A Hydrodynamic Model for Bubble-Particle Attachment

G. H. LUTTRELL AND R.-H. YOON 1

Virginia Center for Coal and Minerals Processing, Virginia Polytechnic Institute and State University,
Blacksburg, Virginia 24061
Received November 21, 1991; accepted April 16, 1992

The process of bubble-particle attachment has been modeled by simulating the trajectory of a small
particle as it approaches a bubble. The particle trajectory has been determined using a dynamic force
balance between the force pressing the particle against the bubble and the force resisting the viscous film
thinning. From the simulated trajectories, the closest approach distance between the particle and the
bubble has been determined. The model assumes that bubble-particle adhesion occurs when the approach
distance becomes smaller than the critical thickness at which the disjoining liquid film ruptures spon-
taneously. The model predictions have been verified experimentally using different sizes of bubbles and
particles. © 1992AcademicPress,Inc.

INTRODUCTION row range of particle sizes, typically between

10 and 100 ~zm (4).
Froth flotation is a widely used industrial
The simplest models for predicting Pc are
process in which air bubbles are introduced
based on calculating the volume swept by a
into an aqueous slurry to collect hydrophobic
bubble in a flotation pulp under quiescent
particles and to remove them from the pulp.
conditions (2-8). The results of the various
Flotation can be used for separating one type
hydrodynamic models may be summarized in
of particle from another, as is done in the
the form
minerals industry, or for simply removing
finely dispersed particulate matter from in-
dustrial waste streams.
Pc=a{Dv]" ' [21
The probability (P) of a particle being col-
lected by an air bubble during flotation may in which A is a function of the Reynolds num-
be described by ( 1, 2) ber of the bubble and n = 2 for the bubbles
used in flotation (8).
P = PcPa(1 - Pa), [11 In predicting Pa, it is usually assumed that
in which Pc is the probability of bubble-par- adhesion occurs when the bubble and particle
ticle collision, Pa the probability of adhesion, are in contact long enough to rupture the dis-
and Pa the probability of detachment. Pa joining water film between them (2, 3, 9-12).
should increase with increasing particle size The minimum time required for thinning and
because of the increased detachment force, but rupturing the film is referred to as induction
it may be neglected for very small particles time (ri). Thus, Pa is given as the fraction of
(3). Pc, on the other hand, decreases with de- particles whose contact times are longer than
creasing particle size because of the low prob- ri. The methods of determining it experimen-
ability of collision (4). Therefore, froth flo- tally have been reported ( 13-15 ). In general,
tation is usually effective over a relatively nar- the more hydrophobic a particle becomes, the
shorter the ri. Hydrodynamic models for pre-
dicting Pa as a function of ri, particle size, and
To whom correspondence should be addressed. bubble size have been developed (11 ). It
129
0021-9797/92 $5.00
Copyright© 1992by AcademicPress,Inc.
Iournalof Colloidand InterfaceScience, Vol. 154,No. 1, November1992 All rightsof reproductionin any formreserved.
130 LUTTRELL AND YOON

should be noted, however, that neither the ex- communications, however, models combining
perimental technique for determining ri nor hydrodynamic forces and surface forces will
the models for predicting Pa account for the be described.
likelihood that T i is also a function of the force
with which a particle strikes a bubble. MODEL DEVELOPMENT
Another method of predicting Pa has re-
cently been suggested, in which bubble-par- Bubble-particle interaction has been mod-
ticle adhesion is considered to occur when the eled in the present work by considering the
kinetic energy of the particle (Ek) is larger than case of an isolated bubble rising through a sus-
the energy barrier (E) (16). This criterion for pension of particles under quiescent condi-
bubble-particle adhesion may be expressed in tions. As the bubble rises, a flow pattern that
the form can be represented by an infinite series of
Pa = e x p ( - E / E k ) . [3] streamlines develops. The mathematical de-
scription of the streamline flow pattern is
E is determined by the surface forces involved known as the stream function. Due to their
in the bubble-particle interaction, which may small mass, fine particles (e.g., <50 gm in di-
be viewed as a heterocoagulation process. ameter for particles of 2.5 specific gravity) will
The potential energy (V) of the bubble- follow the streamlines. In such cases, gravi-
particle interaction varies as a function of the tational and inertial forces can be neglected,
distance (H) separating the two. Vmay consist and the particle trajectory can be determined
of three major components, from an appropriate stream function. For
V= VE+ VD+ V., [41 flows characterized by very small (Stokes flow)
or very large (potential flow) Reynolds num-
in which VE, Vo, and VH represent the elec- bers, the stream functions can be derived di-
trostatic, dispersion, and hydrophobic inter- rectly from the Navier-Stokes equations.
action energies, respectively. E is equal to the However, bubbles commonly used in mineral
V maximum, and the distance at which it oc- flotation systems have Reynolds numbers less
curs is referred to as the critical rupture thick- than 400, for which analytical solutions do not
ness (He). Thus, the criterion for bubble-par- exist.
ticle adhesion that Ek >~E is equivalent to the In the present work, an empirical stream
condition that H ~<He. If the closest approach function has been developed by analyzing
distance is known from the particle trajectory, streamline flow patterns reported in the liter-
it can be used for determining whether adhe- ature for rigid spherical obstructions placed in
sion will occur. The method of determining quiescent flow. For the geometry defined in
Hc has been described (17-19). For very hy- Fig. 1, the stream function (~) applicable for
drophobic particles, He is typically in the range the intermediate Reynolds numbers has been
100-150 nm and decreases with decreasing derived as (8)
hydrophobicity of the solid ( 19-23 ).
2 2 [1
In the present work, a model for predicting = UbRbSin 0]~ X 2 - - 3_ Y + 1
P has been developed. In this model, the tra- 4X
jectory of a particle has been simulated by
considering only the hydrodynamic forces +--~
Re(1
~ - - Z + ~ - + X - 1 , [51
l )]
acting on the particle. The surface forces,
which may also affect the particle trajectory in which Ub is the bubble rise velocity, Rb the
in close proximity to the bubble, have not been bubble radius, X = R / R b , Re the Reynolds
considered. It has been assumed that attach- number of the bubble, and R and 0 are the
ment occurs when a particle approaches a cylindrical coordinates of the streamline. It can
bubble within a distance of Hc. In ensuing be shown that Eq. [ 5 ] is reduced to the ana-
Journal of Colloid and Interface Science, Vol. 154, No. l, November 1992
 BUBBLE-PARTICLE ATTACHMENT 13 1

Streamline that the resistance force acts only in the radial

direction, the following relationship holds,
Particle Trajectory
u~p = ut, [8]
in which Utp is the tangential velocity of the
\ particle. The radial velocity of the particle
( U s ) , on the other hand, is reduced clue to
the resistance it encounters as it approaches
the bubble surface.
The resistance force (Fr) may be estimated
from lubrication theory,
Fr = 67rtzRpUrp3, [9]
where tz is the liquid viscosity, R p the particle
radius, Uqo the radial velocity of the particle,
and 3 the Stokes correction factor. For the case
of a sphere approaching an infinitely flat plate,
3 can be estimated using the Reynolds (24)
lubrication theory,
Kb ~ 3 = Rp/H, [10]
FIG. 1. Trajectories of a particle approaching a bubble in which H is the distance between the bubble
with (dashed line) and without (solid line) the influence and particle surfaces. Equations [ 9 ] and [ 10 ]
of the resistance due to viscous film thinning.
suggest that Fr increases with decreasing H. It
has been shown, however, that Eq. [10] is valid
lytical solution of the Navier-Stokes equations only when H is very small (less than approx-
for Re = 0. imately 10% of the radius of the particle). This
Once an appropriate stream function has problem has been corrected by Maude (25)
been selected, the radial (Ur) and tangential and Brenner (26). For the case of a small
(Ut) velocities of the fluid flowing past a rigid sphere approaching a larger sphere, which is
sphere can be obtained as follows: more applicable to bubble-particle interac-
tions, Goren and O'Neill (27) showed that/3
1 0~ is significantly smaller than for the case of a
U~ R2sin 0 00 [6] sphere approaching a flat plate. Figure 2 com-
and pares the 13 values predicted for the two geo-
metric considerations.
1 0~ The force pressing (Fp) a particle against a
Ut - R sin~ OR" [7] bubble must be equal to its hydrodynamic drag
force (Fd). For a spherical particle carried
At distances far ahead of the bubble, Eqs. [6 ]
along in the streamline flow, it follows, there-
and [7 ] can be used to determine the instan- fore, that
taneous trajectory of the particle. It is usually
assumed that the particle velocity is equal to Fp = Fd = 6zr#RpUr. [11]
the velocity of the streamline passing through Also, a dynamic force balance requires that at
its center (5). any point in time, Fp must be equal to the
As a particle approaches the bubble surface, force resisting (Fr) the film thinning, i.e., Fp
its velocity will become retarded by the resis- = Ft. One can then equate Eqs. [ 9 ] and [ 11 ]
tance to thinning of the viscous water film be- to obtain
tween the bubble and the particle. Assuming Urp = U r / t 3 , [12]
Journal of Colloid and InterfaceScience, Vol. 154,No. 1, November1992
 132 LUTTRELL AND YOON

1000 ........ I ........ I ........ k .....

that as the particle approaches the bubble
- - - Reynolds (1886)
- - - Maude
equator, the particle trajectory shifts further
(1961), Brenner (1964)
nd O'Neill (1971) away from the fluid streamline (Fig. 1 ).
100 "
Once the particle trajectory is known, the
distance (H) between the surfaces of the par-
ticle and bubble can be determined. For a
10 given bubble size and particle size, this dis-
tance should be a function of the initial po-
sition (Ro) of the particle relative to the bubble.
It is assumed that in a given trajectory, H be-
0.001 0.01 0.1 1 10

H/R~ comes smallest when the particle passes the

bubble equator (Fig. 1). It has been shown,
FIG. 2. Relationshipsbetween/3and H/Rpdetermined however, that the closest approach distance
by various investigators. may actually occur slightly before the particle
reaches the bubble equator (7). As has already
been suggested, the disjoining water film rup-
which shows that U~ is smaller than Ur by a tures, i.e., bubble-particle attachment occurs,
factor of 1//3. when H ~< He. The magnitude of Hc is deter-
Using Eqs. [8] and [12] for Utp and Uro, mined by the various surface forces involved
respectively, one can determine the trajectory in the bubble-particle interaction, and it in-
of the particle as it moves past the bubble un- creases with increasing particle hydropho-
der the combined influence of streamline flow bicity.
and hydrodynamic resistance to film thinning. The above criterion for bubble-particle at-
In the present work, the trajectory has been tachment is essential for determining the
determined numerically by sequentially cal- probability (P) of particle collection, which is
culating the positions of the particle as it ap- defined as the fraction of particles in the path
proaches the bubble. The following set of of a bubble that are actually collected. In ref-
equations has been used for describing the po- erence to Fig. 1, P can be calculated from the
sitions as functions of time, ratio of the area circumscribed by Ro to the
area circumscribed by (Rb + Rp + He), i.e.,
One w = 0ol d + UtAt/R [13]

and P = Rb + Rp + Hc [15]
Rnew = Rold + UrAt/13, [14]
where Rp and Rb are the radii of the particle
where At is the time step used in the numerical and bubble, respectively. Equation [15 ] shows
simulation. The values of/3 have been deter- that P is determined by both physical param-
mined using Goren and O'Neill's analysis for eters (bubble size and particle size) and surface
the sphere-sphere interaction (Fig. 2). Far chemistry (H~). Although Ho is usually very
ahead of the bubble,/3 ~ 1 and the particle small compared to Ru and Rp, P is sensitive
trajectory is the same as fluid streamline. As to changes in He. The reason is that a small
the particle approaches to within approxi- change in He requires a large change in Ro in
mately twice the particle radius, fl begins to order for the particle still to be collected by
increase due to the resistance to the viscous the bubble. For example, if He decreases from
film thinning. When the particle approaches 100 to 10 nm, R0 should decrease by 14,200
the bubble within approximately 0.1% of the nm for Rb = 500 #m and Rp = 1 #m.
particle radius,/3 becomes as large as 100. The The values of P obtained using Eq. [15] are
net result of increasing/3 with decreasing H is plotted in Fig. 3 as a function of He for a bub-
Journal of Colloid and Interface Science, Vol. 154, No. 1, N o v e m b e r 1992
 BUBBLE-PARTICLE ATTACHMENT 133

have been referenced to the values given in

Fig. 3 by a normalization factor o f f , i.e.,
0,1
P =fPol, [16]
in which P0.1 is the probability of collection
0.01
for Db = 0.1 mm. As shown in Fig. 4, f a n d
hence P increases exponentially with decreas-
0.001 ing bubble size. Thus, for a given particle size
and hydrophobicity, smaller bubbles result in
a higher collection efficiency. For example, the
0.0001
1 10 100 1000 value of P for a 0.1-mm bubble is more than
Critical Film Thickness (nm) 13 times higher than that obtained using a 0.6-
m m bubble.
FIG. 3. Predictedrelationshipbetween P and Hc for a
bubble diameter of 0.1 mm and particle diameters of 1- The effect of bubble size is more clearly
40 #m. demonstrated in Fig. 5, in which P is plotted
as a function of Hc and Dp for two different
bubble diameters. As shown, the P values for
the 0.1-mm-diameter bubbles are substantially
ble diameter (Oh) of 0.1 m m over a range of
larger than those of the 1-mm bubbles. For
particle diameters (Op). As shown, P decreases
the cases considered, it may be more beneficial
with Dp and increases with He. The plot shows
to increase P by decreasing bubble size than
that the difficulty in floating fine particles,
by adding more reagents to increase He. In
which is a well-recognized industrial problem,
fact, the advantages of utilizing small bubbles
can be corrected by increasing the particle hy-
to improve the recovery of fine particles have
drophobicity (He). This can be achieved either
been demonstrated in laboratory experiments
by using a stronger collector (a surfactant that
(28-32), and commercial flotation machines
renders the particle hydrophobic) or by in-
have recently been developed on the basis of
creasing the reagent dosage during flotation.
this concept.
Simulations similar to those shown in Fig.
3 have been performed for a range of bubble EXPERIMENTAL
sizes. The results are summarized in Fig. 4 to To validate the model developed in the
show the effect of bubble size on P. In this foregoing section, a series of experiments was
figure, the P values for any given bubble size

g
•~ 0.1

O
o t~ 0.01
Lr~
=o
0.001

0.1
Z
o.oool
1 10 100 iO00

Critical Film Thickness (nm)

0.2 0,4 0.6 0.8
Bubble Diameter (ram) FIG. 5. Predicted relationship between P and Hc for
particle diameters of 5 and 20/~m and bubble diameters
FIG. 4. Effectof bubble size on P. of 0.1 and 1 rnm.
Journal of Colloid and Interface Science, Vol. 154, No. 1, November 1992
 134 LUTTRELL AND YOON

carried out in which the probability of collec- (~0.2% solids) by means of a small syringe.
tion (P) was determined at different bubble Ultrasonic agitation coupled with gentle stir-
and particle sizes. The Buller seam coal from ring was employed to provide a completely
New Zealand, containing only 0.13% ash by dispersed suspension, which was examined
weight, was used as a model hydrophobic ma- under a microscope to ensure that only single
terial in this study. Samples of different size particles were present. As the bubble was al-
fractions were prepared by dry-grinding in an lowed to rise through the observation tube,
agate mortar and pestle, followed by screening the downward flow of suspension was adjusted
or by centrifugal sedimentation, depending on by the two-way stopcock to counterbalance the
the size of particles desired. To minimize ox- upward velocity of the bubble. In this manner,
idation, the samples were stored under a ni- the bubble was "trapped" for a desired period
trogen atmosphere in airtight containers at of time. By closing the stopcock, the bubble
ambient temperature. was then allowed to rise along the column
Figure 6 shows the apparatus used for de- height and to enter the inverted tube filled with
termining the probability of collection. It was clear water. The number of particles attached
similar in design to the one used by King et to the bubble was determined visually with a
aL (33) for the measurement of bubble load- microscope viewer. Particles entrained in the
ing, but the technique of measurement was wake of the bubble were allowed to settle and
identical to that employed by Anfruns and return to the column prior to particle counting.
Kitchener (34, 35). In each measurement, a The diameter of the bubble was measured by
single bubble was injected into a dilute slurry means of a graduated eyepiece in the micro-
scope.
The probability of collection (P) was de-
termined by dividing the number of particles
U microscope viewer collected (Arc) by the total number of particles
~caotured bubble in the path of the bubble, i.e.,
spillway IJ I J i n v e r t e d tube filled 4No
~1 with clear water P = ~rD~tUbN' [17]
t ~constant head tank
in which Db is the diameter of the bubble, t
the residence time of the bubble in the tube,
Ub the velocity of the particle suspension
moving past the bubble, and N the number of
particles per unit volume of suspension. The
)bservatlon tube. value of Ub, which is equal to the bubble rise
velocity, was determined experimentally by
measuring the time required for a bubble to
microsyringe rise a given distance in the column. A good
/ correlation was obtained between the mea-
sured bubble diameter and that calculated
)cock
from the bubble rise velocity. The value of N
was determined on the basis of the mean di-
ameter of each narrowly sized sample.
\
variable speed pump RESULTS A N D DISCUSSION
FIG. 6. Schematic representation of the apparatus used
for the measurement of the probability of particle collection The bubble-particle interaction model de-
by a single air bubble. veloped in the present work can predict the
Journal of Colloid and Interface Science, Vol. 154,No. 1, November1992
 BUBBLE-PARTICLE ATTACHMENT 135

probability of collection (P) as functions of

particle diameter (Dp), bubble diameter (Ob),
and the critical rupture thickness (He) of the
film between the particle and the bubble.
While the values of Dp and Db can be readily
determined, Hc is more difficult to determine
experimentally. In most cases, the Ho values z
reported in literature have been obtained using
the bubble-against-the-plate technique, in
which monochromatic light is passed through
the film and the substrate to obtain Newton
tings ( 17-19). However, this technique is Bubble Diameter (mm)

useful only for isotropic, transparent sub-. FIG. 8. C o m p a r i s o n o f the p r e d i c t e d a n d e x p e r i m e n t a l

strates, such as fused silica and mica, and not P values as a f u n c t i o n Of Db f o r Dp = 40.1, 31.0, a n d 11.4
for the coal sample used in the present work. tzm, a n d H c = 150 n m .

For this reason, model predictions have been

made in the present work using an Ho value particles (40.1/zm) show slightly higher P val-
determined by trial-and-error. It has been ues than predicted at the larger bubble sizes.
found that the predicted P values can be best This discrepancy may be expected since the
correlated with the experimental data when Stokes number, which is a measure of the ten-
He = 150 nm for the different bubble and par- dency for a particle to deviate from the
ticle sizes studied (see Fig. 7). Similar values streamline, increases with both particle and
of Hc were observed for very hydrophobic bubble sizes. Also, the experimental values
particles of methylated silica (18), indicating obtained with the finest particles (11.4 izm)
that the coal sample used in the present work are slightly lower than predicted. Two reasons
was very hydrophobic. may be given: First, this particular size fraction
Using the value of Hc determined as such, was obtained using a homemade centrifugal
the P values predicted from Eq. [ 15 ] are com- sedimentation device, which could have pro-
pared with the experimental values in Fig. 8. duced a sample with a wider size distribution
There is a general agreement between the pre- than those produced by screening. Second, the
dicted (solid lines) and experimental (data finer particles may have been oxidized to some
points) values. Note, however, that the larger extent.
The model predictions suggest that small
changes in Hc (as compared to Rb and Rp)
result in substantial changes in P (Figs. 3 and
A 31.0 /zm 5 ). For example, when He is increased from
I 114. I 10 to 100 nm, P increases from 0.0007 to
0.1
0.0030 for Dp = 2/~m and Db = 0.1 mm. An
increase in Hc by 90 nm is extremely small
compared to the bubble and particle sizes, but
0.01 it has a profound impact on P. This finding
ClL~[]
simply reflects the fact that as the particle nears
the bubble equator, it experiences increasing
0.001 difficulty approaching the bubble surface be-
0.001 0.01 0.1
cause of an increase in the resistance to viscous
Predicted P
film thinning. As shown in Fig. 2,/3, which is
FIG. 7. C o m p a r i s o n o f p r e d i c t e d a n d m e a s u r e d P values a measure of the resistance force, increases
a t H ¢ = 150 rim. rapidly with a decrease in the film thickness.
Journal of Colloid and Interface Science, Vol. 154, No. 1, November 1992
136 LUTTRELL AND YOON

At the same time, the radial velocity of the mined from the particle trajectory. Forces
streamline (Ur) decreases as the particle ap- arising from both the streamline flow and the
proaches the bubble equator. According to Eq. resistance to viscous film thinning have been
[12], both of these factors contribute to a de- considered in simulating the particle trajec-
crease in the radial velocity of the particle (Urn) tory.
toward the bubble surface. The results given 3. The hydrodynamic model developed in
in Figs. 3 and 5 also show that P becomes more the present work can predict the probability
sensitive to changes in Hc as the particle size of collection as functions of particle size, bub-
is reduced. An implication of this finding is ble size, and the critical rupture thickness of
that changes in particle hydrophobicity have the disjoining film. The model predictions
a more profound impact on the flotation of suggest that the problems associated with re-
finer particles. covering fine particles during flotation can be
In effect, the bubble-particle interaction minimized either by using smaller air bubbles
model developed in the present work considers or by increasing the critical film rupture thick-
both the probability of collision (Pc) and the ness. The latter can be achieved by increasing
probability of adhesion (Pa) simultaneously. the particle hydrophobicity.
Pc varies according to the hydrodynamic con- 4. The model predictions have been verified
ditions of the system, while Pa depends on the experimentally over a range of bubble and
various surface forces involved in the bubble- particle sizes. A highly hydrophobic coal sam-
particle interaction. He has been used as a pa- ple has been used for the determination of the
rameter representing the net result of the sur- probability of collection. The model validation
face forces interacting with each other. In is based on the assumption that the coal
general, Hc increases with increasing particle sample has the critical rupture thickness of
hydrophobicity, but it also depends on the 150 nm.
electrostatic and dispersion forces operating
during bubble-particle interaction ( 16, 19). ACKNOWLEDGMENTS
The effects of the various surface forces in de- The authors expressappreciation for financial support
termining Hc and the energy barrier (E), from the U.S. Department of Energy (No. DE-FG22-
which determines the kinetics of bubble-par- 83PC60806 and No. DE-AC22-86PC91221), and for the
ticle attachment, will be discussed in a forth- fruitful discussions with ProfessorsG. T. Adel and M. D.
coming publication. It is noted here that the Pritzker.
authors have recently carried out a series of
REFERENCES
trajectory simulations by considering both the
hydrodynamic and surface forces, the results 1. Schuhman, R., J. Phys. Chem. 46, 891 (1942).
of which will also be presented in ensuing 2. Sutherland,K. L., J. Phys. Chem. 52, 394 (1948).
3. Dobby, G. S., and Finch, J. A., J. Colloid Interface
communications.
Sci. 109, 493 (1986).
4. Gaudin, A. M., "Flotation," 2nd ed. McGraw-Hill,
SUMMARY AND CONCLUSIONS New York, 1957.
1. A bubble-particle interaction model has 5. Flint, L. R., and Howarth, W. J., Chem. Eng. Sci. 26,
been developed by simultaneously considering 11155 (1971).
6. Reay, D., and Ratcliff,G. A., Can. J. Chem. Eng. 51,
the probabilities of collision and of adhesion. 178 (1973).
It assumes that under quiescent conditions, 7. Weber,M. E., J. Sep. Process TechnoL 2, 29 ( 1981).
bubble-particle attachment occurs when a 8. Yoon, R.-H., and Luttrell,G. H., in "Frothingin Flo-
particle approaches a bubble surface to within tation" (J. S. Laskowski, Ed.), p. 101. Gordon &
a critical distance, at which the disjoining wa- Breach, New York, 1989.
9. Schulze, H. J., in "Physico-Chemical Elementary
ter film ruptures spontaneously. Processes in Flotation, Developmentsin Mineral
2. The closest approach distance between Processing"(D. Fuerstenau,Ed.), Vol. 4. Elsevier,
the bubble and the particle has been deter- New York, 1984.
Journal of ColloM and Interface Science, Vol, 154, No. 1, November 1992
 BUBBLE-PARTICLE ATTACHMENT 137

10. Dobby, G. S., and Finch, J. A., Int. J. Miner. Process. Warsaw, June 4-9, 1979" (J. Laskowski, Ed.), p.
21, 241 (1987). 21.
11. Luttrell, G. H., and Yoon, R.-H., in "Production and 24. Reynolds, O., Philos. Trans. R. Soc. London 177, 157
Processing of Fine Particles" (A. Plumpton, Ed.), (1886).
p. 159. Pergamon, Toronto, 1988. 25. Maude, A. D., Br. J. Appl. Phys. 12, 242 (1961).
12. Crawford, R., and Ralston, J., Int. ,L Miner. Process. 26. Brenner, H., Chem. Eng. Sci. 19, 703 (1964).
23, 1 (1988). 27. Goren, S. L., and O'Neill, M. E., Chem. Eng. Sci. 26,
13. Eigles, M. A., and Volova, M. L., in "Proceedings, 325 ( 1971 ).
5th International Mineral Processing Congress," 28. Bennett, A. J., Chapman, W. R., and Dell, C. C., in
p. 271. Inst. Mining & Metall., London, 1960. "Proceedings, 3rd International Coal Preparation
14. Ye, Y., and Miller, J. D., CoalPrep. 5, 147 (1988). Congress, Brussels-Leige, June, 1958," paper E2.
15. Yordan, J. L., and Yoon, R.-H., J. Colloid Interface 29. Brown, D. J., in "Froth Flotation" (D. Fuerstenau,
Sci. 141, 374 (1990). Ed.), p. 518. AIME, New York, 1962.
16. Yoon, R.-H.,Aufbereit. Tech. 32, No. 9, 474 ( 1991 ). 30. De Vivo, D. G., and Karger, B. L., Sep. Sci. 5, 145
17. Derjaguin, B. V., and Kussakov, M., Acta Physico- (1970).
chim. URSS 10( 1), 26 (1939). 31. Yoon, R.-H., Min. Cong. J. 68(12), 76 (1982).
18. Blake, T. D., and Kitchener, J. A., J. Chem. Soc. Far- 32. Yoon, R.-H., and Luttrell, G. H., Coal Prep. Int. J.
aday Trans. 1 68, 1435 (1972). 2, 179 (1986).
19. Yoon, R.-H., and Yordan, J. L., J. Colloid Interface 33. King, R. P., Hatton, T. A., and Hulbert, D. G., Tech.
Sci., in press ( 1991 ). Note, Trans. Inst. Min. Metall. Sect. C., March,
20. Platikanov, D., J. Phys. Chem. 68, 3619 (1964). C9 (1974).
21. Schulze, H. J., and Cichos, C., J. Phys. Chem. 251, 34. Anfruns, J. P., and Kitchener, J. A., in "Flotation"
145 (1972). (M. Fuerstenau, Ed.), A. M. Gaudin Memorial
22. Blake, T. D., Chem. Eng. 182, 117 (1973). Vol. 2 p. 625. AIME, New York, 1976.
23. Derjaguin, B. V., and Dukhin, S. S., in "Proceedings, 35. Anfruns, J. P., and Kitchener, J. A., Trans. Inst. Min.
13th International Mineral Processing Congress, Metall. Sect. C 86, C9 (1977).

Journal of Colloid and Interface Seb'nce, Vol. 154, No. 1, November 1992