42,l1m,
were chosen to provide a wide density range; polystyrene latex. (spherical; s.g.
1.05), quartz (s.g. 2.65) and zircon (s.g. 4.56).
Figure 5 to Figure 7 show the flotation rate constants as a function of the particle
diameters at various levels of agitation for latex, quartz and zircon. With no
agitation it was not possible to measure the flotation rates of the latex particles with
the two largest bubbles. With an increase in agitation, the rates increase and at
300 rpm bubbles of all sizes are effective. With further increase in agitation, the
flotation rate increases but the rate of increase is inversely related to the bubble
90 N. AHMED AND G. J. JAMESON

size, and directly to the particle size, for a particular bubble size. Consequently,
agitation has a significant effect on the rate constant-particle diameter relationship.
For latex with no agitation, it has the approximate form k DC d~·5. However, as seen
in Figure 5b and Figure 5c the exponent decreases with an increase in agitation and
becomes nearly the same « I) at 600 rpm for all bubble sizes. Similar effects are
observed for the quartz particles. For the heavier ziron particles (s.g. 4.65) the
rate constant passes through a maximum as the particle size increases, as usually
observed in flotation cells.
The effect of bubble size can be seen in Figures 5- 7. For quartz and zircon, at
the minimum suspension speed, there is almost a fifty-fold difference between the
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

flotation rates with the smallest and the largest bubbles. With increased agitation
this ratio gradually decreases because of the unequal increase in flotation rates for
the larger bubbles. At 600 rpm, for quartz and zircon, the rates for the three larger
bubbles actually fall below that obtained at minimum suspension speed. As
mentioned above, with the neutrally-buoyant latex, larger bubbles are effective
only under agitation and the flotation rates are a direct function of the bubble size
with increasing agitation.
Expressing the bubble size-flotation rate constant dependence as k DC dZ, it is
observed that II was never nearly-equal to 3 as suggested by theoretical considera-

2
10- ,--- - - - - - - - - - ,--- - - - - - - - - -,- - - - - - - --;-;;-:7;1
(1.]0)

(1. 'i0)
,
~

(1.51\)
~ 10-)
s...
......
z '.

'"oz
o
~
.
a:
10-
10

a no agitation b 300 rpm C 600 rpm

10 20 30 10 20 Jo 10 20 .10

PARTICLE DIAMETER (MICROMETRESI

FIGURE 5 Latex particles: the measured rate constants plotted ad a function of particle diameter, for
the stirrer speeds and bubble diameters shown. The braeketted figure gives the slope of the straight line
of best fit, through an individual set of data points. (Bubble size: fc.-75 I'm; 0-165 I'm; 17-360
I,m; 0-655 I,m.) .
Note: the flotation rates for the two larger bubble sizes (360 I'm and 655 I'm) were too small to be
measured. with no agitation.
FLOTATION KINETICS 91

~:
Y:7
~\O
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

a
10 20 40 5 10 20 40 ID 20 40
PARTICLE ClAM ETER (MICROMETRES)

FIGURE 6 Quartz: the measured rate constants planed as a function of particle diameter, for the
stirrer speeds and bubble diameters shown. The bracketted figure gives the slope of the straight line of
best fit. through an individual set of data points. (Bubble size: /:,-75 pm; 0-165 pm; '\7-360 pm:
Q-655 pm).

10- 1 , , , , ,
l- -

~
~

r ~-
CJ
~ 10-
2
~
I-
~
I-
, I-
0 -
'"
z

V r
0

/"
CJ

~ 10- 1 I-
<
a:
I-

]0- 4 l..-.
a r? to
30,0 r~m
zo 4U
b
,
4
I

'0
500 r p,:,
zo 4"
c
,
10
600 rpm
2" 4"
PARTICLE DIAMETER (MICROMETRES)

FIGURE 7 Zircon: the measured rate constants planed as a function of particle diameter, for the
stirrer speeds and bubble diameters shown. (Bubble sizes: /:,-75 pm; 0-165 .um; '\7-360 pm;
Q-655 pm).
92 N. AHMED AND G. J JAMESON

tion for quiescent systems. For quartz and zircon the exponents are of the same
order of magnitude at the minimum suspension speed. But decrease with increasing
agitation and are below unity for all the solids at 600 rpm.
It was observed that the system agitation altered the relationships between the
flotation rate constant and the particle and bubble sizes. The reasons become more
apparent from Fiugre S and Figure 9 which show flotation rates as a function of
impeller speed for some selected particle sizes of latex and quartz respectively. For
zircon the range of speed is too small for a graphical representation of significance
and the rate constant data are used directly from reference?",
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

In the case of light latex particles there is an increase in rates up to 600 rpm for all
the particle sizes with bubbles of all sizes. Another factor, also observed earlier,
which is now clearly indicated is that the dependency of rate on the impeller speed
N is directly related to the bubble size. For example for 26,um diameter latex
particles, there is a small dependence of the flotation rate (k oc No. 56 , Figure Sa) on
impeller speed with 75 ,urn bubbles. As the bubble size increases, this dependency
on the impeller speed increases to such an extent that for 655 ,urn bubbles it is
nearly four times (k oc N1.96 ) that with 751tm bubbles. This observation also seems
to be an indication of the unbalanced increase with agitation, in the attachment and
detachment forces on the particle-bubble aggregates with changing bubble size,
especially for the larger particles. It will appear that with agitation, the probability
of attachment of the particles to bubbles increases due to the enhanced momentum
of the particles. The detachment probability or force increase with a reduction in
the bubble size for the same level of agitation. For the larger bubbles therefore,

! lIO 200 40\1 hOD I UO 200 400 soc IUU

IMPELLER SPEED (rpm)

FIGURE 8 Latex: The measured rate constants plotted as a function of the impeller speed for the
particle and bubble sizes shown. The figures in brackets give the slope of the best fit through the data
points. (Bubble sizes: t:.-75 I'm: 0-165 I'm; '7-360 I'm; 0-655 I'm).
FLOTATION KINETICS 93

,
~

ow

.
~ 10 2

.
Z
.. 5

'"o
Z
o
w
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

:t 10]
a:

dp
5.1 I'm
I I ,
200 400 /)00

IMPELLER SPEED (rpm)

FIGURE 9 Quartz: The measured rate constant plotted as a function of the impeller speed for the
particle and bubble sizes shown. (Bubble sizes: !'>-75 I'm; 0-165 I'm; \J -360 I'm; 0-655 I'm).

increased collision and attachment chances result in a sharp increase in the flotation
rates with agitation. For the smaller bubbles, although the detachment forces
cannot predominate at this density, they mitigate the effects of the increased
collisions and attachments, resulting in a lowering of the dependency on the
impeller speed and hence the difference in the slopes of the k ~ N plots with
various bubble sizes.
With a decrease in the particle size (10 ,urn and below; Fiugres 8c, 8d), the
detrimental effect of agitation is gradually reduced as indicated by the growing
dependence of the rate constants on impeller speeds, even for the smaller bubbles.
This is most probably because the enhanced momentum of the very fine particles
gain advantage over the disrupting influences. In other words the detachment force
weakens with decreasing particle size.
A study of k-n curves for quartz (Figures 9a-9d) is very rewarding as they
demonstrate clearly and simply the manner in which the flotation rates behave with
agitation for the denser particles. The role of the bubble size is also very
conspicuous. Attention is drawn to the wider particles size range, the highest
particle diameter in this case is 41.2,um instead of 26,um for latex.
For particles of 42 ,urn diameter (Figure 9a) there is an increase in rates for all
bubble sizes up to 300 rpm, above which rates begin to decrease for bubble sizes up
to 360,um diameter. The amount of decrease is inversely related to the bubble size,
that is, it IS more acute with the 75,um bubbles, little less with 165.um bubble and
so on. For 75 ,urn bubbles the rate at 600 rpm is less than at 100 rpm. But on the
other hand, the rates keep increasing up to 600 rpm for 655 ,urn bubbles and the
relation between the rate constant and the impeller speed is approximately linear.
94 N. AHMED AND G. J. JAMESON

With a decrease in particle diameter, the basic behaviour of the k-N relation
does not change considerably. It may be noticed though, that for particle sizes
20.6.um and below (Figures 9b-9d), the fall in rates for bubble diameters 75 .urn
to 360.um above 300 rpm start becoming relatively less severe but this fall in rate
is still inversely related to the bubble size. The behaviour of 655 .urn bubbles
remains approximately the same for all particle sizes.
For zircon, the trend was similar to those for quartz. For 31.6.um particles there
was a decrease in rates with agitation for 75.um and 165.um bubbles above 300 rpm,
which in this case was the minimum speed for suspension. With 655.um bubbles the
rates increased systematically with agitation for all particle sizes. With a reduction
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

in particle size, the effect of agitation was reduced in that the decrease in rates with
agitation becomes less intense as for quartz.
These observations would suggest that:
i) With increased agitation, the particle-bubble collision probability increases in
order of increasing bubble size. For the same gas flow rate and agitation, however,
the actual number of collisions increase with decreasing bubble size because of the
increase in the number of bubbles.
ii) Bubble-particle attachment probability is inversely proportional to the
bubble diameter and increases with particle density and size and agitation.
iii) Detachment probability is inversely proportional to the bubble diameter
and directly proportional to the particle density, particle size and agitation.
iv) The observations offer support of a qualitative nature to the theoretical
predictions taking turbulence into consideration. Models for quiescent systems are
irrelevant in agitated conditions.

BEST OPERATING CONDITIONS FOR A FLOTATION CELL

It appears that in agitated conditions, the ultimate flotation rate depends on a


balance between the attachment and detachment effects and the bounds which
define the best impeller speed and bubble size to use, are strong functions of the
particle size and density.
For low-density or fine particles there is an appreciable increase in flotation rates
with agitation for all bubble sizes but there is still an order or magnitude difference
in the rates obtained with 75 .urn and with 655/lm diameter bubbles, at the highest
speed. Low density fines, e.g., in coal flotation show very slow recovery. The
results suggust that in such systems, the use of smaller bubbles at moderate to high
agitation would be beneficial.
For coarse or high density particles the advantage of using smaller bubbles is
reduced, when the agitation level is above that required for complete suspension of
the particles. In processing of such particles, as would occur in most ore dressing
circuits, the agitation requirements would have to be optimised with regard to the
bubble size for most efficient operation.
FLOTATION KINETICS 95

DETACHMENT

NO LOW

LIFT RECOVERY
o
w
W
D.
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

Ul SEDIMENT AT ION
c:
W DETACHMENT
...J
...J
W
D.
~ NO LOW
.:FINE:· ..
L1FT+ ----11- RECOVERY

SEDIMENTATION

BUBBLE SIZE

FIGURE 10 Zones of best flotation performance for particle of different sizes.

The existence of an optimum field of operation for various particle sizes has been
demonstrated also by Schubert and Bischofberger?" and Schubert et al. 77 in their
study of the hydrodynamics of flotation cells. In experiments with cassiterite and
sylvite flotation, they found that for fine particles «125 ,urn), an increase in speed
above the minimum suspension value resulted in increased recovery. On the other
hand, for the coarse particles (especially above 500 ,urn), recovery was a maximum
at minimum suspension speed, and decreased with further agitation. As they were
using a standard flotation cell we may assume the average bubble size to be in the
vicinity of 600 pm 78.70 and the increased recovery of fines with agitation agrees
with our results for 655 pm bubbles. In Schubert and Bischofberger's experiments,
the change in recovery behaviour starts at an impeller speed above the minimum
suspension value and with a particle size of 125 ,urn, i.e., with dp/d b of 0.2
approximately. In the experiments reported here, for smaller bubbles and zircon
particles, the rates begin to decrease beyond the speed of minimum suspension.
Taking the average particle size to be 15 pm and considering the 75,um bubbles,
dp/d b is 0.2, similar to the case of Schubert and Bischofberger. Thus, it is likely that
the optimum recovery point is the minimum suspension speed at this dp/d b ratio.
The conflicting requirements of the different particle sizes can be illustrated
diagrammatically as in Figure 10. With flotation feeds containing a wide range of
particle sizes, it is unlikely that a single flotation cell can be devised to give
optimum recovery over the whole particle size range. This gives strong support to
the views of Trahar'" and Adorjan'", who suggested that serious consideration
96 N. AHMED AND G. J. JAMESON

should be given to the design of separate circuits to handle "fine" and "coarse"
particles, using conventional flotation cells with large bubbles (-650 ,urn
diameter). Suppose for example we had a dense ore of s.g. around 5.0 to be floated,
with a size range of 5,um to 125,um. If the incoming feed could be split so that the
values were in two size ranges, e.g.. 5 lim to 30 ,urn; and 30,um to 125 ,urn, the
observations would suggest that in conventional floation cells, the fines should be
floated with moderate to high agitation, whereas the coarse fraction would respond
best to stirring at just above the minimum speed for complete suspension. If on the
other hand, a bubble size distribution could be generated in the volumes required,
with a considerable proportion less than 100,um in diameter, particles of all sizes
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

could be floated at high rates of recovery, with agitation at the minimum speed for
suspension. (Large bubbles of order 500,um diameter will still be needed for the
largest particles).

CONCLUDING REMARKS

Looking at the probabilistic model of flotation, and considering the pulp phase
only, it would appear that in quiescent systems. the limiting factors for success-
ful Rotation are the bubble-particle collision and attachment steps. For fine or
low density particles the chances of attachment are reduced due to their low
momentum. On the other hand, coarse or high density particles are more easily
attached to the bubbles and detachment of particles from stable bubble particle
aggregates is insignificant. The upper particle size that may be floated is therefore
limited by the buoyancy of such aggregates. For such systems the observed effects
of the physical variables agree reasonably well with the theoretical predications and
would be applicable in the special case of column flotation.
Industrial flotation tanks however, operate under intense agitation. Under such
conditions the detachment effects become important. Broadly speaking, the same
physical parameters favour both attachment and detachment so that the ultimate
flotation rate is a fine balance between these two competing mechanisms. For
example, the rate can be considerably enhanced by the use of small bubbles,
because they produce a greater surface area for particle collection, per unit volume
of air. On the other hand. reducing the bubble size can have deleterious effects, in
that forces tending to disrupt the particle bubble aggregates in a turbulent fields,
tends to increase as the bubble size decreases. A similar argument applies to the
particle size and density and the system turbulence.
In general, an optimum bubble and particle size can be found to maximise
recovery, which is a compromise between the various phenomena. The conclusion
is that in one flotation cell it may not be possible to obtain recovery of all particles
if the feed contains a wide range of sizes. This agrees well with conclusions reached
from performance studies of flotation cell, and support Barbery's view"! that con-
ventional machines are too far from their "optimum average" to provide the best
combination of factors for coarse and tine particles. The challenge for the mineral
processing engineer therefore, is to bridge the gap and obtain an "optimum" for all
particle sizes. In this context, three options deserve further investigation.
FLOTATION KINET[CS 97

1. Splitting of feed. For a feed containing a wide spread of particle sizes, it may
be advisable to classify the feed and treat the fine and coarse fractions
separately, at their optimum recovery conditions.
2. Mechanical methods of generating a broad range of bubble sizes. If it were
possible to generate bubbles of known size and distribution, independent of
the system turbulence, it would be possible to float particles of all sizes at their
highest recovery rate by adjusting the bubble size distribution.
3. Improved tank and impeller geometry. For coarse and high density particles,
the optimum recovery rate is achieved at impeller speeds just sufficient to
provide full suspension of the particles. It has been shown that the flotation
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

tank and impeller geometry may be modified to provide suspension at lower


speeds?". This area requires serious consideration since it could lead to
significant improvements in the overall energy consumption in flotation,
owing to lower running costs, improved recovery and energy consumption
associated with grinding for ores that are coarsely mineralised.

References
I. J. A. Kitchener, "The Froth Flotation Process: Past. Present and Future-s-In Brief", in The
Scientific Basis of Flotation (K. J. Ives, ed.). Martinus Nijhoff Publishers. The Hague (1984)
pp.3-51.
2. N. Arbiter and C. C. Harris. "Flotation Kinetics", in Froth Flotation, 50th Anniversary Volume
(D. W. Fuerstenau, ed.), AIME, New York (1982) pp. 215-246.
3. G. Barbery, "Engineering Aspects of Flotation in the Minerals Industry: Flotation Machines,
Circuits and their Simulation", op. cit. (I), Martinus Nijhoff Publishers, The Hague (1984)
pp. 289-348.
4. T. Inoue. "Flotation Process Modelling", in Principles of Mineral Flotation (M. H. Jones and J. T.
Woodcock, eds.), The Australasian Inst. Min. Metall., Parkville. Victoria (1984) pp. 255-271.
5. A. J. Lynch, "The Technology of Flotation", op. cit. (4). The Australasian Inst. Min. Metall.,
Parkville, Victoria (l984)-pp 233-253.
6. T. Inoue, M. Nonaka and T. Irnaizumi, "Flotation Kinetics-Its Macro and Micro Structure", in
Advances in Mineral Processing (P. Somasundaran ed.) Soc. Mining Engineers. Inc., Colorado
(1986) pp. 209-228.
7. J. Leja, Surface Chemistry of Froth Flotation, Plenum Press, New York (1982).
8. G. J. Jameson. S. Nam and M. Moo Young, Minerals Sci. Engng, 9, 103 (1977).
9. W. J. Trahar and L. J. Warren, Int. J. Miner. Process., 3,103 (1976).
10. H. G. Zuniga, Boletin Minero de 10 Sociedat Nacional de Minero, Santiago. 47. 84 (1935).
II. A. M. Gaudin, J. O. Groh and H. B. Henderson, AIME, T.P. 414 (1931), pp. 3-23.
12. O. S. Bogdanov, B. V. Kizevalter and V. Ya Khayman, Non-ferr. Metals, No.4 (1954).
13. V. I. Klassen and V. A. Mokrousov, An Introduction to the Theory of Flotation, Butterwortns.
London (1963).
14. N. Arbiter, Trans. AIME, 190,791 (L951).
15. P. L. de Bruyn and H. J. Modi, Mining Engg., Trans. AIME, 205. 415 (1956).
16. T. Imaizumi and T. Inoue, "Kinetic Considerations of Froth Flotation", in Proc. Sixth lnternation
Mineral Processing Congress- Cannes (A Roberts, ed.) Pergamon, Oxford (t963) pp. 581-605.
17. E. C. Dowling, R. R. KlimpeL and F. F. Aplan, Minerals and Metallurgical Processing. 2, 87
( (985).
18. I. Huber-Panu, E. Ene-Danalache and D. G. Cojocariu, "Mathematical Models of Batch and
Continuous Flotation". in Flotation-s-A, M. Gaudin. Memorial Volume (M. C. Fuerstenau, ed.)
Vol. 2, AIME, New York (1976) pp. 675-724.
19. T. H. Naylor, et 01. Computer Simulation Techniques. WiLey, New York (1966). Mentioned in
reference 17.
20. A. J. Lynch, et 01., Mineral and Coal Flotation Circuits- Their Simulation and Control,
Developments in Mineral Processing (D. W. Fuerstenau. ed.) vol. 3, Elsevier, Amsterdam (1981).
98 N. AHMED AND G. J. JAMESON

21. R. J. Gochin, "Flotation", in Solid-Liquid Separation (L. Svanovsky, ed.) 2nd Ed., Butterworths,
London (1981) pp. 503-524.
22. R. P. King, Trans. Soc. Min. Engr..., AIME, 258, 287 (1975).
23. A. F. Taggart, Handbook of Mineral Dressing, 2nd Edition, Wiley, New York (1945).
24. O. S. Bogdanov and M. Filanovski, 1. Phy. Chem. USSR .. 14,224 (1940).
25. H. R. Spedden and W. S. Hannan, Min. Technol. 12, No.2, T.P. 2354 (1948).
26. P. F. Whelan and D. J. Brown, Trans. Instil. Min. Metall., 65,181 (1956).
27. R. Schuhamnn, J. Phys. Chem. 46,891 (1942).
28. K. L. Sutherland, J. Phys. Chem. 52,394 (1948).
29. H. S. Tomlinson and M. G. Fleming. "Flotation Rate Studies", in op. cit. (16), Pergamon, Oxford
(1963) pp. 563-573.
30. L. Spielman and S. L. Goren, Env. Sci. Tech., 2, 279 (1968).
31. A. M. Gaudin, Flora/ion. 2nd ed., McGraw Hill, New York (1957).
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

32. B. V. Dcrjaguin and S. S. Dukhin, Trans. lnst, Min. Metall., 70, 221 (1961).
33. L. R. Flint and W. J. Howarth. Chem. Eng. Sci., 26,1155 (1971).
34. D. Reay and G. A. Ratcliff, Can. J. Chem. Engng., 51, 178 (1973).
35. G. L. Collins, "Dispersed Air Flow/ion of Fine Particles"; Ph.D. Thesis, Imperial College,
University of London (1975).
36. M. E. Weber. J. Separ. Proc. Technol .. 2, 29 (1981).
37. M. E. Weber and D. Paddock, J. Call. tnt. Sci., 94. 328 (1983).
38. Z. W. Jiang and P. N. Holtham, Trans. Instil. Min. Me/all.. Sect. C, 95, CI87 (1986).
39. V. G. Lcvich, Physicochemical Hydrodynamics. Prentice-Hall, Inc., New Jersey (1962).
40. P. G. Saffman and J. S. Turner, J. Fluid Mechanics, I, 16 (1956).
41. M. Smoluchowski, Z. Phys. Chem., 92.1219 (1917).
42. A. N. Kolmogoroff, Comptes rendus (Doklady) de l'Academie des Sciences de I' U.R.S.S., 30. 301;
31, 538; 32, 16. Translations of the original in Turbulence, Classical Papers on Statistical Theory
(S. K. Friedlander and L. Topfer, eds.), Inter Science Publishers, Inc., New York (1961).
43. T. S. Mika and D. W. Fuerstenau, "A Micro-scopic Model of the Flotation Process", in Proc.
8/h International Mineral Processing Congress, vol. 2, Institute Mekhanbor, Leningrad (1968),
pp. 246-269 (Russian text); Paper S-4 (English text).
44. H. Schubert and C. Bischofberger, "On the Optimization of Hydrodynamics in Flotation
Processes", in Mineral Processing. Proc. 13th International Miner. Process. Congr. (J. Laskowski,
ed.) Warsaw, 1979. Part B, Elsevier and PWN-Polish Scientific Publishers, Warsaw (1981)
353-377.
45. J. Abrahamson. Chem. Engg. Sci. 30, 1371 (1975).
46. T. R. Camp and P. C. Stein, J. Boston Soc. Civil Eng., 30, 219 (1943).
47. M. Nonaka. T. Inoue and T. Imaizumi, "A Micro Hydrodynamic Flotation Model and its
Application to the Flotation Process", Preprints of the I4/h bu. Miner. Process. Cong., Toronto,
Vol. 3, Paper 9 (1982).
48. A. Jowett, "Formation and Disruption of Particle-Bubble Aggregates in Flotation", in Fine
Panicles Processing (P. Somasundaran, ed.), Vol. 1. AIME, New York (1980) pp. 720-754.
49. 1. Sven-Nilsson. Kolloid Zeischrift, 69. 230 (1934).
SO. W. Philippoff, Min. Engng., 4, 386 (1952).
51. L. F. Evans, Ind. Engng. Chem., 46. 2420 (1954).
52. J. M. W. Mackenzie and G. H. Matheson, Trans. Soc. Min. Engrs., AIME. 226, 68 (1963).
53. T. P. Meloy, "The Treatment of Fine Particles During Flotation", in op. cit. (2), AIME, New
York (1962) pp. 247-257.
54. V. A. Glembotsky, Bull. Acad. Sci. S.S.S.R., No. 11(1953).
55. G. L. Collins and G. J. Jameson, Chem. Eng. Sci., 31, 985 (1976).
56. A. Scheludko, B. V. Toshev and D. T. Bojadjiev, J. Chem. Soc. Farad. Trans., 72, 2815 (1976).
57. J. Laskowski, Minerals Sci. Engng., 6, 223 (1974).
58. T. M. Morris, Min. Engng., 4, 794 (1952).
59. H. J. Schulze, Int. J. Miner. Process., 4, 241 (1977).
60. H. J. Schulze, E. Topfer , G. Gottschalk, Aufbereitungs-Technik, 22,465 (1981).
61. H. J. Schulze, lnt, J. Miner. Process, 9, 321 (1982).
62. A. D. Gaudin, R. Schuhmann and A. W. Schlechten, J. Phys. Chem., 46, 902 (1942).
63. K. L. Sutherland and J. Wark, Principles of Flotation, Australasian Inst. Min. Metall., Melbourne
(1955).
64. C. H. G. Bushell, Trans. Soc. Min. Engrs., A/ME, 223, 266 (1962).
FLOTATION KINETICS 99

65. W. J. Trahar, Int. J. Miner. Process., 8, 289 (1981).


66. N. Ahmed, Fine-Particle Flotation in a Stirred Vessel, Ph.D. Thesis, University of Newcastle,
N.S.W., Australia (1983).
67. D. Reay and G. A. Ratcliff. Can. J. Chern. Engng., 53, 481 (1975).
68. C. C. Collins and G. J. Jameson, Chern. Engg. Sci., 32, 239 (1977).
69. J. J. Anfruns and J. Kitchener, Trans. Inst. Min. Metall., 86, Section C, C9 (1977).
70. E. A. Cassell, K. M. Kaufman and E. Matijevic, Waler Research 9, 1017 (1975).
71. A. J. R. Bennell, W. R. Chapman, W. R. and C. C. Dell, "Froth and Flotation of Coal", 3rd
International Coal Preparation Congress, Brussels-Liege (1958) pp. 452-462.
72. D. G. De Vivo and B. L. Karger, Separ. Sci., 5, 145 (1970).
73. R. J. Yoon and G. H. Luttrell, Coal Preparation, 2,179 (1986).
74. R. Varbanov, Trans. 1/lS1II. Min. Metall .. Section C, 93, C6 (1984).
75. N. Ahmed and G. J. Jameson, 1111. J. Mineral Process., 14, 195 (1985).
Downloaded by [University of Toronto Libraries] at 03:53 24 December 2014

76. H. Schubert and C. Bischotberger, Int. J. Miner Process., 5,131 (1978).


77. H. Schubert, C. Bischotberger and P. Koch, Aubereitungs-Technik, 6, 306 (1982).
78. C. C. Harris, "Flotation Machines", op. cit. (18), Vol. 2, AIME, New York (1976) pp. 753-815.
79. W. Grunder and J. F. Kauffman, Errmemll, 9, 559 (1956).
80. L. A. Adorjan, Min. Annual Rev., 257 (1982).
81. G. Barbery, "Recent Progress in the Flotation and Hydrometallurgy of Sulphide ores", in
Flotation of Sulphide Minerals (K. S. Eric Forssberg, ed.) Developments in Mineral Processing,
Vol. 6, Elsevier, Amsterdam (1985) pp. 1-38.

You might also like