Article

Cite This: Langmuir 2019, 35, 8294−8307 pubs.acs.org/Langmuir

Modeling Bubble Collisions at Liquid−Liquid and Compound

Interfaces
Travis S. Emery† and Satish G. Kandlikar*,†,‡
†
Microsystems Engineering Department and ‡Mechanical Engineering Department, Rochester Institute of Technology, 76 Lomb
Memorial Dr., Rochester, New York 14623, United States
*
S Supporting Information

ABSTRACT: The collision of a bubble at liquid−liquid, solid−liquid−liquid, and

See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

gas−liquid−liquid interfaces, the latter two of which are referred to as compound

Downloaded via INDIAN INST OF TECH DELHI- IIT on January 16, 2021 at 03:53:25 (UTC).

interfaces, is modeled to predict the bubble’s velocity profile and the pressure buildup
and drainage rate of the film(s) formed at impact. A force balance approach,
previously outlined for bubble collisions at solid and free surfaces, is employed, which
takes into account four forces acting on the bubble: buoyancy, drag, inertia of the
surrounding liquid through an added mass force, and a film force resulting from the
pressure buildup in the liquid film formed between the bubble and the interface upon
impact. The augmented Young−Laplace equation is applied to define the pressure
buildup in the film(s), while lubrication theory is employed to define the film drainage rate(s) through the use of the Stokes−
Reynolds equation. This is the first time this modeling technique has been implemented for bubble collisions with these
interface types as all previous models have relied only on grid-based simulations. The models were validated through
experiments conducted here with water and silicone oils of various viscosities and from data found in literature. A reasonable
agreement is observed between the theoretical and experimental velocity profiles found for these liquid combinations under
varying conditions of impact velocity and top film thickness. The spatiotemporal film thickness and pressure profile evolution,
features not yet able to be captured through experiment, are also presented and discussed.

■ INTRODUCTION
Bubble interactions at multifluid interfaces find relevance in
numerous industrial applications and environmental phenom-
ena. Bubble collisions with free and solid surfaces have been
studied in relation to processes such as mineral flotation, foam
formation, and direct contact evaporation.1−3 Bubble passage
through a liquid−liquid interface has been explored primarily
due to its importance in metallurgical applications, though
other processes such as nuclear reactor accident scenarios and
the ascent of plumes through the Earth’s mantle have also been
noted as motivators. The collisions of bubbles with gas−
liquid−liquid or solid−liquid−liquid interfaces are associated
with a newly developed technique for nanoemulsion formation
and natural phenomena such as bubble collisions with the sea
surface microlayer or the collision of liquid-encapsulated
bubbles.4,5 In these applications, the flow dynamics associated
with the bubble collision play a crucial role in governing the
overall process.
The collision of bubbles at solid and free surfaces has been
studied extensively, and comparisons are often made between
them due to their similarity. A review of the rise and impact of
bubbles with solid surfaces is offered by Manica et al.6 They
considered spherical bubbles in Stokes flow and deformable
bubbles with higher Reynolds numbers. In either case, a
balance of forces was successfully used to predict the bubble
motion from rest to its terminal velocity and its impact with a
solid surface. The forces considered are buoyancy, drag, added
mass, history, film, van der Waals, and electrical double layer.
However, certain assumptions allow some of these to be
neglected, as will be discussed later in The Force Balance
Model section of the paper. Previously derived numerical
models from this same group for the impact of bubbles with
mobile7 and immobile8 surface conditions are also discussed.
These models couple the force balance model with the
Stokes−Reynolds−Young−Laplace equations to predict the
bubble motion and the pressure buildup and drainage rate of
the film formed between the bubble and the interface at
impact. The Young−Laplace equation has wide reaching
applications including describing the interfaces of compound
sessile drops and drops floating at a free surface.9,10 The solid
surface bubble collision models have been shown to be highly
accurate in predicting not only the motion of the bubble but
also the spatiotemporal evolution of the film thickness.11 A
similar approach was also recently applied to model a bubble
collision with a free surface.12 Although the film thickness
evolution has not yet been validated for such collisions, the
success of the previous solid surface collision model and the
excellent agreement of this model in the prediction of the
bubble motion provide a strong basis for the model’s accuracy.
Additionally, high-speed monochromatic light interferometry
was recently introduced as a means to quantify the film
drainage during the collision of a 365 μm-diameter bubble and
Received: April 24, 2019
Revised: May 23, 2019
Published: May 29, 2019

© 2019 American Chemical Society 8294 DOI: 10.1021/acs.langmuir.9b01209

Langmuir 2019, 35, 8294−8307
Langmuir
a free surface.13 This technique could be used in the future to
provide further validation of the free surface collision model
with respect to its film drainage predictions. A number of other
experimental studies further examine various aspects of bubble
collisions with free and solid surfaces such as the influence of
surfactants,2,14,15 oscillating interfaces,1,16,17 and bubble kinetic
energy.3,18 Surfactants in particular introduce an added level of
complexity to the bubble collision process by influencing the
mobility of the liquid−gas interface(s), which in turn affects
the liquid film drainage rate. In a recent work from Liu et al.,19
it was shown that the interface velocity is determined by a
balance of shear and Marangoni stresses at the interface. As the
stresses evolve continuously during the collision, this implies
that the drainage rate could change continuously during the
collision depending on the surfactant concentration and initial
impact velocity. Other works have focused on bubble−solid
surface interactions in relation to their application in flotation
separation.20−23 Niecikowska et al.21 examined the effect of
surface charge and hydrophobicity in three-phase contact
(TPC) formation and found that when surfaces are highly
hydrophobic, the effects of surface charge are negligible and
surface roughness defines the rupture and TPC formation.
Hydrophilic surfaces were shown to enable the stability of the
wetting films even in the presence of attractive electrostatic
forces. Krasowska and Malysa23 found that TPC formation can
vary by over an order of magnitude depending on surface
roughness, with surfaces having a roughness below 1 μm
demonstrating four to five bubble collision cycles prior to
attachment and surfaces having a roughness of 50 μm or larger
inducing rupture on the first collision.
The collision of a bubble with a liquid−liquid interface has
received much less research focus compared to a solid−liquid
or gas−liquid interface. Most experimental studies focused on
bubbles that pass through the liquid−liquid interface with a
volume of the lower liquid entrained around or behind the
bubble.24−34 These works have focused on the influence of
upper and lower liquid properties,26 the formation of
encapsulated bubbles, 24 stages of film rupture, 31 and
identifying and classifying associated flow regimes.34 Numer-
ical models of this process have relied on grid-based simulation
techniques, either using commercial software such as ANSYS
Fluent or other customized schemes to solve the flow
equations.27,28,35 Confining previous studies to bubbles that
do not pass through the interface but instead experience a
bouncing behavior similar to that seen with solid and free
surface collisions significantly reduces the breadth of relevant
previous works. In the grid-based simulations of such collisions
from Bonhomme et al.,27 this was the only situation that the
computational predictions differed dramatically from exper-
imental observations. While the experiments showed the
bubble becoming trapped at the interface, the simulation
predicted the bubble to pass through the interface. The noted
reason for this is that the flow in the film between the bubble
and the interface is poorly described due to its small thickness
in comparison to the grid size. Vakarelski et al. 36
experimentally compared the impact of varying size bubbles
or water droplets with a PP11−water interface and a PP11−air
interface to study the influence of interface mobility on
coalescence dynamics. The liquid−gas interfaces of the system
were fully mobile, while the liquid−liquid interfaces were
considered immobile. Concerning bubbles, they found that the
coalescence induction time increased by 2 to 3 orders of
magnitude for PP11−water interface collisions as compared to
8295
Article
PP11−air interface collisions. This is due to the two mobile
interfaces of the free surface collision, which offer little
resistance to film thinning. The previously mentioned work by
Liu et al.19 extended this work to further understand the role of
surfactants on film drainage.
Just recently, the collision of a bubble at a gas−liquid−liquid
interface, therein labeled a compound interface, has begun to
be explored.4,5,37,38 Specifically, bubble collisions with an air−
oil−water surfactant interface were studied as a means of
dispersing submicrometer oil droplets in water to form
nanoemulsions.4,37 In this, a thin layer of oil, 0.1−2.0 mm in
thickness, coats the top of a water bath. Bubbles released in
water collide with the multiliquid interface and, upon rupture
of the oil and water films ahead of the bubble, generate
microdroplets of oil in water. The addition of a lipophilic
material to the oil has also been studied as a scalable platform
for creating functional nanoemulsions for use in areas such as
drug delivery, material science, biology, functional foods, and
nutraceuticals.4 The stability of the bilayer film formed at
impact was also studied through the derivation of a numerical
model considering van der Waals intermolecular attractions to
provide parametric regimes where rupture behavior is
expected.38 A finite-difference/front-tracking method was
applied to model bubble collisions with this type of compound
interface by solving the Navier−Stokes equations on a fixed
uniform Eulerian back grid and tracking the interface motion
on a Lagrangian front grid.5 A close agreement is found in
comparison to experiments conducted with water as the
bottom liquid and various viscosity silicone oils as the top
liquid.
In the present work, the force balance approach previously
employed for modeling bubble collisions with solid and free
surfaces is extended to model bubble collisions at liquid−liquid
and compound interfaces. In addition to the gas−liquid−liquid
compound interface, we also consider a bubble collision at a
solid−liquid−liquid compound interface. Previous numerical
models of these phenomena have relied on grid-based
numerical simulations. As noted for solid interface collisions,6
a full solution of the Navier−Stokes equations to model such
systems is complex to implement and computationally
expensive. Furthermore, it is noted that very fine mesh sizes
are needed to capture the film drainage, which is orders of
magnitude smaller than the bubble size. The force balance
approach offers a simplified method that overcomes these
challenges while further providing valuable insight into the
flow dynamics. In comparison to the previous solid and free
surface collision models employing this approach, changes
must be made in their derivation to account for the variation in
interfacial versus bubble surface tension and the presence of a
viscous and non-massless fluid above the interface. The present
work outlines this derivation and also provides the
experimental data to be used for validation.
■
EXPERIMENTAL SETUP
The experimental setup used to capture bubble collisions with liquid−
liquid and compound interfaces is shown in Figure 1. The primary
housing consisted of a rectangular polycarbonate column with an
inner cross section of 5 cm × 5 cm and height of 15 cm. A glass
capillary tube with an inner diameter of 0.05 mm was positioned at
the bottom of the container to generate single bubbles of radius 0.65
± 0.03 mm in water. This is in perfect agreement with the theoretical
bubble radius of 0.65 mm predicted by Tate’s law for a capillary tube
of this size.39 Wall effects on rising bubbles have been previously
shown to influence the bubble motion when the container radius is
DOI: 10.1021/acs.langmuir.9b01209
Langmuir 2019, 35, 8294−8307
Langmuir
Figure 1. Experimental setup used to capture bubble collisions with
liquid−liquid and compound interfaces.
less than three times the bubble diameter.40 Since the hydraulic radius
of the column used here was over 19 times the bubble diameter, these
effects were deemed negligible. A peristaltic pump was used to control
air flow to the capillary to form single bubbles. In this testing,
deionized water was always used as the bottom liquid and 0.82, 4.59,
or 9.35 mPa·s silicone oil was used as the top liquid. In order to attain
various film thicknesses, a prescribed volume of the top liquid, as
determined by the cross-sectional area of the container multiplied by
the desired film thickness, was carefully added to the container. The
influence of the meniscus formed at the container edge was neglected
due to the relatively large cross-sectional area of the container. Film
thicknesses of 0.25, 0.50, and 1.00 mm were tested as compound
interfaces. For solid−liquid−liquid compound interfaces, a solid
polypropylene surface was additionally placed above the top liquid.
Polypropylene was chosen due to its oleophilic and hydrophobic
properties, which promoted the stability of the oil layer formed
between the water and the solid surface. For liquid−liquid interface
collisions, the top liquid layer thickness was increased to ∼10 mm, a
value much larger than the expected interface deformation.
High-speed videos of the collision were captured using a Photron
FASTCAM camera at 3000 fps. From these videos, the bubble size
and location were measured. A calibration for converting pixels to
millimeters was determined prior to recording using an object of
known size, such as the capillary tube outer diameter as shown in
Figure 2a. The equivalent undeformed bubble radius, R, was
Figure 2. (a) 4 mm outer diameter capillary used for spatial
calibration and horizontal and vertical bubble diameter measurement
and (b) bubble displacement used in velocity calculation.
calculated as 2R = (Dh2Dv)1/3,18 where Dh and Dv are the horizontal
and vertical bubble diameters, respectively, as shown in Figure 2a. The
bubble trajectory was determined by tracking the location of the
bubble center or side using the Photron FASTCAM Analysis software.
The instantaneous bubble velocity was determined from the trajectory
data using a second-order central finite difference scheme:
y − yi − 1
Vi = i + 1
2Δt (1)
where y refers to the vertical position of the bubble as shown in Figure
2a,b, the subscript i refers to the corresponding frame or time, and Δt
is the time between frames.
Between experiments, the system container was thoroughly washed
using hot water and dish soap to remove any contaminants or leftover
liquid. Once clean, the container was rinsed several times with hot
8296
Article
water and was then placed in a hot water bath for several hours. The
container was rinsed several more times with hot water to remove any
remaining contaminants. The system cleanliness can significantly alter
the behavior of the bubble during its rise and collision. Specifically,
the bubble surface mobility is very sensitive to system contamination
levels. For a “clean” bubble, the surface is fully mobile, while a
“contaminated” bubble has an immobile surface. These surface
conditions are reflected in the bubble’s terminal velocity in that clean
bubbles have significantly higher terminal velocities compared to
contaminated bubbles. A theoretical terminal velocity, VT, can be
determined by equating the buoyancy and drag forces to yield VT2 =
8Rg/3CD, where g is the gravitational constant, and CD is the drag
coefficient. The drag coefficient may be found for a clean or
contaminated bubble using the theory compiled by Loth,41 which is
given in the Supporting Information. Based on this compiled theory, a
0.65 mm radius bubble in water has a terminal velocity of 34.0 cm/s if
it is clean and just 14.2 cm/s if it is contaminated. The terminal
velocity reduces very quickly once the contaminant concentration
reaches a critical threshold, and intermediate terminal velocities can
only be attained with very precise control of the surface-active
substance concentration.42 The bubble terminal velocity found in the
present experiments coincided with that theoretically predicted for a
clean bubble, as demonstrated in the Experimental Validation section.
As such, it was concluded that the bubbles rose with mobile surface
conditions and the cleaning process was sufficient. Additionally, it was
noted that the bubbles rose in a rectilinear rise path as opposed to a
zigzag pattern, which occurs with bubbles rising in water with radii
larger than ∼0.91 mm.43,44
The experimental procedure went as follows: after cleaning, the
container was filled with water to produce the desired distance
between the capillary and liquid−liquid interface, L. The specific
volume of the top liquid was then carefully added to form the desired
top layer thickness. The polypropylene surface was then lowered into
place if we were testing for the solid−liquid−liquid interface
configuration. The peristaltic pump was then actuated to form a
single bubble at the capillary orifice. The bubble collision at the
interface was captured using the high-speed camera. Video analysis
was then used to determine the bubble size, trajectory, and velocity.
There are three primary liquid properties that govern the behavior
of the systems considered here: density (ρ), viscosity (μ), and surface
tension (σ). The subscripts B and T will be used to designate these
properties in association with either the bottom or top liquid,
respectively. The experiments conducted here employed liquid
combinations made up from four different liquids: water and three
different silicone oils (SO) with viscosities ranging from 0.82−9.35
mPa·s. The pertinent properties for each of these liquids are listed in
Table 1. The properties of the silicone oils were taken as specified by
Table 1. Properties of Experimental Liquids at 25 °C
density, ρ viscosity, μ surface tension, σ
liquid (kg/m3) (mPa·s) (mN/m)
water 1000 1.00 72.0
SO-1 818 0.82 17.4
SO-5 918 4.59 19.7
SO-10 935 9.35 20.1
the manufacturer,45−47 while the properties of water are found in the
literature.12 In the present experiments, water was always used as the
bottom liquid, while one of the three silicone oils was used as the top
liquid. The interfacial tension of each of these combinations was
determined using pendant drop analysis. Full details of this testing
and the analysis employed to determine the interfacial tension are
discussed in the Supporting Information. The measured values for the
interfacial tension between water and 0.82, 4.59, or 9.35 mPa·s
silicone oil were 50, 51, and 48 mN/m, respectively.
DOI: 10.1021/acs.langmuir.9b01209
Langmuir 2019, 35, 8294−8307