Biography xiii

flotation data. The graduate students who contributed to the development of the model and simulator included
Laoquin Mao, Ian Scherrell, Brain Schimmoller, Hyunsun Do, Kyle Kelley, Aaron Noble, and Serhat Kelles.
Yoon’s group generated numerous publications on the measurement of hydrophobic forces between
collector-coated surfaces. Most recently, the measurements were conducted between gold surfaces hydropho-
bized with a long-chain (C16) alkyl thiol at different temperatures to monitor the changes in thermodynamic
functions. The results showed that both the excess entropy and enthalpy of the TFW decreases as two hydro-
phobic surfaces approach each other, which is contrary to the molecular-scale hydrophobic interaction (e.g.,
self-assembly). Further, the enthalpy term (ΔHf ) is slightly larger than the entropy term (TΔSf ) in magni-
tude in the macroscopic hydrophobic interaction (Wang et al., Journal of Colloid and Interface Science 2011,
361:321). This finding is opposite what is known of the molecular-scale hydrophobic interactions, which are
driven by entropy changes. These new findings suggest that the hydrophobic interactions at macroscopic scale
are caused by the changes in the structure of the TFW confined between hydrophobic surfaces. This suggestion
is consistent with the most recent findings that low-density liquid species are present in water at temperatures
as high as 300 K (Mallamace, et al., Proceedings of the National Academy of Sciences of the United States of
America 2008, 105:12725). The graduate students who worked on various aspects of the surface force mea-
surements included Jorge Yordan, S.A. Ravishankar, S. Vivek, Rajesh Pazhianur, Darren Flinn, Jialin Wang,
Jinhong Zhang, Nini Ma, Juoli Li, and Lei Pan. Professor Jacob Rabinovich and Dr. Dave Guzonas were instru-
mental in helping the surface forces research started at Virginia Tech.
Aside from his efforts to prove the existence of the hydrophobic force in TFW and to better understand its
role in flotation, Yoon explored the possibility of developing new separation processes from the basic informa-
tion. He found that hydrophobic particles such as bituminous coal self-coagulate in the absence of chemical
additives when the slurry is gently agitated. The kinetic energies required to induce the coagulation was less
than predicted by the DLVO theory by orders of magnitude, indicating that the coagulation is caused by the
hydrophobic force. The process, which is referred to as hydrophobic coagulation, can be applied for separat-
ing ultrafine coal from clay as disclosed in a U.S. patent (No. 5,161,694). The key people who initiated this
research included Zhenghe Xu and Rick Honaker, who have become outstanding university faculty members.
Yoon’s group has also developed methods of enhancing the hydrophobicity of collector-coated mineral
particles by using various hydrophobicity-enhancing reagents and thereby increasing the rate of flotation and
dewatering fine particles. In addition, Yoon developed a reagent package that can be used to increase the hydro-
phobicity of small diamonds that are in the range of millimeters, which is well beyond the range of particles
floated in the minerals and coal industries. These technologies have been patented and licensed to Nalco for
commercial use. Further, Yoon has developed a novel separation and drying process using recyclable hydro-
phobic liquids, which is also based on hydrophobic interaction. The graduate students who contributed to the
technology development included Ramazan Asmatulu, Jingmin Zhang, Dongchul Shin, and Kerem Eraydin.
In the coal industry, fine coal is cleaned by flotation, with clean coal products being dewatered using
mechanical and/or thermal drying methods. The latter is costly, while the former is inefficient. As a result,
many companies are forced to discard a significant portion of the fine coal to numerous impoundments or to the
environment. A consequence of this problem was the black water witnessed by Yoon when he was on his way
to his first summer job in Korea. His group has recently developed a hyperbaric centrifuge technology, which
has been licensed to Decanter Machine, Inc. The first commercial units installed at Jim Walters Resources,
Brookwood, Alabama, demonstrated that the new technology can help coal companies recover high-value
products from waste coal while at the same time improving the environment. The students who contributed to
this technology development were Ramazan Asmatulu and Serhat Kelles.
At Virginia Tech, Yoon is a University Distinguished Professor and Nicholas T. Camicia Professor. He
established the Center for Advanced Separation Technologies in 2001 to help the minerals and coal industries
through technology development and by producing highly trained personnel. During the course of his 33 years
of service at the university, Yoon has advised 32 doctoral students and 34 master-of-science-degree students
to successful completion, and has worked closely with 18 post-doctoral fellows and many visiting scholars.
Twelve of his former graduate students are teaching in major research universities. To date, he has more than
300 technical papers, 13 book chapters, one edited book, and 50 patents to his credit.

© 2012 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2012.
© 2012 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2012.
Effect of Frother Mobility on the Bubble Behavior in Bubble Rise
Velocity Measurements at the Initial Stages of Bubble Formation

K. Heiskanen, Z. Javor, G. Wierink, B. Omelka, and N. Schreithofer

Department of Materials Science, Aalto University, Aalto, Finland

ABSTRACT
Bubble behaviour in static and turbulent fields show that the physical model of bubble behaviour in flotation
is not complete. The stagnant cap model cannot be considered as a generic model. The Marangoni stress at the
gas-liquid interface disclosing as surface rigidity will depend on the properties of the surfactant. Experiments
in stagnant liquid do not support the hypothesis of bubble shape primarily affecting the terminal velocity. The
results highlight the static characteristics of the assumptions made by Sutherland (1948). Bubble rigidity and
momentun uncoupling cannot be sustained in dynamic systems. This leads to the conclusion that the vector
fields (velocity, pressure, concentration gradients etc) causing the energy barrier cannot be symmetric making
the attachment equations spatially local.

Introduction In the paper flotation was subdivided into sub-pro-

cesses, namely collision, attachment and detachment.
The flotation process is a very complex process
In the work Sutherland considered the rate constant
combining both chemical and physical processes.
to be a function of the number of encounters mul-
Modeling of flotation has proven to be a very dif-
tiplied by the probabilities of these sub-processes.
ficult challenge. Flotation research has advanced on
For our purposes we can repeat here the assumptions
two fronts, namely flotation hydrodynamics and flo-
used by used by Sutherland in developing his equa-
tation chemistry, both of which are very difficult and
tions, but originally put forward by Gaudin (1932):
demanding due to the complexity encountered. In
• Viscous forces can be neglected,
order to contain this complexity, the historical start-
• Particle and bubbles are very large compared
ing point was, like in so many other areas as well,
to the water molecules,
a practical engineering need. Doing batch experi-
• Both particles and bubbles behave as rigid
ments was an obvious first choice also for modeling
spheres,
studies. There is lots of information to be collected
• Bubbles and particles are the only disturbing
but the drawback is that a batch experiment is by its
factors in the flow field,
nature a very integrative experiment.
• Motion of particles and bubbles is irrota-
Early simple ideas of similarity between chemi-
tional, and
cal reactions and flotation behavior, stemming from
• Particle inertia can be neglected (added by
the flotation experiment results, lead to the extensive
Sutherland, 1948).
use of first-rate kinetic equations with a lumped flo-
The assumptions can be described as a continuum
tation rate constant k. The approach had and still has
model in a slow potential flow in an ideal Eulerian fluid.
its engineering appeal. In its simplicity, it is easy to
All of these sub-processes led to the development
fit into existing data and can therefore be used to pre-
of several analytical models in recent efforts that dif-
dict plant performance. By dividing the rate constant
fered in their initial assumptions. The most discussed
into several subcategories, say fast, slow and non-
models for collision probability Pc are the Generalized
floating, the curve fitting capability can be improved
Sutherland equation and the Yoon equation:
further. One way to contain the complexity and over-
come experimental difficulties is to use experimen-
tal set-ups and model systems with as low dynamic  d p2 
component as possible. (
9 vb ρ p − ρ f ) db

 
All the above mentioned has had a severe conse-  
quence to our ability to draw a detailed picture of the dp 2   d   
Pc = 3 sin θ t exp   cos θ t  ln b − 1.8   (1)
interplay of physics and chemistry in flotation. Much db   dp  
of the work now ongoing is based on the ideas put  
  − 2 + cos θ t − 3cos θ t db  
3

forward in the seminal paper of Sutherland (1948).

  2 sin θ t
2
3d p  

© 2012 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2012.
2 Surface Forces and Hydrophobicity

and
gdb2 ∆ ρ  1 + κ 
vb =   (4)
2 6 µ  2 + 3κ 
 3 4 Re 0,72   d p 
Pc =  +
2 15   db  When a bubble rises in a quiescent fluid a convective
(2) mass transfer takes place. Surfactant molecules then
 E   W + E1  
exp  − 1  1 − exp  − A  slide towards bottom of the bubble. This has lead to
 Ek    Ek ′   a stagnant cap model, where surfactant form a rigid
layer at the “bottom” of the particles up to level
respectively. Yoon and his group (Yoon 1991, Lutrell above which the surface is freely moving.
and Yoon 1992, Yoon and Mao 1996, Mao and Yoon If the desorption rate of the surfactant is much
1997, Yoon 2000) were the first to formulate a model smaller than the adsorption rate, there are several
of one of these sub-processes, namely attachment, studies about the effect of surfactants on the terminal
by first principles and related measurements. They rising velocity giving proof to the stagnant cap model
assumed that the attachment is proportional to the including, for example, Ybert and di Meglio (2000)
ratio between the energy barrier E1 and the kinetic and Krzan et al. (2004) who noted the cap angle is
energy of the approaching particle Ek. dependent on the adsorption-desorption characteris-
tics of the surfactant and observed oscillations in the
terminal velocity. In other studies Tomiyama et al.
 E 
Pa = exp  − 1  (3) (2002) finds three distinct regimes depending on the
 Ek  Eotvös number of the bubble:

The determination of the energy barrier requires an

estimate of the different interaction energies. An
Eo =
(
gd 2 ρ f − ρ g ) (5)
implicit assumption made has been that the energy σ
barrier is symmetric over the bubble surface. It is
noted that the common denominators to these mod- For low Eo (Eo<0.25) the viscous forces domi-
els is they are all continuum models with decoupled nate and the surface is rigid. For intermediate Eo
momentum transfer and they treat forces such as numbers (0.25<Eo<40) the bubbles are surface ten-
hydrodynamic resistance, inertial, gravitational, sion dominated. For high Eo numbers (Eo>40) the
short-range (surface), and capillary forces as either bubble movement is inertia governed. Tomiyama et
scalars or simple vectors maintaining their magni- al. (2002) observe a large scatter of terminal veloci-
tude and direction. ties in the literature. They claim that the prime cause
is the difference in the initial shape deformation, not
Aim the difference in surfactant concentration. In their
opinion the bubble motion, shape and velocity are
Some of the issues that are difficult or impossible to
sensitive to initial shape deformation.
capture are the multiple momentum coupling effects
Kracht and Finch (2010) have observed bubble
taking place at the boundaries, the dynamic changes
terminal velocities at short times finding a strong
at the boundaries and also the changes in the bound-
correlation with velocity retardation and aspect ratio.
ary layer properties. This paper gives some experi-
However, there was an effect of the surfactant type
mental evidence that the changes in the boundary
to be seen as well. They found different frequency
layers will affect the bubble movement and the dis-
oscillations. The superimposed oscillations were
tribution of surfactants.
typically around 5% in amplitude having a frequency
As a bubble moves in a fluid, surfactants will be
around 100 Hz, while larger oscillations had a fre-
adsorbed and desorbed onto its surface causing the
quency around 10 Hz and initially amplitudes up to
fluid dynamics to change. Early in the mathemati-
50%.
cal formulation of fluid dynamics of bodies moving
in a fluid, Hadamard and Rybzynski argued sepa-
Experimental
rately in 1911 that a physical requirement that no
energy jump can exist at the boundary requiring that Another line of experiments has been the rela-
an equal boundary velocity exists on both sides of tive movement of a bubble when surfactants are
the surface. This must lead to a non-zero tangential adsorbed onto the surface. The experiments were
boundary velocity if the boundary is not rigid. They performed in a simple device where bubble were
developed a model that equates the ratio of viscosities formed into ultrapure water (resistivity 18.2 MΩ,
κ (µb/µf) to the final rising velocity a bubble obtains Millipore Direct-Q system) and then released via a
in a quiescent fluid: set of valves to the fluid with surfactant. In the pre-
vious experiments we produced bubbles directly to

© 2012 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2012.
 Frother Mobility on the Bubble Behavior in Bubble Rise Velocity Measurements 3

Table 1. Frothers tested

Nasfroth 240 (NF 240) Dowfroth 200 (DF 200) Dowfroth 250 (DF 250)
CH3 CH3
C 4H 9 O CH2 CH2 OH
H 3C O CH2 CH OH H 3C O CH2 CH OH
3
3 4

206.29 g/mol 206.29 g/mol 264.37 g/mol

analyzed with Matrox Inspector software. The mea-

sured parameters were the area (A), the horizontal
(dh) and vertical (dv) diameters and the middle point
position (x, y) of the bubbles. The bubbles were also
observed with the high-speed camera in turbulent
fields having varying rates of turbulent energy dis-
sipation (Omelka et al. 2010).
Rising velocities were calculated from the
change of the middle point position of the bubble
along the y-axis in consecutive pictures (0.185 ms).
The aspect ratio of a bubble was defined as the
ratio between the horizontal (dh) and vertical (dv)
diameter.

Results
At the lower concentration, the early velocities were
rather similar irrespective of the frother used (see
Figure 1). It can be concluded that no retardation
of the bubble velocity took place even if the aspect
ratio fluctuated and increased during the first 50 ms
from spherical to over 2 in ultrapure water and to
2 with DF 200 and NF 240. This indicated that the
Marangoni stress at the surface was not enough to
prevent bubble deformation.
When the concentration was increased to
25 ppm, DF 250 started to show some retardation
Figure 1. Results with 8 ppm. Bubble rising of rising velocity (see Figure 2). The Marangoni
velocity and aspect ratio dh/dv. stresses were already so high that bubbles in DF 250
solution maintained their shape almost spherical,
very close to the value predicted by Tomiyama. With
the surfactant containing fluid. The bubble size in NF 240 the increased Marangoni stresses kept the
the experiments was 2.2–2.3 mm. Reynolds num- aspect ratio at 1.6. With DF 200 no marked effect
bers varied in the experiments from 50 to 200 during could be seen even at this higher concentration.
the first 50 ms. Eotvös numbers in the work were The results indicate that fast adsorption-desorp-
0.5–0.6. This is within the surface tension dominated tion kinetics of DF 200 (Javor 2010) caused the sur-
area. According to Tomiyama et al. (2002), the cor- factant molecules to move on the surface without a
responding aspect ratio would be around 1.1 for fully marked Marangoni stress build-up. The results do
contaminated bubbles. not support the claim of Tomiyama et al. (2002) that
Dowfroth 200 and 250 and Nasfroth 240 were the shape primarily affects the terminal velocity.
the frother studied (see Table 1). The baseline results
were obtained with ultra-pure water as in the previous Discussion
study (Javor et al., 2010). The concentrations used
were 8 and 25 ppm (0.04 mmol/L and 0.12 mmol/L It can be numerically shown that the adsorption of
for Dow 200 and Nasfroth and 0.03 mmol/L and surfactant onto the surfaces in dynamic conditions
0.1 mmol/L for Dow 250, respectively). includes and generates a complex momentum trans-
All bubbles were recorded with a high-speed fer (see Figure 3) as the adsorbed surfactant layer
camera (5400 fps). The recorded images were will not remain stationary at the rear of the bubble

© 2012 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2012.
4 Surface Forces and Hydrophobicity

Figure 3. CFD simulation of a bubble with

adsorption-desorption estimated with a
simple Freundlich isotherm (Wierink 2009,
unpublished).

surface. Of course, the stagnant cap model is not

valid in turbulent flows.
Figure 2. Results with 25 ppm. Bubble rising It is evident from these simple frother studies
velocity and aspect ratio dh/dv. that the physical picture created from static experi-
ments and described with analytical equations is not
complete enough. We must conclude that Equation 3
but will oscillate. This oscillation is a complex inter- must be interpreted in the future as a local variable.
action between the internal and external flow. The We have previously thought that the kinetic energy,
interaction is affected by the amount of surfactant Ek, is the simpler part of the energies. We must fur-
adsorbed at any given time on the surfaces. ther conclude that the complex momentum interac-
A different kind of picture starts to emerge tion at the boundary, which is also affected strongly
when looking at the results obtained in a turbulent by the frother behavior, will make the estimation of
field. In a pure system (15 MΩ water), the bubbles Ek as tough as is the estimation of E1. We have not
tend to wobble and vibrate violently when they pass even started to look at the effects of interacting elec-
the turbulent eddies. The highest frequencies ana- tric double layer and the fluid boundary layer causing
lyzed with Fast Fourier Transformation (FFT) from the electric double layer to become non-symmetric.
the videos corresponded with the Kolmogorov time Finally, we have to come to a conclusion
scale turnover frequency computed from the tur- regarding the assumptions Sutherland (1948) made
bulent energy dissipation. Highest recorded values such that we cannot accept the bubble rigidity. We
were between1500–2000 Hz with a turbulent energy have shown that bubbles do not behave like rigid
dissipation of 100 m2s–3. The ratio between the lon- spheres even as an approximation. It is also evident
gest and shortest Feret diameter varied from one to that decoupling the momentum transfer is not a good
well over two in time-spans less than 0.5 s. Frequent approximation of the reality.
break up into smaller bubbles was also observed.
The behaviour of bubbles in a fluid containing 5 ppm References
Dow 250 was different. The FFT did not show any
bubble fluctuations faster than 300 Hz. The bubbles Gaudin, A., Flotation, McGrawHill, New York,
are typically also erratically rotating at varying but 1939.
slow angular velocities around x, y and z axles. Javor, Z., N. Schreithofer and K. Heiskanen, Fast
The fast turbulent impacts cause local deforma- adsorption phenomena at air/liquid interfaces,
tions and create local surface excess, which in turn Proc XXV IMPC, AusIMM, 2010.
creates a Marangoni stress stiffening the particle

© 2012 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2012.
 Frother Mobility on the Bubble Behavior in Bubble Rise Velocity Measurements 5

Kracht, W., Finch, J.A., Effect of frother on ini- Sutherland, K.L., Kinetics of the flotation process, J.
tial bubble shape and velocity, Int. J. Miner. Phys. Chem. 52, 1948.
Process., 2010. Tomiyama, A., G.P. Celata, S. Hosokawa and Y.
Krzan, M., K. Lunkenheimer and K. Malysa, On Yoshida, Terminal velocity of single bubbles in
the influence of the surfactant’s polar group surface tension force dominant regime, Int. J. of
on the local and terminal velocities of bubbles. Multiphase Flow 28, 2002.
Colloids and Surfaces A: Physicochem. Eng. Wierink, G., Personal communication, 2009.
Aspects, 250, 2004. Ybert, C. and di Meglio, J.-M., Ascending air bub-
Luttrell, G.H. and R.H. Yoon, A hydrodynamic bles in solutions of surface-active molecules:
model for bubble-particle attachment, J. Colloid influence of desorption kinetics, Eur. Phys. J. E
Interface Sci. 154, 1992. 3, 143–148, 2000.
Mao, L. and R.H Yoon, Predicting flotation rates Yoon, R.-H., Hydrodynamic and surface forces in
using a rate equation derived from first prin- bubble-particle interactions, Aufbereitungs
ciples, Int. J. Miner. Process., 51, 1997. Technik 32, 1991.
Omelka, B., N. Schreithofer, G. Wierink and K. Yoon, R.-H. and L. Mao, Application of extended
Heiskanen, Particle detachment in flotation, DLVO theory, IV: derivation of flotation
Proceedings of the VI International Minerals rate equation from first principles, J Colloid
Processing Seminar, Procemin. 2009. Interface Sci., 181, 1996.
Sutherland, K.L., Physical chemistry of flotation Yoon, R.-H., The role of hydrodynamic and surface
XI. Kinetics of the flotation process, J. Phys. & forces in bubble-particle interaction. Int. J.
Coll. Chem. 52, 1948. Miner. Process., 58, 2000.

© 2012 by the Society for Mining, Metallurgy, and Exploration. All rights reserved. Electronic edition published 2012.